An Index-based FAQ Retrieval System for HIV/AIDS

Sister G. Steyn, Yirsaw Ayalew, and Gontlafetse Mosweunyane

Keywords

FAQ retrieval, FAQ-based question answering, automated FAQ retrieval

Abstract

Automated Frequently Asked Questions (FAQ) answering systems are becoming popular due to their ability to provide answers to user questions in restricted domains using pre-defined FAQ database. According to the literature on FAQ retrieval systems, there are many FAQ retrieval systems that have been developed and are being used in various domains. However, there are very limited FAQ-based question answering systems in the area of HIV/AIDS. This paper presents a discussion of the implementation of an index-based FAQ answering system on HIV/AIDS by adapting different techniques and algorithms from existing FAQ retrieval systems. For this purpose, we conducted an evaluation of 10 FAQ retrieval systems which are already deployed in other domains. For the implementation of the system, we used the open source Apache LUCENE information retrieval software library which provides Java-based indexing and search technology. Our FAQ repository contains the FAQs from the MASA booklet (which has 205 HIV/AIDS question-answer pairs) and IPOLETSE call center manual (which has 150 HIV/AIDS question-answer pairs). To evaluate the effectiveness of our system, we conducted experiments with two categories of user queries: queries which match with questions in the FAQ repository and queries which are not questions on HIV/AIDS. The performance of the system is encouraging with a success rate of 94.2% for the first category of questions and a rejection rate of 100% for the second category of questions. Moreover, we observed that the system performs better when user queries are compared with questions in the FAQ repository than when user queries are compared with question-answer pairs in the FAQ repository.

Important Links:

Go Back