Given a sentence, how can I generate similar sentences of that particular sentence but not with the same meaning using machine learning?

Question

I want to generate similar sentences of a particular sentence but not with the same meaning from the text corpus I have. How can I do it using machine learning? Any suggestions?

Answer 1

generative adversarial networks (GANs) are a good selection for it. (May you need remove generated sentences with high similarity measure to remove same meaning sentences). By this way u can use other similar sentence generation methods and remove sentences with high similarity measures to remove sentences with same meaning. Besides I suggest using mean of word2vec representation of words of each sentence to compute similarity of sentences as a simple method. Also, you can use parser to replace objects, verbs, subjects or other part of the sentence.

As u asked your question in a general way, I answered it in a simple and easy to implement method. Other better methods are available if you want to know them plz mention me.