Building an Automated Causal Inference Tool for IR Questions

Session Abstract: As IR professionals, we often receive questions like whether a specific program can improve students’ GPA and if a certain policy may increase matriculation rates. Answering such questions usually requires conducting causal inference analysis. In this session, we will present a web-based tool we have built that can automatically answer all similar questions by simply uploading relevant data onto the tool. We will demonstrate how the tool works with two typical cases in IR. Statistical models behind the tool and IT resources to build such tools will also be introduced. The session will help audience gain better understanding of causal inference in IR and how to build automated tools for a certain type of analysis. Note: this session is NOT about generative AI.

Keywords: Causal Inference, Automated Tool, IR Analytics, Self-service