Similarity Search - Setup


Category: Artificial Intelligence


Learn to activate Similarity Search in Mindbreeze to deliver smarter, context-aware results and enhance your search capabilities with natural language understanding.

Introduction to Similarity Search

To start, let's have a look at an example for Similarity Search by asking questions about products from the Shure website. We'll be using the SM58, Shure's flagship microphone, for a lot of our questions going forward.

For example, when entering the search term, "Why should you choose the SM58 over the SM57?", we get specific answers. These answers are found using Similarity Search.

Steps to Activate Similarity Search

1. Open the Index

Next, we will activate Similarity Search. We start by opening the index of your choice. In our case, we select the Shure index with advanced settings.

2. Configure Semantic Text Extraction

In the index settings, scroll down to the section called Semantic Text Extraction. Here, check the box for language detection to enable the language detector. Additionally, select the option to use only pre-built models for named entity recognition.

3. Enable Sentence Transformation

Scroll down further to the section for Sentence Transformation. Once we enable the checkbox for this feature, Similarity Search will be activated automatically.

Save and Restart

After activating Similarity Search, scroll to the top of the menu and save the settings. Please note that this requires re-inverting the data, so the restart may take some time.

Using Similarity Search

After the restart is complete, we can start using Similarity Search directly within our Client and all other Insight Apps.

For example, when searching for "What is the SM58?", the answers appear immediately.