Search

Information extraction is a field of computer science research which explores the problem of detecting and retrieving desired information from textual data. This paper proposes a two-steps method that enables the detection of relevant information within a corpus of textual data. The first phase consists of observing the most recurrent structures through the study of textual co-occurrences and collocations, while the following phase consists of deriving rules from these structures which make it possible to create an inventory of all the expressions that identify a particular concept of interest, that is, the desired information.

In order to delineate the state of the art of the main TM applications a two-step strategy has been pursued: first of all, some of the main European and Italian companies offering TM solutions were contacted, in order to collect information on the characteristics of the applications; secondly, a detailed search on the web was made to collect further information about users or developers and applications. On the basis of the material collected, a synthetic grid was built to collocate, from more than 300 cases analysed, the 100 ones that we considered most relevant for the typology of function and sector of activity. The joint analysis of the different case studies has given an adequate picture of TM applications according to the possible types of results that can be obtained, the main specifications of the sectors of applications and the type of functions. Finally it is possible to classify the applications matching the level of customisation (followed in the tools development) and the level of integration (between users and developers). This matching produces four different situations: standardisation, outsourcing, internalisation, synergism.

Text Mining (TM) is a competitive statistical technology to extract relevant information from huge textual unstructured databases (document warehousing). In this paper, from an immense linguistic archive such as that coming of 10 years of daily “La Repubblica”, we describe several examples on the language productivity and the changes of language in the Nineties, with a particular attention of the use evolution of declining of verb mood, tense and person.

In the first part of this paper text mining (TM) is examined following three main dimensions: kind of applications, from customer relationship management (CRM) and market analysis to technology watch (TW) and patent analysis (PA); sectors of activity, from financial domain to health sector, from media and communication to public administration; scheme of strategy (document pre-processing, lexical and TM processing). We also report some results of a survey on text analysis traditions in Italy and some of the most relevant Italian company experiences in the domain of production of linguistic technology and TM solutions. In the second part of the paper, we describe some examples of TM applications in particular for CRM, TW, and PA, mined from the Internet.