Rayson, Paul Edward (2018) Increasing Interoperability for Embedding Corpus Annotation Pipelines in Wmatrix and other corpus retrieval tools. In: 6th Workshop on the Challenges in the Management of Large Corpora : Proceedings of the 11th Edition of the Language Resources and Evaluation Conference - Miyazaki, Japan. UNSPECIFIED, JPN, pp. 33-36. ISBN 9791095546146
wmatrix_interoperability_cmlc.pdf - Accepted Version
Available under License Creative Commons Attribution-NonCommercial.
Download (233kB)
Abstract
Computational tools and methods employed in corpus linguistics are split into three main types: compilation, annotation and retrieval. These mirror and support the usual corpus linguistics methodology of corpus collection, manual and/or automatic tagging, followed by query and analysis. Typically, corpus software to support retrieval implements some or all of the five major methods in corpus linguistics only at the word level: frequency list, concordance, keyword, collocation and n-gram, and such software may or may not provide support for text which has already been tagged, for example at the part-of-speech (POS) level. Wmatrix is currently one of the few retrieval tools which have annotation tools built in. However, annotation in Wmatrix is currently limited to the UCREL English POS and semantic tagging pipeline. In this paper, we describe an approach to extend support for embedding other tagging pipelines and tools in Wmatrix via the use of APIs, and describe how such an approach is also applicable to other retrieval tools, potentially enabling support for tagged data.