Much work in recent years has gone into the construction of large knowledge bases (KBs), such as Freebase, DBPedia, NELL, and YAGO. While these KBs are very large, they are still very incomplete, necessitating the use of inference to fill in gaps. Prior work has shown how to make use of a large text corpus to augment random walk inference over KBs. We present two improvements to the use of such large corpora to augment KB inference. First, we present a new technique for combining KB relations and surface text into a single graph representation that is much more compact than graphs used in prior work. Second, we describe how to incorporate vector space similarity into random walk inference over KBs, reducing the feature sparsity inherent in using surface text. This allows us to combine distributional similarity with symbolic logical inference in novel and effective ways. With experiments on many relations from two separate KBs, we show that our methods significantly outperform prior work on KB inference, both in the size of problem our methods can handle and in the quality of predictions made.
Incorporating Vector Space Similarity in Random Walk Inference over Knowledge Bases
Matt Gardner,P. Talukdar,Jayant Krishnamurthy,Tom Michael Mitchell
Published 2014 in Conference on Empirical Methods in Natural Language Processing
ABSTRACT
PUBLICATION RECORD
- Publication year
2014
- Venue
Conference on Empirical Methods in Natural Language Processing
- Publication date
2014-10-01
- Fields of study
Computer Science
- Identifiers
- External record
- Source metadata
Semantic Scholar
CITATION MAP
EXTRACTION MAP
CLAIMS
CONCEPTS
- compact graph representation
A graph construction that merges multiple evidence sources into a smaller unified graph for inference.
Aliases: compact unified graph
- knowledge base inference
The task of predicting missing facts or relations in a knowledge base.
Aliases: KB inference
- knowledge base relations
Structured relation edges from a knowledge base that connect entities and support inference.
Aliases: KB relations, KB edges
- prediction quality
The accuracy or correctness of inferred candidate facts produced by the system.
Aliases: prediction accuracy
- prior work
Earlier published methods used as the comparison baseline in the abstract.
Aliases: previous work
- random walk inference over knowledge bases
A graph-based inference procedure that uses random walks on a knowledge-base representation to score candidate facts.
Aliases: KB random walk inference, random walk KB inference
- surface text
Textual evidence from a large corpus that links entities through observed co-occurrence and lexical patterns.
Aliases: text corpus evidence, textual evidence
- symbolic logical inference
Inference that uses explicit symbolic relations and logical structure rather than only statistical similarity.
Aliases: logical inference
- vector space similarity
A distributional similarity signal computed from vector embeddings or other continuous representations of terms or entities.
Aliases: distributional similarity, embedding similarity
REFERENCES
Showing 1-18 of 18 references · Page 1 of 1