predict essay website

The secret weapon for awesome text. Unique writing, every time.

🚀 speedwrite is insanely fast. it creates new, unique, best-in-class writing, from any source text..

✨Trusted by hundreds of thousands of users, Speedwrite has written millions of lines of text.

️🕑  Start with any creative-commons source. Use Speedwrite to predict new writing based on that source. Create an essay, article, or report, in just minutes.

Speedwrite In 12 Seconds

Video: Speedwrite In 60 Seconds

⚠️ Other AI-writers don't make unique text...

The only text generator with great style.

Speedwrite makes high-quality writing for:

• 🎒 University Students
• 💁 Marketers
• 🎨 Creatives
• 👩‍💼 Professionals
• 🥷 Ninjas (SEO & Social)

It's for anyone who needs to write in English.

It's for anyone who needs unique, well-written text.

But what does that really mean?

UNIQUE writing that won't get flagged 👀 by plagiarism detectors.

Writing that Google has never seen before.

If you are doing SEO, you want Google to rank your page. Google HATES duplicate content.

So. Do you need to write something right now? Here's your job:

Everyone wants writing that is NEW, high-quality, and not duplicate.

You don't want DUPLICATE text.

You need UNIQUE text.

But wait. Can you just use any AI-writer?

No, it won't work.

This is not a good idea.

There are "cops" who are now "reporting" GPT usage to schools, employers, and Google...

And, there is a bigger problem:

These other tools only summarize information that they have found on the internet.

There is NO guarantee of originality. 🤔

These tools might copy entire sentences or even entire paragraphs , directly from the internet. These tools are not safe.

There are thousands of examples of people using AI writing tools like this, and ending up with text that exactly matched some other text, that was already on the internet. 🫢

And, get this: These tools are now "watermarking" their output. 😟 To try to catch anyone who copies the text. Also not safe!

And. Think rationally for a second. What if someone else at your company, or at your school, uses the same tool? You're both going to end up with the same text! ☠️ Not safe!

So what can you do?

Hm... if only there was a secret weapon...

Let's try a little test.

Let's copy just the first paragraph that was generated by the AI-assisted writer, and paste it into... 🚀

....and use the "Predict" button a few times.

OMG. What just happened there.

It looks like EACH time we used the "Predict" button, we got a new, unique, (and very well-written) paragraph! 💡

What is this wizardry?! 🧙

It seems that Speedwrite directly solves our most important problem: It fully randomizes text, so it's totally new, totally unique, every time. 🙂

You need Speedwrite, because...

But that's not the only reason you need it....

Your thoughts, not just bland summaries...

When you start using any of the supposedly "advanced" AI-assisted writers, you'll find something curious.

It seems they want you to enter the topic of your document, toggle a few settings, and then press "generate."

This is just stupid. 🤦‍♂️

The only information that this AI writer has is information that it got from the internet!

And let's be real: Your school or your work doesn't just want recycled info.... ♻️

Your school, or your boss, wants you to use your brain 🧠, to come up with something ORIGINAL.

And they want YOUR research. The research that YOU have done, that you have added to Speedwrite:

They want YOUR ideas. The ideas you have typed into Speedwrite Creative:

Unlike the other AI-writers, Speedwrite doesn't just regurgitate bland, well-known information. Instead, Speedwrite reacts carefully to YOUR input, to YOUR ideas, and to YOUR research. 😀

Yes. But that's not the only reason you need Speedwrite. Also... 🚀

📅 Always up-to-date...

Most AI-assisted writers have a big, ugly problem. They don't know anything new!

Let's say your boss comes to you with an assignment...

Your boss: "I need marketing text for a brand-new robotic mopper... use these details..."

All-New RoboMooper: The Chenzu RoboM is Re-engineered to boast our strongest ever suction power (2000Pa Max) while maintaining quiet operation and a super-slim design (2. 85”). Wi-Fi Convenience: The Chenzu RoboM Alexa and the Google Assistant voice control-services.... [more boring stuff]...

First, let's try our "super powerful" AI-assisted writer:

We only have to try this once.

It's completely obvious: this AI-writer has NO knowledge of the "Chenzu RoboM" at all!

It doesn't know anything useful about this product. 🙄

That's because this product is NEW, and there's no information about it on the internet yet!

So this AI-writer is just making stuff up. Making up random, bland stuff, that has no meaning. 😕

Let's try something else.

Let's try... 🚀

Paste in the details that our boss provided...

Wow. That is really a LOT better. 😁

We've used the info that our boss gave us, which is the newest, up-to-date info about the product.

Then, Speedwrite created some awesome new writing. We can be sure this text isn't "duplicate" with anything else on the internet. 😏

So with Speedwrite, we can use NEW information, based on OUR OWN research, that is NOT DUPLICATE with other stuff on the internet, and generate great new writing, each time. 👍

But that's not the only reason you need Speedwrite. Also...

 Speedwrite is crazy smart.

When you add a source to Speedwrite, and hit "Predict," you get re-written, unique, university-level output, every time.

But, unlike the many low-quality & out-dated "automatic thesaurus" or "phrasing" websites, Speedwrite also *fixes* input text, often in an astonishingly fluent and intelligent manner.

Here's an example -- read this input/output pair carefully and you will see that Speedwrite has taken an error-filled input text, and fixed it in an amazing way:

It cannot be late because the complaint can be show in the social media and if we late, the other customers can see really fast and make our image become bad. Wonderful, positive tone can influence the clients; by this strategy it isadditionally conceivable that the other followers might guard you within the open.Don't erase negative comments since it'll as it were disturb the clients and take thediscussion someplace else where the venue cannot handle it. In this way, theykeep the matter in their hands as it were.

It's also important that the complaints are taken seriously as they can affect the image of the business. If the complaint is not resolved in time, the other customers might see it as an indication that the business is not taking the necessary steps to address the issue. One of the most effective ways to improve a customer's perception of a company is by having a positive tone. This can be done through the use of social media. However, it's also important to avoid deleting negative comments as it can make the customers feel unwelcome and disturb them.

 How can it be so good?

Speedwrite is totally unique, and has no direct competitors.

Our technology is completely different from the many low-quality "automatic thesaurus" solutions on the internet.

For comparison's sake, we did some web searches, to test Speedwrite against other services on the internet, which claim to do something similar.

 Speedwrite performed well. There are serious problems with the others..

For each of these websites, we used this input text, from Wikipedia...

This is the result from the first website we tried...

It kind of looks like English, right? But let's dig into some sentences....

Uh... what is that?

What is this inner thing?

Hm..... This makes no sense at all...

That's because this website is basically working like an advanced thesaurus . It's just changing some words around. And that leads to a very serious problem: The text that it outputs does not make sense.

 And even when the text is semi-OK...

We found that there were other, obvious problems. For example, this website, google hit #2, produces text which is exactly the same length as the input text. Not only that, but each sentence of the input clearly corresponded with another sentence in the output:

 Speedwrite: a breath of fresh air....

And, when we check this text for originality...

...It's determined to be 100% original.

 Speedwrite: Writes better, more original text...

 Other automatic writers don't modify the length of your input. You get paragraphs that are nearly exact duplicates of your input paragraphs. It's obvious that the output paragraphs are derived exactly from the input paragraphs...

 Speedwrite is the only text generator that intelligently interprets your input, to make new, original text....

 Other services write text that is only good for computers to read....

Look what happened when we put this text into Google's #2 search result ....

  Compare that to what Speedwrite creates, from the same input...

Creative Mode: Your Ideas & Opinions

Writing free-form fiction or fantasy? Try the text synthesizer.

Wordblast was created by the Speedwrite team.

Want to learn more? Let's learn HOW SPEEDWRITE WORKS...

Ready try Speedwrite? Take it for a test drive...

We appreciate hearing your ideas as we constantly strive to improve

If you have any questions, please visit  our faq page, - send us a message via lightkey's facebook page (recommended)​, - send us a message using the contact form below.

Thanks! Message sent.

• Jan 5, 2022

How to Write a Research Paper Fast Using Predictive Text

When you're assigned a research paper, the last thing you want to think about is how long it's going to take you to write it. But if you want to get the assignment done on time, and done well, you need to have a plan. In this blog post, we'll give you tips for how to write a paper fast, without sacrificing quality. So read on, and get started!

Writing Fast vs. Writing Well

Before we get into the details of how to write a paper fast, let's first talk about what we mean when we say “fast.” Even though you might be tempted to crank out ten pages in an hour, writing fast doesn't necessarily mean writing badly. You can still produce quality work if you know how to use your time efficiently and have the right tools throughout the writing process.

Work smart, not hard . The key is consistency — put in 30 minutes every day rather than 2 hours sporadically throughout the week, and you'll be surprised by how much progress you can make.

Without further ado, let’s discuss the best tips for how to write a paper fast.

7 Tips To Help You Write a Paper Fast

Now that you know that both speed and good quality are possible, let's talk about some strategies for achieving both!

1. Set a timer and stick to it.

One of the best ways to ensure that you write a paper quickly is to set a timer and stick to it. When you know that you only have a certain amount of time to work on the assignment, you'll be forced to focus and get the job done. And if you're worried about meeting deadlines, try using predictive text technology like Lightkey , which can help you type up papers much faster than usual by predicting content as far as 12 words ahead and correcting spelling errors in real-time.

