SUMMARY OF IDEA
This idea uses the existing Australian Ad Observatory dataset and the WAIST(why am I seeing this?) data collected alongside statistical data associated with postcodes to identify discriminatory patterns such as proxy and price discrimination.
Another layer will be built on top of the existing visualisation tool to represent the data using Bayesian Networks engineered from a combination of ML (Machine Learning) and Expert Knowledge which then:
- Allows researchers to quickly identify discrimination (and other poor advertising practices) through the process of causal inference and counterfactual assessment, and
- Provides the same tools through a public dashboard that empower the public to search this data and expose bad behaviours themselves. This approach can be expanded to other sources of data beyond the Australian Ad Observatory dataset.
Download a PDF version of the Presentation: Identifying Discriminatory Patterns in Online Advertising Data
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Credit: Dr Kelly Lewis, Grant Nicholas, Ross Pearson, Alec Sathiyamoorthy, Vikram Sondergaard, Mingqiu Wang, Guangnan (Rio) Zhu, Dr Abdul Obeid and Xue Ying (Jane) Tan.
Suggested Citation: Lewis, K., Nicholas, G., Pearson, R., Sathiyamoorthy, A., Sondergaard, V., Wang, M., Zhu, G., Obeid, A. and Tan, X-Y. (2022). Tech for Good: ADM+S Dark Ads Hackathon. Using Postcodes to Identify Discriminatory Patterns in Online Advertising Data. Australian Research Council Centre for Automated Decision-Making and Society.