Web-Based Sources for an Annotated Corpus Building and Composite Proper Name Identification

摘要

Nowadays, collections of texts with annotations on several levels are useful resources. Huge efforts are required to develop this resource for languages like Spanish. In this work, we present the initial step, lexical level annotation, for the compilation of an annotated Mexican corpus using Web-based sources. We also describe a method based on heterogeneous knowledge and simple Web-based sources for the proper name identification required in such annotation. We focused our work on composite entities (names with coordinated constituents, names with several prepositional phrases, and names of songs, books, movies, etc.). The preliminary obtained results are presented.