The automatic coding of Economic Activities descriptions for Web users

ACTR (Automatic Coding by Text Recognition) system was introduced in Istat in ’90s to code data collected in statistical surveys through a batch process, producing good results. ACTR was used in different contexts and with different classifications. Being a generalised system, to be used it is necessary to build up the informative base for each classification; as larger the dictionary is as better the results are. In a lot of surveys the economic activity is one of the most important collected variables, its classification is very complex and its structure has several hierachical levels. The coding activity perfomed manually was so heavy that it was decided to build an automatic coding application which was widely used with very good results both from the quantitative and the qualitative point of view. With the last revision the new classification (Nace rev.2) is deeply different from the previous one, so updating the coding application was a very heavy task. On the other hand, the classification is more suitable to the present economic context and is unique for all the users. For this reasons it was decided to implement a new tool to support the Istat Web site users in identifying the code corresponding to the activities perfomed by their enterprises. This tool integrates the potentialities of ACTR in textual matching with the web technologies, guaranteeing a more performing solution than simply navigating in the official classification manual