2. Outline your paper before you start writing.

If you want to make sure your essay is well-organized and flows logically, outline it before you start writing. This will help you stay on track while composing the paper and will make the revision process much easier.

3. Identify your action words, and use them.

If you want to learn how to write a paper quickly, start with using more action words in your writing. These are words like "conduct," "analyze," "perform," or "collect." They're strong verbs that will help you get into the meat of the paper, instead of spending too much time describing what occurred during an experiment or what happened when you interviewed people.

4. Get specific with language whenever possible.

For instance, if you're telling a story about how something occurred, use phrases like "I observed," "he/she said," "the results indicated," etc. These specific details will make your paper sound more authoritative and convincing.

5. Use bullet points to organize your thoughts.

If you find that you are struggling to keep your thoughts organized, try using bullet points. This will help you keep track of the main points you want to make in your essay.

6. Edit as you go.

Rather than waiting until the end to edit your work, edit as you go. This will help you catch errors early on and prevent them from snowballing into bigger problems down the line. Lightkey's real-time spelling correction capability, for example, identifies errors and makes editing suggestions as you type, so you don't waste time later.

7. Utilize predictive text technology.

Predictive text technology can help you write your essay significantly faster and more accurately . This is because the software anticipates the word or phrase you are likely to type next, allowing you to quickly input it without having to type the entire thing out yourself. This can be a major time-saver, especially when writing longer essays.

Write a Research Paper Fast Using Lightkey

Individuals who struggle with writing tasks such as research papers may benefit from using language-processing software. Lightkey is a text prediction software that can help you build presentations and type up papers much faster than usual. To learn more about predictive text technology and how it can help you write papers faster, visit the Lightkey website.

Want to write papers fast right now? Download the Lightkey free version to explore the tool or sign up for any of its premium plans to take advantage of all its benefits.

Recent Posts

Crafting Keyword-Rich Headlines: Balancing SEO Impact With Creative Expression

How to Ensure Accessibility When Incorporating Media Into Your Content

How to Overcome Written Miscommunication: 16 Examples

The Read Time Speed Reading Extension

Read upto 3x faster without distractions, words to time converter, – accurately estimate reading times for any reading material, text, speech or voice-over script., why use our chrome extension.

The Read Time is a powerful chrome extension to speed read articles, blog posts, news, emails or any text on a clutter & ad free window on your desktop browser.

Boost Productivity & Unlock Your True Potential!

Transform webpages like this…

into plain text without ads & other distractions…

& read fast at your own set speed by switching to speed read mode…

Distraction-Free Reading

Speed read webpages by transforming them into clean text or just paste any text into our speed reading interface after installing the extension.

Dual Reading Modes

Use read mode for a traditional reading experience or switch to speed read mode to read at higher speeds via Rapid Serial Visual Presentation.

Customizable Settings

Control reading speed by adjusting words per minute and words at a time. Personalize your reading experience with an array of fonts, themes, and advanced settings.

More on Reading Time Estimation…

What is reading time.

Reading time is the time taken for an average person to silently read a piece of text while maintaining reading comprehension. Based on the meta-analysis of 100’s of studies involving over 18000 participants, the average silent reading speed for an adult individual has been estimated to be approximately 238 words per minute (Marc Brysbaert,2019) .

The average reading time for a piece of text can thus be deduced by dividing the total word count by 238.

Mathematical formula for calculating reading time in minutes:

Reading Time = Total Word Count / 238

What is Speech Time?

Speech time is the time taken for an average person to read aloud a piece of text. Based on the meta-analysis of nearly 80 studies involving 6000 participants, the average oral reading speed for an adult individual is considered to be 183 words per minute (Marc Brysbaert,2019) .

The average speech time of a piece of text can thus be deduced by dividing the total word count by 183.

Marc Brysbaert’s paper further states that reading speeds are generally lower for older adults, children and readers with english as a second language.

How To Use Our Read Time Estimation Tool?

Input text by copying & pasting or uploading PDF, DOCX or TXT files, or just directly enter the word count in number format.

The tool instantly calculates both read aloud and silent reading times based on the word count and average reading speeds of 183 WPM and 238 WPM.

Adjust the speed sliders to match your personal reading speed to see updated reading times in real time. If you aren’t sure of your reading speed, get it tested with our free reading speed test .

How Long Does It Take To Read 1000 Words?

Assuming the average reading speed of an adult is 238 words per minute, it takes approximately 4 minutes and 12 seconds to read 1000 words.

How Long Does It Take To Read 100 Pages?

Assuming a page consists of 500 words, it approximately takes 3 hours and 30 minutes to read 100 pages.

How Long Does It Take To Speak 1000 Words?

Given that the average oral reading speed of an adult individual is 183 words per minute, it takes approximately 5 minutes and 28 seconds to orate 1000 words.

Reading Time For Popular Word Counts

Reading Time For Popular Page Counts

International Baccalaureate Diploma Programme

Predicted score calculator, welcome to ib predict, last updated may 14, 2024.

Getting into the university of your dreams won't be easy, but IB Predict is here to help you along the way. IB exams are hard for a reason. They separate the skilled from the unskilled, the prepared from the unprepared, and the knowers from the throwers. In many cases, teachers are more than willing to inflate their students' predicted grades, deluding them into a dangerous state of overconfidence. The IB Predict calculator absolutely does not lie. It uses grade boundary data from past IB examinations to ensure precision. With IB Predict, you'll know exactly what is needed in order to score a 4, 5, 6, or 7. No more, no less.

I have updated the site with November 2023 grade boundaries. Compared to M23 and N22, N23 boundaries have somewhat increased.

• There have been significant point increases in the following major subjects: SL/HL English LAL, SL/HL English Lit, SL/HL Math AA, HL Math AI, SL Physics, ESS, SL/HL SEHS, SL Visual Arts, SL BM, SL Economics, SL Geography, SL/HL Global Politics.
• There have been significant point decreases in the following subjects: SL/HL Dance.
• Extended essay boundaries were raised by 1 point to pre-covid levels.

Take this information with a grain of salt. The N23 cohort only had about 20k candidates, as opposed to the approximate 120k candidates during M23. Due to the small sample size, it may be the case that N23 exihibited relatively higher signs of intelligence. N23 boundaries therefore should not be taken as a strong indicator of M24 boundaries - use M23 instead.

Please report all errors and concerns to [email protected] or join the Discord Server. Thank you and good luck to all M24 candidates!

Individual Grade Calculators

Upcoming May 2024 Syllabus Changes (old syllabuses archived as of April 9)

Select the grade boundary.

Select the timezone.

Group 1: Studies In Language And Literature

Please provide more details, group 2: language acquisition, group 3: individuals and societies, group 4: sciences, group 5: mathematics, group 6: the arts, theory of knowledge.

Theory Of Knowledge Essay

Weight: 67%

Theory Of Knowledge Exhibition

Weight: 33%

Extended Essay

Weight: 100%

The most accurate college admissions calculator, ever

Factor your extracurriculars, intended major, background, and more for the most accurate chancing available. All for free.

What is chancing?

How accurate is it, what current seniors are saying, understanding your chances can be difficult.

CollegeVine takes the guesswork out of college admissions. Students have many misconceptions about average acceptance rates and their personal admissions chances. Here are important things you need to understand about your chances:

The acceptance rate on a school’s website is not your chance of getting in

Schools calculate acceptance rates by dividing the number of accepted students by the total number that applied. Among all the students who apply, some have a 99% chance of acceptance, while others have less than a 1% chance.

There are factors outside your control that have an impact on your chances

Colleges and universities build each freshman class to include a diverse array of students, and that means selecting for diverse racial, economic, and personal backgrounds.

Everything you do in high school can impact your admissions outcomes

Grades matter, but so does the difficulty of the classes you took. Extracurricular activities count, both in terms of what positions you have held and how you have improved your community.

We take all factors into account, just like admissions officers do

Most websites only consider GPA and test scores when determining what sort of applicant you are. We consider your coursework, GPA, extracurriculars, demographics, intended major, class rank, and test scores (if you have them).

We predict your admissions chances based on real data

CollegeVine is the only free college guidance company that offers data-driven chancing, then works with students to help optimize their profiles. We use thousands of real acceptance results to fine-tune our algorithm. We also explain your chancing results and teach you how to strengthen your profile.

How hard is it to get into a selective school?

Applying to Stanford, Harvard, Princeton, and similar schools is exciting, but only a few students can reasonably expect to be admitted. With over 600+ schools in our database, CollegeVine helps you build a balanced school list to maximizes the chance you’ll get in somewhere that you love and meets all your goals.

The free, all-in-one guidance platform to help you with every step of the college process

Your account unlocks all these free tools to help you apply to college with confidence

Advising livestreams

Join interactive livestreams about nearly every topic in the college process, hosted by college admissions experts.

Essays guidance and peer review

Submit your own essay for a review in less than 6 hours on the world’s first entirely free college essay review system.

Q&A with experts

Ask questions and get quick and helpful answers from CollegeVine experts and a community of supportive peers.

How is CollegeVine free?

We believe that every student deserves expert guidance. To make that possible, access to the CollegeVine platform is free for students. We partner with colleges that pay to join our ecosystem and interact with students via virtual events and 1-1 connections.

