Semantic Search Fundamentals

Vectara lets you build a semantic, LLM-powered search application. Semantic search is not just about finding data, but about understanding data and helping you answer questions about your data. This topic outlines what Vectara can do for this use case as well as why and how to employ these features for the best overall end-user experience. By integrating advanced features such as metadata filtering, reranking, and Retrieval Augmented Generation, Vectara not only simplifies the search process but also enriches the quality of information retrieved.

Large Language Models (LLMs)

LLMs are deep neural nets that are built with the task of specifically understanding human language. These models can be a great asset to many different use cases, including search and language generation.

These models generally work by reading immense amounts of text to build the model and then using that model to convert text into vectors, both at index and at query time. For many use cases, this obviates the need for many language rules of traditional keyword systems like synonym management, stemming, and phrase parsing because the LLM can inherently understand what the user is asking.

The team behind Vectara has built LLMs that work across a wide variety of languages and verticals. When you index data into Vectara or perform a search, also known as retrieval, the text is converted to one or more vectors via a LLM and then used to answer questions that your users have.

Zero-shot models

Zero-shot models have an excellent understanding of language in general. They can understand and respond to the semantic meaning of questions without any additional tuning. This obviates much of the need for fine-tuning and specialized training on a particular dataset or in a particular vertical.

The Vectara platform makes extensive use of zero-shot models that have been developed by the team to allow your end users to query using the language and verbiage of their choosing and find the right documents, regardless of the domain your documents are in.

Hybrid search

While zero-shot LLMs work very well in the vast majority of search use cases, there are some occasions where they struggle. In particular, many zero-shot LLMs don't work as well when users perform queries for things which have little semantic meaning.

For example, a UPC code, barcode number, or particular named configuration setting has little to no semantic meaning, and if you expect your users to perform this type of search, it's best to look into our hybrid search documentation to learn about how to blend neural search and keyword search. The ability to toggle between neural and keyword search methodologies enhances the effectiveness of search results in these use cases.

Advanced query configurations

Application builders can define specific query parameters for their searches, including context, pagination, metadata filters, and semantics. This flexibility empowers users to tailog queries to specific use cases, ensuring that the search results are as relevant and precise as possible.

Query request and response

Developers can specify the query text and manage pagination through the offset and limit parameters. This structured approach helps in managing the flow of search results effectively.

Metadata filtering

Vectara supports enhanced metadata filtering, which allows users to restrict
searches to specific parts of the corpus based on defined criteria, using common SQL syntax.

Reranking

Vectara enhances the relevance of search results through its reranking configurations. The reranker object has different types that can be used to adjust the relevance of search results based on specific needs, such as diversity or precision.

Consider a scenario where a user queries about the latest advancements in medical research. You can configure advanced query settings to pull relevant documents from specified corpora, apply metadata filters to focus on recent publications, and then use RAG to generate a concise, informative summary that directly answers the user’s query.