Towards Extractive Text Summarization Using Multidimensional Knowledge Representation

Multidimensional knowledge representation (MKR) is the result of integrative text mining. Analysis results from individual text mining methods such as named entity recognition, sentiment analysis, or topic detection are represented as dimensions in a knowledge base to support knowledge discovery, visualization or complex computer-aided writing tasks. Extractive text summarization is a content-oriented task which uses available information from text to shorten its length in order to summarize it. In this regard, a MKR knowledge base provides a structure which is applicable as an innovative selection instrument for text summarization. This paper introduces cross-dimensional text summarization based on dimensional selection and filtering of results retrieved from MKR knowledge base.