A Learning Agent that Assists the Browsing of Software Libraries

  • Authors:
  • Christopher G. Drummond;Dan Ionescu;Robert C. Holte

  • Affiliations:
  • Univ. of Ottawa, ONT, Canada;Univ. of Ottawa, ONT, Canada;Univ. of Ottawa, ONT, Canada

  • Venue:
  • IEEE Transactions on Software Engineering
  • Year:
  • 2000

Quantified Score

Hi-index 0.01

Visualization

Abstract

Locating software items is difficult, even for knowledgeable software designers, when searching in large, complex, and continuously growing libraries. This paper describes a technique we term active browsing. An active browser suggests to the designer items it estimates to be close to the target of the a search. The novel aspect of active browsing is that it is entirely unobtrusive: It infers its similarity measure from the designer's normal browsing actions, without any special input. Experiments are presented in which the active browsing system succeeds 40 percent of the time in identifying the target before the designer has found it. An additional experiment indicates that this approach does, indeed, speed-up searches.