Optimal Community-Generation Methods for Acquiring Extensive Knowledge on Twitter.

This paper proposes a method to generate an optimal community on Twitter by grouping users using a solution to the knapsack problem. Most past studies have proposed methods to recommend one user or some candidates, rather than recommending multiple users as candidates. It is quite difficult to recommend candidates as a single group that considers the balance of user characteristics, because grouping in terms of relationships among users is a combinatorial problem, especially on Twitter, as a huge number of users must be handled, so combinatorial explosion occurs easily. Although the combination optimization problem is difficult, with the knapsack problem, it is possible to obtain a solution of good quality within a practical calculation time. In this paper, we calculate the combination of users whose total amount of knowledge is maximized by using “amount of knowledge acquired” as a community evaluation item. In addition, we conduct a subject experiment to evaluate the information obtained from the generated community and the performance of the proposed method.