Distance Teaching and Mobile Learning (DTML) had the honor of sharing one of its most cutting-edge developments with an elite audience recently, presenting at the Fall 2018 edition of the Machine Learning, AI & Data Science (MLADS) conference.
DTML’s Dr. Aleksey Sinyagin, VIkas Mohandoss, and Ben Pham dazzled the assembled crowd with their presentation on "Learning English Through Intelligent Games on DTML.org platform”.
The presentation revealed DMTL’s series of experiments using words extracted from children’s movies, cartoons, and books. By illustrating various characteristics, DMTL hopes to develop better educational programs by evaluating the reading skill of students.
How It Works
The experiment was run to detect if frequency groups could be used to project student language proficiency. Using constructed graphs, frequency, and POS tables, DMTL developed a set of learning activities on its web portal to analyze student proficiency based on what words students could recognize. One of the experiment’s hypothesis is that word frequency index in children’s books and movies can serve as an indicator to assess individual language proficiency; the assumption being that words in more frequent groups should be more easy to recognized than those in less frequent groups. Validation from the hypothesis came in the form of a game called Words Battle in which students are presented English words and challenged to translate them into the same word of their native language. Words were placed into 11 frequency groups and students were randomly assigned words from all 11 groups.
The results of 1,900+ students were recorded and the data speaks to the original hypothesis. Students were able to answer correctly to 42% of the Level 1 words - the ones categorized as most easy to recognize - compared with just 28% of the Level 11 words - the ones deemed the toughest to recognize. The expeiment showed higher words retention rate and increased lenght of user learning session.
The Fall 2018 Machine Learning, AI & Data Science (MLADS) Conference was the eighth in a popular series that was launched in 2014. More than 3,500 turned out in Redmond, Washington, for the event, which was also live streamed around the world. The Fall 2018 conference packed in 95 talk sessions, 20 tutorials, and 65 poster/demo sessions. Topics covered ranged from Azure Machine Learning to TIme Series Methods to Cognitive Services to Forecasting. The very latest AI tools and applications were on hand. Every session, a call for for content goes out and several hundred submissions are sent in. The DTML presentation was one of just a handful to be selected for a live showing.