Integrating Cultural Knowledge into Artificially Intelligent Systems: Human Experiments and Computational Implementations

Authors

Anurag Acharya, Florida International UniversityFollow

Document Type

Dissertation

Degree

Doctor of Philosophy (PhD)

Major/Program

Computer Science

First Advisor's Name

Mark A. Finlayson

First Advisor's Committee Title

Committee Chair

Second Advisor's Name

Monique Ross

Second Advisor's Committee Title

Committee Member

Third Advisor's Name

Fahad Saeed

Third Advisor's Committee Title

Committee Member

Fourth Advisor's Name

Leonardo Bobadilla

Fourth Advisor's Committee Title

Committee Member

Fifth Advisor's Name

Ronald Fisher

Fifth Advisor's Committee Title

Committee Member

Keywords

artificial intelligence, cognitive science, natural language processing, culture, ritual, motifs, computational linguistics

Date of Defense

5-31-2022

Abstract

With the advancement of Artificial Intelligence, it seems as if every aspect of our lives is impacted by AI in one way or the other. As AI is used for everything from driving vehicles to criminal justice, it becomes crucial that it overcome any biases that might hinder its fair application. We are constantly trying to make AI be more like humans. But most AI systems so far fail to address one of the main aspects of humanity: our culture and the differences between cultures. We cannot truly consider AI to have understood human reasoning without understanding culture. So it is important for cultural information to be embedded into AI systems in some way, as well as for the AI systems to understand the differences across these cultures.

The main way I have chosen to do this are using two cultural markers: motifs and rituals. This is because they are both so inherently part of any culture. Motifs are things that are repeated often and are grounded in well-known stories, and tend to be very specific to individual cultures. Rituals are something that are part of every culture in some way, and while there are some that are constant across all cultures, some are very specific to individual ones. This makes them great to compare and to contrast.

The first two parts of this dissertation talk about a couple of cognitive psychology studies I conducted. The first is to see how people understood motifs. Is is true that in-culture people identify motifs better than out-culture people? We see that my study shows this to indeed be the case. The second study attempts to test if motifs are recognizable in texts, regardless of whether or not people might understand their meaning. Our results confirm our hypothesis that motifs are recognizable.

The third part of my work discusses the survey and data collection effort around rituals. I collected data about rituals from people from various national groups, and observed the differences in their responses. The main results from this was twofold: first, that cultural differences across groups are quantifiable, and that they are prevalent and observable with proper effort; and second, to collect and curate a substantial culturally sensitive dataset that can have a wide variety of use across various AI systems.

The fourth part of the dissertation focuses on a system I built, called the motif association miner, which provides information about motifs present in input text, like associations, sources of motifs, connotations, etc. This information will be highly useful as this will enable future systems to use my output as input for their systems, and have a better understanding of motifs, especially as this shows an approach of bringing out meaning of motifs specific to certain culture to wider usage.

As the final contribution, this thesis details my efforts to use the curated ritual data to improve existing Question Answering system, and show that this method helps systems perform better in situations which vary by culture. This data and approach, which will be made publicly available, will enable others in the field to take advantage of the information contained within to try and combat some bias in their systems.

Identifier

FIDC010715

ORCID

0000-0002-3883-5287

Previously Published In

Anurag Acharya, Kartik Talamadupula, and Mark A. Finlayson. Toward an atlas of cultural commonsense for machine reasoning. In Proceedings of the Workshop on Common Sense Knowledge Graphs (CSKGs), Online, February 2021. URL https://usc-isi-i2.github.io/AAAI21workshop/ papers/Acharya_CSKGsAAAI-21.pdf. Held in conjunction with the 35th AAAI Conference on Artificial Intelligence (AAAI 2021).

Recommended Citation

Acharya, Anurag, "Integrating Cultural Knowledge into Artificially Intelligent Systems: Human Experiments and Computational Implementations" (2022). FIU Electronic Theses and Dissertations. 5083.
https://digitalcommons.fiu.edu/etd/5083