AQUA: A Closed-Domain Question Answering System

Abstract

This article describes AQUA, an experimental question answering system. AQUA combines Natural Language processing (NLP), Ontologies, Logic, and Information Retrieval technologies in a uniform framework. AQUA makes intensive use of an ontology in several parts of the question answering system. The ontology is used in the refinement of the initial query, the reasoning process and in the novel similarity algorithm. The similarity algorithm is a key feature of AQUA. It is used to find similarities between relations/concepts in the translated query and relations/concepts in the ontological structures. The similarities detected then allow the interchange of concepts or relations in a logic formula corresponding to the user query.

Keywords:

• question answering
• ontologies
• similarity algorithm

Notes

1. http://www.ask.com/

2. The goal of the Semantic Web is to help users or software agents to organize, locate and process content on the WWW.

3. The AKT ontology contains classes and instances of people, organizations, research areas, publications, technologies and events (http://akt.open.ac.uk/ocml/domains/akt-support-ontology/).

4. Focus is a word or a sequence of words that defines the question and disambiguates it in the sense that it indicates what the question is looking for.

5. The pattern-matching algorithm tries to find an exact match with names in the ontology.

6. http://www.google.com

7. Armadillo is an information extraction engine that uses resources such as CiteSeer to find a limited range of information types such as publications.

8. http://citeseer.nj.nec.com/cs

9. As a reminder, the Dice coefficient measuring the similarity of two vectors X, and Y is defined by 2 | X ∩ Y ||X | + |Y| It is normalized for length by dividing by the total number of non-zero entries and doubled so that we get a measure that ranges from 0.0 to 1.0. A coefficient of 1.0 indicates identical vectors.

10. S_c indicates how many concepts G_1 and G_2 have in common.

11. S_r indicates how similar the relation between the same concepts are.

12. Grounded argument means instantiated argument.

13. Intersection means to find a subgraph in G_2 that contains all concepts contained in graph G_1.

14. If more than one relation is suggested to be similar, AQUA rewrites the query and tries to generate the proof using one at the time until all alternatives are considered.

15. AQUA has been implemented in Sicstus Prolog, C, OCML and PHP.