Generality evaluation of automatically generated knowledge for the japanese conceptnet

Authors:
Rafal Rzepka;Koichi Muramoto;Kenji Araki
Affiliations:
Graduate School of Information Science and Technology, Hokkaido University, Sapporo, Japan;Graduate School of Information Science and Technology, Hokkaido University, Sapporo, Japan;Graduate School of Information Science and Technology, Hokkaido University, Sapporo, Japan
Venue:
AI'11 Proceedings of the 24th international conference on Advances in Artificial Intelligence
Year:
2011

Citing 5
Cited 0

Open Mind Common Sense: Knowledge Acquisition from the General Public

On the Move to Meaningful Internet Systems, 2002 - DOA/CoopIS/ODBASE 2002 Confederated International Conferences DOA, CoopIS and ODBASE 2002
Learner: a system for acquiring commonsense knowledge by analogy

Proceedings of the 2nd international conference on Knowledge capture
Large-scale extraction and use of knowledge from text

Proceedings of the fifth international conference on Knowledge capture
Deriving generalized knowledge from corpora using WordNet abstraction

EACL '09 Proceedings of the 12th Conference of the European Chapter of the Association for Computational Linguistics
Exploiting probabilistic knowledge under uncertain sensing for efficient robot behaviour

IJCAI'11 Proceedings of the Twenty-Second international joint conference on Artificial Intelligence - Volume Volume Three

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we introduce three methods for automatic generality evaluation of commonsense statements candidates generated for Open Mind Common Sense (OMCS), which is the basis of ConceptNet, a commonsense knowledge base. By using sister terms from Japanese WordNet, our system generates new statements which are automatically evaluated by using WWW co-occurrences and hit number retrieved by a Web search engine. These values are used in three generality judgment methods we propose. Evaluation experiments show that the best of them was "exact match ratio" which achieved accuracy of 62.6% when evaluating general sentences and "co-occurrences in snippets" method scored highest with 48.6% when judging unnatural phrases. Compared to the data without noise elimination, the "exact match ratio" achieved 38.2 points increase in accuracy.