0

I am looking into extracting the meaning of expressions used in everyday speaking. For an instance, it is apparent to a human that the sentence The meal we had at restaurant A tasted like food at my granny's. means that the food was tasty.

How can I extract this meaning using a tool or a technique?

The method I've found so far is to first extract phrases using Stanford CoreNLP POS tagging, and use a Word Sense Induction tool to derive the meaning of the phrase. However, as WSI tools are used to get the meaning of words when they have multiple meanings, I am not sure if it would be the best tool to use.

What would be the best method to extract the meanings? Or is there any tool that can both identify phrases and extract their meanings?

Any help is much appreciated. Thanks in advance.

  • 3
    I dispute your premise. If said with a certain tone of voice, that sentence would mean that the food was tasteless and overcooked. The level of meaning that you are looking to extract does not reside in the words. – Colin Fine Mar 8 '18 at 11:09
  • 1
    Unless you own Google you probably don't have the computing power to extract the meaning from random English sentences. – curiousdannii Mar 8 '18 at 13:04
  • 1
    To do this generally would take full AI. If you can live with a solution for a specific niche like restaurant reviews then you could train a classifier with a few niche-specific labels (eg tasty...). – Adam Bittlingmayer Mar 8 '18 at 17:08
2

This is very difficult. I'll recommend three things:

  1. Use the U of I CogComp shallow parser to get phrases (not CoreNLP), see: http://nlp.cogcomp.org/ It's much better at picking up phrases, IMO.

  2. If you google around, you'll find several pre-built list of phrases (idioms, fixed expressions, etc.); use the ones that meet your needs for example, https://github.com/WithEnglishWeCan/generated-english-idioms/blob/master/idioms.build.json

  3. Do old fashion RegEx matching between the list of phrases and your input sentence. Later, you can look at fancier machine learning classifiers. For example, swap out their "sarcasm" data with your colloquial expressions: https://github.com/surajr/SarcasmDetection

Also, there's a couple organizations that specialize in this task. It's worth checking out places like: http://afflatus.ucd.ie/

Best of luck!

| improve this answer | |
  • Thank you very much! The resources you have directed to seem rather helpful. – Fleur Mar 9 '18 at 3:58

Your Answer

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

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