Intelligent information tasks such as summarization and categorization are performed based on criteria driven by users' perceptions.   However, conventional information systems determine such criteria in a static manner so that the results are often different from what users expect, i.e. that users' perceptions are enforced to adapt these systems.  The reason is that symbolic and discrete approach (hard computing) is often used. However, it does not get along with users' perceptions well because such static way treats tolerance as a negative factor.

Soft computing, a new emerging computational paradigm consisting of technologies such as fuzzy sets, neural networks, and genetic algorithms, arises as antithesis of hard computing in the extent of nature of human beings such as tolerance and granulation.   One of the main features of soft computing is its adaptive behavior so that systems adapt users' perceptions.  It has been successfully applied in the continuous domain such as image recognition, signal processing, and market forecasting. Fuzzy sets play core roles of soft computing in order to provide a map between linguistic and numeric representation.

In this thesis a simple intelligent information processing method based on users' perception, named perceptual information processing (PIP), is proposed. This generic method has an adaptive and feedback behavior consisting of the following tasks: conceptualization to form an initial perceptions, recognition to perform a required intelligent task such as summarization and categorization, and refinement to adopt users' perception by comparing results coming out of the system and the user.  Fuzzy sets are used as the representation of perception, and a soft computing method called Mass Assignment Theory (MAT) is used as the computational core in order to keep consistency between evidential data sets and perceptions as well as granules. As an application of PIP, text categorization based on word frequencies is studied because (1) there are not many works of the similar approach in discrete domain such as text, and (2) varieties of applications in Internet societies are available.

The challenges of this research are (1) to implement a novel method with adaptive behavior by handling perception and granule consistently and (2) to
explore its applicability in the new domain, i.e. text.

KEYWORDS: Perceptual Information Processing, Fuzzy Sets, Mass Assignments, Soft Computing, Text Categorization.

Mr. INOUE received both BA and MA in Computer Science from Eastern Washington University, Cheney, WA in 1989 and 1990 respectively.  He was culturally diversified since 1984 when he came to the U.S.A. from Japan.  He was a researcher of CAD/CAM/CAE at Hitachi Research Laboratory of Hitachi Ltd., IBARAKI, JAPAN from 1990 to 1996 where he was involved in research and development of various intelligent engineering information systems.  He took on-leave from 1993 to 1995 and joined Laboratory for International Fuzzy Engineering Research (LIFE) to study fuzzy theory in order to handle fuzziness as a positive, fundamental feature for information processing.   He is currently pursuing his Ph.D. degree in Computer Engineering and Science at the Department of Electrical and Computer Engineering and Computer Science of the University of Cincinnati, Cincinnati, OH from 1996.  At the same time, he is working as a full-time software engineer at EVOLV Adaptive Technology, Cincinnati, OH to explore commercial application of his dissertation research, i.e. Perceptual Information Processing.

His research interest is focused on concern of handling semantics in information processing. It includes topics from both theoretical and practical aspects such as soft computing, natural language processing, database, and client-server systems for the Internet. He is a member of Japan Society for Fuzzy Theory and Systems (SOFT) since 1993.

For more information, contact Venu Dasigi.

This page is maintained by Venu Dasigi.