Using natural language processing tools to assist semiotic analysis of information systems

1. Abstract Semiotic Analysis has been used to aid understanding of information or communication systems, providing information that can be used during requirements engineering. The MEASUR approach begins by analysing short, natural language problem statements and manually extracting the key themes involved. As the process is scaled up and applied to longer problem statements, as found in many real life circumstances, the manual effort required increases. When the starting point for Semiotic Analysis is a large document describing the information system, such as an ethnographic report, assistance in the analytical process is necessary. This paper investigates how statistical Natural Language Processing Tools can aid this analysis. Natural Language Processing Tools can assist the analyst by directing them to the central themes in the document. Comparing a frequency list of the document with a frequency list from a large corpus of text such as the British National Corpus reveals the key words in the document. Collocation analysis of these keywords enables the creation of a lexical network and then closer investigation of the collocates in context allows the analyst to add semantic information to the model.