The IIRIS team support Informa marketing teams to promote 1.5k trade shows by collecting, enriching, and mining c2.5b customer interactions. In our journey to better understand our audience we have encountered several challenges that we tried to solve by using a fusion of traditional ML and state of the art Generative AI and Large Language Models (LLMs).

The biggest user cases are:

  1. Understanding audience interests: how we extract topics from content and derive the area of interests for users.
  2. Improving data quality: mapping client event registration information to a master reference data list of registration terms.

During this talk we will expand on the two techniques (Topic Modelling & Classification problem using LLM): what the data solution architecture and how we fine-tuned open source LLMs.

Technical level: Technical practitioner

Session length: 40 minutes