Information retrieval in linked data: A model based on concept maps and complex networks analysis

Abstract

This article presents a model for information retrieval in linked open data using methods and complex network operations for ranking and selecting information, and concept maps for presenting the retrieved information to the user. The model shows the relationships between query terms that represent an informational need and presents them as concept maps. The underlying hypothesis is that the user’s relationship to the retrieved information occurs in the light of Brookes’ fundamental equation of information science. The cognitive structure of the cognoscente is a complex network that is modulated by the retrieved information which, in turn, is derived from a complex network. The final complex network is mapped into a resulting concept map enhanced by heuristics, such as the application of controlled vocabulary. The first study conducted, with qualitative characteristics and using an exploratory approach, was an information retrieval pilot test. It allowed the assessment of the algorithms used in the ranking and selection of the intermediate information networks and provided the framework for the implementation of a prototype. The prototype used a knowledge base of linked open data, derived from DBpedia, on which complex network analysis were carried out. The validation of the model presented relevant recall and precision when applied to a group of 17 users. The results are promising for the use of complex network operations and concept maps for information retrieval, especially linked data. Further research should observe the demand for more interactive actions and conduct experiments in other knowledge bases.

Keywords
Concept maps; Information retrieval; Information networks; Semantic web