Named Entity Recognition Using Web Document Corpus

Abstract

This paper introduces a named entity recognition approach in textual corpus.\nThis Named Entity (NE) can be a named: location, person, organization, date,\ntime, etc., characterized by instances. A NE is found in texts accompanied by\ncontexts: words that are left or right of the NE. The work mainly aims at\nidentifying contexts inducing the NE's nature. As such, The occurrence of the\nword "President" in a text, means that this word or context may be followed by\nthe name of a president as President "Obama". Likewise, a word preceded by the\nstring "footballer" induces that this is the name of a footballer. NE\nrecognition may be viewed as a classification method, where every word is\nassigned to a NE class, regarding the context. The aim of this study is then to\nidentify and classify the contexts that are most relevant to recognize a NE,\nthose which are frequently found with the NE. A learning approach using\ntraining corpus: web documents, constructed from learning examples is then\nsuggested. Frequency representations and modified tf-idf representations are\nused to calculate the context weights associated to context frequency, learning\nexample frequency, and document frequency in the corpus.\n