Tuesday, February 19, 2013

Mind maps of #textanalytics - automatic generation

What I want to show you today is "snippet" of a project I'm working during my spare time.
The project has the aim to gather, provide and represent the knowledge in a different way.
For the time being I don't want to disclose too much info, but the main idea is to leverage my concept of the graph entropy to capture information in a document (or in a data set) to create a new kind of indexer.

One of the output of the algorithm is the generation of the mind map.

Through the API provided by Twitter I downloaded the last 200 tweets from the hashtag #textanalytics.
I processed the tweets and here is the results:
#textanalytics mind map

In red have been depicted the first 10 relevant words extracted through the graph entropy criteria.
It's interesting notice how the map shows intuitively the links among the words.
Of course the map can be enriched with more links or more relevant words: the choice depends on the user choice.
In the next posts I'll show you how the clustering techniques discussed in the former posts can be profitably used to create homogeneous group of words related each others.
what do you think?
Stay  tuned


  1. Very interesting, as usual. I have been reading your blog from time to time and always enjoy your creative and nontraditional text mining projects.
    I will re-read them all again in the following few weeks to get ideas for my text mining class project.

  2. Informatics Outsourcing is an Offshore Data Management service company. Data Management Service includes all types of Data Conversion, File Conversion, XML Conversion, HTML Conversion,SGML Conversion, Document Conversion,Data Entry, Data Extraction and Validation,OCR and ICR Services with affordable price. Our team to give the solution quickly and given requirements.

  3. Hi Cristian

    Just stumbled across your blog - intriguing. And good to see someone coming at NLP from a commercial rather than academic angle!

    What kind of use did you envisage for the visualisation above - is it purely for 'exploration' purposes, or did you have something more in mind?

    1. Hi Tony
      actually I'm working on a project (in my spare time) to extract, store and provide "knowledge" from unstructured content as documents, web pages, tweets and so on.
      The mind map is just one of the possible outputs that can be obtained.

  4. cảm ơn bài viết rất hay

  5. this is a great article! i have learnt so much about mining from this interesting article