Document Summary System™ - About Us - Biographies
(back)
Dr. Chua Tat-Seng obtained his PhD from the
University of Leeds, UK and is currently the
Professor at the School of Computing, National University of Singapore.
Dr. Chua was the Acting and Founding Dean of Computing from 1998-2000 and spent three
years as a research staff member at the
Institute of Systems Science (now IR2) in late 1980s.
His main research interest is in multimedia information processing, in particular, on
the extraction, retrieval and question-answering (QA) of video and
text information. He focuses on the use of relations between entities
and external information & knowledge sources to enhance information
processing.
Current projects include: news video retrieval and
tracking, question answering (QA), video QA, and information extraction
on the web and regularly participates in TRE-QA and
TRECVID news video retrieval evaluations.
He is also active in the international research community where he
has organized and served as program committee member
of numerous international conferences in the areas of computer graphics,
multimedia and text processing. Chua is the
Conference Co-Chair of CIVR' 2005, ACM Multimedia 2005, and ACM SIGIR 2008.
Dr. Chua serves in the editorial
boards of: The Visual Computer(Springer-Verlag) and Multimedia Tools
and Applications (Kluwer).
He is the member of Steering Committee of Computer Graphics Society and
Multimedia Tools and Applications (international), and in Review Panel to
a Research Institute in Europe. In the industry front, Dr Chua serves as Chair of Board of Assessor of Certified
IT Project Management (CITPM) and "Certification in Outsourcing Management for IT (COMIT).
Research Activities and Interests:
Dr. Chua's research interests mainly includes multimedia information
processing, in particular, on the extraction, retrieval and question-answering
(QA) of video and text information. Currently Dr. Tat-Seng is focusing
on the use of relations between entities and external information and
knowledge sources to enhance information processing.
Current projects:
Dr. Chua's current projects include news video retrieval and tracking,
personal photo album, QA,
video QA, and information issues in the related areas to facilitate practical
deployment of technologies to solving real-life problems.
Video Modeling and Retrieval:
Dr. Chua's long term goal is to develop automated techniques
to index an input video stream to facilitate the retrieval, summarization
and personalization of video. Current work targets include news video retrieval,
in particular on TRECVID data. Work also includes sports, movies
and paintings. The common focus is on use of external information
sources and domain knowledge to perform ontology-based learning for
event detection, indexing and retrieval.
Topics included under research:
News Video Retrieval: This research aims to develop a fully automated news video retrieval system
that is capable of retrieving relevant video shots using a multi-media
query. The system exploits the domain models for news, together with
speech (in terms of ASR or Automatic Speech Recognition output) and
various audio-visual (AV) features inherent in a news video stream.
Research focuses on the following aspects of news modeling.
-
-
Photo Album: This research explores automated techniques to annotate the "who" information
in family photos. The system exploits the fact that in family album,
the same groups of people tend to appear in similar events, in which
they tend to wear the same clothes within a short time duration and
in nearby places. Therefore the use of social context and visual
context (body) information to estimate the probability of the person's
presence and identify other examples of the same recognized person.
The system first performs face recognition. However, its performance
is limited by the uncontrolled condition of family photos. Next,
social context information is used to cluster photos into events.
Within each event, the body information is clustered. The three sources
of information are eventually fused for person identification.
Experiments on a photo album containing over 1500 photos demonstrate
that this approach is effective.
Current work focuses on the use of social network information and AV analysis
techniques for the general problem of concept annotation and propagation.
-
Other Domains: Along the same theme of utilizing domain knowledge models and
external information, exploration of the problems of
event detection in movie and sports video, and the automated annotation
of paintings with high level concepts.
The key research issues explored
are: ontology-based learning; and
transductive learning as training data are hard to come by.
In the case of movie, current development includes an
automated system to index movies based on ASR, video OCR, and concepts
and events extracted from multi-modal analysis.
-
Processing - Passage Retrieval and Question Answering: Current focus pertains to question answering (qa) from open domain (free-)
text corpus, mainly in news. Depending on the type of questions, development
is targeted toward techniques to return precise answers either at
the phrase level, sentence level, or as query-focused summary.
Exploration into the techniques to answer deductive and exploratory questions
is currently being researched.
Currently focus is toward extending the technique for precise
information retrieval and also applying the technique to vertical
domains of education and legal search.
-
Question-Answering: Question-answering (QA) aims to find exact answers to a user's natural
language queries, instead of ranked lists of documents as is done in
current search engines. It is a major step toward information retrieval
instead of document retrieval.
This QA system employs a pipeline structure
that consists of several modules to obtain short and precise answers
to a user's questions.
It searches for answers at increasingly finer-grained
units of:
(1) locating the relevant documents,
(2) retrieving passages that may contain the answer, and
(3) pinpointing the exact answer from candidate passages.
-
Publications:
-
View Professor Chua's
Web Site
for his many Publications.