Canonicalizing Open Knowledge Bases

Luis Galárraga (1) , Geremy Heitz , Kevin Murphy (2) , Fabian M. Suchanek (1)

1 LTCI - Laboratoire Traitement et Communication de l'Information
2 Google Inc.

Open information extraction approaches have led to the creation of large knowledge bases from the Web. The problem with such methods is that their entities and relations are not canonicalized, leading to redundant and ambiguous facts. For example, they may store Barack Obama, was born in, Honolulu and Obama, place of birth, Honolulu. In this paper, we present an approach based on machine learning methods that can canonicalize such Open IE triples, by clustering synonymous names and phrases. We also provide a detailed discussion about the different signals, features and design choices that influence the quality of synonym resolution for noun phrases in Open IE KBs, thus shedding light on the middle ground between " open " and " closed " information extraction systems.

Domaines: Web; Base de données [cs.DB]

Luis Galárraga, Geremy Heitz, Kevin Murphy, Fabian M. Suchanek. Canonicalizing Open Knowledge Bases. CIKM, Nov 2014, Shanghai, France. ⟨10.1145/2661829.2662073⟩. ⟨hal-01699884⟩