Entity recommendation and search in heterogeneous information networks

Yu, Xiao

Permalink

https://hdl.handle.net/2142/49374 Copy

Description

Title

Entity recommendation and search in heterogeneous information networks

Author(s)

Yu, Xiao

Issue Date

2014-05-30T16:40:46Z

Director of Research (if dissertation) or Advisor (if thesis)

Han, Jiawei

Doctoral Committee Chair(s)

Han, Jiawei

Committee Member(s)

• Zhai, ChengXiang
• Huang, Thomas S.
• Ma, Hao

Department of Study

Computer Science

Discipline

Computer Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Date of Ingest

2014-05-30T16:40:46Z

Keyword(s)

• entity recommendation
• entity search
• information network

Abstract

With the rapid development of social media and information network-based web services, data mining studies on network analysis have gained increasing attention in recent years. Many early studies focus on homogeneous network mining, with the assumption that the network nodes and links are of the same type (e.g., social networks). However, real-world data in many domains and applications are often multi-typed and interconnected, forming heterogeneous information networks. The objective of my thesis is to study effective and scalable approaches to help users explore and discover useful information and knowledge in heterogeneous information networks. I also aim to advance the principles and methodologies of mining heterogeneous information networks through these studies. Specifically, I study and focus on entity recommendation and search related problems in heterogeneous information networks. I investigate and propose data mining methodologies to facilitate the construction of entity recommender systems and search engines for heterogeneous networks. In this thesis, I first propose to study entity recommendation problem in heterogeneous information network scope with implicit feedback. Second, I study a real-world large-scale entity recommendation application with commercial search engine user logs and a web-scale entity graph. Third, I combine text information and heterogeneous relationships between entities to study citation prediction and search problem in bibliographical networks. Fourth, I introduce a user-guided entity similarity search framework in information networks to integrate users' guidance into entity search process, which helps alleviate entity similarity ambiguity problem in heterogeneous networks. The methodologies proposed in this thesis are critically important for information exploration in heterogeneous information networks. The principles and theoretical findings in these studies have potential impact in other information network related research fields and can be applied in a wide range of real-world applications.

Graduation Semester

2014-05

Permalink

http://hdl.handle.net/2142/49374

Copyright and License Information

Copyright 2014 Xiao Yu

Owning Collections