Modelling the system of receiving quick answers for e-government services: Study for the crime domain in the Republic of Serbia

Abstract

© 2017, Budapest Tech Polytechnical Institution. All rights reserved. Today, e-Governments services operate quick answer systems in many different domains in order to help citizens receive answers to their questions at any time and within a very short time. The automatic mapping of relevant documents stands out as an important application for automatic questions-documents classification strategies. This paper presents a contribution to the identification concepts in text comprehension in unstructured documents as an important step towards clarifying the role of explicit concepts in information retrieval in general. The most common representational scheme in text categorization is the Bag of Words approach when the dictionary, as incorporating background knowledge, is large. The authors introduce a new approach to create concept based text representations, and apply it to a text categorization collection to create predefined classes in the case of short text document analysis problem. Further in this paper, a classification-based algorithm for questions matching topic modelling is presented. The paper also describes the weighting of concepts that present terms with high frequency of occurrence in questions is based on their similarity relevance in the predefined classes of documents. The results of the experiment within the criminal law domain in the present case study show that the performance of concept-based text representations has proven to be satisfactory in the case when there is no dictionary for this domain.