#community-help

Troubleshooting Semantic Search with Typesense

TLDR Koushik asked for help on semantic search with Typesense. Jason suggested ensuring spaces between field values and trying different models.

Powered by Struct AI
5
4mo
Solved
Join the chat
Aug 03, 2023 (4 months ago)
Koushik
Photo of md5-237fc2ec43c199f75556faf6aa5d6a0b
Koushik
03:41 PM
Can anyone suggest how typesense works while we would like to do Semantic search ? Reason behind this question is I would like to use use typesense for our movie search where I may keep a combination of Title, Description, Language, Genres, keywards etc as my vector creation String.
Creating a vector with Sentence transformers "all-mpnet-base-v2" and adding a vec key with the json and uploaded it to typesense.

While querying with different type of sentence it looks like I am not getting closest match. I thing I need to do something more. Do I need to train database or need to use different vector creation model ?
Any suggestion will be highly appreciated.
-Thanks
Jason
Photo of md5-8813087cccc512313602b6d9f9ece19f
Jason
04:39 PM
Could you make sure you have spaces between each of the field values when you concatenate them when you send them to the model?
04:39
Jason
04:39 PM
Also you might want to try these other models: https://huggingface.co/typesense/models/tree/main
Aug 04, 2023 (4 months ago)
Koushik
Photo of md5-237fc2ec43c199f75556faf6aa5d6a0b
Koushik
08:49 AM
Yes. I placed space between words and put comma between keywords.
Aug 06, 2023 (4 months ago)
Koushik
Photo of md5-237fc2ec43c199f75556faf6aa5d6a0b
Koushik
07:55 AM
Ok ... I am trying other models too. Thanks for the reference.