Bitcoin User Network Analysis and Mining – BITUNAM

We use a combination of machine learning and network science approaches, both to make the data interpretable (for instance, to de-anonymize actors) and to interpret it.

Although the project is still in its infancy, we already present a general overview of the data we collected and analyse in a prototype website: bitunam.sci-web.net

Several theoretical articles have already published on how to analyse large dynamic networks.
In the future, we plan to release enriched dataset and analysis

1. Remy Cazabet, Souâad Boudebza, Giulio Rossetti, Evaluating community detection algorithms for progressively evolving graphs, Journal of Complex Networks, Volume 8, Issue 6, 1 December 2020, cnaa027, doi.org/10.1093/comnet/cnaa027
2. Vaudaine, R., Cazabet, R., & Largeron, C. (2020, April). Comparing the preservation of network properties by graph embeddings. In International Symposium on Intelligent Data Analysis (pp. 522-534). Springer, Cham doi.org/10.1007/978-3-030-44584-3_41
3. Duvivier, L., Robardet, C., & Cazabet, R. (2019, December). Minimum entropy stochastic block models neglect edge distribution heterogeneity. In International Conference on Complex Networks and Their Applications (pp. 545-555). Springer, Cham. doi.org/10.1007/978-3-030-36687-2_45
4. Boudebza, S., Cazabet, R., Nouali, O., & Azouaou, F. (2019, September). Detecting Stable Communities in Link Streams at Multiple Temporal Scales. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases(pp. 353-367). Springer, Cham. doi.org/10.1007/978-3-030-43823-4_30
5. Duvivier, L., Cazabet, R., & Robardet, C. (2020, December). Edge Based Stochastic Block Model Statistical Inference. In International Conference on Complex Networks and Their Applications (pp. 462-473). Springer, Cham. doi.org/10.1007/978-3-030-65351-4_37

Crypto-currencies, a technology introduced a few years back, are met with a large success, as seen for instance with Bitcoin reaching a capitalization of nearly 150 billion dollars (March 2018).
The information of all transactions done in bitcoin is freely available to all, stored in a public distributed ledger called the blockchain. However, little is known about who the actors of the bitcoin economy are, and what are the main usages of digital currencies. This is mainly due to the anonymous scheme of Bitcoin, that forbids to directly track the activity of a particular user. Recently published works have shown that it is nevertheless possible to retrieve the various transaction of the same actor using network analysis techniques. We propose to develop methods to characterize the nature of actors and of their activities, requiring contributions in large network mining. By quantitatively analyzing uses, we will be able to identify how important speculation, illegal activities and genuine transactions are in the overall Bitcoin usage, thus answering to the controversial question of the nature of crypto-currencies as a form of money. The project will involve two socio-economists working on money and alternative currencies to help us answer such questions.

An objective of the project is also to develop and shape a scientific community around the analysis –using knowledge from Big Data and Data Science– of these new sociotechnical systems that have potential to bring disrupting changes to economy and society.