Interpreting Reputation Through Frequent Named Entities in Twitter

Nacéra Bennacer Seghouani (1, 2) , Francesca Bugiotti (1, 2) , Moditha Hewasinghage (1) , Suela Isaj (1) , Gianluca Quercini (1, 2)

1 CentraleSupélec
2 LRI - Laboratoire de Recherche en Informatique

Twitter is a social network that provides a powerful source of data. The analysis of those data offers many challenges among those stands out the opportunity to find the reputation of a product, of a person , or of any other entity of interest. Several tools for sentiment analysis have been built in order to calculate the general opinion of an entity using a static analysis of the sentiments expressed in tweets. However, entities are not static; they collaborate with other entities and get involved in events. A simple aggregation of sentiments is then not sufficient to represent this dynamism. In this paper, we present a new approach that identifies the reputation of an entity on the basis of the set of events it is involved into by providing a transparent and self explanatory way for interpreting reputation. In order to perform this analysis we define a new sampling method based on a tweet weighting to retrieve relevant information. In our experiments we show that the 90% of the reputation of the entity originates from the events it is involved into, especially in the case of entities that represent public figures.

Mots clés: Reputation; Frequent Itemsets; Sampling; Opinion Mining

Domaines: Base de données [cs.DB]

Nacéra Bennacer Seghouani, Francesca Bugiotti, Moditha Hewasinghage, Suela Isaj, Gianluca Quercini. Interpreting Reputation Through Frequent Named Entities in Twitter. Web Information Systems Engineering - WISE, Oct 2017, Puschino - Moscow, Russia. pp.49-56, ⟨10.1007/978-3-319-68783-4_4⟩. ⟨hal-01611593⟩