Identifying context information in datasets

Abstract

Datasets are used in various applications assisting in performing reasoning and grouping actions on available data (e.g., clustering, classification, recommendations). Such sources of information may contain aspects relevant to context. In order to use to the fullest this context and draw useful conclusions, it is vital to have intelligent techniques that understand which portions of the dataset are relevant to context and what kind of context they represent. In this work we address the above issue by proposing a context extraction technique from existing datasets. We present a process that maps the given data of a dataset to a specific context concept. The prototype of our work is evaluated through an initial collection of datasets collected from various online sources. © Springer International Publishing Switzerland 2015.