Flexible recommendations over rich data

Abstract

CourseRank is a course planning tool aimed at helping students at Stanford. Recommendations comprise an integral part of the system. However, implementing existing recommendation methods leads to fixed, pre-specified recommendations that cannot adapt to each particular student's changing requirements and do not help exploit the full extent of the available learning opportunities at the university. In this paper, we describe the concept of a flexible recommendation workflow, i.e., a high-level description of a parameterized process for computing recommendations. The input parameters of a flexible recommendation process comprise the "knobs" that control the final output and hence support flexible recommendations. We describe how flexible recommendations can be expressed over a relational database and we present our prototype system that allows defining and executing different, fully-parameterized, recommendation workflows over relational data. Finally, we describe a user interface in CourseRank that allows students to make use of two flexible recommendation workflows.

Evaluating Recommender Systems for Technology Enhanced Learning: A Quantitative Survey

“…Functional testing of prototypes, as performed e.g. in [38], [39], [40] and [41] were not considered as an evaluation of the recommender system. Some publications covered several evaluation methods, for example, an offline experiment and a complementary user study in [82].…”

Recommendation Systems for Personalized Technology-Enhanced Learning

“…Once sequences have been discovered and saved, all related recommended links are shown to the teacher for eventual validation in order to select which links should be used by a recommendation engine to the student. Koutrika et al (2008) presented an example of a closed-community social system named CourseRank, which is an educational and social site where students can explore learning materials. Students can search for courses of interest, rank the accuracy of each other's comments and receive personalized recommendations.…”

FlexRecs

“…Designing, implementing and experimenting with new methods can be time-consuming and counterproductive [15]. Furthermore, hard-wired algorithms generate only a predefined and fixed set of recommendations, which cannot capture the real-time user information needs.…”

“…The inherent limitations of recommendation systems have been acknowledged [5,15] and some extensions have been recently proposed, such as incorporating multi-criteria ratings into recommendations [3]. The language RQL has been the first proposal that allows users to formulate recommendations in a flexible manner [4].…”