Student research opportunities

Improving Recommendation Quality Through Entity Resolution

Project Code: CECS_775

This project is available at the following levels:
Honours, Summer Scholar, Masters, PhD

Keywords:

Recommender Systems, Data Quality, Entity Resolution, Data Matching, Web 2.0, Web Mining

Supervisors:

Assoc Professor Peter Christen
Dr Huizhi (Elly) Liang

Outline:

Recommender Systems are a kind of effective tool to help users solve the information overload issue. They have been popularly used in various kinds of websites such as Amazon.com to suggest most relevant items (e.g., books, movies) to users. Accuracy and diversity are two important measures of recommendation quality. In web 2.0, the read only web becomes a read and write web. Users can publish their opinions and reviews, share resources (e.g., book marks, photos, videos) and experiences, and build social networks. However, the existence of various kinds of versions of the same item (e.g., video clips, books, and photos) bring challenges to make quality recommendations. Entity resolution, also known as data matching, or record linkage, is the process of identifying which records in databases refer to the same real-world entity. This project will explore how to detect the items that belong to the same entity, and how to improve recommendation quality through adopting advanced entity resolution approaches.

Goals of this project

The objectives of this project are to design algorithms to detect item records that belong to the same entity, design algorithms to make quality recommendations, conduct experimental evaluations on large scale datasets crawled from real-world web 2.0 applications such as Delicious.com, and Twitter.com.

Requirements/Prerequisites

Having attended courses in algorithms and data structures
Good programming skills such as Java or Python.
knowledge about data mining or machine learning is a plus.

Student Gain

Students will learn about recommender systems, data mining and entity resolution techniques, web mining and personalization approaches

Background Literature

1.Badrul Sarwar et. al., Item-based Collaborative Filtering Recommendation Algorithms, WWW’01.
2.Gediminas Adomavicius et. al., Toward the Next Generation of Recommender Systems: A Survey of the State-of-the-Art and Possible Extensions. IEEE Transactions on Knowledge and Data Engineering.
3.Peter Christen, et. al., Similarity-Aware Indexing for Real-Time Entity Resolution,CIKM'09.