Multi-granular fusion for social data analysis for a decision and intelligence application

Nowadays, social data is a common research field in Computer Science. It is based on Social Networks (SNs), people and links among them and Social Software enabling to send messages. Social data is easy to gather on the Web as there are plenty of them. As people are also authors in such social platforms, information is not always reliable due to abbreviations, acronyms, misspellings, nicknames, misleading information. Furthermore, people are not aware of the structure of information and few are semantically annotating their information. Social Networks Analysis (SNA) is interested in the structure of the network but misspellings or misleading information are collapsing analysis. We present here algorithms for multi-granular fusion enabling to re-structure social information for use of SNAs and reduce data imperfection.