Visualising Inconsistency and Incompleteness in RDF Gene Expression Data using FCA

Abstract

The integration of data from different data sources can result to the existence of inconsistent or incomplete data (IID). IID can undermine the validity of information retrieved from an integrated dataset. There is therefore a need to identify these anomalies. This work presents SPARQL queries that retrieve from an EMAGE dataset, information which are inconsistent or incomplete. Also, it will be shown how Formal Concept Analysis (FCA) tools notably FcaBedrock and Concept Explorer can be applied to identify and visualise IID existing in these retrieved information. Although, instances of IID can exist in most data formats, the investigation is focused on RDF dataset.

Mining and visualising contradictory data

“…There are calls for visual analysis applications with interactive capabilities [24,25]. Nwagwu and Orphanides [26,27] demonstrate how FcaBedrock [28] and Concept Explorer [29] can be used to visually analyse and identify gene expressions contradicting the mutually exclusion rule in gene expression dataset where a gene in a particular tissue of a particular Theiler stage is expected to be associated to only one type of expression. They applied formal concept analysis (FCA) tools and techniques in visualising the contradictory data in a large dataset.…”

“…More so, ConTra's approach enables the evaluation of the identified contradictions. Unlike existing approaches such as in [26,27], ConTra improves on the use of traditional visualisation tool (pie chart) in visual analysis of contradictory data, by mining and evaluating only the contradicting data. This enables the use of traditional visualisation tools in visually analysing the contradictory data in a large CSV dataset.…”

Visual Analysis of a Large and Noisy Dataset

“…Also, its tools and techniques enable the analysis of a large dataset. For example, Andrews and Orphanides (2010) demonstrate how some open-source FCA tools notably FcaBedrock 5 and In-Close 6 are used to analyse large formal (Dau, 2013a;Melo, et al 2013;Nwagwu, 2014), how different FCA tools and techniques are used to identify and visualise IID existing in a dataset.…”

“…It is shown in (Nwagwu, 2014;Dau 2013a), how IID in a large and noisy dataset is identified by visualising the attributes of the IID in the dataset. Jiang et al (2009) used a node without a label for its own object in a concept lattice to show an anonymous node.…”

“…These works present examples of descriptive or qualitative approach to visually analysing IID. It is also shown in (Nwagwu, 2014), how the amount of IID existing in a dataset can be evaluated and also depicted in a concept lattice. This is achieved by measuring the IID existing in each object of a dataset and depicting the amount of IID, its corresponding attributes and the associated object in a concept lattice.…”