There are zero ads on our site and you can rest assured that you are always in control of your personal data. Connections with colleges are student-initiated, meaning your profile is only shared if and when you opt-in.

• Infographics
• Check Your Text

For large texts, please install the app first

To enable large texts, please confirm

To check your text, copy and paste or write directly into the online editor above.

To make sure your sentences are clear and your word choice is perfect, double-check your text.

Yes. This tool checks not only grammar, but also spelling, punctuation, word choice, and even style.

This tool is for anyone who writes in English, from students and learners to professionals, teachers, and even proofreaders. After all, even native speakers of English aren't immune to mistakes.

• Online Editor
• Privacy Policy

Popular Infographics

© 2024 GrammarCheck

Essay Papers Writing Online

Websites to write essays easily and effortlessly with these top platforms.

Are you a student struggling with your essay assignments? Look no further! In today’s fast-paced world, finding reliable and efficient websites for essay writing can be a daunting task. With so many options available, it’s crucial to find the platform that suits your specific needs and guarantees top-notch quality. This guide will help you navigate through the vast sea of online essay writing services and provide you with a list of exceptional websites that will save you time, effort, and deliver outstanding results.

When it comes to writing essays, it’s imperative to find platforms that offer a seamless user experience, have a wide range of subject expertise, and boast professional writers with in-depth knowledge. Additionally, accessibility and affordability play a significant role in your decision-making process. You want a website that is available 24/7, easy to navigate, and offers competitive prices. These factors, combined with a commitment to originality, confidentiality, and prompt delivery, are the key components of a reliable essay writing website.

Whether you are a high school student working on your first essay or a graduate student seeking assistance with complex research papers, finding the right website can make a world of difference in your academic success. With the help of this guide, you will discover the top-notch essay writing platforms that have gained a stellar reputation for their exceptional services. Say goodbye to hours spent on endless research and stress – get ready to find the website that will turn your essay writing experience into a breeze!

Best Websites to Write Essays

When it comes to online platforms for academic writing, there are several outstanding options available. These platforms provide a space where individuals can showcase their writing skills and earn money by offering their services to students and professionals in need of essays or other written content. Whether you are a seasoned writer looking for new opportunities or a student in search of assistance with your assignments, these websites offer a wide range of features and benefits to meet your needs.

1. WritingHub

One of the top websites for essay writing is WritingHub. This platform offers a user-friendly interface and a large pool of talented writers specializing in various disciplines. With WritingHub, you can submit your essay requirements and receive bids from qualified writers. You can then choose the writer who best fits your needs based on their experience, ratings, and price. The platform also provides a messaging system for direct communication with your chosen writer, ensuring a personalized and efficient writing process.

2. AcademicPro

If you are seeking a platform that guarantees high-quality essays, AcademicPro is a great choice. This website prides itself on having a rigorous screening process to ensure that only the most skilled and experienced writers are accepted. AcademicPro also offers a money-back guarantee, so you can be confident that you will receive a well-written essay that meets your requirements. Additionally, the platform provides 24/7 customer support to address any concerns or questions you may have throughout the writing process.

3. EssayExperts

For those searching for a website that offers a wide range of writing services beyond essays, EssayExperts is an excellent option. In addition to essays, this platform offers assistance with research papers, dissertations, thesis statements, and more. EssayExperts boasts a team of highly qualified writers who have expertise in various academic fields. The platform also offers free revisions to ensure that you are completely satisfied with the final product.

4. ProEssayWriter

ProEssayWriter is a website that stands out for its efficiency and affordability. This platform offers a quick turnaround time without compromising the quality of the writing. ProEssayWriter also provides competitive prices, making it an attractive option for those on a budget. Additionally, the platform has a customer rating system, allowing you to see the feedback and ratings of previous clients to help you make an informed decision when choosing a writer.

5. EssayMasters

If you are looking for a website that guarantees original and plagiarism-free essays, EssayMasters is a top choice. This platform emphasizes the importance of authenticity and ensures that all essays are written from scratch. EssayMasters also offers a team of highly skilled writers who are experienced in various citation styles and can meet even the strictest formatting requirements. With their dedication to quality and originality, EssayMasters aims to exceed your expectations with every essay they deliver.

In conclusion, these websites provide a platform for individuals to showcase their writing skills and offer their services to those who need assistance with essays or other written content. With their user-friendly interfaces, talented writers, and array of features, these websites are the go-to options for anyone in search of high-quality essays and academic writing services. Whether you are a writer looking for new opportunities or a student seeking help with your assignments, these platforms offer the necessary tools and support to ensure your success.

In the world of online essay writing services, EssayPro stands out as a reliable and efficient platform for students and professionals alike. With its user-friendly interface and a wide range of features, EssayPro provides a seamless experience for those seeking high-quality essays on various topics.

One of the key advantages of EssayPro is its team of experienced writers who are well-versed in different academic disciplines. They can tackle any subject and deliver well-researched and original essays that meet the highest standards of quality. Additionally, EssayPro allows users to choose their preferred writer based on their ratings, reviews, and area of expertise, ensuring that each assignment is in capable hands.

Another noteworthy feature of EssayPro is its efficient customer support system. The platform provides 24/7 assistance to address any inquiries or concerns that users may have. The support team is highly responsive and knowledgeable, ensuring a smooth communication process between the writers and the clients.

Furthermore, EssayPro values confidentiality and privacy, ensuring that all personal and payment information is protected. The platform uses secure encryption protocols to safeguard user data, giving users peace of mind when accessing the service.

• Quality essays written by experienced professionals
• Ability to choose preferred writer based on ratings and expertise
• 24/7 customer support for immediate assistance
• Strict confidentiality and privacy measures
• User-friendly interface for a seamless experience

Overall, EssayPro is a top-notch platform that offers a reliable and efficient solution for those in need of essay writing services. With its experienced writers, user-friendly interface, and excellent customer support, EssayPro is a go-to choice for students and professionals who value high-quality work.

PapersOwl is a well-established online platform that offers a range of services for academic writing. It provides a convenient and reliable solution for students looking to get their essays written professionally.

One of the key advantages of PapersOwl is the quality of the papers they deliver. Their team of experienced writers ensures that each essay is well-researched, properly structured, and written to a high standard. Students can expect to receive well-written and well-argued essays that demonstrate a thorough understanding of the topic.

Another advantage of PapersOwl is the professionalism and reliability of their writers. They have a rigorous selection process, which ensures that only the best writers are hired. This means that students can be confident that their essays will be handled by experts in their field. Additionally, PapersOwl provides a messaging system that allows students to communicate directly with the assigned writer, ensuring that any questions or concerns can be addressed promptly.

Timely delivery is also a key feature of PapersOwl. They understand the importance of meeting deadlines, and their writers are committed to delivering essays on time. This is particularly useful for students who may have tight deadlines and need their essays urgently.

However, there are a few drawbacks to using PapersOwl. One of these is the cost. The high quality and professionalism of their writers come at a price, and PapersOwl can be quite expensive compared to other platforms. Additionally, PapersOwl may not be suitable for urgent orders, as they prioritize quality over speed.

In conclusion, PapersOwl is a reputable platform for essay writing that offers high-quality papers, experienced writers, and timely delivery. While it may be more expensive than other options, it provides a reliable solution for students who value quality and professionalism in their academic writing.

EduBirdie is an online platform that provides essay writing services to students. It is a popular option for those who are looking for professional assistance and guidance in their academic writing tasks. With a team of experienced writers, EduBirdie offers a wide range of essay writing services to cater to the diverse needs of students. Whether you need help with research papers, term papers, or any other type of academic writing, EduBirdie has got you covered.

One of the key features of EduBirdie is its easy-to-use interface, which allows students to place their orders quickly and conveniently. The platform also offers a user-friendly messaging system, which enables students to communicate directly with their assigned writers. This ensures clear and effective communication, resulting in high-quality essays that meet the specific requirements and expectations of each student.

EduBirdie takes pride in its team of writers, who are highly qualified and knowledgeable in various academic disciplines. They have a deep understanding of different writing styles and formats, ensuring that each essay is well-written and properly structured. The writers at EduBirdie also conduct thorough research to gather relevant and up-to-date information, providing students with well-researched and informative essays.

Another important aspect of EduBirdie is its commitment to delivering plagiarism-free content. The platform has strict policies in place to ensure that all essays are original and written from scratch. This ensures that students receive unique and authentic essays that are free from any form of plagiarism. EduBirdie also provides plagiarism reports upon request, giving students peace of mind and confidence in the originality of their essays.

Overall, EduBirdie is a reliable and trustworthy platform for essay writing services. With its user-friendly interface, experienced writers, and commitment to plagiarism-free content, it is a top choice for students seeking professional writing assistance. Whether you need help with a single essay or multiple assignments, EduBirdie is here to help you achieve academic success.

EssayShark is a popular platform that offers students a convenient and reliable solution for their essay writing needs. With its user-friendly interface, vast pool of talented writers, and efficient system, EssayShark has earned a reputation as a go-to website for students seeking assistance with their academic writing tasks.

When it comes to finding the best essay writing service, EssayShark stands out for its commitment to quality and professionalism. The platform carefully selects its writers, ensuring they have the necessary expertise and qualifications to deliver top-notch essays. Whether you need help with an essay on literature, history, science, or any other subject, EssayShark has writers who can handle it all.

