IndoCulture: Exploring Geographically-Influenced Cultural Commonsense Reasoning Across Eleven Indonesian Provinces
It addresses cultural bias in AI for diverse Indonesian populations, but is incremental as it builds on prior work by using manual data creation instead of templates or scraping.
The paper tackles the problem of Anglocentric bias in commonsense reasoning by introducing IndoCulture, a dataset exploring geographically-influenced cultural commonsense across eleven Indonesian provinces, and finds that open-weight Llama-3 is competitive with GPT-4 while other models struggle with accuracies below 50%, and location context enhances performance for larger models.
Although commonsense reasoning is greatly shaped by cultural and geographical factors, previous studies have predominantly centered on cultures grounded in the English language, potentially resulting in an Anglocentric bias. In this paper, we introduce IndoCulture, aimed at understanding the influence of geographical factors on language model reasoning ability, with a specific emphasis on the diverse cultures found within eleven Indonesian provinces. In contrast to prior work that has relied on templates (Yin et al., 2022) and online scrapping (Fung et al., 2024), we create IndoCulture by asking local people to manually develop a cultural context and plausible options, across a set of predefined topics. Evaluation of 27 language models reveals several insights: (1) the open-weight Llama-3 is competitive with GPT-4, while other open-weight models struggle, with accuracies below 50%; (2) there is a general pattern of models generally performing better for some provinces, such as Bali and West Java, and less well for others; and (3) the inclusion of location context enhances performance, especially for larger models like GPT-4, emphasizing the significance of geographical context in commonsense reasoning.