Community

Cheating on Chegg? How the Company Aims to Catch Tutoring Requests That Go Too Far

By Tina Nazerian     Mar 1, 2018

Cheating on Chegg? How the Company Aims to Catch Tutoring Requests That Go Too Far
Sanghamitra Deb gives her presentation at the 2018 Bay Area Learning Analytics Conference.

“I need a 10 page essay written on the deforestation of the Amazon rainforest. Must have 7 resources.”

That’s an example of a request that can be found on online tutoring services like Chegg, as shared by a senior data scientist at the company, Sanghamitra Deb. And you don’t need to be an expert in academic honor codes to know that it constitutes as cheating.

Yet this is the kind of “help” that Chegg’s tutoring service can sometimes be used for. And it’s what the Santa Clara, Calif.-based company wants to thwart.

Speaking to an audience of researchers from the education industry last week at the Bay Area Learning Analytics Conference, held at the University of California at Berkeley, Deb shared how Chegg uses a Stanford-built tool to try to keep out students who just want an online tutor to do their homework.

Among the services that the publicly-traded digital education company offers is Chegg Tutors, which matches students with tutors who help them one-on-one in a variety of subjects, such as physics and French. To get a match, though, a student must send a message describing what they need help with to either an individual tutor, or to a general form.

The company, however, does not want students to use its site to get tutors to do their work for them. So to flag such ethically dubious requests, the company uses machine learning. Deb said that while Chegg has a lot of content, it's not well structured for training machine learning algorithms. So, to solve that, the company relies on Stanford-built Snorkel, a system for creating, modeling and managing training data.

Deb said their data scientists ask subject matter experts what cheating looks like in their area of expertise. The data scientists then use that knowledge to create linguistic "rules" that have hallmarks of cheating requests in different subject areas. Snorkel looks at both context and sentence structure and applies those "rules" to generate labeled training data for building a machine learning model. When the model detects a student asking to cheat, it alerts someone who will make a final determination whether there was, indeed, a cheating request.

The company’s tutors are instructed by the company not to engage if they encounter such solicitations. If they do, they can be kicked out of the system. But Chegg doesn’t generally take action against the person making the request. “You can come here and ask [someone to do your work for you], we can’t stop you from that,” said Deb.

Universities usually have strict honor code policies about students cheating on exams, homework and other assignments. Take Stanford University’s honor code policy, which states, in part, that students “will not give or receive aid in examinations,” and “will not give or receive unpermitted aid in class work, in the preparation of reports, or in any other work that is to be used by instructor as the basis of grading.”

But the internet has given students an assortment of digital tools to get around those rules. As The Chronicle of Higher Education reported in 2016, there are plenty of websites online where students can pay someone else to complete academic work—and it can be difficult for colleges to find out. One source in the story called online tutoring services a “‘gateway drug’” to cheating.

There are opportunities for students to get answers outside of the tutoring section of the site—Chegg’s “Textbook Solutions” claims to give step-by-step solutions for over 22,000 ISBNs, and it’s “Expert Q&A” service claims a student can “ask any study question and get expert answers in as little as two hours.”

However, the Chegg Tutors honor code policy states that if a student’s institution contacts Chegg to investigate his or her academic integrity, Chegg is authorized to fully cooperate and commonly does so.

Deb said that in the long-term, Chegg doesn’t want university professors to associate it with cheating and tell students not to use the product.

“So it’s not just [about the] brand, there’s a very, very good long-term business reason to be a product that helps in learning rather than cheating.”

Update (March 2, 2018): The company has responded with the following statement about its policy towards cheating:

“Chegg does not tolerate cheating. When users violate our Honor Code policies, whether they are within the Chegg Tutor experience, while using Chegg Study, or across any of our products and services, we remove those users from the platform. We also work with school administrations on any type of honor code investigation and respond promptly.”