One of the key features that sets EssayShark apart is its bidding system. Rather than assigning writers to projects, the platform allows students to post their essay requirements and receive bids from available writers. This gives students the opportunity to choose the writer that best fits their needs and budget. It also promotes competition among writers, driving them to deliver their best work.

EssayShark offers a transparent and secure payment system, ensuring that students only release payment once they are satisfied with the completed essay. The platform also provides a messaging system that allows direct communication between students and writers, enabling them to collaborate effectively throughout the writing process.

With its reputation for delivering high-quality essays, personalized service, and reliable support, EssayShark has become a trusted platform for students around the world. Whether you need help with a simple essay or a complex research paper, EssayShark is a platform you can rely on to meet your essay writing needs.

GradeMiners

When it comes to finding a reliable platform for academic writing services, GradeMiners is a highly recommended option to consider. With a strong reputation in the industry, GradeMiners offers a wide range of writing services to meet the needs of students at various academic levels.

GradeMiners understands the importance of delivering high-quality essays that meet the academic standards and requirements of each client. Their team of professional writers possesses the expertise and knowledge to tackle a diverse range of topics and subjects.

One of the key benefits of using GradeMiners is their commitment to delivering plagiarism-free papers. Each essay is thoroughly checked using advanced plagiarism detection technology to ensure originality and authenticity.

Moreover, GradeMiners places a strong emphasis on meeting deadlines. Time management is crucial for students, and GradeMiners understands the importance of timely delivery. They strive to complete each order within the specified timeframe, allowing students to submit their essays without worry.

In addition to their writing services, GradeMiners also offers a range of other useful features. These include free revisions, 24/7 customer support, and a user-friendly interface that makes the ordering process simple and efficient.

Whether you need assistance with a research paper, an essay, or any other academic writing assignment, GradeMiners is a platform that can be trusted to deliver high-quality results. Their professionalism and dedication to customer satisfaction make them a top choice for students seeking reliable essay writing services.

ProEssayWriting

ProEssayWriting is a reliable online platform that offers professional essay writing services. If you are looking for assistance with your academic assignments, ProEssayWriting is here to help. With a team of experienced writers, this platform guarantees high-quality essays that meet all your requirements.

ProEssayWriting stands out from other essay writing services due to its attention to detail and commitment to customer satisfaction. Each essay is carefully crafted to ensure originality and relevance to the topic. The writers at ProEssayWriting are highly skilled in various subjects, allowing them to provide customized essays that are tailored to your needs.

When you choose ProEssayWriting, you can expect timely delivery and excellent customer support. The platform has a user-friendly interface that makes it easy to place an order and track its progress. The customer support team is available 24/7 to address any concerns or questions you may have.

ProEssayWriting also prioritizes your confidentiality and privacy. Your personal information and order details are kept secure and confidential, ensuring that your identity remains protected. With ProEssayWriting, you can rest assured that your essay will be handled with professionalism and confidentiality.

Whether you need help with an essay, research paper, or any other academic assignment, ProEssayWriting is the platform to choose. Their experienced writers, attention to detail, and commitment to customer satisfaction make them a top choice for students seeking reliable essay writing services.

In the vast world of online essay writing platforms, BestEssays stands out as a top choice for students seeking high-quality papers.

BestEssays combines the best features of various online essay writing platforms to create a unique and comprehensive experience for users. With a team of talented writers and an easy-to-use website, BestEssays offers a seamless process from start to finish.

When it comes to essay writing services, BestEssays excels in providing personalized and customized solutions. Whether you need assistance with a simple essay or a complex research paper, BestEssays ensures that your requirements are met with precision and expertise.

One of the key strengths of BestEssays is its commitment to delivering plagiarism-free content. With a rigorous quality control process, every essay produced by BestEssays is thoroughly checked for originality and adherence to academic standards.

In addition to their commitment to quality, BestEssays understands the importance of meeting deadlines. With a track record of timely delivery, students can rely on BestEssays to receive their completed essays well before the submission date.

Moreover, BestEssays offers reliable customer support that is available 24/7. Whether you have a question about the progress of your essay or need assistance with the ordering process, the BestEssays support team is ready to assist you at any time.

Overall, BestEssays is a top-notch platform that combines excellent writing services, personalized solutions, plagiarism-free content, timely delivery, and reliable customer support. With BestEssays, students can trust that their academic needs will be met with professionalism and excellence.

Related Post

How to master the art of writing expository essays and captivate your audience, step-by-step guide to crafting a powerful literary analysis essay, convenient and reliable source to purchase college essays online, unlock success with a comprehensive business research paper example guide, unlock your writing potential with writers college – transform your passion into profession, “unlocking the secrets of academic success – navigating the world of research papers in college”, master the art of sociological expression – elevate your writing skills in sociology.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Entropy (Basel)

Improving Automated Essay Scoring by Prompt Prediction and Matching

1 School of Artificial Intelligence, Beijing Normal University, Beijing 100875, China

Tianbao Song

2 School of Computer Science and Engineering, Beijing Technology and Business University, Beijing 100048, China

Weiming Peng

Associated data.

Publicly available datasets were used in this study. These data can be found here: http://hsk.blcu.edu.cn/ (accessed on 6 March 2022).

Automated essay scoring aims to evaluate the quality of an essay automatically. It is one of the main educational application in the field of natural language processing. Recently, Pre-training techniques have been used to improve performance on downstream tasks, and many studies have attempted to use pre-training and then fine-tuning mechanisms in an essay scoring system. However, obtaining better features such as prompts by the pre-trained encoder is critical but not fully studied. In this paper, we create a prompt feature fusion method that is better suited for fine-tuning. Besides, we use multi-task learning by designing two auxiliary tasks, prompt prediction and prompt matching, to obtain better features. The experimental results show that both auxiliary tasks can improve model performance, and the combination of the two auxiliary tasks with the NEZHA pre-trained encoder produces the best results, with Quadratic Weighted Kappa improving 2.5% and Pearson’s Correlation Coefficient improving 2% on average across all results on the HSK dataset.

1. Introduction

Automated essay scoring (AES), which aims to automatically evaluate and score essays, is one typical application of natural language processing (NLP) technique in the field of education [ 1 ]. In earlier studies, a combination of handcrafted design features and statistical machine learning is used [ 2 , 3 ], and with the development of deep learning, neural network-based approaches gradually become mainstream [ 4 , 5 , 6 , 7 , 8 ]. Recently, pre-trained language models have gradually become the foundation module of NLP, and the paradigm of pre-training, then fine-tuning, is also widely adopted. Pre-training is the most common method for transfer learning, in which a model is trained on a surrogate task and then adapted to the desired downstream task by fine-tuning [ 9 ]. Some research has attempted to use pre-training modules in AES tasks [ 10 , 11 , 12 ]. Howard et al. [ 10 ] utilize the pre-trained encoder as a feature extraction module to obtain a representation of the input text and update the pre-trained model parameters based on the downstream text classification task by adding a linear layer. Rodriguez et al. [ 11 ] employ a pre-trained encoder as the essay representation extraction module for the AES task, with inputs at various granularities of the sentence, paragraph, overall, etc., and then use regression as the training target for the downstream task to further optimize the representation. In this paper, we fine-tune the pre-trained encoder as a feature extraction module and convert the essay scoring task into regression as in previous studies [ 4 , 5 , 6 , 7 ].

The existing neural methods obtain a generic representation of the text through a hierarchical model using convolutional neural networks (CNN) for word-level representation and long short-term memory (LSTM) for sentence-level representation [ 4 ], which is not specific to different features. To enhance the representation of the essay, some studies have attempted to incorporate features such as prompt [ 3 , 13 ], organization [ 14 ], coherence [ 2 ], and discourse structure [ 15 , 16 , 17 ] into the neural model. These features are critical for the AES task because they help the model understand the essay while also making the essay scoring more interpretable. In actual scenarios, prompt adherence is an important feature in essay scoring tasks [ 3 ]. The hierarchical model is insensitive to changes in the corresponding prompt for the essay and always assigns the same score for the same essay, regardless of the essay prompt. Persing and Ng [ 3 ] propose a feature-rich approach that integrates the prompt adherence dimension. Ref. [ 18 ] improves document modeling with a topic word. Li et al. [ 7 ] utilizes a hierarchical structure with an attention mechanism to construct prompt information. However, the above feature fusion methods are unsuitable for fine-tuning.

The two challenges in effectively incorporating pre-trained models into AES feature representation are the data dimension and the methodological dimension. For the data dimension, the use of fine-tuning approaches to transfer the pre-trained encoder to downstream tasks frequently necessitates sufficient data, and there has been more research on both training and testing data from the same target prompt [ 4 , 5 ], but the data size is relatively small, varying between a few hundred and a few thousand, and pre-trained encoders cannot be fine-tuned well. In order to solve this challenge, we use the whole training set, which includes various prompts. In terms of methodology, we employ the pre-training and multi-task learning (MTL) paradigms, which can learn features that cannot be learned in a single task through joint learning, learning to learn, and learning with auxiliary tasks [ 19 ], etc. MTL methods have been applied to several NLP tasks, such as text classification [ 20 , 21 ], semantic analysis [ 22 ] et al. Our method creates two auxiliary tasks that need to be learned alongside the main task. The main task and auxiliary tasks can increase each other’s performance by sharing information and complementing each other.

