A Large Language Model Technique to Summarize Interaction Histories

What is it about?

We describe a pipeline for interaction log summarization using large language models (LLM). The technique essentially converts analysis interactions into sentences. Adding key topics to the sentences allows LLMs to generate reports automatically.

Featured Image

Photo by National Cancer Institute on Unsplash

Photo by National Cancer Institute on Unsplash

Why is it important?

We show how we can summarize interactions by repeatedly prompting an LLM to summarize segments of the interaction history. This is an automatic technique that shows that LLMs can be used to write factual summaries of interactions automatically, making reporting easier.