Tool-Box for Climate Change Research

In today's digital age, images play a pivotal role in shaping our understanding of complex phenomena like climate change. To empower the social science communities in their quest for deeper insights, we're thrilled to introduce the Image Analysis Toolbox — a cutting-edge resource designed to contribute to the study of climate change through the lens of imagery. This tool-box will provide open-access to researchers, including:

  • Automated Image Analysis
  • Image Clustering
  • Detection of Manipulated Images
  • Automated Text Analysis

    • By establishing image analysis as a computational social science method, our toolbox allows researchers to unlock new dimensions of understanding in the field of climate change. Stay tuned for the official release of the Image Analysis Toolbox, and together, let’s embark on a transformative exploration of climate change imagery.


    Interactive Image Analysis Tool

    This Streamlit application allows users to explore the capabilities of OpenAI's CLIP model for interactive image analysis. The tool enables users to upload images and receive predictions about their content, bridging the gap between visual and textual understanding.

    Key Features:

    • Image Upload Interface: Easily upload images directly through the user-friendly interface.
    • Visual Content Analysis: Gain insights into the content of uploaded images through predictions generated by the CLIP model, which is trained on a diverse set of image-text pairs.

    Applications in Climate Research:

    This tool can be utilized in climate research to analyze visual data, such as identifying patterns or elements in climate-related imagery. For example:

    • Analyzing satellite imagery to identify deforestation or urban expansion.
    • Understanding public engagement by analyzing visuals shared on social media related to climate events.
    • Supporting education and outreach efforts by providing insights into the connotative and denotative framing of climate imagery.

    For further details and access to the source code, visit the GitHub repository.

    Automated Text Analysis

    We are excited to introduce TextAnalyzer, a R package designed to simplify text analysis through an intuitive Shiny app. This tool is particularly suited for researchers and individuals with varying levels of programming experience who wish to explore insights from textual data effectively.

    Key Features

    TextAnalyzer offers a comprehensive suite of features to support your text analysis needs:

    • Flexible Data Input: Analyze text by pasting content directly into the app or uploading files in formats like .xlsx, .csv, .txt, or .json.
    • Dataset Overview: Examine your dataset with frequency distributions, graphical visualizations, and entity recognition tools.
    • Sentiment Analysis: Assess sentiment using various dictionaries, including Bing, NRC, Loughran-McDonald, and Afinn.
    • Emotion Detection: Identify emotions present in your data with tools like the NRC and Loughran-McDonald dictionaries.
    • Arousal & Valence Metrics: Explore arousal, valence, and dominance levels within your text.
    • Topic Modeling: Utilize Latent Dirichlet Allocation (LDA) to uncover optimal topics and visualize their distribution.
    • SeededLDA: Define custom topics and analyze your dataset with precision.

    Applications in Climate Research

    TextAnalyzer is a versatile tool that can support research efforts in various fields, including climate change studies. For example, researchers can use the app to:

    • Analyze climate-related news articles to understand public sentiment.
    • Detect emotions in social media posts related to environmental activism.
    • Perform topic modeling on climate policy documents to uncover recurring themes.

    Learn More

    Discover more about TextAnalyzer by visiting its GitHub repository: https://github.com/IsaacBravo/TextAnalyzer

    This repository includes detailed instructions, documentation, and examples to help you maximize the tool's potential.