Concept comparison engines: A new frontier of search

摘要

In a traditional search engine interaction scenario, a user begins with a certain concept and finds documents that are similar to their concept. However, the user may wish to compare alternatives and a search capability should compare concepts and present the best alternatives. This task can be difficult without proper decision aids. We propose a concept comparison engine as a decision support tool that may be used to compare attributes of different alternatives and aid in making an informed selection. We describe an architecture and an interaction scenario and implement a prototype. We propose a number of evaluation metrics for measuring the viability of different terms for the purpose of comparing concepts. In scripted experiments, orderings for candidate terms from the prototype are compared to gold standard ranking lists from structured external sources. Our results indicate that a Rankor analysis may be promising as a measure of the differentiating power of candidate terms a user might choose to support concept comparison.