ROSE

It is often the case that we use specific language when referring to an entity for the first time, but when it is already established in context we use less specific language. For example, you might first refer to an injury on your calf, and later say that you put a bandage on your leg, referring to the same injury (see the rational speech act framework: https://michael-franke.github.io/probLang/chapters/01-introduction.html). As we use LLMs to collect information about the state of the world, the LLM will likely follow the same behavior as it is modeled off of human language. Therefore, we would like to use some structured knowledge graph to understand when the same entity is being referred to using different language. We apply an ontology to this, with our use case being in medical treatment.

This project is in progress. I am the co-PI on this project with one other person. This project was an internally funded research project that I pitched and won a grant for.