6533b82dfe1ef96bd129110d

RESEARCH PRODUCT

Médias sociaux et gestion de communautés - applications dans le domaine de la gestion de la relation client

subject

description

Customer relationship management (CRM) is a term that emerged in the middle of the 1990's, and that is often used to describe computerized tools that provide services to consumers before, during and after a sale.These consumers have followed the transformation of the Web that has happened in the last few years, where each user becomes a supplier of content using tools like, amongst others, social networks by sharing resources, content, and annotating. General use social networks, such as Facebook or Twitter, are now used daily by a very large number of users.Companies have to follow this evolution and include social networks as a new communication channel in order to interact with their clients and prospects, and vice versa. Social CRM, or s-CRM, is the CRM evolution linked to the integration of social networks in the tools used to customer relationship management. This evolution has to take into account the richness of the social networks data and take advantage of them.Social network data have a great potential of adding value to CRM. Several tools have been developed to analyse them.However, in order to take into account the diversity, in terms of data, algorithms and use, a generic platform that can be used by experts has to be developed.This platform will have to manage the collection, storing and analysis and visualisation of social network data.In order to answer the multiple needs related to the social network data analysis (community analysis, influential users identification, event detection, etc.), and to reduce the data formatting time for real time analysis, while quickly accessing large amount of data, several data modelling paradigms (so as to store them) have to be considered.Semantics, which we see as the contextualization of data or the results of the algorithms in relation to the domain knowledge, and also social network data modelling with graphs or user profiles, are the two central themes of our research.We submit a community detection and event characterization that doesn't use any specific parameter, except for a time window used to detect long or short events. The events are characterized by contextualizing with the most representative hashtags on the detected period.We also submit an influence detection method using weighted and directed graph and the HITS algorithm. The various algorithms are validated with real world data.A proof of concept of platform, called SNFreezer, has been developed and used to validate several features through various projects. We validated the data collection and storing scalability of the platform. This platform was the starting of a more industrial implementation. This was linked to the DisCoCRM project that is part of the will, initiated with the beginning of my thesis, of eb-Lab and Teletech International to integrate social networks into their CRM tools.