Regresar

Using domain ontologies for text classification. A use case to classify computer science papers

Abstract:

The web facilitates the creation, publication, and exchange of a wide variety of information. Particularly, the dramatic growth of unstructured web content makes text classification in a basic task to automate. Although well-known classification methods come from natural language processing and the machine learning fields, in this paper, the authors address the classification task as the ability to recognize topics or concepts in a text. To achieve this goal, we use a domain ontology as a driver of this process. The main motivation behind this work is to take advantage of the existing domain ontologies to classify and analyze the scientific production of a certain area of knowledge. Preliminary findings obtained by classifying a subset of Computer Science papers encourage us to remain researching in this area. Also, we will continue to discover a better way to improve results by combining it with other well-known approaches for text classification.