IMPROVING COLLECTION PROCESS FOR SOCIAL MEDIA INTELLIGENCE: A CASE STUDY
L. Franchina, F. Greco, A. Lucariello, A. Socal, L. Teodonno
Social Media Intelligence (SOCMINT) is a specific section of Open Source Intelligence. Open Source Intelligence (OSINT) consists in the collection and analysis of information that is gathered from public, or open sources. Social Media Intelligence allows to collect data gathering from Social Media web sites (such as Facebook, Twitter, YouTube etc…). Both OSINT and SOCMINT are based on the Intelligence Cycle. This Paper aims to illustrate advantages gained by applying text mining to collection phase of the intelligence cycle, in order to perform threat analysis. The first step for detecting information related to a specific target is to define a consistent set of keywords. Web sources are various and characterized by different writing styles. Repeating this process manually for each source could be very inefficient and time consuming. Text mining specific software have been used in order to automatize the process and to reach more reliable results. A partially automatized procedure has been developed in order to gather information on specific topic using the Social Media Twitter. The procedure consists in searching manually a set of few keywords to be used for a specific threat analysis. Then TwitteR of R Statistics was used to gather tweets that were collected in a corpus and processed with T-Lab software in order to identify a new list of keywords according to their occurrence and association. Finally, an analysis of advantages and drawbacks of the developed method.