Extraction and Approximation of Numerical Attributes from the Web

Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, pages 1308–1317, Uppsala, Sweden, 11-16 July 2010. c⃝2010 Association for Computational Linguistics Extraction and Approximation of Numerical Attributes from the Web Dmitry Davidov ICNC The Hebrew University Jerusalem, Israel dmitry@alice.nc.huji.ac.il Ari Rappoport Institute of Computer Science The Hebrew University Jerusalem, Israel arir@cs.huji.ac.il Abstract We present a novel framework for auto- mated extraction and approximation of nu- merical object attributes such as height and weight from the Web. Given an object-attribute pair, we discover and ana- lyze attribute information for a set of com- parable objects in order to infer the desired value. This allows us to approximate the desired numerical values even when no ex- act values can be found in the text. Our framework makes use of relation defining patterns and WordNet similarity information. First, we obtain from the Web and WordNet a list of terms similar to the given object. Then we retrieve attribute values for each term in this list, and infor- mation that allows us to compare different objects in the list and to infer the attribute value range. Finally, we combine the re- trieved data for all terms from the list to select or approximate the requested value. We evaluate our method using automated question answering, WordNet enrichment, and comparison with answers given in Wikipedia and by leading search engines. In all of these, our framework provides a significant improvement. 1 Introduction Information on various numerical properties of physical objects, such as length, width and weight is fundamental in question answering frameworks and for answering search engine queries. While in some cases manual annotation of objects with numerical properties is possible, it is a hard and labor intensive task, and is impractical for dealing with the vast amount of objects of interest. Hence, there is a need for automated semantic acquisition algorithms targeting such properties. In addition to answering direct questions, the ability to make a crude comparison or estimation of object attributes is important as well. For ex- ample, it allows to disambiguate relationships be- tween objects such as X part-of Y or X inside Y. Thus, a coarse approximation of the height of a house and a window is sufficient to decide that in the ‘house window’ nominal compound, ‘win- dow’ is very likely to be a part of house and not vice versa. Such relationship information can, in turn, help summarization, machine translation or textual entailment tasks. Due to the importance of relationship and at- tribute acquisition in NLP, numerous methods were proposed for extraction of various lexical re- lationships and attributes from text. Some of these methods can be successfully used for extracting numerical attributes. However, numerical attribute extraction is substantially different in two aspects, verification and approximation. First, unlike most general lexical attributes, nu- merical attribute values are comparable. It usually makes no sense to compare the names of two ac- tors, but it is meaningful to compare their ages. The ability to compare values of different objects …

Content truncated to 500 words