Classification of Under-Resourced Language Documents Using English Ontology

Abstract

Automatic documents classification is an important task due to the rapid growth of the number of electronic documents, which aims automatically assign the document to a predefined category based on its contents. The use of automatic document classification has been plays an important role in information extraction, summarization, text retrieval, question answering, e-mail spam detection, web page content filtering, automatic message routing , etc.Most existing methods and techniques in the field of document classification are keyword based, but due to lack of semantic consideration of this technique, it incurs low performance. In contrast, documents also be classified by taking their semantics using ontology as a knowledge base for classification; however, it is very challenging of building ontology with under-resourced language. Hence, this approach is only limited to resourced language (i.e. English) support. As a result, under-resourced language written documents are not benefited such ontology based classification approach. This paper describes the design of automatic document classification of under-resourced language written documents. In this work, we propose an approach that performs classification of under-resourced language written documents on top of English ontology. We used a bilingual dictionary with Part of Speech feature for word-by-word text translation to enable the classification of document without any language barrier. The design has a concept-mapping component, which uses lexical and semantic features to map the translated sense along the ontology concepts. Beside this, the design also has a categorization component, which determines a category of a given document based on weight of mapped concept. To evaluate the performance of the proposed approach 20-test documents for Amharic and Tigrinya and 15-test document for Afaan Oromo in each news category used. In order to observe the effect of incorporated features (i.e. lemma based index term selection, pre-processing strategies during concept mapping, lexical and semantics based concept mapping) five experimental techniques conducted. The experimental result indicated that the proposed approach with incorporation of all features and components achieved an average F-measure of 92.37%, 86.07% and 88.12% for Amharic, Afaan Oromo and Tigrinya documents respectively.

Keywords: under-resourced language, Multilingual, Documents or text Classification, knowledge base, Ontology based text categorization, multilingual text classification, Ontology.

DOI: 10.7176/CEIS/10-6-02

Publication date:July 31^st 2019