Research
I have conducted research in the areas of Educational Data Mining, Learning Analytics and Recommender Systems.
I have developed methods that were inspired by recommendation techniques but tailored to address problems in the educational domain such as predicting student grades in within-course activities, predicting final course grades for students prior to taking these courses, and generating personalized top-n course rankings to help college students find relevant courses.
We showed how leveraging domain knowledge about student-course enrollments and extracting features from the students interactions with the online learning management systems lead to more accurate prediction and ranking results.
I have also worked on combining ideas from neighborhood-based and latent-based methods to address the new item recommendation problem; which is also know as item cold-start recommendation.
I developed methods generate user-specific models that estimate how much a user would be interested in a new item.
These models utilize the item features that describe the new item and the items that the user has previously liked, as well as the global user preferences patterns in order to benefit from the similarities among the different users.
