Despite recent progress, large language models (LLMs) still face the challenge of appropriately reacting to the intricacies of social and cultural conventions. Several works have focused on augmenting them with cultural commonsense knowledge (CCSK) [1,2,3], yet, have uniformly divided cultures along national, linguistic or ethnic boundaries. Yet important cultural differences manifest themselves also within generations of the same society (e.g., teenagers, young adults, retirees, or baby boomers/Gen Z/ …). The goal of this thesis is to investigate CCSK compilation and exploitation for problems that require intergenerational knowledge.

References

[1] Tuan-Phong Nguyen, Simon Razniewski, Aparna Varde, and Gerhard Weikum. 2023. Extracting cultural commonsense knowledge at scale. In Proceedings of the Web Conference 2023, WWW
[2] Yi Fung, Ruining Zhao, Jae Doo, Chenkai Sun, and Heng Ji. 2024. Massively multi-cultural knowledge acquisition & LM benchmarking. CoRR
[3] Awantee Deshpande, Dana Ruiter, Marius Mosbach, and Dietrich Klakow. 2022. StereoKG: Data-driven knowledge graph construction for cultural knowledge and stereotypes. In Proceedings of the Sixth Workshop on Online Abuse and Harms (WOAH)