1 min read

Vector search and explainability

The why and what behind semantic models

Empathy Platform’s vector search uses approximate nearest neighbor (AAN) algorithms to create associations between search terms and product data, to understand shopper intent and semantic similarities, using that understanding to empower traditional keyword search.


Learn more about Vector Search in Empathy Platform and how it works on the Why vector search page.

As Empathy Platform is built on a cloud-native stack, the semantic models used to calculate semantic similarities are trained within each customer’s specific domain, ensuring privacy.

Vector search & explainability

How semantic models work and calculate vector similarities are usually difficult to explain and understand, but Empathy's Play Vector gives you a visual explanation of the similarities between queries and product-related information in the form of vector models. Since semantic models can return vector points to be represented in a high-dimensional space, you can locate a specific query and then visually navigate and explore the related terms.


Want to understand how Empathy Platform trains semantic models? Check the Semantic models in Empathy Platform.

Play Vector closes the explainability gap by offering visual insight into the what and why behind semantic models. It ensures that you understand shoppers' intent and have a clear view of what shoppers are looking for. Then, you can make informed decisions and take action to adapt the search and discovery experience accordingly, based on your brand and business needs.