Documentare e comunicare l'attività di trasferimento tecnologico. Analisi testuale della comunicazione dei poli di innovazione

There is an increasing attention on the needs to support SMEs in enhancing their innovation op-portunities and capabilities. Through a policy measure to foster the regional innovation system, 12 innovation poles were active in Tuscany in the period 2011-2014 to provide to their members (af-filiation is needed) a range of knowledge-intensive services such as knowledge and technology mapping, R&D partnership formation, technical assistance in R&D projects, technology transfer. Each pole was created as a consortium of organizations operating as public or private research cen-tres and service centres (universities, innovation centres or technology transfer centres and firms). In this paper we adopt a statistical analysis of textual content produced by the innovation poles to identify distinctive or common elements in the various texts they produced in three years of activi-ty and to draw some assessment of their communication on their activities. Documents under analysis are of different types: designed as written texts (on Smart Specializa-tion Strategy and monitoring the activities of the poles), transcripts of spoken language (the re-cordings of interviews); web communication. Through and automatic analysis we propose a sys-tematic comparison of all these documents that would not be possible through direct reading of texts: on the whole it is over 56,000 graphic forms, for a total of over two million occurrences. To compare both the intra diversity across the same type of document and across the different types of documents, first we analyse each of the four body separately, in order to identify the specific con-tent and the four languages used by the poles of innovation: "report", texts structured in the format of the monitoring; "design", the documents on smart specialization strategy; "reflection and analy-sis", in the transcription of interviews; and "communication", that characterizes the web sites. For this analysis, each document is associated with one or more categories (such as, for example, pole' band category, date of the document) that allow us to group or isolate relevant content in different contexts. In this work we first introduce the set of processing of texts aimed at the selection of graphic forms on which we focus our analysis. Then, we present for each corpus the description of the analysed documents, the results of calculations performed for the treatment of the text and the analysis of the main components that explain the variability of language within each corpus. These analyses (represented by the factorial of two main components) interpret the selection of graphic forms be-ing analysed with respect to categorical variables, defined for each document in each of the corpo-ra. The analysis concludes