In this paper, we propose an essay scoring model based on fine-tuning that utilizes multi-task learning to fuse prompt features by designing two auxiliary tasks, prompt prediction, and prompt matching, which is more suitable for fine-tuning. Our approach can effectively incorporate the prompt feature in essays and improve the representation and understanding of the essay. The paper is organized as follows. In Section 2 , we first review related studies. We describe our method and experiment in Section 3 and Section 4 . Section 5 presents the findings and discussions. Finally, in Section 6 , we provide a conclusion, future work, and the limitations of the paper.

2. Related Work

Pre-trained language models, such as BERT [ 23 ], BERT-WWM [ 24 ], RoBERTa [ 25 ], and NEZHA [ 26 ], have gradually become a fundamental technique for NLP, with great success on both English and Chinese tasks [ 27 ]. In our approach, we use the BERT and NEZHA feature extraction layers. BERT is the abbreviation of Bidirectional Encoder Representations from Transformers, and it is based on transformer blocks that are built using the attention mechanism [ 28 ] to extract semantic information. It is trained on two unsupervised tasks using large-scale datasets: masked language model (MLM) and next sentence prediction (NSP). NEZHA is a Chinese pre-training model that employs functional relative positional encoding and whole word masking (WWM) rather than BERT. The pre-training then the fine-tuning mechanism is widely used in downstream NLP tasks, including AES [ 11 , 12 , 15 ]. Mim et al. [ 15 ] propose a pre-training approach for evaluating the organization and argument strength of essays based on modeling coherence. Song et al. [ 12 ] present a multi-stage pre-training method for automated Chinese essay scoring that consists of three components: weakly supervised pre-training, supervised cross-prompt fine-tuning, and supervised target-prompt fine-tuning. Rodriguez et al. [ 11 ] use BERT and XLNET [ 29 ] for representation and fine-tuning of English corpus.

The essay prompt introduces the topic, offers concepts, and restricts both content and perspective. Some studies have attempted to enhance the AES system by incorporating prompt features in many ways, such as by integrating prompt information to determine if an essay is off-topic [ 13 , 18 ] or by considering prompt adherence as a crucial indicator [ 3 ]. Louis and Higgins [ 13 ] improve model performance by expanding prompt information with a list of related words and reducing spelling errors. Persing and Ng [ 3 ] propose a feature-rich method for incorporating the prompt adherence dimension via manual annotation. Klebanov et al. [ 18 ] also improve essay modeling with topic words to quantify the overall relevance of the essay to the prompt, and the relationship between prompt adherence scores and total essay quality is also discussed. The methods described above mostly employ statistical machine learning, prompt information is enriched by annotation and the construction of datasets, as well as the construction of word lists and topic word mining. While all of them are making great progress, the approaches they are employing are more difficult to directly transfer to fine-tuning. Li et al. [ 7 ] propose a shared model and an enhanced model (EModel), and utilize a neural network hierarchical structure with an attention mechanism to construct features of the essay such as discourse, coherence, relevancy, and prompt. For the representation, the paper employs GloVe [ 30 ] rather than a pre-trained model. In the experiment section, we compared our method to the sub-module of EModel (Pro.) which incorporates the prompt feature.

3.1. Motivation

Although previous studies on automated essay scoring models for specific prompts have shown promising results, most research focuses on generic features of essays. Only a few studies have focused on prompt feature extraction, and no one has attempted to use a multi-task approach to make the model capture prompt features and be sensitive to prompts automatically. Our approach is motivated by capturing prompt features to make the model aware of the prompt and using pre-training and then the fine-tuning mechanism for AES. Based on this motivation, we use a multi-task learning approach to obtain features that are more applicable to Essay Scoring (ES) by adding essay prompts to the model input and proposing two auxiliary tasks: Prompt Prediction ( PP ) and Prompt Matching ( PM ). The overall architecture of our model is illustrated in Figure 1 .

The proposed framework. “一封求职信” is the prompt of the essay, the English translation is “A cover letter”. “主管您好” means “Hello Manager”. The prompt and essay are separated by [SEP].

3.2. Input and Feature Extraction Layer

The input representation for a given essay is built by adding the corresponding token embeddings E t o k e n , segment embeddings E s e g m e n t , and position embeddings E p o s i t i o n . To fully exploit the prompt information, we concatenate the prompt in front of the essay. The first token of each input is a special classification token [CLS], and the prompt and essay are separated by [SEP]. The token embedding of the j -th essay in the i -th prompt can be expressed as Equation ( 1 ), E s e g m e n t and E p o s i t i o n are obtained from the tokenizer of the pre-train encoder.

We utilize the BERT and NEZHA as feature extraction layers. The final hidden state corresponding to the [CLS] token is the essay representation r e for essay scoring and subtasks.

3.3. Essay Scoring Layer

We view essay scoring as a regression task. To enable data mapping regression problems, the real scores are scaled to the range [ 0 , 1 ] for training and rescaled during evaluation, according to the existing studies:

where s i j is the scaled score for i -th prompt j -th essay, and s c o r e i j is the actual score for i -th prompt j -th essay, m a x s c o r e i and m i n s c o r e i are the maximum and minimum of the real scores for the i -th prompt. The input is essay representation r e from the pre-trained encoder, which is fed into a linear layer with a sigmoid activation function:

where s ^ is the predicted score by AES system, σ is the sigmoid function, W e s is a trainable weights, and b e s is a bias. The essay scoring (es) training objective is described as:

3.4. Subtask 1: Prompt Prediction

The definition of prompt prediction is giving an essay to determine which prompt it belongs to. We view prompt prediction as a classification task. The input is essay representation r e , which is fed into a linear layer with a softmax function. The formula is given by Equation ( 5 ):

where u ^ is the probability distribution of classification results, W p p is a parameter matrix, and b p p is a bias. The loss function is formalized as follows:

where u k is the real prompt label for the k -th sample, p p p k c is the probability that the k -th sample belongs to the c -th category, C denotes the number of prompts, which in this study is ten.

3.5. Subtask 2: Prompt Matching

The definition of prompt matching is giving a pair of a prompt and an essay, and to decide if the essay and the prompt are compatible. We consider prompt matching to be a classification task. The following is the formula:

where v ^ is the probability distribution of matching results, W p m is a parameter matrix, and b p m is a bias. The objective function is shown in Equation ( 9 )

where v k indicates whether the input prompt and essay match. p p m k m is the likelihood that the matching degree of k -th sample falls into category m. m denotes the matching degree, 0 for a match, 1 for a dismatch. The distinction between prompt prediction and prompt matching is that as the number of prompts increases, the difference in classification targets leads to increasingly obvious differences in task difficulty, sample distribution and diversity, and scalability.

3.6. Multi-Task Loss Function

The final loss function for each input is a weighted sum of the loss functions for essay scoring and two subtasks: prompt prediction and prompt matching, with the loss formalized as follows:

where α , β , and γ are non-negative weights assigned in advance to balance the importance of the three tasks. Because the objective of this research is to improve the AES system, the main task should be given more weight than the two auxiliary tasks. The optimal parameters in this paper are α : β = α : γ = 100:1, and in Section 5.3 , we design experiments to figure out the optimal value interval for α , β , and γ .

4. Experiment

4.1. dataset.

We use HSK (HSK is the acronym of Hanyu Shuiping Kaoshi, which is Chinese Pinyin for the Chinese Proficiency Test). Dynamic Composition Corpus ( http://hsk.blcu.edu.cn/ (accessed on 6 March 2022)) as our dataset as in existing studies [ 31 ]. HSK is also called “TOEFL in Chinese”, which is a national standardized test designed to test the proficiency of non-native speakers of Chinese. The HSK corpus includes 11,569 essays composed by foreigners from more than thirty different nations or regions in response to more than fifty distinct prompts. We eliminate any prompts with fewer than 500 student writings from the HSK dataset to constitute the experimental data. The statistical results of the final filtered dataset are provided in Table 1 , which comprises 8878 essays across 10 prompts taken from the actual HSK test. Each essay score ranges from 40 to 95 points. We divide the entire dataset at random into the training set, validation set, and test set in the ratio of 6:2:2. To alleviate the problem of insufficient data under a single prompt, we apply the entire training set that consists of different prompts for fine-tuning. We test every prompt individually as well as the entire test set during the testing phase and utilize the same 5-fold cross-validation procedure as [ 4 , 5 ]. Finally, we report the average performance.

HSK dataset statistic.

4.2. Evaluation Metrics

For the main task, we use the Quadratic Weighted Kappa (QWK)approach, which is widely used in AES [ 32 ], to analyze the agreement between prediction scores and the ground truth. QWK can be calculated by Equations ( 11 ) and ( 12 )

where i and j are the golden score of the human rater and the AES system score, and each essay has N possible ratings. Second, calculate the QWK score using Equation ( 12 ).

where O i , j denotes the number of essays that receive a rating i by the human rater and a rating j by the AES system. The expected rating matrix Z is histogram vectors of the golden rating and AES system rating and normalized so that the sum of its elements equals the sum of its elements in O . We also utilize Pearson’s Correlation Coefficient (PCC) to measure the association as in previous studies [ 3 , 32 , 33 ], which quantifies the degree of linear dependency between two variables and describes the level of covariation. In contrast to the QWK metric, which evaluates the agreement between the model output and the gold standard, we use PCC to assess whether the AES system ranks essays similarly to the gold standard, indicating the capacity of the AES system to appropriately rank texts, i.e., high scores ahead of low scores. For auxiliary tasks, we consider prompt prediction and prompt matching as classification problems and use macro-F1 score (F1), and accuracy (Acc.) as evaluation metrics.

