Chatbot Capstone Help

Hello,

I’ve attempted to build a retrieval chatbot for the capstone however i’ve run into some issues with my code:

I chose a closed-domain architecture to keep the responses true to the source and not allow for deviation from the source.

Use case(s) for chatbot:

Get random aphorisms on philosophy from schools of thought

I had difficulty implementing the bag of words for intent selection, when i added the functions it took too long to return a response and crashed multiple times due to insufficient memory. The word2vec model was changed from spacy.loan(‘en’) to spacy.load(“en_core_web_md”). Lastly, the selection of the response is random and not using the intent and entity similarity function.

I was thinking that I could pregenerate the BOW matrix for the corpus before and save the result in a dictionary for future use.

Any tips or reccomendations to complete this?

Here are a few tips which i learn after building a chatbot for myself:

  1. Simplify your bag of words approach to avoid crashes.
  2. Upgrade to a better word2vec model like spacy.load("en_core_web_md").
  3. Consider matching user queries with relevant aphorisms for more tailored responses.
  4. Precompute your BOW matrix for efficiency.
  5. Keep testing and refining your chatbot. Good luck!