Research Article
pyDNetTopic: A Framework for Uncovering What Darknet Market Users Talking About
@INPROCEEDINGS{10.1007/978-3-030-63086-7_8, author={Jingcheng Yang and Haowei Ye and Futai Zou}, title={pyDNetTopic: A Framework for Uncovering What Darknet Market Users Talking About}, proceedings={Security and Privacy in Communication Networks. 16th EAI International Conference, SecureComm 2020, Washington, DC, USA, October 21-23, 2020, Proceedings, Part I}, proceedings_a={SECURECOMM}, year={2020}, month={12}, keywords={Darknet market forums Topic modeling FBTM pyDNetTopic}, doi={10.1007/978-3-030-63086-7_8} }- Jingcheng Yang
Haowei Ye
Futai Zou
Year: 2020
pyDNetTopic: A Framework for Uncovering What Darknet Market Users Talking About
SECURECOMM
Springer
DOI: 10.1007/978-3-030-63086-7_8
Abstract
Although Dark Net Market (DNM) has attracted more and more researchers’ interests, we found most works focus on the markets while ignore the forums related with them. Ignoring DNM forums is undoubtedly a huge waste of informative intelligence. Previous works usually utilize LDA for darknet data mining. However, traditional topic models cannot handle the posts in forums with various lengths, which incurs unaffordable complexity or performance degradation. In this paper, an improved Bi-term Topic Model named Filtered Bi-term Model, is proposed to extract potential topics in DNM forums for balancing both overhead and performance. Experimental results prove that the topical words extracted by FBTM are more coherent than LDA and DMM. Furthermore, we proposed a general framework named pyDNetTopic for content extracting and topic modeling uncovering DNM forums automatically. The full results we apply pyDNetTopic to Agora forum demonstrate the capability of FBTM to capture informative intelligence in DNM forums as well as the practicality of pyDNetTopic.
