From entity extraction to network analysis: a method and an application to a Portuguese textual source

This paper reports advances in the entity extraction task (named entity identification) of a text mining process that aims at unveiling non-trivial semantic structures, such as relationships and interaction between entities or communities. We proposed a 3-phase method that is applicable to the Portuguese language and potentially applicable to other languages as well. The method relies on flexible pattern matching, part-of-speech tagging, lexical-based rules and distance-based entity name merging. All steps are implemented using free software and taking advantage of various existing packages. Evaluation of the efficacy of the entity extraction method on part of a book written in portuguese indicates improved F1 results. For further evaluation and illustration of the usefulness of the proposed method, it is applied to a book on Freemasonry and observe the differences in the entity word clouds produced. We also define a social network of named entities solely from information contained in the book and extract structural insights that reveal connections, relationships and communities between entities.