13-03-2012, 04:35 PM
Mining Web Graphs for Recommendations
05680907 (1).pdf (Size: 1.78 MB / Downloads: 76)
INTRODUCTION
With the diverse and explosive growth of Web information,
how to organize and utilize the information effectively and
efficiently has become more and more critical. This is especially
important for Web 2.0 related applications since usergenerated
information is more free-style and less structured,
which increases the difficulties in mining useful information
from these data sources. In order to satisfy the information
needs of Web users and improve the user experience in
many Web applications, Recommender Systems, have been
well studied in academia and widely deployed in industry.
Typically, recommender systems are based on Collaborative
Filtering [14], [22], [25], [41], [46], [49], which is a technique
that automatically predicts the interest of an active user
by collecting rating information from other similar users or
items. The underlying assumption of collaborative filtering
is that the active user will prefer those items which other
similar users prefer [38].
RELATED WORK
Recommendation on the Web is a general term representing
a specific type of information filtering technique that attempts
to present information items (queries, movies, images, books,
Web pages, etc.) that are likely of interest to the users. In this
section, we review several work related to recommendation,
including collaborative filtering, query suggestion techniques,
image recommendation methods, and clickthrough data analysis.
2.1 Collaborative Filtering
Two types of collaborative filtering approaches are widely
studied: neighborhood-based and model-based.
The neighborhood-based approaches are the most popular
prediction methods and are widely adopted in commercial
collaborative filtering systems [37], [47]. The most analyzed
examples of neighborhood-based collaborative filtering include
user-based approaches [7], [21] and item-based approaches
[15], [37], [50]. User-based approaches predict the
ratings of active users based on the ratings of their similar
users, and item-based approaches predict the ratings of active
users based on the computed information of items similar
to those chosen by the active user.
2.2 Query Suggestion
In order to recommend relevant queries to Web users, a
valuable technique, query suggestion, has been employed by
some prominent commercial search engines, such as Yahoo!3,
Live Search4, Ask5 and Google6. However, due to commercial
reasons, few public papers have been released to reveal the
methods they adopt.
The goal of query suggestion is similar to that of query
expansion [11], [13], [56], [61], query substitution [31] and
query refinement [35], [57], which all focus on understanding
users’ search intentions and improving the queries submitted
by users. Query suggestion is closely related to query expansion
or query substitution, which extends the original query
with new search terms to narrow down the scope of the search.
But different from query expansion, query suggestion aims
to suggest full queries that have been formulated by previous
users so that query integrity and coherence are preserved in the
suggested queries [18]. Query refinement is another closelyrelated
notion, since the objective of query refinement is
interactively recommending new queries related to a particular
query.