E-mail Prioritization through Social Network Analysis

Context. Trust and reliability are important issues in online communication. By rapid growth of online social networks (OSNs), online communication becomes richer by the integrating of social interaction into the communication model. However, E-mail communication systems concern about unsolicited messages. Objectives. In this thesis the aim is to investigate how to prioritize E-mails between recipients and senders by using information from OSNs. Methods. An algorithm is presented for computing trust by measuring users‟ interaction and similarity in online social networks and this trust is used by another algorithm for prioritizing the E-mail inbox. Results. An evaluation of the proposed method is performed via a case study and the prediction error of the method is compared with the prediction error of the random feedback. The error of the method is significantly lower than random feedback and is relatively low, given the small number of observations. Conclusions. This thesis contributes in its review and categorization of existing trust models. Furthermore, it provides an analysis on how to use social information for E-mail prioritization. Based on the analysis, a method is presented for improving the reliability of E-mail communication by extracting information from OSNs. The information is used for computing the trust score between two OSN friends. In this thesis, it is suggested that, inbox prioritization is achievable using the selected method.This thesis has addressed E-mail prioritization through social network by using social information. The task has been done by focusing on the interaction and similarity between friends in the OSN. A theoretical analysis has been performed in order to identify the characteristic of suitable trust model. An algorithm (Algorithm 1) has been suggested to estimate weights of different criteria of social information. In order to have the trust predictions based on the user‟s preferences, the algorithm adjusted the weights based on the user‟s feedback. In addition, another algorithm (Algorithm 2) has been proposed to compute trust scores and prioritize E-mails inbox. Finally, an algorithm (Algorithm 3) has been presented to evaluate the error of the computed (predicted) trust scores. In order to display the applicability of the method as well as to motivate the theoretical foundation, a case study was reported in which the proposed method was applied to Facebook. The analysis showed that the proposed method was feasible to be used, and it provided users a mean to prioritize E-mail inboxes based on the social information extracted from Facebook. The analysis indicated that least squares method was a suitable approach to estimate weights that were used in computing trust scores and thus prioritizing E-mails inbox