Information Technology Reference
In-Depth Information
Chapter 4
Personalized Multimedia Search
Abstract
Given results from multimedia content analysis and user modeling,
personalized multimedia services are developed to satisfy customized needs. In
this chapter, we introduce a general solution framework for personalized multi-
media search. We first propose a multimodal generative model to simultaneously
address tasks of multimedia content analysis and user modeling, and then present
the risk minimization-based theoretical framework for personalized image search.
The framework considers the noisy tag issue and enables easy incorporation of social
relation.
4.1 Introduction
Keyword-based search has been the most popular search paradigm in today's search
market. Despite simplicity and efficiency, the performance of keyword-based search
is far from satisfying. Investigation has indicated its poor user experience: on Google
search, for 52% of 20,000 queries, searchers did not find any relevant results [
29
].
This is due to two reasons. (1) Queries are in general short and nonspecific, e.g.,
the query of “IR” has the interpretation of both information retrieval and infrared.
(2) Usersmay have different intentions for the same query, e.g., searching for “jaguar”
by a car fan has a completely different meaning from searching by an animal spe-
cialist. One solution to address these problems is
personalized image search
, where
user-specific information is considered to distinguish the exact intentions of the user
queries and re-rank the list results. Given the large and growing importance of search
engines, personalized search has the potential to significantly improve searching
experience.
Compared with non-personalized search, in personalized search, the rank of a
document (web page, image, video, etc.) in the result list is decided not only by the
query, but also by the preference of user. Combined with the discussion in previous
chapters, practical personalizedmultimedia services need to integrate the results from
multimedia content analysis and user modeling. In this chapter, for multimedia con-
tent analysis, we consider multimodal topic modeling, to obtain the photos' semantic
topic distribution according the associated textual metadata as well as visual informa-
tion. For user modeling, we consider the social influence, one type of social relation
