I have a big database that contains two table CLIENT_DESCRIPTION and REPORTER_DESCRIPTIPN. This database is for a study on how people answer a specific question How do you describe your experience with our product in your country?.

For every client in this questionnaire, a client will provide us with a paragraph. And a reporter listening to the client should provide us a paragraph containing no more than 3 sentences that shorts the client opinion. The reporter will write the description as he understand it from the client.

For example:
Client description: Your product is very good compared to other similar products. I can found it easily in the store, and the price is accepted. I think that the cover color is very bright. I guess you should use a dark color instead. The product broke with me once, is it possible to fix it?
Reporter description: The client lives in LA. The sales man should contact him to promote our premium plan. He want to have a dark copy of the product cover. he has a broken one and needs to fix it.

I have thousands of similar entries in my two table, and for 70% of them, every client entry is matched with it's reporter entry.

Is it possible to train the dataset to match the remaining entries?

70% of the entries are matched(Client descriptions matched with their equivalent reporters descriptions).
We have 30% of client descriptions entries from CLIENT_DESCRIPTION that for some reason are not matched with their equivalent reporter description from REPORTER_DESCRIPTIPN.

What algorithm or library I can use to train the already matched entries to match the unmatched ones.

Your Answer

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

Browse other questions tagged or ask your own question.