LLooM Examples
LLooM can assist with a range of data analysis goals—from preliminary exploratory analysis to theory-driven confirmatory analysis. Analysts can review LLooM concepts to interpret emergent trends in the data, but they can also author concepts to actively seek out certain phenomena in the data.
Check out the examples in this section to see what LLooM can do, or visit the What is LLooM? page to learn more about the motivations and methods behind LLooM.
Political Social Media
Example Inputs
Example
Outputs
Government Critique
Criteria: Does this text criticize government actions or policies?
Summary: Critique of government actions, policies, and officials, advocating for accountability, transparency, and reform.
Trust in Institutions
Criteria: Does this text address trust or distrust in social or governmental institutions?
Summary: Emphasizing trust in institutions through healthcare access, equality, disaster preparedness, combat readiness, and justice initiatives.
Social and Economic Inequality
Criteria: Does this text discuss social or economic disparities?
Summary: Advocating for social justice, economic equality, healthcare access, and accountability in government and society.
Policy and Healthcare Concerns
Criteria: Does this text express concerns about healthcare policies or costs?
Summary: Advocating for healthcare access, protecting abortion rights, lowering drug prices, and investigating federal agency corruption.
Analysis
➡️ Try out analyzing this data with LLooM on this Colab notebook.
Task: Investigate how social media amplifies political polarization
Political polarization is a dominant concern in the United States, and it poses potential existential risks to democracy. If social media algorithms play a role in amplifying partisan animosity, how might we redesign social media algorithms to mitigate this effect? We can use LLooM to investigate political social media posts to explore whether we can detect and downrank content that amplifies partisan animosity.
Dataset: Political social media posts
We use a dataset of public Facebook posts from Jia et al.'s CSCW paper on Embedding Democratic Values into Social Media AIs via Societal Objective Functions. This dataset was generated by filtering for political posts on CrowdTangle using politics-related page categories such as “politics,” “politician,” “political organization,” and “political party.” The dataset consists of 405 posts that were randomly sampled and manually annotated for partisan animosity.