Abstract:

As a means of communication, social networks have taken root in people’s hearts. The user data of social networks has a lot of value in this big data era. With the opening of Twitter Application Programming Interface (API), Twitter, as a social networking site, has become a popular research object, especially the user influence. The PageRank algorithm has  long been in use to calculate users’ influence, however, it is too dependent on the following relationship between users, so the ranking of users does not have strong timeliness. We introduce user activity to improve the PageRank algorithm, which has a certain degree of timeliness, but not convincing and accurate. We propose a new algorithm called PageRank activity based (ABP) algorithm according to the time distribution of user activity, and corresponding ageing weight factors are applied to the active degree of different periods of time. Finally we taking Twitter as the research object and combining with the social relationship graph, we prove that the ABP algorithm is more efficient and persuasive through an example analysis, and it can be more accurate in improving the ranking of active users and reducing the ranking of inactive users.

Key words: social network, data acquisition, user influence, ABP algorithm