4.3. Comparisons

Our model is compared to the baseline models listed below. The former three are existing neural AES methods, and we experiment with both character and word input when training for comparison. The fourth method is to fine-tune the pre-trained model, and the rest are variations of our proposed method.

CNN-LSTM [ 4 ]: This method builds a document using CNN for word-level representation and LSTM for sentence-level representation, as well as the addition of a pooling layer to obtain the text representation. Finally, the score is obtained by applying the linear layer of the sigmoid function.

CNN-LSTM-att [ 5 ]: This method incorporates an attention mechanism into both the word-level and sentence-level representations of CNN-LSTM.

EModel (Pro.): This method concatenates the prompt information in the input layer of CNN-LSTM-att, which is a sub-module of [ 7 ].

BERT/NEZHA-FT: This method is used to fine-tune the pre-trained model. To obtain the essay representation, we directly feed an essay into the pre-trained encoder as the input. We choose the [CLS] embedding as essay representations and feed them into a linear layer of the sigmoid function for scoring.

BERT/NEZHA-concat: The difference between this method and fine-tune is that the input representation concatenates the prompt to the front of the essay in token embedding, as in Figure 1 .

BERT/NEZHA-PP: This model incorporates prompt prediction as an auxiliary task, with the same input as the concat model and the output using [CLS] as the essay representation. A linear layer with the sigmoid function is used for essay scoring, and a linear layer with the softmax function is used for prompt prediction.

BERT/NEZHA-PM: This model includes prompt matching as an auxiliary task. In the input stage of constructing the training data, there is a 50% probability that the prompt and the essay are mismatched. [CLS] embedding is used to represent the essay. A linear layer with the sigmoid function is used for essay scoring, and a linear layer with the softmax function is used for prompt matching.

BERT/NEZHA-PP&PM: This model utilizes two auxiliary tasks, prompt prediction, and prompt matching, with the same inputs and outputs as the PM model. The output layer of the auxiliary tasks is the same as above.

4.4. Parameter Settings

We use BERT ( https://github.com/google-research/bert (accessed on 11 March 2022)) and NEZHA ( https://github.com/huawei-noah/Pretrained-Language-Model/tree/master/NEZHA-TensorFlow (accessed on 11 March 2022)) as pre-trained encoder. To obtain tokens and token embeddings, we employ the tokenizer and vocabulary of the pre-trained encoder. The parameters of the pre-trained encoder are learnable during both the fine-tuning and training phases. The maximum length of the input is set to 512 and Table 2 includes additional parameters. The baseline models, CNN-LSTM and CNN-LSTM-att, are trained from scratch, and their parameters are shown in Table 2 . Our experiments are carried out on NVIDIA TESLA V100 32 G GPUs.

Parameter settings.

5. Results and Discussions

5.1. main results and analysis.

We report our experimental results in Table 3 and Table A1 (Due to space limitations, this table is included in Appendix A ). Table A1 illustrates the average QWK and PCC for each prompt. Table 3 shows QWK and PCC across the entire test set and the average results of each prompt test set. As shown in Table 3 , we can find that the proposed auxiliary tasks (PP, PM, and PP&PM) (line 8–10 & 13–15) outperform other contrast models on both QWK and PCC, PP&PM models with the pre-trained encoder, BERT, and NEZHA, outperform PP and PM on QWK. In terms of the PCC metric, PM models exceeded the other two models except for the average result with the NEZHA encoder. The findings above indicate that our proposed two auxiliary tasks are both effective.

QWK and PCC for the total test set and Average QWK and PCC for each prompt test set; † denotes input as a character; ‡ denotes input as word. The best results are in bold.

On Total test set, our best results, a pre-trained encoder with PM and PP, are higher compared to fine-tuning method and EModel(Pro.), exceed the strong baseline concat model by 1.8% with BERT and 2.3% with NEZHA on QWK, and get a generally consistent correlation. It is shown from Table 3 that our proposed models also yield similar results to the Average test set, 1.6% of BERT and 2% of NEZHA on QWK of PP&PM models compared to concat model, 2% of BERT and 2.5% of NEZHA on QWK of PP&PM models compared to fine-tuning model, and competitive results on PCC metric. Using the multi-task learning approach and fine-tuning comparison, our proposed approach outperforms the baseline system on both QWK and PCC, indicating that better essay representation can be obtained through multi-tasking learning. Furthermore, when compared to the concat model with fused prompt representation, our proposed approach outperform the baseline in QWK scores, but line 10 and line 15 in Table 3   Total track PCC values are lower within 1% of the baseline. It demonstrates that our proposed auxiliary task is effective in representing the essay prompt.

We train the hierarchical model (line 1–4) using character and word as input, respectively, and the results show that using the character for training is generally better, with the best results in Total and Average being more than 4% lower than those with the pre-training method. The results indicate that using pre-trained encoders both BERT and NEZHA for feature extraction works well on the HSK dataset. The pre-training model comparison reveals that BERT and NEZHA are competitive, with NEZHA delivering the best results.

Results of each prompt with BERT and NEZHA are displayed in Figure 2 . The results of our proposed models (PP, PM, and PP&PM) have made positive progress on several prompts. Among them, the results of PP&PM, in addition, to prompt 1 and prompt 5, extend beyond the two baselines of fine-tuning and concat . The results indicate that our proposed auxiliary tasks to incorporate prompt is generic and can be employed with a range of genres and prompts. The primary cause of the results of individual prompts being suboptimal is that the hyperparameters of loss function α , β , and γ are not adjusted specifically for each prompt and we will further analyze the reasons for this in Section 5.3 .

( a ) Results of each prompt with BERT pre-trained encoder on QWK; ( b ) Results of each prompt with NEZHA pre-trained encoder on QWK.

5.2. Result and Effect of Auxiliary Tasks

Table 4 depicts the results of the auxiliary tasks (PP and PM) on validation set, the accuracy and F1 are both greater than 85% for BERT and 90% for NEZHA, and the model is well trained in the auxiliary task, when compared to both pre-trained models BERT and NEZHA, the latter produces better. The results of auxiliary tasks with NEZHA perform better as feature extraction modules.

Accuracy and F1 for PP and PM on validation set.

Comparing the contribution of PP and PM, as shown in Table A1 and Table 3 and Figure 3 , the contribution of PM is higher and more effective. Figure 3 a,b illustrate radar graphs of various pre-trained encoders of PP and PM across 10 prompts utilizing QWK metrics. Figure 3 a shows that the QWK value of PM is higher than PP in all but prompt 9 with BERT encoder, and Figure 3 b demonstrates that the results of PM are 60% better compared to those of PP, implying that PM is also superior to PP for a specific prompt. The PM and PP comparison results for the Total and Average datasets are provided in Figure 3 c,d. Except for the PM model with the NEZHA pre-trained encoder, which has a slightly lower QWK than the PP model, all models that use PM as a single auxiliary task perform better, further demonstrating the superiority of prompt matching in prompt representing and incorporating.

( a ) Radar graph of BERT-PP&BERT-PM; ( b ) Radar graph of NEZHA-PP&NEZHA-PM; ( c ) Results of PP and PM on QWK; ( d ) Results of PP and PM on PCC.

5.3. Effect of Loss Weight

We examine how the ratio of loss weight parameters β and γ affects the model. Figure 4 a shows that the model works best when the ratio is 1:1 on both QWK and PCC metrics. Figure A1 depicts the QWK results for various β and γ ratios, as well as revealing that the model produces the greatest results at around 1:1 for different prompts, except for prompts 1, 5, and 6, and the same is true for the average results. Concerning the issue of our model being suboptimal for individual prompts, Figure A1 illustrates that the best results for prompts 1, 5, and 6 are not achieved at 1:1, suggesting that it is inappropriate for such parameters in these prompts. Because we disorder the entire training set and fix the β and γ ratio before testing it independently, the parameters of the different prompts cannot be dynamically adjusted within a single training procedure. The reasons are to address the lack of data and also to focus more on the average performance of the model, which also prevents the model from overfitting for specific prompts. Compared to the results in Table A1 , NEZHA-PP and NEZHA-PM both outperform the baselines and the PP&PM model for prompt 1, indicating that both PP and PM can enhance the results when employed separately. For prompt 5, NEZHA-PP performs better than NEZHA-PM, showing that PP plays a greater role. The PP&PM model is already the best result for prompt 6, even though the 1:1 parameter is not optimal in Figure A1 , demonstrating that there is still potential for improvement. The information above reveals that different prompts have varying degrees of difficulty for joint training and parameter optimization of the main and auxiliary tasks, along with different conditions of applicability for the two auxiliary tasks we presented.

( a ) The effect of PP&PM in different β / γ ratios of QWK and PCC on Total dataset, we fix the value of α in this section of the experiment.; ( b ) The smoothing results for training losses across all tasks; ( c ) The results of different α : β (PP), α : γ (PM), and α : β : γ (PP&PM) ratios on QWK.

