We are pleased to share the recording of our webinar we held on April 4, 2024 on RAG question answering of Enron emails.
Use case: RAG on the Enron e-mail corpus
During this product update webinar we showed the perspective of an end-user using the Enron e-mail corpus as an example:
- Ask a question or select from pre-configured questions (convenient for support or onboarding)
- A summary of the generated answer with links to the right chunks of information
- Access the original source and view the relevant chunk immediately in context, a huge time-saver for everyone who searches for information in documents
- Select basic options: search method, vectorizer and Large Language Model (LLM)
- Filter on metadata to obtain better results
Behind the scenes: RAG
Afterwards we have a quick look behind the scenes using different projects on news items and patents:
- Easy and quick question-answering project creation
- Customize components with dedicated pipelines for conversion (speech, video or information contained in spreadsheets such as audit reports can be used), chunking and vectorizers
- Create metadata contained in the source or through purpose-build models for classification or named entities
- Modify and enrich prompts, add and compare LLMs
If you want the presentation in PDF format you can find it here:
See also: RAG for forensics (kairntech.com)