CohesionNet: Software for network-based textual cohesion analysis

Abstract

Cohesion is one of the main defining characteristics of a text. CohesionNet, an R app with a Shiny interface, processes raw text to calculate network-based cohesion indices. The indices are based on stem repetition and on the analysis of synonymy and hypernymy. The app also constructs a network representation of the text that can be saved in the Pajek NET format. CohesionNet facilitates the assessment of potential applications of the indices, like text classification, automatic summarization, and readability improvement. Currently supporting English texts only, upcoming versions will include additional language support.