Customizing the RAG application further
There are various ways to improve RAG performance on unstructured data. In the next chapters, we’ll take a look at improving document parsing and ingestion of documents. Here are the main things to try while improving your RAG application:
- Improve parsing by experimenting with different parsers. We’ll look at this in Chapter 5.
- Improve parsing by experimenting with different chunking strategies; that is, the ways you split your documents into parts. We’ll look at this in Chapter 8.
Use different embedding techniques and customize your retrieval. We’ll look at this in the next chapter.
Query expansion
Query expansion is a technique of reformulating queries to get a larger number of relevant documents from a search engine. The key idea is simple – our generation is good only if we have found relevant context (in other words, ALL relevant documents in our corpus that are needed to answer...
