NFmcp @ ECML-PKDD 2018
7th International Workshop on
New Frontiers in Mining Complex Patterns
Amedeo Napoli
Université de Lorraine, CNRS, Inria, LORIA, 54000 Nancy, France
Title:
Contributions to Exploratory Knowledge Discovery based on Formal Concept Analysis
Abstract.
Knowledge
discovery (KD) in complex datasets and especially the mining of
interesting patterns can be read along several dimensions among which
data, knowledge and problem-solving dimensions. These dimensions have a
direct impact on the whole Knowledge Discovery process, and for guiding
the exploration of the data and the pattern spaces. In this
presentation, we will discuss Exploratory Knowledge Discovery in the
framework of Formal Concept Analysis (FCA) considering the dimensions of
data, knowledge and problem-solving.
FCA starts with a binary context and outputs a concept lattice, which can be visualized, navigated and interpreted by human agents, and, as well, processed by software agents. FCA can be extended into Pattern Structures, where objects may have complex descriptions such as numbers, sequences, trees, and graphs. Descriptions can be compared and a pattern concept lattice can be built accordingly.
In this framework, we will detail two important data mining problems, namely the mining of definitions in RDF data (Linked Data) and the search for functional dependencies. The former depends on the discovery of implications within RDF data and can be used to detect missing information and to complete RDF data. The second one is also based on implication discovery and partition pattern structures which can also be applied to biclustering.
We will discuss how pattern structures provide a general basis for solving such data mining problems. Finally, we will conclude by analyzing the benefits of FCA and pattern structures for exploratory knowledge discovery.
FCA starts with a binary context and outputs a concept lattice, which can be visualized, navigated and interpreted by human agents, and, as well, processed by software agents. FCA can be extended into Pattern Structures, where objects may have complex descriptions such as numbers, sequences, trees, and graphs. Descriptions can be compared and a pattern concept lattice can be built accordingly.
In this framework, we will detail two important data mining problems, namely the mining of definitions in RDF data (Linked Data) and the search for functional dependencies. The former depends on the discovery of implications within RDF data and can be used to detect missing information and to complete RDF data. The second one is also based on implication discovery and partition pattern structures which can also be applied to biclustering.
We will discuss how pattern structures provide a general basis for solving such data mining problems. Finally, we will conclude by analyzing the benefits of FCA and pattern structures for exploratory knowledge discovery.