VORTRAG
*******
Oesterreichisches Forschungsinstitut fuer Artificial Intelligence(OFAI)
Freyung 6/6, A-1010 Wien
Tel.: +43-1-53361120, Fax: +43-1-5336112-77, Email: sec@ofai.at
-------------------------------------------------------------------------
Dr. Marcin Skowron
OFAI
A WEB BASED APPROACH TO FACTOID AND COMMONSENSE
KNOWLEDGE RETRIEVAL
In recent years, the amount of machine readable resources has been
rapidly increasing, including the textual content of the WWW. Although,
new generations of the Internet search engines employed much more
sophisticated methods and took the advantage of the constantly
increasing computer processing power for many Internet users,
especially for those who do not possess sufficient knowledge or
experience with using the currently available solutions, the problem
of accessing relevant and reliable information in a timely manner
awaits for its proper solution. Since language along with its
communication function is used to represent knowledge and human beings'
recognition of the world, one can also perceive the growing size of
textual resources as an asset usable for providing commonsense and
general knowledge to computer systems. Without such knowledge machines
cannot reason in a way similar to human beings, which slows down the
spread of AI systems.
The talk attempts to addresses the challenges and opportunities that
appeared along with the growth of the machine readable textual
resources by introducing methods for factoid and commonsense knowledge
retrieval from the WWW. These include the Interactive Clustering-Driven
Approach to Question Answering, the query formation method based on the
Query Generation Patterns, Support Vector Machines based question
classification with the extended set of features, as well as the
extensive utilization of the question category information in the
documents retrieval, answer candidates extraction and verification
processes of the Question Answering System. The presentation
discusses also the method for the automatic retrieval of commonsense
and general knowledge concepts from the WWW, usable by the computer
systems. The methods were applied and evaluated in the Web Based
Question Answering, Interactive Clustering-Driven Question Answering
and Web Based Knowledge Retrieval Systems.
*********
Zeit: Montag, 29. Mai 2006, 18:30 Uhr pktl.
Ort: Oesterreichisches Forschungsinstitut
fuer Artificial Intelligence, OFAI
Freyung 6, Stiege 6, 1010 Wien.
OESTERREICHISCHES FORSCHUNGSINSTITUT
FUER ARTIFICIAL INTELLIGENCE
o.Univ.-Prof. Ing. Dr. Robert Trappl