Level structure of $^{123}$Cs observed from $^{123}$Ba decay and described using the IBFM and CQPC models
Résumé
With the steady growth of business and scientific activities and the recent advances in Information Technology, huge amounts of electronically available but unstructured data have to be dealt with. New tools able to analyse and structure textual data need to be developed so that non-expert users can understand and evaluate the contents of their documents. This paper describes the first results of a R&D program set up by TEMIS Company, whose core business is Text Mining, i.e. information extraction, information processing, visualisation and valorisation of all the data issued by or received in a company in its field of activity. The aim of the program is to build a multilingual knowledge station, called K-Station, in order to build knowledge components (terms and extraction rules) in an iterative way. The knowledge station is independent of the field of activity. The K-Station allows the linguist to create knowledge components for a determined industrial sector (cars or chemistry for example) or for a specific job (e.g. journalism or computer science). We particularly insist on the process of information extraction from a corpus, by detailing the methodology used. Our approach is based on the recycling of existing terms databases and knowledge discovery from a corpus, according to an iterative process which includes extraction rules. All the components are part of a client/server architecture. We explain the various levels stages of analysis integrated within this architecture and we illustrate them with results taken from the field of Competitive Intelligence. Knowledge components are included in a software suite called Insight Discoverer