Q&A

One of the most common use-cases for an LLM application is to answer questions about a set of documents. LlamaIndex has rich support for many forms of question & answering.

Types of question answering use cases

Q&A has all sorts of sub-types, such as:

What to do

• Semantic search: finding data that matches not just your query terms, but your intent and the meaning behind your question. This is sometimes known as “top k” search.

  • Example of semantic search

• Summarization: condensing a large amount of data into a short summary relevant to your current question

  • Example of summarization

Where to search

• Over documents: LlamaIndex can pull in unstructured text, PDFs, Notion and Slack documents and more and index the data within them.

  • Example of search over documents

  • Building a multi-document agent over the LlamaIndex docs

• Over structured data: if your data already exists in a SQL database, as JSON or as any number of other structured formats, LlamaIndex can query the data in these sources.

  • Searching Pandas tables

  • Text to SQL

How to search

• Combine multiple sources: is some of your data in Slack, some in PDFs, some in unstructured text? LlamaIndex can combine queries across an arbitrary number of sources and combine them.

  • Example of combining multiple sources

• Route across multiple sources: given multiple data sources, your application can first pick the best source and then “route” the question to that source.

  • Example of routing across multiple sources

• Multi-document queries: some questions have partial answers in multiple data sources which need to be questioned separately before they can be combined

  • Example of multi-document queries

Further examples

For further examples of Q&A use cases, see our Q&A section in Putting it All Together.