Extracting new knowledge from the Web of Data using traditional data mining techniques is challenging for many reasons, including: the rich and complex graph structure; the large variety of link types; the availability of domain knowledge expressed as schemas and ontologies; the heterogeneity in the modelling goals behind individual datasets. This talk discusses the implications of Linked Data for a specific branch of descriptive data mining known as "Pattern Mining". A use case involving a recommender engine for tourist attractions is presented to illustrate the principles.
Keywords: Web of Data, Machine learning, Data mining, Concept analysis
Author: Valtchev, Petko
Publisher: Institut des sciences cognitives, UQAM, Montreal, Canada
Date created: 2014-07-16 07:00:00.000
Time required: P50M
- Articulates differences between the RDF abstract data model and the XML and relational models.
- Understands the RDF abstract data model as a directed labeled graph.
- Understands that Linked Data (2006) extended the notion of a web of documents (the Web) to a notion of a web of finer-grained data (the Linked Data cloud).