About Catalog Semantic Encoder (CSE)

This is for:

System Administrator
Important

Contact your Coveo representative to enable Coveo Machine Learning (Coveo ML) Catalog Semantic Encoder (CSE) in your Coveo organization.

Coveo Machine Learning (Coveo ML) Catalog Semantic Encoder (CSE) enhances product discovery by understanding the meaning of queries. When used in a Coveo-powered commerce search interface, CSE uses vector search to retrieve products from your index based on their semantic similarity to a query. Unlike traditional keyword-based search, which relies heavily on exact matches and predefined synonyms, CSE leverages natural language processing (NLP) and similarity search to deliver more relevant results.

In commerce, product descriptions and attributes are often structured in ways that don’t align with how shoppers describe products in the queries they enter.

By using CSE in a Coveo-powered commerce search interface, you:

  • Improve relevance of search results: CSE enhances the ranking of results by interpreting user intent rather than relying solely on exact keyword matches.

  • Improve handling of complex queries: Verbose or vague queries are better understood with CSE, which can interpret the meaning of the query and retrieve relevant products.

  • Reduce manual tuning: CSE automatically learns the relationships between products and queries, reducing the need for manual tuning, such as creating synonyms or boosting rules.

How CSE works

CSE uses multilingual semantic encoders to create a vector representation of the textual information found in your catalog data. At query time, CSE converts the query into a vector in the same high-dimensional space as the product vectors. It then computes the similarity between the query vector and the product vectors to retrieve the most relevant products.

Here’s how the process works:

  1. When the model is trained, it uses the information contained in your catalog data to place the products into a high-dimensional vector space where similar products are close together, and dissimilar products are far apart.

  2. When a user enters a query, CSE transforms the query into a vector in the same high-dimensional space.

  3. The CSE model then computes the similarity between the query vector and the product vectors to retrieve the products that are closest to the query in the vector space. This allows CSE to retrieve products that are semantically similar to the query, even if they don’t contain the exact keywords.

  4. Finally, CSE works with Coveo ranking algorithms to optimize the ranking of the results based on both semantic and keyword relevance.

Note

CSE is designed to work seamlessly along Coveo’s AI ranking models, such as Automatic Relevance Tuning (ART) and Intent-Aware Product Ranking (IAPR).

Use case example

A visitor enters high-definition display in the search box. In your catalog data, you only have products that are described as 4K monitor. With traditional keyword-based search, the visitor wouldn’t necessary find the products they’re searching for because the query doesn’t match the keywords in the product’s metadata. However, with CSE, the model can recognize that high-definition display and 4K monitor are semantically similar, returning relevant 4K screen products.