Graph Co-Attentive Session-based Recommendation

Session-based recommendation aims to generate recommendations merely based on the ongoing session, which is a challenging task. Previous methods mainly focus on modeling the sequential signals or the transition relations between items in the current session using RNNs or GNNs to identify user’s intent for recommendation. Such models generally ignore the dynamic connections between the local and global item transition patterns, although the global information is taken into consideration by exploiting the global-level pair-wise item transitions. Moreover, existing methods that mainly adopt the cross-entropy loss with softmax generally face a serious over-fitting problem, harming the recommendation accuracy. Thus, in this article, we propose a Graph Co-Attentive Recommendation Machine (GCARM) for session-based recommendation. In detail, we first design a Graph Co-Attention Network (GCAT) to consider the dynamic correlations between the local and global neighbors of each node during the information propagation. Then, the item-level dynamic connections between the output of the local and global graphs are modeled to generate the final item representations. After that, we produce the prediction scores and design a Max Cross-Entropy (MCE) loss to prevent over-fitting. Extensive experiments are conducted on three benchmark datasets, i.e., Diginetica, Gowalla, and Yoochoose. The experimental results show that GCARM can achieve the state-of-the-art performance in terms of Recall and MRR, especially on boosting the ranking of the target item.

Journey of a Thousand Miles Begins with a Single Step: Transparity A Web-based Job Portal

Gender based discrimination is still common and one of the biggest issues in Pakistan. Transgender are ignored in almost every lifestyle. They face biasedness while finding for a job and are forced to earn money using unrespectable ways such as begging and dancing. Although, Pakistani law declares them equal citizen and many government as well as private organizations offer them jobs. Most of the transgender people remain unaware of these job offers due to the lack of an existing information propagation channel. This paper has proposed Transparity; which a dedicated online job portal for transgender. Transparity is also equipped with supportive features such as CV creation, online courses and motivational profiles to make it effective and helpful. Our study reveals that no such dedicated portal exists in Pakistan. This study focuses on the improvement of employment issues of transgender through.

TPSL: A propagation source localization method based on limited observation nodes

The research on localization of propagation sources on complex networks has farreaching significance in various fields. Many source localization methods have been proposed. However, the assumptions of some existing methods are too ideal, which means they cannot be widely deployed on realistic networks. In this paper, we propose a multi-source localization method TPSL based on limited observation nodes and backward diffusion-based algorithm with the consideration of heterogeneity of the propagation probabilities between nodes. Specifically, given a network topology with time and probability distributions, TPSL can infer the sources of propagation by comprehensively considering the time and probability factors in a way that accords with the characteristics of information propagation in reality. The experiments on artificial and empirical networks demonstrate that TPSL has excellent performance on these networks. We also explore the influence of different strategies of choosing observation nodes on TPSL, and find out that choosing the nodes with larger closeness centrality as observation nodes performs better. Moreover, the performance of TPSL does not be affected by the number of sources.

SOCIAL MEDIA IN INFORMATION WARFARE. ASSAULT WEAPON WITH HIGH RECOIL

Beyond the benefits or risks of individual or institutional communication through social media, we must note that it is the perfect environment for fake news and propaganda because of the speed of information propagation, the unfriendly environment for checking sources, algorithms behind social networks and, last but not least, the extremely low cost. In other words, the Internet and web 2.0 have created the favorable framework for the conduct of the war "for minds and hearts", as it can be called the information war waged through social media. Beyond these considerations, the non-regulation of the online domain - the lack of rules, be they deontological, make social media a powerful weapon of attack in this type of war. At the same time, the use of this space by state actors should be done with caution because it involves risks that could result in the loss of the most important action capacity: credibility. This article aims to analyze social media as a tool in information warfare

Injurious information propagation and its global stability considering activity and normalized recovering rate

This paper establishes a compartment model describing the propagation of injurious information among a well-mixed population. We define the information’s injuriousness as the people practicing the information being injured and leaving the system. Some informed people practice the information and are active, while others do not practice and are inactive. With the recovery resources fixed, the two groups of informed people’s recovering rates are normalized considering the information features. The stability of the nonlinear system is thoroughly studied. Analyzing the reproduction number of the injurious information, we find that in general parameter space, when there are people in an informed compartment, it is not always necessary to consider their recovery resource allocation. Instead, only when their proportion reaches a critical point should it be allocated. Unless the people in an informed compartment form a certain proportion, we can take a laissez-faire attitude towards them. In a more realistic parameter space, once inactive informed people exist, they should be allocated recovery resources. On the one hand, when the recovering rate rises, the focus on both groups of informed people is necessary for more situations. On the other hand, when the rate of active informed people leaving the system rises, ignoring active informed people benefits removing the injurious information in more cases. The model provides qualitative ways in the scenarios of removing injurious information.