We also measure the effect of α on the model, where we fix the β / γ ratio constant at 1:1. Figure 4 c demonstrates that the PP, PM, and PP&PM models are all optimal at α : β = α : γ = 100:1, with the best QWK values for PP&PM, indicating that our suggested method of combining two auxiliary tasks for joint training is effective. The observation of [ 1 , 100 ] shows that when the ratio is small, the main task cannot be trained well, the two auxiliary tasks have a negative impact on the main task, but the single auxiliary task has less impact, indicating that multiple auxiliary tasks are more difficult to train concurrently than a single auxiliary task. In addition, future research should consider how to dynamically optimize the parameters of multiple tasks.

The training losses for ES, PP, and PM are included in Figure 4 b, and it can be seen that the loss of the main task decreases rapidly in the early stage, and the model converges around 6000 steps. The reason for faster model convergence in PM is that the task is a dichotomous classification compared to PP, which is a ten classification, and additionally, among the ten prompts, prompt 6 “A letter to parent” and prompt 9 “Parents are children’s first teachers” are more similar, making PP more difficult. As a result, further research into how to select the appropriate weight ratio and design more matching auxiliary tasks is required.

6. Conclusions and Future Work

This paper presents a pre-training and then fine-tuning model for automated essay scoring. The model incorporates the essay prompts to the model input and obtains better features more applicable to essay scoring by multi-task learning with two auxiliary tasks, prompt prediction, and prompt matching. Experiments demonstrate that the model outperforms baselines in results measured by the QWK and PCC on average across all results on the HSK dataset, indicating that our model is substantially better in terms of agreement and association. The experimental results also show that both auxiliary tasks can effectively improve the model performance, and the combination of the two auxiliary tasks with the NEZHA pre-trained encoder yields the best results, with QWK enhancing 2.5% and PCC improving 2% compared to the strong baseline, the concatenate model, on average across all results on the HSK dataset. When compared to existing neural essay scoring methods, the experimental results show that QWK improves by 7.2% and PCC improves by 8% on average across all results.

Although our work has enhanced the effectiveness of the AES system, there are still limitations. Regarding the data dimension, this research primarily investigates fusing prompt features in Chinese; other languages are not examined extensively. Nevertheless, our method is more convenient for migration than the manual annotation approach, and other languages can be directly migrated. Furthermore, other features in different languages can use our method to create similar auxiliary tasks for information fusion. Moreover, as the number of prompts grows, the difficulty of training for prompt prediction increases, and we will consider combining prompts with genre and other information to design auxiliary tasks suitable for more prompts, as well as attempting to find a balance between the number of essays and the number of prompts to make prompt prediction more efficient. The parameters of the loss function are now defined empirically at the methodological level, which is not conducive to additional auxiliary activities. In future work, we will optimize the parameter selection scheme and build dynamic parameter optimization techniques to accommodate variable numbers of auxiliary tasks. In terms of application, our approaches focus on fusing textual information in prompts, while they do not cover all prompt forms. Our system now requires additional modules for the chart and picture prompt. In future research, we will experiment with multimodal prompt data to improve the application scenarios of the AES system.

Abbreviations

The following abbreviations are used in this manuscript:

QWK and PCC for each prompt on HSK dataset, † denotes input as character; ‡ denotes input as word. The best results are in bold.

The effect of PP&PM in different β / γ ratios of QWK across all dataset, we fix the value of α in this section of the experiment.

Funding Statement

This research was funded by the National Natural Science Foundation of China (Grant No.62007004), the Major Program of the National Social Science Foundation of China (Grant No.18ZDA295), and the Doctoral Interdisciplinary Foundation Project of Beijing Normal University (Grant No.BNUXKJC2020).

Author Contributions

Conceptualization and methodology, J.S. (Jingbo Sun); writing—original draft preparation, J.S. (Jingbo Sun) and T.S.; writing—review and editing, T.S., J.S. (Jihua Song) and W.P. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Data availability statement, conflicts of interest.

The authors declare no conflict of interest.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Advertisement

Tens of thousands of MBTA riders to qualify for half-price fares

Copy the code below to embed the wbur audio player on your site.

<iframe width="100%" height="124" scrolling="no" frameborder="no" src="https://player.wbur.org/news/2024/09/03/boston-mbta-reduced-income-ticket-prices-commuters"></iframe>

• Andrea Perdomo-Hernandez

The MBTA is set to launch a new reduced fare program this week, dramatically widening the pool of people who qualify and potentially cutting transit costs in half for about 60,000 riders.

In the past, the T offered reduced fares only to seniors, people with disabilities or those under age 25 with low incomes. The new program expands access to all adults provided their incomes fall below a certain level.

“We know that public transit is absolutely critical for lower income households,” said Lynsey Heffernan, the transit authority’s chief of policy and strategic planning. She said rider surveys show that “many of them don’t have access to a car, that we are their primary mode and we know affordability is an issue.”

Heffernan said the new fare program, which the T calls the " income-eligible reduced fare program ," will allow for half-price fares on the commuter rail, bus, subway, ferry and The Ride, the T’s door-to-door transit service for people with disabilities.

Adults with incomes at or below 200% of the federal poverty level can qualify — that’s $30,120 a year for a single person or $62,400 for a family of four.

Oscar Torres, 46, is a frequent T user who's among those already benefiting from lower fares because of a disability . He catches the bus on Huntington Avenue near several of Boston's big medical centers and relies on the T to get to church, doctor’s appointments, social services and food pantries. He also takes the bus to his father’s house, where Torres is a caregiver.

“Honestly, I don't make much money. I'm very sick, I have a lot of health issues,” Torres said. “I'm grateful that I have a low-income bus pass. That helps me to survive.”

Transit advocates have for years called on the T to expand its low-income fare program. Jarred Johnson, executive director of the nonprofit TransitMatters, called the program a "hard fought victory," that will "dramatically expand access to public transit for a lot of low-income folks."

Starting Wednesday, Sept. 4, riders who meet the new requirements can apply for the reduced price passes. T officials said they hope thousands will do so. To qualify, applicants must already be signed up for another income-based benefit, like MassHealth, the state's Medicaid program, or the Supplemental Nutrition Assistance Program (SNAP).

Applications can be completed online or in-person at offices of the nonprofit Action for Boston Community Development (ABCD). The organization plans to assist applicants at its offices in downtown Boston, Dorchester, East Boston, Mattapan and Mystic Valley, and hopes to provide similar services at 36 offices throughout eastern Massachusetts.

To check eligibility, the T will rely on data from the Registry of Motor Vehicles and other state agencies that offer benefits.

“If we find their [applicants’] information, both in the RMV's database and the office of Health and Human Services database, they're instantly approved for the program,” said Marissa Rivera, the T’s deputy director of reduced fare programs. “In testing that flow with riders, we’ve seen that it takes about a minute to get approved.” Rivera estimated that about 80% of applicants will be served that way.

Applicants who do not have an RMV-issued form of identification can upload a photo of their identification and submit eligibility documents using the online application. “Those applications are reviewed within two business days,” Rivera said.

Sharon Scott-Chandler, ABCD’s chief executive, encouraged people who think they may be eligible for the MBTA program to visit an ABCD location, even if they are not signed up for any other income-based benefits.

“If folks are not on one of the designated state benefit programs, we can actually get them onto SNAP; we can actually help them access MassHealth,” Scott-Chandler said. “And so thereby making them eligible for the T pass as well.”

Some riders, like East Boston resident Lissi Guerrero, plan to apply immediately.

Speaking in Spanish, she told WBUR she was excited to learn the T was creating a reduced fare program she would qualify for. A mother of four, Guerrero said her family travels downtown for doctor's appointments and other errands, and she often has just enough money to cover their trip.

“Making transportation more affordable will help me a lot,” she said. “Not having to worry about finding a ride or hailing an expensive taxi to get around will be great.”

This segment aired on September 3, 2024.

• MBTA will offer half-price fares to low-income riders. Here's how the program would work
• Long live the passback: MBTA changes course on fare payment practice
• The MBTA's new contactless fare readers are now live. Here are some dos and don'ts

Andrea Perdomo-Hernandez Transportation Reporter Andrea Perdomo-Hernandez is a transportation reporter for WBUR.

More from WBUR

• Latest Latest
• The West The West
• Sports Sports
• Opinion Opinion
• Magazine Magazine

Opinion: For Sweden, Russian aggression is a daily worry

By Jay Evensen

For most Americans, the war in Ukraine takes on a distant hue that makes it easy to talk about in broad terms. For some, distance makes the war easier to dismiss.

To many people in Europe, however, it’s a much more vivid and constant worry.

I just returned from another two-week stay in Sweden, my third within the past 12 months. My wife and her siblings are inheriting property there.

Each time I have visited, Ukraine has been top-of-mind in the media and in casual conversation, in ways that conveyed an emotion Americans have a hard time feeling thousands of miles away. This time was no different.

“Russian plot: ‘Gotland target for nuclear weapons’” was a blaring headline less than a week ago in Expressen , a national daily newspaper I bought off the stand. Gotland is Sweden’s largest island, located in the middle of the Baltic Sea. It has been a tourist destination for many years, and it contains some of the nation’s most intriguing archeological finds (in 1999, the world’s largest collection of Viking silver treasure was discovered in a farmer’s field).

