A method for reusing Web browsing experience to enhance Web information retrieval

Guangfeng Song, Purdue University

Abstract

Human-Web interaction in information retrieval generates rich resources of human knowledge and experience yet to be explored. This study reviews both the human and computer aspects of Web browsing, identifies several problems in the interaction paradigm, and discusses how these problems can be addressed by reusing of web browsing experience. An object-oriented model for Web information retrieval is proposed in an attempt to provide a theoretical framework for analyzing web browsing experience. A method is provided for modeling, learning, and reusing Web browsing experience. Keyword-based algorithms are designed for reuse of browsing experience in predicting users' browsing actions. The reuse of browsing experience as instructions is discussed and a system is created for automatic generation of text-based narrative and graphic-based guided instructions from the learning of web browsing experience. ^ Four hypotheses are proposed to test the developed methods of reusing browsing experience. Two experiments have been conducted with a total of 60 subjects to test the hypotheses. Testing of the hypotheses indicates the following: (1) the accuracies of keyword-based prediction programs are positively correlated with the subjectively perceived similarities between the browsing experience and prediction targets; (2) using guided instructions rather than narrative instructions results in faster performance time and fewer errors; (3) narrative instructions created by human experts results in faster performance time than narrative instructions generated by machine only in complex tasks. (4) Performance time and errors for guided instructions generated by machine are not significantly different from those created by experts. The results demonstrates the potential benefits which may be derived to increase web browsing effectiveness by developing methods for reusing web browsing experience. ^

Degree

Ph.D.

Advisors

Major Professor: Gavriel Salvendy, Purdue University.

Subject Area

Engineering, Industrial|Computer Science

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS