What is the name of a text analysis method where one builds a network of sentences that are connected when they contain words that occur on a predefined reference list of words?

So, lets say I have a predefined list of the following words: "tree","forest","wood". Then I have a random text that I want to compare to my predefined list to build a network. Let's say this random text has the following sentences: "He was climbing trees", "She was in the forest". Both sentences contain words that match the predefined list. In the network, these two sentences should be represented by two nodes. As the sentences share words from the reference list, the nodes should be connected.

I'm really new to text analysis and have no idea how this approach is called. Any help would be appreciated.

  • Welcome to Linguistics SE! – Alenanno Jun 19 '15 at 8:40
  • What you're looking for seems akin to "indexing" that's done for books. I can't give a more concrete answer without an application in mind. – prash Jun 19 '15 at 9:42
  • Where did you hear about this method? What was the context? Why was no name mentioned? – Dominik Lukes Jun 19 '15 at 21:33

I'm not aware of such technique and don't see really what this network would look like.

What you describe is classification or categorization, or topic extraction if you like. In your case it's dictionary-based - you use a dictionary to predict the class or set of classes.

The network part sounds as your way to represent this information. You'll end up with a set of clusters, each node connected to each other in the same cluster (i.e. all sentences containing "tree"). You'll also have some links between clusters, for sentences that contain words from both dictionaries. This however is no longer an NLP problem.

Here are some links for text classification: Coursera course on NLP, Introduction to Information Retrieval book

  • Welcome to Linguistics SE! – Alenanno Jun 19 '15 at 8:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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