Recommendation Algorithms for E-Commerce and Education

Major Research Products 

Sparse linear methods for top-n recommender systems

Xia Ning and George Karypis. SLIM: Sparse linear methods for top-n recommender systems. In Proceedings of the 2011 IEEE 11th International Conference on Data Mining, ICDM’11, pages 497–506, Dec 2011.

Abstract: This paper focuses on developing effective and efficient algorithms for top-N recommender systems. A novel Sparse Linear Method (SLIM) is proposed, which generates top-N recommendations by aggregating from user purchase/rating profiles. A sparse aggregation coefficient matrix W is learned from SLIM by solving an ℓ 1 -norm and ℓ 2 -norm regularized optimization problem. W is demonstrated to produce high quality recommendations and its sparsity allows SLIM to generate recommendations very fast. A comprehensive set of experiments is conducted by comparing the SLIM method and other state-of-the-art top-N recommendation methods. The experiments show that SLIM achieves significant improvements both in run time performance and recommendation quality over the best existing methods.

The SLIM code is available here.

SLIM has been used by Mendeley, Alibaba, Ebay, etc., and has been re-implemented in R, Python, C# and Java, and included in various libraries (librec, mymedialite,  mrec) and textbooks.

Recommendation algorithms for educational data mining

Zhiyun Ren, Xia Ning, and Huzefa Rangwala. Grade prediction with temporal course-wise influence. In Proceedings of the 10th International Confernece on Educational Data Mining, EDM’17, pages 48–56, 2017.

Abstract:  There is a critical need to develop new educational technology applications that analyze the data collected by universities to ensure that students graduate in a timely fashion (4 to 6 years); and they are well prepared for jobs in their respective elds of study. In this paper, we present a novel approach for analyzing historical educational records from a large, public university to perform next-term grade prediction; i.e., to estimate the grades that a student will get in a course that he/she will enroll in the next term. Accurate next-term grade prediction holds the promise for better student degree planning, personalized advising and automated interventions to ensure that students stay on track in their chosen degree program and graduate on time. We present a factorization-based approach called Matrix Factorization with Temporal Course-wise Influence that incorporates course-wise influence effects and temporal eects for grade prediction. In this model, students and courses are represented in a latent “knowledge” space. The grade of a student on a course is modeled as the similarity of their latent representation in the “knowledge” space. Course-wise influence is considered as an additional factor in the grade prediction. Our experimental results show that the proposed method outperforms several baseline approaches and infer meaningful patterns between pairs of courses within academic programs.