ImageSieve: exploratory search of museum archives with named entity-based faceted browsing

  • Authors:

  • Yiling Lin;Jae-wook Ahn;Peter Brusilovsky;Daqing He;William Real

  • Affiliations:

  • University of Pittsburgh, Pittsburgh, PA;University of Pittsburgh, Pittsburgh, PA;University of Pittsburgh, Pittsburgh, PA;University of Pittsburgh, Pittsburgh, PA;Carnegie Museum of Art, Pittsburgh, PA

  • Venue:
  • Proceedings of the 73rd ASIS&T Annual Meeting on Navigating Streams in an Information Ecosystem - Volume 47
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Over the last few years, faceted search emerged as an attractive alternative to the traditional "text box" search and has become one of the standard ways of interaction on many e-commerce sites. However, these applications of faceted search are limited to domains where the objects of interests have already been classified along several independent dimensions, such as price, year, or brand. While automatic approaches to generate faceted search interfaces were proposed, it is not yet clear to what extent the automatically-produced interfaces will be useful to real users, and whether their quality can match or surpass their manually-produced predecessors. The goal of this paper is to introduce an exploratory search interface called ImageSieve, which shares many features with traditional faceted browsing, but can function without the use of traditional faceted metadata. ImageSieve uses automatically extracted and classified named entities, which play important roles in many domains (such as news collections, image archives, etc.). We describe one specific application of ImageSieve for image search. Here, named entities extracted from the descriptions of the retrieved images are used to organize a faceted browsing interface, which then helps users to make sense of and further explore the retrieved images. The results of a user study of ImageSieve demonstrate that a faceted search system based on named entities can help users explore large collections and find relevant information more effectively.