Time series; crypto-currency; blockchain; gradual pattern mining; graph mining; knowledge discovery; imbalanced data
Résumé :
[en] Last few years have witnessed a steady growth in interest on crypto-currencies and blockchains. They are receiving considerable interest from industry and the research community, the most popular one being Bitcoin. However, these crypto-currencies are so far relatively poorly analyzed and investigated. Recently, many solutions, mostly based on ad-hoc engineered solutions, are being developed to discover relevant analysis from crypto-currencies, but are not sufficient to understand behind crypto-currencies. In this paper, we provide a deep analysis of crypto-currencies by proposing a new knowledge discovery approach for each crypto-currency, across crypto-currencies, blockchains, and financial stocks. The novel approach is based on a conjoint use of data mining algorithms on imbalanced time series. It automatically reports co-variation dependency patterns of the time series. The experiments on the public crypto-currencies and financial stocks markets data also demonstrate the usefulness of the approach by discovering
the different relationships across multiple time series sources and insights correlations behind crypto-currencies.
Disciplines :
Sciences informatiques
Auteur, co-auteur :
LAGRAA, Sofiane ; University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT)
CHARLIER, Jérémy Henri J. ; University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT)
STATE, Radu ; University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT)
Co-auteurs externes :
no
Langue du document :
Anglais
Titre :
Knowledge Discovery Approach from Blockchain, Crypto-currencies, and Financial Stock Exchanges
Date de publication/diffusion :
20 août 2018
Nom de la manifestation :
2018 ACM SIGKDD International Conference on Knowledge Discovery and Data mining conference (KDD 2018)
