Leveraging Causal Inference in Text and Image Analysis to Provide Interpretable Digital Content Personalization
Ganesh Iyer, Professor
Business, Haas School
Closed. This professor is continuing with Fall 2023 apprentices on this project; no new apprentices needed for Spring 2024.
With the widespread consumption of online digital content, growing competition among media platforms for people’s attention leads to the demand for innovation in content curation.
To choose which content to popularize, it is common practice in big social media firms (e.g., Facebook, Google, Netflix) to create multiple versions of content and do large-scale A/B tests, resulting in high-volumes of experimental data where treatments are unstructured (e.g., text, image).
How to do meta-analysis of those experiments and utilize the results in the downstream application (e.g., personalized content recommendation) remains a challenge. Our work proposes a new methodology to tackle the challenge.
Qualifications: The research apprentice would involve in various stages of the project potentially: 1) literature review; 2) coding the algorithm; 3) testing the algorithm via simulation; 4) building Python package; 5) real data analysis using the method.
The ideal candidate has intrinsic interest in developing methods to real-world business problem, has strong coding background in Python, has some basic understanding about Bayesian statistics and NLP.
Hours: to be negotiated
Social Sciences