Detection of Incoherences in a Document Corpus Based on the Application of a Neuro-Fuzzy System

The aim of this paper is to detect incoherences in concepts, ideas, values, and others contained in technical document corpora. The way in which document collections are generated, modified or updated generates problems and mistakes in the information coherency, leading to legal, economic and social problems. A solution based on summarization, matching and neuro-fuzzy systems is proposed to dealt with this problem. For this goal, every document (from the electric domain) is summarized by its relevant information in the form of 4-tuples of terms, describing the most relevant ideas and concepts that must be free of incoherences. These representations are then matched using several well-known algorithms (Levenshtein distance and cosine similarity). The final decision about the real existence or not of an incoherence, and its relevancy, is obtained by training a neuro-fuzzy system FasArt in a supervised classification process, based on the previous knowledge of the activity area and domain experts. On the other hand, using this fuzzy approach, it is possible to extract the learnt and expert knowledge from the the neuro-fuzzy system, through a set of fuzzy rules that can support a decision taking system about this complex and non objective problem.