Understanding Land Ownership in Kenya Through NLP to Help Restore Lands

Project completed!

Omdena collaborators, World Resources Institute, Code for Africa, and a team from Stanford University have been building solutions to map land ownership to boost Kenya’s efforts to restore degraded land in an equitable way. 

The problem

Monitoring global environmental change relies both on detecting measurable biophysical changes as well as social changes, such as new policies, changes in governments, grassroots movements, or social barriers to sustainability. Because it is difficult to understand the hidden actors and interests behind land management, the ability of the global development community to develop successful action plans is limited. Understanding the political economy of natural resource management will help understand key bottlenecks for forest and landscape restoration and help fight corruption. 

Code for Africa, collaborators from Stanford University, and the World Resources Institute (WRI) are partnering to understand the network of corporations and persons involved in land transfer and ownership in Africa. Publicly available documents from government institutions in Kenya such as government gazettes, court records, and government procurements will be analyzed using Natural Language Processing techniques. The aim is to understand what networks and connections exist in open-source land documents. This information will be cross-referenced with the ‘Watchlist’ data. The results intend to be used by journalists, researchers, and investigators in Kenya to fight corruption and to promote good governance of land resources by those seeking to restore them. 

The project goals

As a team, the aim is to accomplish the following objectives:

• Creating a knowledge graph of land ownership in Kenya.
• Generating a list of organizations in the form of a ‘Watch List’ to identify if there is any corruption with land ownership.
• Understanding what networks exist in the open-source land documents and eventually map them to polygons. Therefore geo-locate the land and the actors associated with it.

The data

Stanford University, Data Science for Social Good, cohort created a pilot using Kenya’s public land ownership records known as gazettes. Building on this work, the second phase aims to optimize the model for Kenya and create a methodology that can be used for scaling. 

The proposed data sources are a series of initiatives under the connectedAFRICA project, including the sourceAFRICA repository, the gazeti.AFRICA digitization campaign and the connectedAFRICA investigative data initiative. Together these projects seek to digitize key African court/legal, political/electoral, government/public, and company documents, including whistleblower and investigative media evidence, to allow for public-interest analysis and development of an African knowledge graph for entity data. This includes an effort to map politically exposed people (PEPs) to help identify potential connections between politicians and companies regarding deforestation, land management, and forest restoration with a focus on Kenya.