If you are interested, we can discuss a topic of interest to be done in collaboration with AIKO Space
Motivation
Standard information retrieval systems optimize purely for relevance, yet consistently expose certain groups to disparate visibility, a form of algorithmic bias that has real-world consequences in job searches, product discovery, and beyond. The fair-reranker aims to integrate fairness constraints directly into a dense reranking pipeline, processing datasets such as BIOS (occupational bias), ESCI (e-commerce search), and TREC Fair Ranking.
Task
Develop an intersectional fair reranking DL model
Motivation
Consistent and reproducible evaluation enables research advancements. While CrisisFACTs presents a unique challenge in the current retrieval landscape, its evaluation needs to be reconsidered from the ground up to support new research efforts. The objective of the task is to select relevant texts for crisis operators and then create concise summaries. Currently, the evaluation relies on humans, which is impractical for testing any new model.
Task
The thesis aims to develop a new evaluation system and replicate previous solutions to establish a new working benchmark.
References
CrisisFACTS: Building and Evaluating Crisis Timelines
Motivation
In search engines and RAG systems, it is important to find both supporting and non-supporting documents when given a complex query (e.g., social, politics, science, debates), but currently there is a strong bias towards ranking in higher positions only documents that directly support an argument.
Task
We aim to create a first-pass retriever for supporting and contradicting sentences given a query
References