Submissions/QAKiS: natural language question answering from Wikipedia content

From Wikimania 2013 • Hong Kong

After careful consideration, the programme committee has decided not to accept the below submission at this time. Thank you to the author(s) for participating in the Wikimania 2013 programme submission, we hope to still see you at Wikimania this August.

Submission no.
5036
Subject no.
B10
Title of the submission
QAKiS: natural language question answering from Wikipedia content
Type of submission
presentation
Author of the submission
Julien Cojan, Elena Cabrio
Country of origin
France
Affiliation
Inria
E-mail address
julien.cojan@inria.fr
Username
lejuin
Personal homepage or blog
http://www-sop.inria.fr/members/Julien.Cojan
Abstract

QAKiS is a natural question answering system that uses relational patterns to translate questions given in natural language into formal queries run on DBpedia, a database extracted from Wikipedia[1].

Relational patterns represent different ways to express a formal relation from DBpedia in natural language. They are automatically extracted from Wikipedia by matching sentences of the Wikipedia articles with data from DBpedia. This principle could also be applied to Wikidata.

A demo of the system is available at http://dbpedia.inria.fr/qakis.

Detailed proposal

QAKiS (Question Answering wiKiFramework-based System)[2] addresses the task of question answering over structured knowledge-bases (e.g. DBpedia), where the relevant information is expressed also in unstructured forms (e.g. Wikipedia pages). It implements a relation-based match for question interpretation, to convert the user question into a query language (e.g. SPARQL). More specifically, it makes use of relational patterns (automatically extracted from Wikipedia and collected in the WikiFramework repository [3]), that capture different ways to express a certain relation in a given language.

To answer a question of the kind "Which river does the Brooklyn Bridge cross?", QAKiS performs the following steps :

  • Identify the topics of the question, i.e. the Named Entities (NE) that correspond to Wikipedia articles, here "the Brooklyn Bridge"
  • Guess the Expected Answer Type (EAT), e.g. here we are looking for a river (class River from the DBpedia ontology).
  • Build a typed question "[River] does the [Bridge] cross" substituting the NE and the EAT with their classes. Such typed question is compared to the repository of relational patterns collected from Wikipedia pages, and since each pattern is a lexicalization of an ontological relation, the system can return the DBpedia property generated from the attribute "crosses" from "Infobox bridge".
  • The query generated from this property and the resource "the Brooklyn Bridge" returns "East River".

QAKiS currently works for questions in English, it will soon be extended to French and then to other languages. It could also be extended to use data from Wikidata.

In this session we will explain how the system works and we will present future improvements.

Track
Technology and Infrastructure
Length of presentation/talk
25 minutes
Language of presentation/talk
English
Will you attend Wikimania if your submission is not accepted?
Not sure
Slides or further information (optional)
Special requests
  1. ↑ http://dbpedia.org
  2. ↑ E. Cabrio, J. Cojan, A. P. Aprosio, B. Magnini, A. Lavelli, and F. Gandon. Qakis: an open domain qa system based on relational patterns. In Proceedings of the ISWC 2012 Posters and Demonstrations Track,Boston, US, November 2012.
  3. ↑ R. Mahendra, L. Wanzare, R. Bernardi, A. Lavelli, and B. Magnini. Acquiring relational patterns from wikipedia: A case study. In Proceedings of the 5th Language and Technology Conference, Poznan, Poland, 2011.


Interested attendees

If you are interested in attending this session, please sign with your username below. This will help reviewers to decide which sessions are of high interest. Sign with four tildes. (~~~~).

  1. Add your username here.
Retrieved from "https://wikimania2013.wikimedia.org/w/index.php?title=Submissions/QAKiS:_natural_language_question_answering_from_Wikipedia_content&oldid=21388"