Selecting and trust computing for transaction nodes in online social networks

In current social network, the mutual trust relationship is the basis of service transaction. The lack of effective incentive strategy can lead to little incentive and enthusiasm of nodes. Meanwhile, some nodes probably recommend the certain node in evaluating trustworthiness of others because of the interrelated interests, also referred as collaborative cheating, which is likely to cause untrustworthy issues of recommendation. In order to solve this problem, we propose selecting transaction nodes and trust computing between them in social network. First of all, service transaction nodes acquire service opportunity and gain payment through bidding method, which can incentivize enthusiasm of service transaction nodes. Secondly, a novel transaction node selecting method based on entropy weight and TOPSIS to avoid random selecting and select the appropriate node. Then, the transaction influential functions are brought in to fuse direct trust and recommendation trust of nodes. In which, we can avoid malicious recommendation and collaborate cheating by comprehensive consideration of multi-dimensions factors, including recommendation time influential function, transaction content similarity and recommendation acquaintance. Furthermore, we propose a recommendation trustworthiness updating method based on multi-attributions. The experiment results demonstrate good effect of trust computing algorithm in restraining malicious cheating, malicious recommendation and collaborative cheating, and the recommendation trust updating issue can be ensured.