Graph analytics has become instrumental in uncovering insights across various domains, specifically social networks. It serves as a crucial tool for analyzing the relationship between users in different online platforms. On the other hand, the knowledge graph is an organized representation of information where entities are connected through relations. Like human knowledge, it is capable of storing, querying, and analyzing complex relations. In this research, we apply methods of social network analysis to examine the communication patterns among partic- ipants in an online forum recognized for far-right extremism. Our study demon- strates the actors’ relationships and activities through different aspects of applica- tions over networks. In addition to this, with the assistance of Natural Language Processing (NLP) tools, we want to extract the entities to formulate knowledge graphs for a more semantically understanding of the relations between the actors. In extensive analysis, we identify the influential actors and map their relationships throughout the course of 76 monthly networks. Moreover, we illustrate the evo- lution of networks over that period, and their connections with significant events. The findings of this analysis aim to understand the nature of interactions and net- works and to allow practitioners to take necessary precautions to mitigate far-right activities on various online platforms.
Radical group activists capitalize on online and social media platforms to spread information and propaganda among similar-minded personnel, coordinate actions, and recruit new members. They employ public (such as Twitter, Instagram, and internet forums) and private (such as WhatsApp, Telegram, and message boards) platforms. Members share their ideology-related thoughts on public platforms and consistently interact with a community of users. In contrast, to recruit new members, they privately share ideological content, such as articles, quotes, and events information with the users. These platforms often function as echo chambers, where like-minded individuals mutually reinforce each other’s ideas, gradually aligning ideologies without external challenge, facilitating the unchecked spread of extremism. Most of the contents are memes and info-graphics that spread extremist ideologies. Additionally, some participants are especially skillful in deriving and propagating narratives aligned with their beliefs and thus have the potential to wield substantial influence. Hence, the narratives they propagate can be widely disseminated online to both targeted and diverse audiences. Moreover, by leveraging those platforms, these individuals and groups plan and coordinate new events through real-time communication. Besides, they may also raise funds and sell their ideological products through these platforms, further enhancing their capacity.
Semantic networks help machines to understand words’ meaning, text, and lexical analysis from a text corpus. To semantically understand the relation among different entities in a text corpus building a knowledge graph is essential. Knowledge graphs are being employed in many domain-specific tasks such as cultural heritage information preservation, medical terminology integration, keeping drug-drug and protein-protein interaction information. Networks and knowledge graph analyses can play a key role in reducing the misuse of online platforms to spread hate speech and violence. The analyses can also be beneficial for policymakers in designing preventive measures and implementing appropriate strategies to combat far-right activities. In this project, we have already done the basic network analysis on the conversations between dif- ferent group members in the Iron March Web Forum. Our other paramount objective of the project is that we will apply Natural Language Processing (NLP) tools to extract the conservation content and relevant topics discussed in different far-right groups. After getting the topics we create < user, topic, user> and < user, time, user> triplets to form knowledge graphs. To do so, we have extracted the effective named entities using a state-of-the-art fine-tuned Bert model [18]. From the entities, we want to form the knowledge graphs to understand their relations more semantically. In addition to this concrete information, we want to apply ontology in the graph relations to make it more un- derstandable. We embedded knowledge graphs, embedded the text corpus of a month among many users using a miniML sentence transformer. We found some similarities in the 2 −dimensional plottings of the embeddings that show a research direction to apply knowledge graph analysis on more complex network structures.
First, we create a network of users based on common topics in their posts to analyze the Iron- March forums activities. The conversation timestamps between activists are segmented into 30-day windows. In a single window, all the active members are considered as the nodes of the network. In the network, we connect two members with an edge if they post on the same topic thread. Edges are labeled with the topic. If a user posts on a topic and gets no response within that 30-day window, then a self-edge is assigned to that user. Through this process, a total of 76 monthly networks have been formed. Then we apply different functions and algorithms to detect the influential actors, and important events, and analyze the evolution of the networks