Structured Analysis of Personal Housing Stories Using AI
Name
Mohanty-shresmoh-SM-MAS-2026-thesis.pdf
Size
1.86 MB
Format
Adobe PDF
Checksum (MD5)
d9b73f9091d0103f5c92889e2f100e0e
Author(s)
Mohanty, Shrestha
Advisor(s)
Roy, Deb
Date Issued
February 2026
Publisher
Massachusetts Institute of Technology
Abstract
Housing is an issue that touches many aspects of daily life, influencing personal stability, social connections, and economic well-being. In conversations about community challenges, participants sometimes share personal housing stories, narratives that illustrate individual experiences with housing, interactions with local institutions, and the obstacles they face. Such stories can provide insights that are difficult to capture through surveys or quantitative data alone, though their length, unstructured form, and emotional complexity can make systematic analysis challenging. This study presents an LLM-based framework for extracting structured components from personal housing stories to support systematic analysis while preserving individual perspectives. We developed a narrative analysis framework that identifies key story elements including characters and their roles, housing contexts, challenges and desires, emotional tones, and relationship dynamics. Using large language model-guided prompting, we processed housing stories from community conversations, while maintaining human collaboration to help ensure accuracy and build trust in the analytical process. We evaluated the framework through an initial within-subject experiment with five participants who compared structured analyses against original stories. Results suggest that the framework generally improved story comprehension and organization, with participants often finding it useful for research and reference purposes. Participants found the framework particularly helpful for complex, policy-heavy stories that required additional context, while noting that manual analysis would be challenging at scale without LLM assistance. The framework also appeared useful for highlighting recurring themes and challenges across housing narratives. At the same time, participants noted trade-offs between clarity and emotional engagement, emphasizing the importance of balancing systematic structure with the richness of personal accounts. This study offers an initial exploration of structured narrative analysis in housing research and points toward the potential benefits of combining computational tools with human oversight.
MIT Department
Program in Media Arts and Sciences (Massachusetts Institute of Technology)
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
