Towards a Welsh Semantic Annotation System

Towards A Welsh Semantic Annotation System Scott Piao1, Paul Rayson1, Dawn Knight2 and Gareth Watkins2 1School of Computing and Communications, Lancaster University, UK 2School of English, Communication and Philosophy, Cardiff University, UK {s.piao;p.rayson}@lancaster.ac.uk, knightd5@cardiff.ac.uk, gllwatkins@gmail.com Abstract Automatic semantic annotation of natural language data is an important task in Natural Language Processing, and a variety of semantic taggers have been developed for this task, particularly for English. However, for many languages, particularly for low-resource languages, such tools are yet to be developed. In this paper, we report on the development of an automatic Welsh semantic annotation tool (named CySemTagger) in the CorCenCC Project, which will facilitate semantic-level analysis of Welsh language data on a large scale. Based on Lancaster’s USAS semantic tagger framework, this tool tags words in Welsh texts with semantic tags from a semantic classification scheme, and is designed to be compatible with multiple Welsh POS taggers and POS tagsets by mapping different tagsets into a core shared POS tagset that is used internally by CySemTagger. Our initial evaluation shows that the tagger can cover up to 91.78% of words in Welsh text. This tagger is under continuous development, and will provide a critical tool for Welsh language corpus and information processing at semantic level. Keywords: Welsh semantic tagger, corpus annotation, Welsh language, Welsh corpus, CorCenCC 1. Introduction Automatic semantic annotation and analysis is an impor- tant task for Natural Language Processing (NLP), and se- mantic taggers have been developed and used for carrying out semantic analysis of language data on a large scale. A major tool built for such a purpose is USAS (UCREL Se- mantic Analysis System)1 (Rayson et al., 2004; Piao et al., 2017), which is designed to annotate each word or phrase in text with lexical semantic categories derived from Tom McArthur’s Longman Lexicon of Contemporary English (McArthur, 1981), such as “Food” or “Time”. The USAS semantic scheme contains 21 major semantic fields that are further divided into 232 sub-categories. Initially developed for processing English text, it has been extended to tag texts of a number of languages, including Italian, Chinese, Span- ish etc. (Piao et al., 2015). In the CorCenCC Project2, we have been developing a Welsh semantic tagger, named CySemTagger, modelled on the USAS framework, which employs a translated USAS semantic classification scheme and tagset for Welsh lan- guage. During the course of development, we have first constructed large Welsh semantic lexicons containing ap- proximately 136,468 Welsh word entries, which provide a lexical knowledge base for the semantic tagging system3. Based on the lexicons, we have developed an initial version of the tagger software system that is designed to accommo- date different Welsh POS taggers and tagsets that exist or will be developed. In the CorCenCC project, CySemTagger will be mainly based on a new Welsh POS tagger, named CyTag, which has been developed in this Project (Neale et al., 2018). In this paper, we describe the CySemTagger sys- tem and report on its initial evaluation. 1For further details of USAS, see website …

Content truncated to 500 words