Automatic Biomedical Term Polysemy Detection

Automatic Biomedical Term Polysemy Detection Juan Antonio Lossio-Ventura1,2, Clement Jonquet1,2, Mathieu Roche1,2,3, Maguelonne Teisseire1,4,5 LIRMM1, University of Montpellier2, Cirad3, Irstea4, TETIS5 Montpellier, France juan.lossio@lirmm.fr, jonquet@lirmm.fr, mathieu.roche@cirad.fr, maguelonne.teisseire@teledetection.fr Abstract Polysemy is the capacity for a word to have multiple meanings. Polysemy detection is a first step for Word Sense Induction (WSI), which allows to find different meanings for a term. The polysemy detection is also important for information extraction (IE) systems. In addition, the polysemy detection is important for building/enriching terminologies and ontologies. In this paper, we present a novel approach to detect if a biomedical term is polysemic, with the long term goal of enriching biomedical ontologies. This approach is based on the extraction of new features. In this context we propose to extract features following two manners: (i) extracted directly from the text dataset, and (ii) from an induced graph. Our method obtains an Accuracy and F-Measure of 0.978. Keywords: Polysemy Detection, Biomedical Polysemy Detection, BioNLP, Disambiguation, Classification 1. Introduction The Web is by far the largest information archive avail- able worldwide evolving. This resource contains impor- tant information about several domains. This is the case for biomedicine, that brings knowledge through numerous publications (El-Rab et al., 2013). In this context, there are several methods to extract relevant information tackling the disambiguation problem (El-Rab et al., 2013; Zhong and Ng, 2012). This issue has been also recently addressed by the research of concepts, analyzing text to extract in- stances of concepts associated with user queries (Agirre et al., 2014). The ontologies are very useful for the identi- fication of concepts; the main objective is the creation of knowledge in a domain. They must be regularly enriched by the introduction of new terms. So, to enrich ontolo- gies/vocabularies with new terms, it is necessary to know the possible senses of a term, this is the well known Word Sense Induction (WSI) domain. One preliminary step is to detect if a term is polysemic (binary decision). If the term is polysemic, then to make a deep search of their senses. To our knowledge, there are no studies for the same purpose. Therefore, in order to meet the challenge, we propose a novel methodology to detect if a term is polysemic by defin- ing new features, extracted directly from the textual dataset and from an induced graph, as described after. In turn, our methodology uses two dictionaries allowing us to de- termine the use of a same term in different domains (i.e., biomedical and agronomy). To the best of our knowledge, graphs have never been used to define features for classifi- cation purpose. In this work the main idea is to capture the dataset characteristics from the structural shape and size of graph induced from the dataset. This approach enables to obtain excellent results, with 97.8% for Accuracy and F- Measure. The paper is organized as follows. First, the methodology is detailed in Section 2.. Results are presented in Section 3.. We discuss related work in Section 4. followed by conclu- sion …

Content truncated to 500 words