I am dealing with textual data. Imagine a certain number of statements (they could be descriptions), each one referring to a certain process, where each process ends in one f three categories: A, B, C.

I would like to characterize the descriptions of type A, the ones of type B, and the ones of type C. One possible way is to focus on the terms contained in the description. How can I represent them? One possibility is to compute a document term matrix and once each term is characterized by a vector, compute the distance among all the terms. Starting from these distances it is possible to find a minimum spannign tree connecting the terms and recostructing someway the concepts contained in the original descriptions. Doing that only for the description of type A, etc., we could reconstruct graphically what the questions of type A, etc. are about.

The question is: once the concepts are reconstructed...is there other evidence that can be extracted by network analysis?


1 Answer 1


One way to model text data using networks is with a knowledge graph. Knowledge graphs model entities as nodes and relationships as edges.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.