Optimization of Faceted Search Interfaces for Complex Querying

Abstract

Aware Facet Ranking (CAFR) model, which is user interactive and makes the most out of the user by dynamically filtering and ranking the facets according to their relevance, user motivations, and the situational contexts of the user. The model ranks the facets based on a composite score of query similarity, frequency of facet use, and result reduction. The efficiency of the optimized interface was determined based on the user-centered assessment framework, based on the user interaction logs, the satisfaction measures, and the rate of completing the tasks. In a compare and contrast experiment, the applicability of the CAFR model to standard Baseline and Static Ranking settings was comparatively tested. The outcome is a massive performance improvement, on which the CAFR model had a Search Performance Index (SPI) of 0.76, compared to 0.58 in the Baseline. Users were faster, interacted on average 2.8 times (vs. 4.2 with the Baseline), and solved a query (88%), better than with the Baseline (62%). These results suggest that dynamically ranked facets can offer a more interesting and intelligent search user experience in a wide range of applications, including e-commerce, online libraries, and enterprise data management.