Leveraging AI for Land Use Conflicts Identification and Government Policies

Background

Land degradation impacts 3.2 billion people globally and costs the world economy 10% of its gross product annually. Efforts to restore 350 million hectares of degraded land face significant challenges due to land disputes, particularly in countries like India. These conflicts hinder landowners’ ability to implement sustainable practices. With over 21 million hectares pledged for restoration and 3 million people affected by land disputes annually in India, a scalable solution is critical. AI and machine learning offer innovative potential to identify land-use conflicts and align them with appropriate policies for resolution.

Objective

To develop an AI-driven solution that identifies land use conflicts through news data and aligns these events with mediating government policies. The project aims to empower policymakers with actionable insights to resolve conflicts efficiently, save resources, and promote environmental sustainability.

Approach

The project leveraged machine learning and natural language processing (NLP) techniques to identify and map land use conflict events:

1. Data Collection: 
   • Scraped and analyzed 65,000 candidate news articles related to land use conflicts using tools like GDELT and Google BigQuery.
   • Extracted full news texts with news-please and manually labeled 1,600 articles as conflict-positive or conflict-negative.
2. NLP Techniques: 
   • Applied coreference resolution using SpaCy and Neuralcoref to analyze text data.
   • Trained models using ELMo, BERT, and Logistic Regression for predicting conflict articles.
3. Topic Modeling and Matching Policies: 
   • Employed CorEx for topic modeling with 93% accuracy.
   • Mapped conflict events to relevant government policies to identify gaps.
4. Visualization App: 
   • Built a machine-learning-driven visualization app to display conflicts and their policy connections, enabling data-driven decisions.

Results and Impact

• Enhanced Decision-Making: Policymakers can now resolve land conflicts faster and more effectively by matching conflicts with relevant policies.
• Scalable Solution: The AI model processes extensive data, offering a replicable framework for other regions and countries.
• Environmental Sustainability: By addressing land disputes, the solution accelerates the implementation of sustainable land-use practices.
• Quantifiable Impact in India:
  • Land disputes affect 7.7 million people and threaten investments worth $200 billion.
  • AI-powered solutions can help resolve disputes impacting 2.5 million hectares of land.

This initiative demonstrates the power of AI in tackling environmental challenges and advancing global sustainability goals.

Future Implications

The findings pave the way for:

• Policy Development: Tailoring government interventions to reduce land disputes and enhance environmental governance.
• Global Applications: Expanding the model to other countries facing similar challenges.
• Research Advancements: Further refining AI and NLP tools to improve predictive accuracy and scalability.
• Sustainability Practices: Encouraging countries to adopt AI-driven solutions for land restoration and conflict resolution.

This project underscores the transformative potential of AI in achieving environmental sustainability and resolving pressing socio-economic issues.

A detailed whitepaper from the project can be downloaded here.