UQ IElab at TREC 2019 Decision Track

2019 | TREC 2019 Decision Track
Share via twitter Share via email Download PDF

Authors:

Abstract

We describe our participation to the TREC 2019 Decision Track. The first year of this track challenges participants to devise search technologies that retrieve correct health advice from web resources, with the intent to better support people’s health decision making.

Our solution addressed this challenge by extending the Entity Query Feature Expansion model (EQFE), a knowledge base (KB) query expansion method. In previous work we showed that Wikipedia and the Consumer Health Vocabulary resource can be effective as basis for consumer health search query expansion, within the EQFE method. To obtain query expansion terms, first, we mapped entity mentions to KB entities by performing exact matching. After mapping, we used the Title of the mapped KB entities as the source for expansion terms. Despite previous evaluation demonstrating the effectiveness of this method, EQFE did not provide the expected gains over not using query expansion, on both relevant-based and credibility-based evaluation measures.