I'm working on a project which is based on Text Content Analytic. It is actually analyzing given text content and we are presenting it in different output format like - PDF report, Graph in Dashboard. For this we are using one open source tools provided by Stanford University named [CoreNLP][1].

There are basically two phase which generate output and save in DB.

1. KeyPhrase Analytics
2. Entity Analytic

As per development team they are apply some algorithm on above two process to generate output and for that they are using `CoreNLP` jar file.
As I am new about the `CoreNLP`, Initially I read content on Stanford Site to actually get idea about these two process so that I can write test cases to verify the Output and do the functional and Non-Functional testing.  But I don’t find such material to for testing. It is all about the development\customization only. 

1. Which kind of test data I should prepare to test this functionality?
2. Which kind of test scenario I have to define?
3. Is their any free tools available to test my output that it is proper & reliable?


  [1]: https://stanfordnlp.github.io/CoreNLP/index.html#about