Making Council Questions Visible
AI Legislative Analysis
Classified and visualized general-question materials with AI assistance to examine the public issues discussed by the council.
Data analysis and visualization
- Original council-question materials
- Organization and topic classification
- AI-assisted classification and human review
- Visualization for the team and public coverage
The public material does not fully specify category definitions, multi-label rules or treatment of ambiguous topics, so this page does not invent those methods.Read the named report ↗
Problem
Council materials are scattered across livestreams, recordings and transcripts. Understanding what different councilors discuss requires substantial manual organization.
Context
The analysis covered 37 Kaohsiung councilors who had completed general questioning by 14 May 2025. This is a defined observation period, not the full session.
My Role
I collected and organized question materials, used AI to assist topic classification, and produced visualizations. The team used and shared the work; its policy outcomes are not solely my output.
Constraints
Transcript timing, overlapping topics and context lost in summaries limit the questions this dataset can answer.
Approach
I organized summaries, transcripts and topic classifications for comparison, then visualized the distribution of issues.
System
Source collection → summary organization → AI-assisted classification → visualization → team reading and public sharing.
Outcome
A named Liberty Times report attributes the analysis and visualization to Hong Lin and records the scope as 37 councilors and 17 issue categories.
Limits
Topic frequency does not measure a councilor’s performance or residents’ needs. The underlying classifications are not published here, so this site cannot independently recalculate the reported percentages.
Evidence & Sources
SOURCE 01Liberty Times / named report, 15 May 2025 ↗Source markers describe cited material; this site has not rechecked every external system.