Saturday, March 6Digital Marketing Journals

Use of NLP in Canvas. Abstract | by Yash Choksi | Dec, 2020

Yash Choksi

In this project we are creating prototype and then working model for Canvas school mmannagement application. As we are students of California State University- Los Angeles we are using canvas as our daily class management application which deals managing assignments, homeworks and even discussions on any specific topics and many more. Now it is really overwelming taks to reach out to each and every demand which is createed by different professors. To tackle that problem we are introcuding one solution for voice assistant in canvas application. When anny student login into his account it clearly shows them what are pennding tasks and many more instructions. Wee are also extending our feeatures to read out discussions, give summary of long discussions and for that we are using topic modelling in NLP and document classification techniques. And this will also help people who can’t able to see or having some vision problems.

Enrolling into a university is a major commitment fraught with many challenges in which students must face to succeed. One of its’ challenge is the management of schoolwork, such as class assignments, readings, discussion boards, exams, and projects that are the responsibility of the student to complete. Fortunately, online learning platforms such as Canvas have tried to make the management of these tasks easier for students, although Canvas still has its share of imperfections.

One area in need of improvement, is that students still have the burden to manage the Canvas platform to stay updated with their assignments. If a student does not routinely check their canvas notifications, they may miss important announcements or due dates and start falling behind in class. Our research has discovered that students still struggle with Canvas, sometimes finding it difficult or cumbersome to keep up with all the latest class discussion, postings, due dates, and other announcements.

We distributed surveys, conducted interviews, and invited other groups of students to join our focus groups to determine our virtual assistant is effective and convenient for CSULA students. We also created personas and ideation through our research and developed a prototype.

Our survey is based of Cal State LA graduate students, college of Business & Economics.

Survey shows that students are NOT satisfied or somewhat. Statisfed with the crreunt notification system in canvas

1. 3 Tips for your Voice and Chatbot Program from Gartner’s Customer Service Hype Cycle 2020

2. Deploying Watson Assistant Web Chat in Salesforce Lightning Console

3. Are Chatbots Vulnerable? Best Practices to Ensure Chatbots Security

4. Your Path to AI — An IBM Developer Series

Students miss assignments/or tests because canvas is hard to navigate. Canvas does not have a reminder. If a reminder is available or easily accessible students can achieve the highest grade in class.

Majority of survey responses indicate that students are missed a school deadline during their career.

Our survey shows that students miss a deadline a lot. > 33% of students want to do the class work, but the way Canvas is structured, it’s marking easier for students to miss out on class work.

According to above chart, in 2017, there are 81% students are retention in a year. In 2013, Students took about 6 years to graduated. By developing virtual assistant will help with the student efficiency and time management.

According to 2019 Cal State Los Angeles statistics data, there are only 16.5% of students were continuing their education after undergraduate. In addition, graduate enrollment has switched from predominantly male to female. There are two major reasons that students are not able to pursue higher education which are financial and time management. With our Virtual assistant, student can be able to listen their project and assignment by hearing while they are busy doing other duties. They can perform multitask.

In order to improve student user experience in Canvas, empathy is required to understand the struggle students face with the learning platform. One of the first obstacles faced by students is when they first log into Canvas, unfamiliar with its features and functions. The time it takes to become familiar with Canvas is long and tedious, forcing students to implement their own strategy to best manage Canvas. Overtime, they might become more familiar with the learning platform, but the responsibility of the student to keep updated with Canvas activities never changes. This responsibility seems to an inefficient use of student’s time, as students spend too much time trying to locate answers on some of the most common questions asked.

What is due today?

What is due next week?

When time is the quiz?

Was there anything new discussed?

Any new assignments posted?

These questions, among others are always racing through the students mind as they try to keep up with their classes. As most CSULA courses implement Canvas, there is a desire to make the student experience better and more time efficient. Doing so allows students to focus in learning the subject matter and not waste time trying to manage Canvas. Fortunately, Team Exploration consist of students who are currently experiencing the struggles mentioned and are passionate about making a better experience for future students. Our work consists of interviews, surveys, and focus groups which helped narrow our project scope. As a team, we created personas that depict the different types of students enrolled in Cal State LA and their experiences with Canvas.

How might we make Canvas more powerful in meeting student challenges and requirements through voice assistance feature, so that Text to Speech(TTS) helps students achieve their academic goals? Conducting the surveys and interviews and creating personas, we realized that the key objectives of the students are not only to obtain a high GPA academic degree, but also to achieve the most learning outcomes and to create networks for job opportunities. The motivations are to meet university’s expectations and fulfill the responsibilities as students with a key term ”high standard”.

We narrowed down the constraints, pain points and desires into three themes. The categorization of themes allows us first to prioritize and tackle the highest impact needs in every context. The contexts are undergraduate student, graduate student either full time or part time and a single mom who studies for master’s degree.

  • How might we make checking the reminders flexible and more efficient with TTS?
  • How might TTS help students with time management for assignments and deadlines?
  • How might the discussions be augmented with TTS, so that students’ activities in discussions be streamlined?

Students would love to use voice assistance but still want to configure features including adjustable volume and speed and scheduling the playback according to the remaining time of the deadline, while enjoying the tone of voice.

Plus, students must be certain that they are informed of any important update and that they will not miss an assignment before its due date.

Discussions are a great potential for sharing useful information and networking, although they tend to get cluttered and excessive or simply no more time is left for students to read the discussions thoroughly.

Last but not least, students want the Canvas in general and Voice assistance in particular to be simple and user friendly and provide students with a tutorial, help and feedback mechanism.

Subsequently, a group of personas emerged that shows students behaviors, their goals, and problems and how voice assistance can improve their experience. The below story board depicts a web of students in three different contexts as undergraduate student (Alex), a graduate student (Stacy) and a graduate student with major commitments (Suzie).

We imported the result of how-might-we’s, combined with our own knowledge and experience, into 2×2 matrix and relational map, and over 30 ideas were developed. Refining the ideas, we selected part of the ideas into bundles for the prototype.

When the user logins to Canvas, number of pending assignments are announced. An acknowledgement from students is used as an assurance that the student is aware of the assignment. A star reward program can be structured to encourage the students to complete the tasks on time and keep the track of their completions. Announcement of assignments with the past-due helps students to delete old assignments more precisely. Students can read the list of tasks and their due dates by clicking on the To-Do list. Also, there is an option to read the assignments selectively based on the time available for due dates, from playing all assignments to playing only the urgent ones. By clicking on the voice tab in calendar, the assignments with the due dates of the same day are read.

By integrating the TTS into discussions, number of discussions, the sender name and the content of the discussion are read. Students have the option to hear just the first sentence of each thread or to hear the entire thread.

The implementation needs to be user friendly and secure. A username and password are required for access control. The tutorial provides students with detailed features and facilities of voice assistance and a send tab for questions and support.

For general appearance, students have options of voice speed, turnoff or mute features.

We are trying to depict scenario and this are 3 versions of issues:

You can find our working prototype on this link

As we have created prototype about this project and my team is very passionate about data science and Machine Learning approaches then I would like to proceed with creating Neural Networks for NLP model which can able to incorporated into the acutal canvas application.

  1. Adrian Maroquin
  2. Fereshteh Mamaghani
  3. Saranna Khaing
  4. Yash Choksi

Leave a Reply