Ranking Oriented Algorithms for Time and Relation Aware Recommendation

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


PhD Thesis Defence


Title: "Ranking Oriented Algorithms for Time and Relation Aware Recommendation"

By

Mr. Nan Liu


Abstract

Recommender systems have become increasingly important due to the ubiquity of 
information overload across various application domains.Unlike search systems 
in which the user would specify their information need, recommender systems 
have to infer user's information needs from observed user activities in order 
to help user discovery interesting and novel items. As the technology and 
application of recommendation is rapidly evolving in these years, traditional 
collaborative filtering algorithms such as nearest neighbor or matrix 
factorization have fallen short in coping with several emerging but critical 
issues in modern systems. Firstly, ranking items, especially identifying a few 
most interesting items out of a huge pool, has become the core task in most 
application scenarios. However, traditional algorithms focus on doing 
regression on the observed user ratings (i.e., explicit user feedback), which 
is a detour towards the end goal of ranking. In this work, we propose a new 
framework for directly solving the personalized ranking problem by representing 
user feedback using pairwise preference based representation. We show that the 
ranking model provides a unified framework for handling both explicit feedback 
(e.g., ratings) and implicit feedback (e.g., clicks, purchases) as well as 
combination of heterogeneous user feedback, which is a setting that commonly 
arises in modern applications. Secondly, we extend the proposed ranking model 
to also consider the temporal context, as time awareness is becoming an 
increasingly important feature in real world applications, which often need to 
cope rich temporal dynamics and provide context aware recommendations. Finally, 
we further the extend the framework to also consider relational information 
about users and/or items. In particular, we consider the social relations among 
users the taxonomical relations between items, which are commonly found in real 
world systems. Our results demonstrate that utilizing these additional 
knowledge could greatly improve upon pure CF algorithms under data sparsity 
conditions.


Date:			Wednesday, 7 December 2011

Time:			1:00pm - 4:00pm

Venue:			Room 3402
 			Lifts 17/18

Chairman:		Prof. Howard Luong (ECE)

Committee Members:	Prof. Qiang Yang (Supervisor)
 			Prof. Dik-Lun Lee
 			Prof. Wilfred Ng
 			Prof. Kwok-Yip Szeto (PHYS)
                      	Prof. Qing Li (Comp. Sci., CityU)


**** ALL are Welcome ****