Semi-Supervised Bootstrapping of Relationship Extractors with Distributional Semantics

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 499–504, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Semi-Supervised Bootstrapping of Relationship Extractors with Distributional Semantics David S. Batista Bruno Martins M´ario J. Silva INESC-ID, Instituto Superior T´ecnico, Universidade de Lisboa {david.batista,bruno.g.martins,mario.gaspar.silva}@ist.utl.pt Abstract Semi-supervised bootstrapping techniques for relationship extraction from text iter- atively expand a set of initial seed rela- tionships while limiting the semantic drift. We research bootstrapping for relationship extraction using word embeddings to find similar relationships. Experimental results show that relying on word embeddings achieves a better performance on the task of extracting four types of relationships from a collection of newswire documents when compared with a baseline using TF- IDF to find similar relationships. 1 Introduction Relationship Extraction (RE) transforms unstruc- tured text into relational triples, each represent- ing a relationship between two named-entities. A bootstrapping system for RE starts with a collec- tion of documents and a few seed instances. The system scans the document collection, collecting occurrence contexts for the seed instances. Then, based on these contexts, the system generates ex- traction patterns. The documents are scanned again using the patterns to match new relation- ship instances. These newly extracted instances are then added to the seed set, and the process is repeated until a certain stop criteria is met. The objective of bootstrapping is thus to expand the seed set with new relationship instances, while limiting the semantic drift, i.e. the progressive de- viation of the semantics for the extracted relation- ships from the semantics of the seed relationships. State-of-the-art approaches rely on word vec- tor representations with TF-IDF weights (Salton and Buckley, 1988). However expanding the seed set by relying on TF-IDF representations to find similar instances has limitations, since the similar- ity between any two relationship instance vectors of TF-IDF weights is only positive when the in- stances share at least one term. For instance, the phrases was founded by and is the co-founder of do not have any common words, but they have the same semantics. Stemming techniques can aid in these cases, but only for variations of the same root word (Porter, 1980). We propose to address this challenge with an approach based on word embeddings (Mikolov et al., 2013a). By relying on word embeddings, the similarity of two phrases can be captured even if no common words exist. The word embed- dings for co-founder and founded should be sim- ilar, since these words tend to occur in the same contexts. Word embeddings can nonetheless also introduce semantic drift. When using word em- beddings, phrases like studied history at can, for instance, have a high similarity with phrases like history professor at. In our approach, we control the semantic drift by ranking the extracted rela- tionship instances, and by scoring the generated extraction patterns. We implemented these ideas in BREDS, a boot- strapping system for RE based on word embed- dings. BREDS was evaluated with a collection of 1.2 million sentences from news …

Content truncated to 500 words