Text summarization is a long-standing research problem that was "ignited" by Luhn in 1958. However, a half century later, we still came nowhere close to solving this problem (abstractive summarization). The reason for this might be because researchers are resorting to statistical (and sometimes linguistic) methods to find & extract the most salient parts of the text.

Is summarization problem solvable using AI (neural networks to be precise)?


The ability to re-frame summarization as a problem for ANN is rather dependent on what kind of output you're looking for: you mentioned 'salient parts of the text'.

One possibly is to use a deep learning approach that first chunks together words that belong in the same phrase as a single 'feature'.

Another possibility is to identify both key words and relations between them. Here is some previous work on using neural nets for relational learning.

  • $\begingroup$ that's interesting, thanks for the info! Is it theoretically possible to build a knowledge base from the text (using deductive reasoning) and to train NNs to generate text from the knowledge base? $\endgroup$ – user220 Aug 3 '16 at 17:36
  • $\begingroup$ @JamhurMustafayev I would imagine so, although deductive reasoning would not be my first choice for finding relational primitives in text: NLP isn't my area, but my understanding is that it has a lot of specific techniques for entity relationship extraction. $\endgroup$ – NietzscheanAI Aug 3 '16 at 17:47

Your Answer

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