An Image-based Global Climate Sentiment Index for 19 Countries, 2015-2024

A Climate Sentiment Index constructed from approximately 3.19 million climate-related images collected from 19 countries between 2015 and 2024. The dataset, created by Xiangyang Zhai, provides globally comparable positive and negative sentiment indices at daily, weekly, monthly, and annual frequencies. It is designed to transcend linguistic boundaries for improved cross-cultural sentiment analysis.

Use Cases

• Tracking public climate attitude trends over time based on the multi-frequency sentiment indices.
• Analyzing cross-country comparability of climate sentiment based on the visual, language-independent methodology.
• Modeling market expectation formation and risk transmission channels based on the long-horizon sentiment index.
• Evaluating the impact of climate policies or events based on shifts in the positive and negative sentiment indices.

Strengths

• Contains approximately 3.19 million data points (images) collected over a 10-year period (2015-2024).
• Provides sentiment indices at four temporal resolutions (daily, weekly, monthly, annual) for granular and aggregate analysis.
• Covers 19 representative countries, enabling cross-national comparison.
• Uses a visual-based methodology intended to improve cross-cultural robustness compared to text-based approaches.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The specific countries covered are not listed, which may limit geographic suitability assessment.
• Row count for the final tabular indices is unknown.

Provenance

Source

figshare

Collection Method

Images collected via Google Images search using defined keywords, with sentiment detected via computer vision models.

Time Range

2015-2024

Freshness

Last updated 2026-04-24 16:13:34; freshness should be verified.

Geography

19 representative countries (global)