It’s also important militarily, and even more so since Sweden joined NATO earlier this year. Sweden has plans to put more advanced missile defense systems and military equipment on the island.

So when Russian military expert Vladimir Prochvatilov wrote on a Russian website that Gotland would become “target number one” for a Russian nuclear attack in the event of a war with NATO, it got people’s attention.

“It would be better for Sweden to remain a neutral country and for the island of Gotland to remain a tourist mecca, a historical museum and a nature reserve, than to become the number one target for Russian nuclear missiles,” Prochvatilov wrote.

More jarringly, he said, “The hopes of the Swedish authorities that in the event of a direct military conflict with Russia this small country will be able to avoid the total destruction of its mainland infrastructure and mass deaths of the population are illusory.”

At the start of such a war, Russia would launch nuclear weapons at Sweden “to prevent NATO cruise missile strikes on ships of the Baltic and Northern Fleets.”

In response to this, Expressen quoted an associate professor of war science at the Norwegian Defense University, who labeled it a scare tactic, or “diplomacy of intimidation,” aimed at getting Sweden to rethink its plans to militarize the island.

Reports such as these are jarring to a nation that managed to escape any conflict in either world war, and that hasn’t fought a war at all since 1814. But they also are evidence that Sweden’s decision to join NATO, and its plans for Gotland, are the right tactics.

Russia is clearly concerned. A separate story in Expressen said a Russian TV station had devoted considerable time to the militarization of Aland, a Finnish island not far from Gotland.

But perhaps more jarring than these reports are the recent statements of Sweden’s own government and military leaders.

At a conference earlier this year, Swedish Prime Minister Ulf Kristersson urged people to prepare to defend themselves “with weapons in hand and our lives on the line.”

The nation’s supreme military commander, Gen. Micael Byden , showed photos of destruction and death in Ukraine that were projected against the backdrop of a snowy Swedish field. He asked, “Do you think this could be Sweden?”

Once again, as in earlier visits, the people I spoke with expressed strong support for Sweden joining NATO.

They also expressed worries over whether Americans, and especially politicians, were as committed to keeping Europe free as they are.

On that score, recent opinion polls, such as one by the University of Maryland and SSRS, showed that support for the U.S. staying the course in its support for Ukraine is surprisingly strong. A report on the poll by the Brookings Institution said that, when asked how long the U.S. should continue military aid, 48% said as long as necessary, with another 39% saying one to two years.

When broken out by political persuasion, the poll found 37% of Republicans saying as long as necessary, with 53% saying one to two years.

This, despite statements by former President Donald Trump and Sen. J.D. Vance in opposition to further military aid.

For those who inevitably will ask, Sweden already spends 2.2% of its annual budget on its military, including an air force that has long been one of the strongest in Europe. That qualifies as exceeding the NATO target level of 2%.

Swedes no doubt would be comforted by the poll results. As I said after earlier visits, defense against Russian President Vladimir Putin is not a theoretical exercise for them, as it often seems to be in this country.

From what I know of history, the same could have been said for the way many Americans felt about belligerent countries before the outbreak of World War II.

Essay Editing Service

Get accepted at your dream school.

Expert essay editors polish your writing to reflect the work you put into it. Professional college essay editing will:

• Make sure that your essay will get read
• Make sure you'll leave a positive impression
• Make you stand out from your peers

• College Essay Editing & Coaching

Professional editors

Same day delivery, 100% happiness guarantee.

College Essay Editing Services

Our college essay editing package.

The first draft is always the hardest. Work with an admissions essay coach who will give you practical, step-by-step guidance on how to develop your essay and make your story come alive.

Your coach will…

• Read your profile
• Review your essay draft
• Share feedback to help you refine your topic, structure, and voice
• Leave in-text feedback and suggestions
• Proofread for grammar and punctuation
• Give you a roadmap to guide your revisions

We’ll help you show your best traits to college admissions officers and prove that you’re college-ready! With help from our essay coaches, you’ll improve your chances of getting accepted – so what are you waiting for?

Meet Scribbr's essay experts

At Scribbr, you can rest assured that only the best editors will work on your college essay.

All our 800+ editors have passed the challenging Scribbr Academy, which has a passing rate of only 2%.

We handpick your college essay editor on several criteria, including field of study.

Janice holds a PhD in German studies from Duke University. As a former professor, she has helped many students refine their application essays for competitive degree programs and study-abroad programs. Today, she helps students improve their writing skills at Scribbr. In her spare time, when she's not writing or editing, she enjoys working on the small family farm that she owns and operates.

Alexandra earned an Erasmus Mundus MA in cultural narratives and a BA in English from Emory University. As a teacher, editor, and writing coach, she has spent nearly a decade helping students find their voice and express their ideas.

Originally from Maryland, Amy headed west to attend Scripps College in California, where she earned a bachelor's degree in music and gender studies. In 2009, she began working for the admissions office of her alma mater, where she focused on reviewing applications and interviewing prospective students.

“Extremely professional and timely”

Extremely professional and timely! Regis did an amazing job bringing my admission essay to life. It went from overly wordy and unclear to very professional and concise. If I don’t get into my desired program, it won’t be because of my essay.

How it works

Get quality feedback in 4 easy steps, upload your college essay.

Upload your college essay , and tell us a little about yourself. We’ll match you with a certified college essay coach.

Sit back and relax

Your coach will edit your essay and share in-depth advice to help you strengthen your structure, content, and narrative.

Receive your feedback

We’ll send back your fully edited essay. Applying to college is nerve-racking, but at Scribbr you never have to wait for long.

Revise and submit your essay

Thanks to your coach’s feedback, you’ll feel clear about the next steps and confident when you hit “send” on your application.

Scribbr & academic integrity

Scribbr is committed to protecting academic integrity. Our proofreading service, our AI writing tools ( plagiarism checker , paraphrasing tool , grammar checker , summarizer,  Citation Generator ) as well as our free Knowledge Base content are designed to help students produce quality academic papers.

We make every effort to prevent our software from being used for fraudulent or manipulative purposes.

Your questions, answered.

We know that waiting for feedback can be nerve-racking. If you select our faster deadline option, you’ll hear from your personal statement editor in 12 hours or less – including on weekends and holidays. 

We never engage in rewriting or adding text at Scribbr. Our personal statement editors will make grammatical corrections and leave suggestions using Word’s track changes and commenting feature. However, we always make sure that your perspective and voice remain front and center in your essay.

We know that waiting for feedback can be nerve-racking. If you select our faster deadline option, you’ll hear from your college essay coach in 12 hours or less – including on weekends and holidays.

Yes. All our college essay coaches are subject-matter experts and can help you make decisions regarding the content of your essay. Your coach will help you make a strong case for your admission.

Our essay coaches will provide feedback on…

• Whether your topic is appropriate and how to strengthen it
• Whether your essay demonstrates positive traits and authenticity
• Whether the narrative flows smoothly and grabs the reader’s attention
• How to dig deeper and show self-reflection and insight
• What content is missing and should be added
• What content is weak and should be removed
• How to shorten your essay to meet a word count limit

However, we will never…

• Write or rewrite content for you
• Research or fact-check your essay’s content

If you have any questions, feel free to email [email protected].

When you submit your essay to our College Essay Editing & Coaching service, you’ll be asked to indicate what type of content feedback you would like to receive. You can choose from three options:

Dig deep and challenge me

Give me the standard treatment

Tread lightly

Whichever option you choose, you’ll receive in-depth content feedback from your essay coach. However, the type of feedback your essay coach provides may differ, and the option you should choose depends on what outcome you want from the review. You can always elaborate on what type of feedback you want in your note to your essay coach.

Learn more about the three options below.

Choose from three options

If you choose this option, we’ll challenge you to write the best possible essay – no content suggestions are off-limits, including reorienting your topic or reframing your story. Students who choose this option aren’t afraid to go back to the drawing board if it’ll increase their chances of getting accepted. You may have to roll up your sleeves and do some more work on your essay, but your essay will be much stronger because of it.

Select this option if your priority is to have the best possible essay, even if it requires some extra work on your side.

We’ll strike a balance between “dig deep and challenge me” and “tread lightly.” We may suggest changes to your structure and narrative, including removing weak content, elaborating on certain sections, and adding content to strengthen your arguments. You may have to rework sentences and paragraphs, but we won’t suggest major content changes.

With this option, we’ll focus on improving your essay in a way that requires little additional work from you. We’ll provide you with targeted suggestions for how to refine your essay’s structure and narrative, but we will work within your current structure and topic. While you may have to make a few tweaks here and there, we’ll aim to return you a polished draft that’s ready to be submitted to your chosen colleges.

Select this option if you are fairly satisfied with your essay and don’t want to make major changes or if you have limited time to revise.

At Scribbr, we promise to make every customer 100% happy with the service we offer. Our philosophy: Your complaint is always justified – no denial, no doubts.

Our customer support team is here to find the solution that helps you the most, whether that’s a free new edit or a refund for the service.

Get in touch, with real people

We answer your questions quickly and personally from 9:00 to 23:00 CET

• Start live chat
• Email [email protected]
• Call +1 (510) 822-8066
• WhatsApp +31 20 261 6040

Knowledge Base

Finish your college essay with scribbr’s top-rated guides.

College Essay

How to Write a College Essay

What colleges look for in an essay, choosing your college essay topic, college essay format & structure.