Community

Cheating on Chegg? How the Company Aims to Catch Tutoring Requests That Go Too Far

By Tina Nazerian     Mar 1, 2018

Cheating on Chegg? How the Company Aims to Catch Tutoring Requests That Go Too Far
Sanghamitra Deb gives her presentation at the 2018 Bay Area Learning Analytics Conference.

“I need a 10 page essay written on the deforestation of the Amazon rainforest. Must have 7 resources.”

That’s an example of a request that can be found on online tutoring services like Chegg, as shared by a senior data scientist at the company, Sanghamitra Deb. And you don’t need to be an expert in academic honor codes to know that it constitutes as cheating.

Yet this is the kind of “help” that Chegg’s tutoring service can sometimes be used for. And it’s what the Santa Clara, Calif.-based company wants to thwart.

Speaking to an audience of researchers from the education industry last week at the Bay Area Learning Analytics Conference, held at the University of California at Berkeley, Deb shared how Chegg uses a Stanford-built tool to try to keep out students who just want an online tutor to do their homework.

Among the services that the publicly-traded digital education company offers is Chegg Tutors, which matches students with tutors who help them one-on-one in a variety of subjects, such as physics and French. To get a match, though, a student must send a message describing what they need help with to either an individual tutor, or to a general form.

The company, however, does not want students to use its site to get tutors to do their work for them. So to flag such ethically dubious requests, the company uses machine learning. Deb said that while Chegg has a lot of content, it's not well structured for training machine learning algorithms. So, to solve that, the company relies on Stanford-built Snorkel, a system for creating, modeling and managing training data.

Deb said their data scientists ask subject matter experts what cheating looks like in their area of expertise. The data scientists then use that knowledge to create linguistic "rules" that have hallmarks of cheating requests in different subject areas. Snorkel looks at both context and sentence structure and applies those "rules" to generate labeled training data for building a machine learning model. When the model detects a student asking to cheat, it alerts someone who will make a final determination whether there was, indeed, a cheating request.

The company’s tutors are instructed by the company not to engage if they encounter such solicitations. If they do, they can be kicked out of the system. But Chegg doesn’t generally take action against the person making the request. “You can come here and ask [someone to do your work for you], we can’t stop you from that,” said Deb.

Universities usually have strict honor code policies about students cheating on exams, homework and other assignments. Take Stanford University’s honor code policy, which states, in part, that students “will not give or receive aid in examinations,” and “will not give or receive unpermitted aid in class work, in the preparation of reports, or in any other work that is to be used by instructor as the basis of grading.”

But the internet has given students an assortment of digital tools to get around those rules. As The Chronicle of Higher Education reported in 2016, there are plenty of websites online where students can pay someone else to complete academic work—and it can be difficult for colleges to find out. One source in the story called online tutoring services a “‘gateway drug’” to cheating.

There are opportunities for students to get answers outside of the tutoring section of the site—Chegg’s “Textbook Solutions” claims to give step-by-step solutions for over 22,000 ISBNs, and it’s “Expert Q&A” service claims a student can “ask any study question and get expert answers in as little as two hours.”

However, the Chegg Tutors honor code policy states that if a student’s institution contacts Chegg to investigate his or her academic integrity, Chegg is authorized to fully cooperate and commonly does so.

Deb said that in the long-term, Chegg doesn’t want university professors to associate it with cheating and tell students not to use the product.

“So it’s not just [about the] brand, there’s a very, very good long-term business reason to be a product that helps in learning rather than cheating.”

Update (March 2, 2018): The company has responded with the following statement about its policy towards cheating:

“Chegg does not tolerate cheating. When users violate our Honor Code policies, whether they are within the Chegg Tutor experience, while using Chegg Study, or across any of our products and services, we remove those users from the platform. We also work with school administrations on any type of honor code investigation and respond promptly.”

Next In Community

GET THE LATEST HIGHER ED NEWS
Be the first to know, with our weekly newsletter.

GET THE LATEST HIGHER ED NEWS
Be the first to know, with our weekly newsletter.