Back to datasets
Dataset assetOpen Source CommunityPublic Opinion SurveyPolitical Elections
YoronChizu 2024 Open Data
YoronChizu 2024 Open Data is a public service released on November 18, 2024 by the Japanese non‑profit organization Mielka. The dataset contains anonymous voting data collected during the 2024 Japanese House of Representatives election, where users voted on specific questions extracted from party manifestos. It is used to visualize opinion distribution and includes votes from 4,403 independent participants.
Source
github
Created
Nov 22, 2024
Updated
Nov 23, 2024
Signals
109 views
Availability
Linked source ready
Overview
Dataset description and usage context
YoronChizu 2024 Open Data
Dataset Overview
- Release Date: November 18, 2024
- Publishing Organization: Mielka, Japanese non‑profit organization
- Data Source: User voting data during the 2024 Japanese House of Representatives election
- Data Types: CSV files and images
- Data Volume: 4,403 independent users participated
- License: CC BY 4.0
Dataset Content
- Voting Data: Anonymous votes on specific issues extracted from party manifestos.
- Image Data: Visualizations of user opinion distribution.
Technical Details
- Service Tool: Polis used as the server‑side deliberation tool.
- Frontend Development: Re‑engineered from scratch for the Japanese context, optimized for mobile access.
- Features:
- Visualize party positions using party icons
- AI‑driven opinion cluster explanations
- Real‑time map updates of user opinion locations
Data Analysis
- Unrepresented Opinion Groups: Certain economic policy opinion groups are not fully represented by existing political parties.
- Vote‑Cluster Relationship: Experiments show that even with 4,400 votes, the four clusters remain clearly distinct.
- Digital Democracy Opinions: Public opinion on digital democracy is divided into four teams: Red (advertising regulation & AI scepticism), Blue (digital integration & efficiency advocacy), Yellow (AI utilization & personalization), Green (privacy protection advocacy).
Future Challenges
- Opinion Extraction: Extracting unbiased opinions from party manifestos requires manual effort; LLM hallucinations pose risks.
- Data Representation: LLMs exhibit errors when explaining complex opinion distributions, necessitating further optimization of data representation.
- User Opinion Submission: Real‑time submission faces quality assurance, misinformation, and dimensionality‑increase challenges.
Need downstream help?
Pair the dataset with AI analysis and content workflows.
Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.