标题：social context-aware trust networkdiscovery in complex contextual social networks
本文介绍了如何在社会网络中计算信任关系，作者总结前人的成果，对信任度进行了重新的定义，并且提出了自己的算法。这个问题在图论中是一个典型的NP完全问题，文中提出的SCAN算法应该来说在一般情况下能对这个问题加以解决。
下面是本文摘抄的要点：
1 发现信任网络的目的
the trust network discovery aims toidentify a trust network between two nonadjacent participants with less nodesand less links but including the important intermediate participants, theirtrust relations, and the social context.
2 social contextualimpact factors:
Trust: the beliefof one participant in another. T’AB[Di] denote the trust value that A assignsto B in domain i. T’AB[Di]=0, it indicates that A completely distrusts B indomain i, while T’AB[Di]=1 indicates A completely believes B’s future action.
Social intimacydegree: SI’AB denote the social intimacy degree between A and B. SI’AB =0 indicatesthat A and B have no social relationship while SI’AB=1 indicates they have themost intimate social relationship.
Role impactfactors: pA[Di] illustrating the impact of participant A’s social role indomain i, it is 1 indicates that A is a domain expert in domain i. pA[Di]=0 indictesthat A has no knowledge in that domain.
Preferencesimilarity:PS’AB[Di]=0, A and B have no similar preference in the domain , PS’AB[Di]=1,theyhave the same preference in that domain.
Residentiallocation distance:RLD’AB=1, the residential location of A and B are the same.RLD’AB=0 indicates that the residential location between them has the largestdistance.
考虑这些因素后，原来的图变成下图：
3 socialcontext-aware social interaction probability
In our model, weassume the probability distribution of a social interaction between any twoparticipants with social contextual impact factors follows the normaldistribution:
4 quality of trustnetwork
Definition qualityof network is the ability of a contextual trust network to guarantee a certainlevel of trust evaluation, taking T, SI, p, PS, RLD as attributes.
5 trust networkutility
The utility is themeasurement of the trustworthiness of an extracted trust network.
6 SCAN
The sourceparticipant vs is regarded as the current expansion node, and SCAN searches allthe neighboring nodes of vs to investigate whether the current node and itscorresponding links satisfy the QTN constraints.
OptimizationStrategy 1: Avoiding Repeated Feasibil-ity Investigations in Simulations.
OptimizationStrategy 2: Avoiding Repeated Probabil-ity Calculations in Simulations.
