Login

Publications  •  Project Statistics

Glossary  •  Schools  •  Disciplines
People Search: 
   
Title/Abstract Search: 

Dissertation Information for Douglas W. Oard

NAME:
- Douglas W. Oard

DEGREE:
- Ph.D.

DISCIPLINE:
- Computer Engineering

SCHOOL:
- University of Maryland (USA) (1996)

ADVISORS:
- None

COMMITTEE MEMBERS:
- None

MPACT Status: Incomplete - Not_Inspected

Title: Adaptive vector space text filtering for monolingual and cross-language application

Abstract: Adaptive Multilingual Text Filtering is an innovative approach for enhancing the usability of the global information infrastructure. With the recent explosive growth of intranets and the Internet, mediated access to networked information is becoming an increasingly important problem. Much of this information is in (or can be converted to) text form, and an increasing amount is expressed in languages other than English. Text filtering is an information access process in which dynamic information sources are matched against relatively stable information needs; adaptive text filtering systems seek to assist users by automatically improving their profile of the users' information needs over time; and adaptive multilingual text filtering systems seek to do so for collections of text which contain more than one language. The distinguishing feature of adaptive multilingual text filtering is that evidence obtained by observing reading behavior on text in one language can be used to predict interest in texts in another language. Existing multilingual text filtering systems are not adaptive, requiring instead that users explicitly specify their information needs using words from each language of interest.

In this dissertation, present practice in adaptive text filtering is reviewed and a simple monolingual technique based on Latent Semantic Indexing with obvious potential for application to multilingual filtering is selected. Further refinement of this monolingual Latent Semantic Indexing filtering technique is found to result in no performance improvement, so that technique is adopted unchanged as the basis for multilingual filtering experiments. The present state of the art in the related multilingual text retrieval problem is then surveyed. Cross-language application of Latent Semantic Indexing is described and a second technique based on fully automatic machine translation is introduced. A "vector translation" technique designed specifically for adaptive multilingual text filtering is then introduced and an experimental methodology which exploits existing test collections is presented. Experimental results reveal a cost-performance tradeoff among the techniques. Implications of these results for future work on adaptive multilingual text filtering are discussed, and the dissertation concludes with an assessment of the present state of the field and the potential for further advances.

MPACT Scores for Douglas W. Oard

A = 2
C = 1
A+C = 3
T = 2
G = 1
W = 2
TD = 2
TA = 0
calculated 2010-09-27 21:17:07

Advisors and Advisees Graph