Recognition and Classification of Numerical Entities in Basque

Proceedings of Recent Advances in Natural Language Processing, pages 764–769, Hissar, Bulgaria, 12-14 September 2011. Recognition and Classification of Numerical Entities in Basque Ander Soraluze, I˜naki Alegria, Olatz Ansa, Olatz Arregi and Xabier Arregi IXA Group. University of the Basque Country ander.soraluze@ehu.es, i.alegria@ehu.es, olatz.ansa@ehu.es, olatz.arregi@ehu.es, xabier.arregi@ehu.es Abstract This paper presents a system based on Finite State Technology that recognises and classifies numerical entities in texts written in Basque. The system deals with a wide range of entities, such as temporal expressions, numbers related to units of measurement, or those that refer to common nouns. The system obtains 86.96% F-measure score following MUC evaluation and 78.82% using IREX and CONLL simple scoring protocol. 1 Introduction Named Entity Recognition and Classification (NERC) has become an important sub-task in the Natural Lan- guage Processing area. It is known that an effective treatment of Named Entities can benefit the perfor- mance of applications like Machine Translation (MT), Information Extraction (IE), Information Retrieval (IR) or Question Answering (QA). In the early stages, NERC systems identified a few types of entities, namely person, organisation and location names. Over time, numerical and temporal expressions have been also considered as identifiable types of entities. Concerning to Basque, there is a NERC system called Eihera (Alegria et al., 2003) that recognises and classifies person, organisation and location names, but it does not deal with numerical entities up to date. The Numerical Entity Recogniser and Classifier for Basque (NuERCB) presented here aims to address this lack. NuERCB identifies the numbers of the text and decides whether they express date or time, or are associated with units of measurement or, otherwise, just refer to common nouns. When numbers are linked to units or symbols of measurement, NuERCB determines which specific property maps with each of them. For instance, units like “square meter”, “meter per second squared”, “second” or “Celsius” are associated with properties like “area”, “acceleration”, “time” and “temperature” respectively. Since numerical expressions, particularly those related to units of measurement, are very common in technical texts, we have used in this work the ZT corpus, a Basque corpus specialized in science and technology (Areta et al., 2007). In this dataset numerical expressions are more likely to appear, so it allows us to test the system on a wide variety of cases. This paper is structured as follows. After reviewing related work, section 3 describes the linguistic features related to numerical expressions in Basque texts. Sec- tions 4 and 5 show the methods for number detection and classification. Section 6 presents the main experi- mental results, which are analysed in section 7. Finally, the conclusions and future work are mentioned. 2 Related Work The set of categories used to classify Named Entities has enriched over the time. As defined in the Message Understanding Conference (MUC) (Chinchor, 1998), Named Entity recognition consists on the identification and categorization of three types of specializations: “ENAMEX” for person, organisation and location, “TIMEX” for time and date, and “NUMEX” for money and percent. Furthermore, TIMEX2 (Ferro et …

Content truncated to 500 words