Daniel Davis asked 2 months ago
My team doesn’t know much about NLP, and they haven’t time to learn it, either. One thing we need to do is to be able to determine how cohesive a text is, and also the readability of that text. I’ve found frameworks that will return the readability of a text, but not which will summarize and provide a cohesion factor with that.
Is this a feature offered by any of your APIs?
Hi Daniel,
That is a good question. The summarization API provided by RxNLP is meant for review/opinion texts. However, if you are looking to summarize text by extracting core topics, the API supports that, but it has no notion on cohesion. One suggestion I have in terms of general text summarization is to look at tools like MEAD, http://www.summarization.com/mead/ and https://pypi.python.org/pypi/sumy, that does extractive summarization. Because MEAD re-uses sentences as summaries, you would not have to worry about readability. To get some ideas on how to apply cohesion on top of this or what you already have you can check out this slide deck: http://romip.ru/russir2008/slides/nlpiaa/RUSSIR-2008-lecture-sum.ppt or try to get some ideas from this paper: http://www.cs.brandeis.edu/~im5/papers/Spring98Symp.pdf