A Brief Analysis of Amharic NLP: From POS Tagging to Question Answering

6 September 2016

on 9 Sept. 2016 at 12:00, Alsterterrasse 1, Room 128.

Abstract:

In the first part of my talk, I will discuss some NLP research and applications done for Amharic so far. I specifically discuss POS tagging [http://aflat.org/files/HLTD201109.pdf ], Morphological processing [HornMorpho, see http://homes.soic.indiana.edu/gasser/L3/horn2.5.pdf ], Spell checking, Named entity recognition, and Questions Answering [http://etd.aau.edu.et/bitstream/123456789/8587/1/Brook%20Eshetu%20Bete.pdf  ]. A brief list of the challenges in Amharic NLP will be also discussed, related to these topics.

In the second part of my talk, I will present some of the NLP researches we have done so far. My first experience in Amharic NLP is the development of a factoid question answering system. The different components of the QA system and its implication will be presented. I will also present our experience on POS tagging, spell checking and the task of text collection (crawling) for Amharic corpora.

The third part of the talk emphasises on the main challenges and bottlenecks for Amharic NLP. While the main challenge in Amharic NLP is the lack of a dedicated research group for Amharic, I will also briefly discuss some of the technical problems such as the morphological complexity of the language and the difficulty of the writing system (script - Fidel) [needs a Keyboard Mapping Tool]. I will also discuss the possible collaborations of our group (Language Technology) with the Ethiopistic department at the University of Hamburg.  In this regard, I present discussion points to the group towards consistent and successful Amharic NLP research in the future.

                                                  .