Chapter 2

Social Networks: Leveraging User Social Data to Empower Collective Intelligence

Online social networks such as Facebook, Twitter or LinkedIn have become extremely popular and ubiquitous. Users are actively connected to these services for discussing and sharing news, events as well as contents of interest. However, they are facing problems such as the disconnected nature of social Websites, user privacy issues and the huge amount of information available for browsing. Consequently, a considerable part of interesting information remains ignored by the users. We present in this chapter a possible answer to this problem based on a new approach consisting of aggregating relevant information from social networks to empower the collective intelligence shared by a given group of users. We have built our solution on a user-centered approach, the main benefit of which is that each user can delegate to the system the tasks of aggregating his/her scattered social data and extracting information relevant to the topics of interest of the group. Moreover, the user is provided with helpful features to make the best decisions for choosing the information he/she is ready to share.

2.1. Introduction

Online social networks, commonly known as social networks, such as Facebook1, Twitter2 or LinkedIn3 have become a very important part of our everyday life. Hundreds of millions of users from around the world are connected to these Websites to freely use their services for both personal and professional purposes. They use them to create, publish and share with others a lot of news, events as well as contents of their respective interests.

Current social networks are different from each other in terms of offered services and privacy policies [KIM 10]. They are also disconnected from each other. Therefore, the users must create and maintain different profiles if they want to be connected to different social networks. Consequently, they have to browse each of the corresponding Websites when they need to search for information.

Moreover, the friends that a given user has on each social network are not necessarily the same and can rapidly grow. Each friend has his/her own interests and may share only a small part with the user. Given the large number of friends of the user, it is not realistic for him/her to manually select information from all information relevant to his/her current interest from all information published by his/her friends.

A major consequence is that a considerable part of information, including those relevant to the user’s interests, remains ignored by the user. They, furthermore, prevent users, who may share a common interest, from an efficient Internet working knowledge sharing scenario.

In response to such a problem, we present in this chapter a new approach consisting of aggregating user data from different social networks, and filtering and indexing information relevant to the center of interests of the members of a given group. Taking into account user privacy issues, our approach is user centered. It means that the users grant the system-specific permission to aggregate their different social profiles into aggregated profiles and decide which part of their aggregated information should be shared or not.

This chapter is organized as follow. In section 2.2, we introduce our main motivations and the expected benefit of what we will define as a collective intelligence based on our user-centered social network aggregation approach. Then, we present some recent related works in section 2.3. In section 2.4, we describe an extensible architecture supporting our approach. In section 2.5, we introduce a special feature of our proposed system for supporting decision-making. A use scenario is described in section 2.6 for illustrating the system operation. We present our primary prototype in section 2.7. Finally, we conclude and present our future work.

2.2. Collective intelligence by user-centered social network aggregation

There are a lot of social Websites available on the Internet ranging from very general to domain-specific networks4. It is very common that the users are simultaneously connected to several Websites. They use them for two main purposes: (1) as a new type of communication platform [CHE 11] and (2) as a new medium of information sharing [KWA 10].

A huge amount of content is generated on social networks every day by the users. A lot of it is conversations between users while others reflect the users’ interests. This content, that we call social data, therefore includes a lot of updated personal and social information. Since the users may discuss and share with others any topics on social networks, social data can also be considered as a multidomain data source.

Thus, it is difficult and time-consuming for the social networks users to keep track of and to select interesting information from the enormous amount of social data published by their friends on different social networks. Some commercial solutions such as FriendFeed5 or Gathera6 allow users to pull their social accounts in one place and subsequently to search information across different accounts. However, this information is not categorized by the users’ center of interests and limited to those belonging to the users and their friends.

On the other hand, some social data may match the scope of interests of a given group, which could be a professional team, a collaborative group or any community of people sharing common interests. It would be very interesting to be able to capture this information and make it available within the group so that its members can easily access the gathered information at a single point. We can also visit different groups for viewing information relevant to one’s different interests. However, the disconnected nature of today’s social networks and user privacy issues prevent groups from an efficient Internetworking information collection in both manual and automated ways.

We have considered both of these problems that a group of users and single users are, respectively, facing, at the same time in order to propose solutions that could benefit both the group and the user. Our approach, collective intelligence by user-centered social network aggregation, consists of allowing the members of a given group to (1) aggregate their different social profiles and (2) share a part of aggregated information with the group according to its center of interests. Each user who is member of a group can therefore access more interesting information, including the information available from other members of the group who are not yet their social network friends.

Collective intelligence: we expect from this approach that the collective intelligence including knowledge sharing and knowledge creation of the group could be collaboratively empowered by the use of its members’ social data. Here are some types of information that we have identified as being useful for the group’s collective intelligence.

– Members’ additional interests/expertise: Each member’s interests/expertise is evolving and changing over time. Social data can be an alternative source for updating and enriching one’s profile by allowing the uncovering of his/her most recent interests and/or expertise [GAO 12, ORL 12]. Therefore, the group will learn more about each member. – New web resources: social networks are intensively used for publishing and spreading news content and web resources. For example, a significant part of tweets, short messages published by Twitter users, can be considered as information sharing [NAA 10]. Most of them contain URLs referring to web pages, thus allowing the discovery of new resources matching the interests of the group. –Emerging topics: by watching recently captured members’ additional interests and new web resources, emerging topics could be identified. – Possible subgroups: each member can be connected to some other members on one or several social networks. These relationships and their interaction degree [VU 12] will provide extra indicators beside the similarity indicators [HOR 10] for efficiently locating some possible subgroups. – Extended membership: some external users, who are not currently members of the group, may be considered for extended memberships when they belong to the lists of contacts of several members of the group. These people could be invited to join the group or lent a certain level of trust if other members would like to reach them for information.

User-centered: our approach depends essentially on the motivation and willingness for sharing personal social data with each of the other members of the group. They are the key components of our approach. We must keep in mind that one of the main tasks is to provide them with helpful means to have the best control over the information that they are ready to share. Therefore, we have adopted a user-centered approach. This means that the users are solely responsible for aggregating their own social profiles and for sharing some part of their aggregated information with a given group. Such an approach enables us to considerably reduce user privacy issues since social data are now authenticated and authorized by the users.

Moreover, it is obviously impossible to ask each member to copy their relevant information manually, already published on social networks, into the group. An automated process is...