#community-help

Boosting Document Rankings Using Search History

TLDR Jason inquired about boosting document rankings based on user search history. Jason explained it's achievable using the vector search feature of Typesense and suggested using a machine learning model before providing an API to demonstrate it.

12mo

Join the chat

Nov 29, 2022 (13 months ago)

Photo of md5-c7ebbbcac16a1763e16d0e8e8177b0ae

Jason

07:00 PM

Is it possible to boost (not filter) document rankings by the presence of certain document values? example: a user searches for restaurants in NYC. We know from their search history that they engage most with Indian and Korean food, the service should apply a ranking boost to any document which contains the words "matar paneer", "japchei, "lasooni gobi" etc... The absence of these terms from a document should not exclude it from results, only "boosting" those documents which contain those terms....

Is this achievable?

Photo of md5-8813087cccc512313602b6d9f9ece19f

Jason

07:11 PM

This is possible using the vector search feature in the latest RC build of Typesense. But to generate these vectors, you would have to build an ML model that takes into account each user’s past searches, clicks, etc and generates these vectors for you, store the vectors in Typesense and then you can then do a “nearest neighbor search”

07:12

Jason

07:12 PM

Here’s an example of how you can use word2vec to generate vectors in a similar scenario: https://towardsdatascience.com/session-based-recommender-systems-with-word2vec-666afb775509

Dec 29, 2022 (12 months ago)

Photo of md5-972da58c82de3b38862220702e852eda

Viktor

02:31 PM

What would the search query look like that achieves this? Assuming vectors are in place.

Jason

04:19 PM

It would look like this: https://github.com/typesense/typesense-website/blob/v0.24.0/docs-site/content/0.24.0/api/vector-search.md#nearest-neighbor-vector-search

Typesense

Lightning-fast, open source search engine for everyone | Knowledge Base powered by Struct.AI

Indexed 3015 threads (79% resolved)

Join Our Community

Similar Threads

Integrating Semantic Search with Typesense

Krish wants to integrate a semantic search functionality with typesense but struggles with the limitations. Kishore Nallan provides resources, clarifications and workarounds to the raised issues.

11mo

Optimizing Dataset of Podcast Feeds for a Searchable Database

Alexander seeks advice on optimizing a podcast database for search. Kishore Nallan suggests data size and stopwords impact RAM usage, and that benchmarking on 1M records would be useful. satish raises the potential need for vector searching. Both recommend feeding user activity data into ML models for relevancy ranking. Collaboration was suggested.

20mo

Typesense Query for Related Documents in Ecommerce

Vlassis asked about finding related documents using a Typesense query. Jason suggested using vector search and recommended ML libraries for vector generation, with 'word2vec' for text content. Vlassis agreed to try it out.

14mo