Social network data analysis and mining applications for the Internet of Data

Abstract

Summary Social network analysis is an interdisciplinary topic attracting researchers from biology, economics, psychology, and machine learning, with an existing long history based on graph theory. It has since attracted interests from both the research and business communities for a strong potential and variety of applications. In addition, this interest has been fueled by the large success of online social networking sites and the subsequent abundance of social network data produced. An important aspect in this research field is influence maximization in social networks. The goal is to find a set of individuals to be targeted with the aim to drive social contagion and generate a diffusion cascade. We provide here an overview of the models and approaches used to analyze social networks. In this context, we also discuss data preparation and privacy concerns. We further describe different kind of approaches based on centrality measures, which express a sociological interpretation of the data, and stochastic influence and information propagation techniques, which aim at modeling the underlying diffusion processes that govern social interactions.

Building University-Industry Co-Innovation Networks in Transnational Innovation Ecosystems: Towards a Transdisciplinary Approach of Integrating Social Sciences and Artificial Intelligence

“…Data mining, as a field of study within ML learning, "is one of the most effective alternatives to extract knowledge from the great volume of data, discovering hidden relationships, patterns and generating rules to predict and correlate data" [118] (p. 687). Such a technique has often been applied in social network analysis [119,120].…”

Social network analysis of an emerging innovation: direct-to-garment printing technology

“…Usergenerated data collection overcomes a number of shortcomings associated with traditional primary data collection. The capture of user-generated data facilitates analysis of dynamic interaction among geographically dispersed respondents and also potentially reduces bias (Cuomo and Maiorano, 2017). Primarily researcher bias, commonly associated with instrument development and data collection, is eliminated among user-generated data.…”

APEX2S: A two‐layer machine learning model for discovery of host‐pathogen protein‐protein interactions on cloud‐based multiomics data

“…11 The major impediments toward the success of scientific workflows may be due to, first, the specific domain knowledge, and second, the distributed data over the internet, although the internet of data paradigm has stimulated numerous data analysis and mining applications. [12][13][14] In our research, we focus on host-pathogen protein-protein interactions, which we anticipate in delivering a workflow to bridge the aforementioned research gaps. This work is a substantial extension of the conference version, 3 which has initially investigated the performance of the two-layer model by one feature representation algorithm.…”

“…Since the nature of different scientific applications of workflow may varies dramatically, performing workflows in a service‐oriented environment and further applying machine learning models for predictive tasks have remained a difficult task in the life science applications 11 . The major impediments toward the success of scientific workflows may be due to, first, the specific domain knowledge, and second, the distributed data over the internet, although the internet of data paradigm has stimulated numerous data analysis and mining applications 12‐14 …”