Representation Learning of Social Survey Data
Project Description

Social scientists have accumulated rich survey datasets across all social domains. In most practices of social research using survey data, scholars often rely on a single dataset to test their hypotheses. This project aims to use graph convolutional networks (GCN) to learn representations of survey respondents as well as their answers. This will potentially help researchers to borrow information across surveys and

Supervisor
ZHANG Han
Quota
3
Course type
UROP1100
UROP2100
UROP3100
UROP4100
Applicant's Roles

- Preprocessing survey data
- Implement GCN algorithms to train the embedding
- Evaluate and apply the resulted embeddings

Applicant's Learning Objectives

- Use python for data preprocessing
- Understand the GCN algorithm
- Apply standard GCN algorithm to a new problem domain

Complexity of the project
Challenging