There is a task. We have a huge database of reviews for certain brands / products. I need to generate meaningful review for a new product based on a trained model for this database. I can parse some input about this product from the network and based on it and based on the trained neural network, generate a review about this product.

  1. Is this real now?
  2. Is it better to create the application myself or are there services that provide solution?
  3. If create the app myself, what technologies are better to use for this?
  4. Is it possible separate the reviews as pros and cons properties of the product?

There is one obvious problem with this task: the result is not a real review, it's a generated text which looks like a review. Given that the point of a product review is usually to provide the reader with some information about the product, it's not clear to me how this task would be useful: if the review can be made without even testing the product, its informative value is zero (it's even misleading).

Similarly one could train a model to generate "news articles": if the model is well trained the generated text would look exactly like real news, but there's very little chance that the content would correspond to what actually happens in the world.

  • $\begingroup$ Thanks for your reply. Yes, we are quite aware that the result will not be as good as the worst review from a real person. And we are not going to deceive anyone - the review will be marked as generated. It is just such a small game for the users' amusement. $\endgroup$
    – Astraport
    Jun 16 at 18:14
  • $\begingroup$ @Astraport these artificial reviews are not only "not as good" as real reviews, they are simply not reviews, at best they are fictions about imaginary products. Imho to avoid any confusion you should use fake product names as well, because otherwise you might get sued by the manufacturers if the model produces a obviously wrong negative review. $\endgroup$
    – Erwan
    Jun 16 at 22:55

Not the answer you're looking for? Browse other questions tagged or ask your own question.