Bisociative Knowledge Discovery for Microarray Data Analysis

Mozetic , I , Lavrac , N , Podpecan , V , Novak , P K , Motaln , H , Petek , M , Gruden , K , Toivonen , H & Kulovesi , K 2010 , ' Bisociative knowledge discovery for microarray data analysis ' , pp. 190-199 05/2011;
Source: OAI


The paper presents an approach to computational knowledge discovery through the mechanism of bisociation. Bisociative reasoning is at the heart of creative, accidental discovery (e.g., serendipity), and is focused on finding unexpected links by crossing contexts. Contextu- alization and linking between highly diverse and distributed data and knowledge sources is therefore crucial for the implementation of bisocia- tive reasoning. In the paper we explore these ideas on the problem of analysis of microarray data. We show how enriched gene sets are found by using ontology information as background knowledge in semantic sub- group discovery. These genes are then contextualized by the computation of probabilistic links to diverse bioinformatics resources. Preliminary ex- periments with microarray data illustrate the approach.

Download full-text


Available from: Petra Kralj Novak
  • Source
    • "Several papers have been published on supporting creativity by discovering links between concepts. In creative biological problem solving, for instance, Mozetic et al. [26] propose a method for finding unexpected links between concepts from different contexts. Examples of methods more directly based on link prediction in heterogeneous networks are given by Eronen and Toivonen [27]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: A fluent ability to associate tasks, concepts, ideas, knowledge and experiences in a relevant way is often considered an important factor of creativity, especially in problem solving. We are interested in providing computational support for such creative associations. In this paper we design minimally supervised methods that can perform well in the remote associates test (RAT), a well-known psychometric measure of creativity. We show that with a large corpus of text and some relatively simple principles, this can be achieved. We then develop methods for a more general word association model that could be used in lexical creativity support systems, and which also could be a small step towards lexical creativity in computers.
    Full-text · Conference Paper · Nov 2012
  • Source
    • "As shown in Section 5, semantic data mining results in more general and semantically more meaningful rules, if compared to standard rule learning. Motivated by the successful applications of SEGS [19] [14], we have decided to generalize SEGS to become domain independent, and developed a new system named g-SEGS (generalized SEGS). From the four main components of SEGS, only the SEGS hypothesis language and the generation and pruning procedure are used unchanged in the new semantic data mining system g-SEGS. "
    [Show abstract] [Hide abstract]
    ABSTRACT: With the expanding of the Semantic Web and the availability of numerous ontologies which provide domain background knowledge and semantic descriptors to the data, the amount of semantic data is rapidly growing. The data mining community is faced with a paradigm shift: instead of mining the abundance of empirical data supported by the background knowledge, the new challenge is to mine the abundance of knowledge encoded in domain ontologies, constrained by the heuristics computed from the empirical data collection. We address this challenge by an approach, named semantic data mining, where domain ontologies define the hypothesis search space, and the data is used as means of constraining and guiding the process of hypothesis search and evaluation. The use of prototype semantic data mining systems SEGS and g-SEGS is demonstrated in a simple semantic data mining scenario and in two real-life functional genomics scenarios of mining biological ontologies with the support of experimental microarray data. KeywordsSemantic data mining–ontologies–background knowledge–relational data mining
    Full-text · Chapter · Sep 2011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We consider data mining methods on large graphs where a set of labels is associated to each vertex. A typical example concerns the social network of collaborating researchers where additional information concern the main publication targets (preferred conferences or journals) for each author. We investigate the extraction of sets of dense subgraphs such that the vertices in all subgraphs of a set share a large enough set of labels. As a first step, we consider here the special case of dense sub-graphs that are cliques. We proposed a method to compute all maximal homogeneous clique sets that satisfy user-defined constraints on the num-ber of separated cliques, on the size of the cliques, and on the number of labels shared by all the vertices. The empirical validation illustrates the scalability of our approach and it provides experimental feedback on two real datasets, more precisely an annotated social network derived from the DBLP database and an enriched biological network concerning protein-protein interactions. In both cases, we discuss the relevancy of extracted patterns thanks to available domain knowledge.
    Full-text · Conference Paper · Sep 2010
Show more