Understanding User's Expectations for Recommander Systems: the Case of Social Media

Abstract : Different sociotechnical recommender systems (personalized advertising engines, commercial research engines, suggestions for activities or content, etc.) are today ever-present in the experience of Internet users. They are the driving forces of Web development, they increase the turnover of e-commerce websites, develop audiences for information websites, direct traffic towards target content, etc. Promotional speeches by commercial players put forward advanced behavioral targeting technologies and aim to reintroduce a new enthusiasm for the new economy [BOU 10]. It is obviously a success, as shown by many projects involving the use of different kinds of data towards the personalization of services such as the ones promoted by the Next Generation Internet Foundation or the regularly reported analysis of the website Internetactu or again the invariably high satisfaction rates, of online shoppers. Social medias are also filled up with various recommender systems. They accompany the user after registration in order to propose link with others, activities to carry out, content to visit and commercial offerings to consume, making the social Web a communicative and informative marketing space [MIL 10, ARN, 09]. Some researchers [STE 09] have deduced that the driving force of participation, as it was highly showcased in promotional speeches for Web 2.0, is none other than the prescription. This seems convergent with the commercial players of social medias strategy, now explicitly assumed [REB 11]. However, this prescription, in the case of social medias, takes a particular form due to the importance associated to pairs by users. This importance is illustrated at the same time by the creativity of the latter when it involves opportunities for participation, by the enthusiasm that these propositions arouse, and by preferences, which are more or less overpowering depending on the website, but still very important, for activities carried out with pairs in comparison with activities proposed by organizations, which struggle to catch the attention of Internet users and engage them in their prescriptions for activities [STE 09]. Although recommendation strategies and tools seem to have reached maturity today in the e-business context, leaving powerful commercial potential to the imagination, adapting them to the social medias context may require a few adaptations to the specifics of this space. This is the hypothesis which has guided us in this article. Indeed, the results of two research projects regarding social networks sites and Twitter showed us practices and opinions related to recommender systems in the social medias context which enquire about the last activities of their users. The first one is dealing with collective activity with sociability purposes on Facebook, the second with the professional use of Twitter in communication and web marketing jobs. We will firstly present the social theories of prescriptions and the way we analyzed recommender systems through a theory of enunciation. The second part will outline the results questioning the perceived use of these systems and associate them with similar results which can confirm these doubts. We will therefore propose several approaches for a sociotechnical understanding of a recommender systems relevance on social medias. In conclusion, we will ask ourselves about the ethics of the intervention of recommender systems in the context of social medias.
Type de document :
Chapitre d'ouvrage
Gérald Kembelec; Ghislaine Chartron; Imad Saleh. Recommender Systems, ISTE, pp.25-52, 2014
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01514316
Contributeur : Jean-Claude Domenget <>
Soumis le : mercredi 26 avril 2017 - 08:52:10
Dernière modification le : jeudi 27 avril 2017 - 01:05:59

Identifiants

  • HAL Id : hal-01514316, version 1

Collections

Citation

Jean-Claude Domenget, Alexandre Coutant. Understanding User's Expectations for Recommander Systems: the Case of Social Media. Gérald Kembelec; Ghislaine Chartron; Imad Saleh. Recommender Systems, ISTE, pp.25-52, 2014. <hal-01514316>

Partager

Métriques

Consultations de la notice

20