Does Algolia have semantic search capability?

I am working on a social media type app and I need semantic search implemented. I see there are a couple of blog posts on Algolia website about semantic search but I can’t find whether Algolia has that capability or not.

Essentially I like to search the index given a phrase and get relevant set of posts back. Not looking for textual comparison.

Thank you!

Hi! You’re catching us at a bad time. We had a semantic search capability called “Answers” we were testing based on open source models, but it wasn’t living up to expectations.

We recently acquired search.io which has strong semantic capabilities via their Neuralsearch technology, but it won’t be available via Algolia until Q1.

Feel free to reach out to me via DM (or at chuck.meyer@gmail.com is you can’t DM yet) and I can get you in touch with the appropriate parties if you want to help test the new products/features.

Hi Chuck,

Hope all is well and you enjoyed your time off.
Just circling back to see if semantic search capability is close to be released.

Additionally, I am very interested in media related products Algolia has to offer.

Can you please share any content or information related to implementation for the following:

Thank you in advance

Hey @atefyamin

The semantic plans are still in progress. Our intention is to have search.io’s Neuralsearch capabilities integrated into our APIs this Spring. Tagging @dustin in case he has more to add.

For building media experiences, there is already a ton of goodness out there to explore. What are you looking to build?

I would start by looking at some of our pre-built media integrations for Netlify, Contentful, and Storyblok.

There’s also the general use Algolia Crawler you can purchase as an add on to crawl any media content.

On the AI/ML side, our Recommendation’s service includes a Related Content model that provides content-based filtering.

For video content, my colleague @michael.king just posted a blog on a video search tool we’ve been working on called A/V Search. Check it out here.

Finally, there are some UI examples in our inspiration library that may give you some insights on improving media search.

Hopefully this gives you some ways to get started while we dig further into the semantic side of things.

Hi @chuck.meyer , spring is coming :slight_smile: We, at Aura Health, are very interested in the semantic search on our Algolia Indices. Do we have an estimate date of release? any beta we could participate in?

1 Like

Hey @jerome

Thanks for following up! Our semantic offering, NeuralSearch is in private beta with its first customers as we speak!

It’s currently a manual setup process, so we can only bring a few customers into the program, but we are hoping to open it up wider very soon!

In the meantime, you can make sure Aura Health is ready! NeuralSearch’s semantic understanding is based on training a model on search and (most importantly) click data for your existing search application.

To be a candidate for the beta, make sure your search application is capturing click data for search results. If you’re using InstantSearch, this means wiring in our Insights middleware, or importing the click events from an external tool like Google Tag Manager or Segment:

You can also sign up to attend DevBit on 3/29 — the first in a series of quarterly Algolia events talking about new features and best practices.

Consider integrating NLP libraries like spaCy or using pre trained languages models like BERT to extract semantic information from user queries and improve the relevance of search results. You can process the user query using these models and then pass the processed query to Algolia for searching within your indexed data. This way, you can combine the power of algolia fast and flexible search infrastructure with semantic understanding to deliver more accurate and meaningful search results.