Hassan Shakeri

Computing trust resultant using intervals

An important problem in trust management area is to evaluate the trust value among two nodes in a web of trust using intermediate nodes. This is widely used when the source node has no experience of direct interaction with the target. If trust is used to support decision making, it is important to have not only an accurate estimate of trust, but also a measure of confidence in the intermediate nodes and in the final estimated value of trust. In this paper we introduce a novel approach for representation of trust and confidence -both together- using intervals. Based on this representation, we then propose a kind of summation operation that is a method for calculating the resultant of trust opinions. We show that this operator is more accurate for evaluation of trust resultant than the usual method of weighted-averaging. We also report and analyze the results of experiments carried out on a well-known trust dataset which show that our proposed method reduces the average absolute difference and increases the correlation among the inferred and direct values of trust comparing to the weighted-averaging method.

A New Graph-Based Algorithm for Persian Text Summarization

Nowadays, with increasing volume of electronic text information, the need for production of summary systems becomes essential. Summary systems capture and summarize the most important concepts of the documents and help the user to go through the main points of the text faster and make the processing of information much easier. An important class of such systems is the ones that produce extractive summaries. This summary is produced by selecting most important parts of the document without doing any modification on the main text. One approach for producing this kind of summary is using the graph theory. In this paper a new algorithm based on the graph theory is introduced to select the most important sentences of the document. In this algorithm the nodes and edges will be assigned with different weights and then the final weight of each one will be defined by combining these values. This final weight indicates the importance of the sentence and the probability of appearing this sentence in the final summary. The results show that considering simultaneous different criteria generate a summary which is more similar to human one.

RTBIMS: Accuracy enhancement in Iterative Multiplication Strategy for computing propagated trust

Trust Management (TM) systems are frameworks for managing security in decentralized environments. However computing the trust value is a challenge when the source entity has no experience of previous direct interaction with the sink. A popular way to estimate the trust value in this case is trust propagation. It suggests evaluating the trust value between two nodes based on the information received from intermediate nodes between them. As we often use a number between 0 and 1 to represent the trust value, the Iterative Multiplication Strategy (IMS) is usually used to estimate the value of propagated trust. In this paper we introduce RTBIMS, an accuracy-enhanced version of IMS. In our algorithm, we first calculate recommendation trust for each intermediate node based on its trustworthiness in judgments and similarity between its views and the source ones. We then use these values in the multiplication chain. We have examined RTBIMS on a popular trust dataset (Advogato) and the results show that its accuracy in evaluating trust is much higher than the basic IMS.

STAR: Improved Algorithm based on Sliding Window for Trust-Aware Routing in WSNs

This research with attention to the establishment of trust in WSNs and with the goal of increment in energy supply and growth in malicious node detection accuracy by using of improved sliding window, is saving energy by using computation of the previous periods. Also, this research calculates the trust in aspect of transferring information, based on subjective logic model and incremented the detection rate of malicious node by proposing two algorithms for identifying these nodes. Then, this method increases the speed of routing. The results of simulation of STAR compared to EDTM (Jiang, et al., 2015) shows 11.99% increment in the residual energy of network and growth of 1.52% in detecting the accuracy of malicious nodes.

A Model for evaluating trust between users in social networks based on subjective logic

In recent years, social networks developed in terms of diversities, capabilities and users. Therefore, researchers in different fields have been attracted to these kind of networks. Trust management between users in these kind of networks is one of the most important research topics. In this article a new model for evaluating trust between users in social networks based on subjective logic is presented. Firstly, trust and confidence are calculated with regard to the criteria and categories have been considered. Then, these calculated opinions are converted to subjective logic model and finally these opinions will be combined to obtain general trust. For accepting or rejecting friend requests a threshold is considered. For evaluating the proposed model, a certain type of Leave-One-Out validation technique is used. The proposed model is evaluated by different kinds of evaluation methods. Precision and Recall metrics that have been used for this model, represent a significant improvement in this method over the other methods.

CATEF: Confidence-aware trust estimation framework

In a trust management system, it is useful to have a measure of confidence in the estimated value of trust. Considering the concept of confidence, results in more accurate estimation of trust. In this paper we introduce a computational framework for confidence-aware trust estimation, CATEF. In this framework, for an atomic trust opinion, the reliability of the opinion is estimated and considered as the confidence value. CATEF also suggests novel operators for aggregation and propagation of confidence-aware trust opinions. These operators take into account the number and consistency level of the aggregated opinions and the length of the propagation chain. The results of experiments carried out on the well-known dataset of Epinions, show that our approach increases the accuracy of trust inference comparing to the existing methods.

Alleviating controversiality problem in trust estimation

Trustworthiness of users has become an indispensable aspect of security in networks. Not having an experience of interacting among nodes is a challenge for this phenomenon. Trust Management(TM) systems brought into play to bring security to any kind of network. Many web-based social networks assign users unique scores based on their interaction with their neighbors. Others, however, argue that trustworthiness of a user varies according to the different views of different users, and consider the so-called local trust metrics. Due to the existence of controversial users, users who are trusted and distrusted by many, in any social community, we have to consider other facts to estimate trusts with better accuracy. There is a strong bond between sociology and trust. Therefore, in this paper, we propose PredSimTrust, a trust estimation algorithm which estimates trust of a target user based on the similarities and trusts of predecessors. We examined the proposed algorithm on a real user community. The results show that our algorithm outperforms other models such as Iterative Multiplication strategy (IMS) and other models which take just one of these notions (trust value and similarity) into consideration.