Development of Amazighe Named Entity Recognition System Using Hybrid Method

The Named Entity Recognition (NER) is very important task revolving around many natural language processing applications. However, most Named Entity Recognition (NER) systems have been developed using either of two approaches: a rule-based or Machine Learning (ML) based approach, with their effectiveness and weaknesses. In this paper, the problem of Amazighe NER is tackled through using the two approaches together to produce a hybrid system with the aim of enhancing in general performance of NER tasks. The proposed system is able of recognizing 5 different types of named entities (NEs): Person, Location, Organization, Date and Number. It was tested on a corpus of Amazigh reports containing 867 diverse articles. Furthermore, a comparison with the baselines of the system based on the case of using just gazetteers and hand-written heuristics is presented. We also provide the detailed analysis of the results.