CyberLab Graduate Research Assistant Produces New Algorithm for CN Word Cloud Analytics

Maziar Bouali, a graduate research assistant and software project manager in the IUPUI CyberLab, has researched and developed a module/framework for the Word Cloud Analytics on CN. This module aims to benefit instructors and students alike and is set to release as an LTI (Learning Tools Interoperability) tool soon.

A traditional word cloud feature counts the number of times a word is mentioned in a given text, and produces a visual compiled of popular words; the more frequently used words appear in larger text and the less frequently used words appear in smaller text. This shows the viewer the key, or most important topics at a glance.

Maziar Aboualizadeh Behbahani
Maziar’s version includes additional “input parameters”, making it a more efficient algorithm while providing meaningful, and appealing word cloud visuals to instructors and learners. His version allows the filtration of “junk words” or common words of the English language, such as articles so the system can make a more valuable selection. This “junk words” filter is customizable based on language and course needs and can be set up by CourseNetworking Admin or an Instructor. Maziar’s algorithm also allows grouping of similar words or tenses of one word, along with a “two-word coefficient”. The “two-word coefficient”, Maziar said, “was actually a request from a professor to Dr. Ali Jafari [CEO of Course Networking]” and the task was given to him. He went on to explain,

“It groups together words that are used next to one another, for example, if the two words ‘social media’ are used next to one another three times, it has more importance than just ‘social’ or just ‘media’ being used five or six times each. This logic is configurable.”

This means that his algorithm recognizes the importance of a repeated two word phrase, subject or topic and treats it as such, instead of separating the two words as a traditional word cloud would. The improvements also include the ability to ignore numbers, case of words, and enforce a minimum frequency a word must appear to be included in the world cloud visual. Along with these additions, his algorithm also allows administrators or instructors to choose the number of selected worlds. Maziar explained, “If the word cloud algorithm filters through posts and finds 100 words that may be used, an administrator can dictate, or possibly, an instructor may select to view or use the top 15 and so on.” This narrows down topics and heightens word cloud value even more.

With all of the additions to the traditional word cloud, within Maziar’s algorithm, there are many benefits to be gained from this project. He described the projected benefits and incentives for creating algorithm by saying, “it will save instructors and learners time, they can look at the word cloud from time to time and quickly gain the idea what the main discussion topics in the course were; it will also help keep them up to date with content and on track with what the major learning interests appear to be [based on size of the words].” This algorithm’s goal is to be faster, smarter, of higher quality, improve analytics and to produce a better course focus.

This update didn’t come easy, however. There is no ‘open source tool’ or template code for this type of algorithm. Maziar Bouali developed the algorithm from scratch based on his own research and use of the platform jQuery. When asked about his process, he admitted 

“Getting [the algorithm] to recognize similar words was, and continues to be the biggest challenge, humans complete tasks like this with ease but computers require artificial intelligence.” 

After his research and analyzation of other website’s capabilities, he was able to complete his development. Soon, Maziar’s LTI Word Cloud will be available at as a beta. As for the future, CourseNetworking plans to conduct extensive testing and then release Maziar’s algorithm as an LTI tool that may be added to other LMS as a plug in, for example, the tool could be added to Canvas, Sakai, Moodle, and others. Maziar says there could even be talk of releasing his project as an open source tool someday. He is set to graduate from IUPUI with a Master’s Degree in Electronic Computer Engineering in May of 2016 and will continue his work as a software developer.

No comments:

CourseNetworking LCC. Powered by Blogger.