Radoslaw Michalski presented a talk titled "Cognition-driven Temporal Social Networks," coauthored by Boleslaw K. Szymanski et al. at the Networks 2021 Conference, remote, on July 6, 2021

Radoslaw Michalski presented a talk titled "Cognition-driven Temporal Social Networks," coauthored by Boleslaw K. Szymanski, Przemyslaw Kazienko, Christian Lebiere, Omar Lizardo, and Marcin Kulisiewicz at the Networks 2021 Conference, remote, on July 6, 2021. The talk focus on human relations are driven by social events - people interact with each other, exchange information, share knowledge and emotions, and gather news from the mass media. These events leave traces in human memory which is an essential part of human cognition. The initial strength of a trace depends on cognitive factors such as emotions or attention span. Each trace continuously weakens over time unless another related memory activity strengthens it. Here, we introduce a novel Cognition-driven Social Network (CogSNet) model that accounts for cognitive aspects of social relation perception. To validate the model, we use the NetSense dataset on social interactions between university students. We apply the CogSNet model to NetSense data accounting for memory impact on social network perception. The results demonstrate that the cognition-driven model improves quality of modeling human interactions in social networks. We introduce a the forgetting function f( t) over time interval t can be of any type, but here, informed by the studies of cognitive processes, we evaluate only two such functions: the exponential and the power functions. To simplify optimal parameters search, we aggregate all three model parameters into the trace life time L which defines the time after which an unreinforced memory trace is forgotten. In the model, L is the time over which the forgetting function reduces the edge weight to the level causing the edge to be removed. For validation of the model, we used data of about 200 university freshmen that were surveyed at University of Notre Dame after each term in order to acquire information about them and their relations with others (the NetSense project). We use this data to study the evolution of two coupled social networks of university students. The first is a behavioral network of interactions between individuals in the form of the records of their mobile calls and text messages. The second network has perceptual edges defined by the personal nominations by students. These nominations are based on students' perception of the corresponding relations as one of the top twenty most interacting peers in the surveys administered to participants. Using the CogSNet model, we also build a network based on recorded mobile phone communications. We compare the list of nominations predicted from the CogSNet network model purely from the communication event data with the list of nominations collected in a given survey. The best precision results, significantly outperforming all the baselines, were achieved for trace life time L up to 1-2 weeks.