Grey Literature for Natural Language Processing: a Terminological and Statistical Approach

This paper presents the results of a study on grey literature (GL) in the field of Natural Language Processing (NLP). Our data has been collected in a corpus of ca 13,000 records corresponding to the titles of papers presented at International Conferences from 1950 to June 2008. A statistical representation of the most significant terms relative to GL in NLP and other interrelated disciplines associates old and new words, highlighting the terminological changes that have taken place in the course of time. Aim of our study is to contribute to the creation of language resources for the extraction of GL coming from the Web in order to help prevent the disappearance of documents containing NLP words that have undergone rapid development over the last decades. This paper is organised as follows: after a general introduction to our work, section 2 provides a historical overview of NLP; sections 3 and 4 offer an account of the most relevant terms used by specialists in different periods, and indicative of the changes that have taken place; section 5 describes the methodology we have used and also contains information on our GL database and a graphical representation of the data. Finally, the conclusions stress the need to integrate pre-existing or obsolete words and expressions, creating NLP synonym relations.