Elements of a corpus semantics for humanities

application to the classification of subjective texts

Mathieu Valette

pp. 141-150

in: Dario Compagno (ed), Quantitative semiotic analysis, Berlin, Springer, 2018

Abstrakt

This article aims to review a possible contribution of corpus semantics to digital humanities, relying on an example of text profiling and text mining based on data from the social Web. We will focus, in particular, on the expression of contrasting values and points of view, allowing us to develop the concept of "agonist". First, we sketch out a critical overview of the relationship between NLP and semantics. Then we quickly outline how to provide text mining tools to corpus semantics. Finally, we illustrate our discussion with an example from applications. We wish, in particular, to highlight the potential benefit of a methodological dialogue between theory (text semantics), textometric methods and current NLP practices in terms of algorithms but, above all, evaluation.