A Large DataBase of Hypernymy Relations Extracted from the Web.

A Large Database of Hypernymy Relations Extracted from the Web Julian Seitner1, Christian Bizer1, Kai Eckert2, Stefano Faralli1, Robert Meusel1, Heiko Paulheim1, Simone Ponzetto1 1Data and Web Science Group, University of Mannheim, Germany jseitner@mail.uni-mannheim.de,{chris,stefano,robert,heiko,simone}@informatik.uni-mannheim.de 2Stuttgart Media University eckert@hdm-stuttgart.de Abstract Hypernymy relations (those where an hyponym term shares a “isa” relationship with his hypernym) play a key role for many Natural Language Processing (NLP) tasks, e.g. ontology learning, automatically building or extending knowledge bases, or word sense disambiguation and induction. In fact, such relations may provide the basis for the construction of more complex structures such as taxonomies, or be used as effective background knowledge for many word understanding applications. We present a publicly available database containing more than 400 million hypernymy relations we extracted from the CommonCrawl web corpus. We describe the infrastructure we developed to iterate over the web corpus for extracting the hypernymy relations and store them effectively into a large database. This collection of relations represents a rich source of knowledge and may be useful for many researchers. We offer the tuple dataset for public download and an Application Programming Interface (API) to help other researchers programmatically query the database. Keywords: Hearst patterns, hypernym extraction, big data, Web crawling, Application Programming Interface. 1. Introduction Hypernymy relations are important in many Artificial In- telligence (AI) and Natural Language Processing (NLP) ap- plications, and in general in data-intensive applications pro- cessing large amounts of unstructured and semi-structured information in order to interpret data and text that are not already semantically annotated, assigning types to entities (be it named entities in text, or entities in semi-structured data) represents a crucial step to understanding the data. For instance, in Paulheim and Fürnkranz (2012), it has been shown that adding precise types of instances (in that case taken from Semantic Web resources such as DBpedia and YAGO) to a data mining problem can lead to a significantly improved performance in many data mining tasks. Also when performing data integration, e.g., of a large collection of tabular datasets, into a large knowledge base, first un- derstanding whether the entities in a table are, e.g., cities, states, or mountains, is a very important step towards a high quality integration result (Ritze et al., 2015). While there are quite a few named entity recognition and disambiguation tools that do serve that purpose and exploit knowledge bases such as Wikipedia, DBpedia, or Freebase, a common problem is dealing with the long tail of less popular entities that are not contained in such knowledge bases. Most of the existing wide-coverage knowledge bases have no problems in identifying, e.g., major cities (“New York is a city”) or celebrities (“Madonna is a singer”), but they show limitations with respect to small villages and less known people. However, the potential of web-scale intel- ligent, data-intensive applications can only be unlocked if they are capable of dealing with the most prominent enti- ties, as well as the long tail. Hence, the necessity of extend- ing existing knowledge bases with hypernymy relations …

Content truncated to 500 words