Metagnostic deductive question answering with explanation from texts

  • Authors:

  • John Kontos;Joseph Armaos;Ioanna Malagardi

  • Affiliations:

  • National and Kapodistrian University of Athens, University Campus, Ano Ilisia, Athens, Hellas;National and Kapodistrian University of Athens, University Campus, Ano Ilisia, Athens, Hellas;National and Kapodistrian University of Athens, University Campus, Ano Ilisia, Athens, Hellas

  • Venue:
  • UAHCI'11 Proceedings of the 6th international conference on Universal access in human-computer interaction: applications and services - Volume Part IV
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

The present paper presents a system called AMYNTAS for "metagnostic" deductive question answering from texts. This system can logically combine information from texts and answer questions generating explanations for its operation exhibiting "self-awareness". The deductions are performed directly with the natural language text without previous translation into a formal representation. The "metagnostic" effect is accomplished by representing and processing the state of linguistic processing and reasoning of the system. The system is implemented in Prolog and uses a text grammar to parse sentences that contain the information being sought. The system uses reasoning rules, lexicon, ontology, prerequisite knowledge and the history of its state. The system may easily be adapted to completely different domains such as biomedical texts and texts of the proofs of theorems of Euclidean geometry. An evaluation performed with real sentences from these two completely different domains gave satisfactory results of accuracy and facility of domain adaptation.