News — 14 June, 2023

Driving Change through Data: Exploring Humanitarian Mapping Research and Analysis Initiatives

As we dive into the second half of 2023, we are thrilled to share the progress made by HOT and the Community Team in leveraging OSM data for informed decision making.

Versión en español

Hello! I’m Rubén Martín, HOT’s Community Strategist and Research Lead, and in this blog post, I aim to highlight our recent achievements and extend an invitation to OSM communities, contributors, organizations, and individuals interested in OSM data to collaborate with us on analysis and research endeavors.

Focusing on Actionable Insights: A Data-Driven Approach

At HOT, our data approach goes beyond satisfying curiosity; we strive to generate actionable insights that enable us to better support the communities utilizing Open Mapping to address local needs. Over the past months we have, together with our dedicated Data and Analysis Intern, Caleb Fagunloye (along with support from Benni Herfort at HeiGIT) made significant progress in two key areas outlined below:

• Better understand OSM Contributor Data and validate assumptions around contributors decline.

• Explore and experiment how to make OSM Data more accessible to non-technical people.

OSM Contributions Decline

Earlier this year, Simon Poole’s observation about a 20% decline in new OSM contributors in 2022 caught our attention. Recognizing the potential implications of this decline on HOT’s support to local communities, our Regional Hub teams were eager to investigate further.

After internal consultations, we embarked on an analysis to validate and understand these claims. Our approach involved examining historical data from previous years, considering not only the number of contributors but also contributions data and editing patterns on the map.

We selected four countries in each region where HOT has an Open Mapping Hub (Asia Pacific - AP, Latin America and the Caribbean - LAC, Western and North Africa - WNA, East and South Africa - ESA) where HOT had a strong support history and four countries with low or no presence. Additionally, we included a few countries in Europe and the USA.

By analyzing the number of active/new contributors and elements added to the map between 2017 and 2022, we aimed to provide a Big Picture Analysis that offers general insights to inform further research. While this initial analysis does not look into the root causes or propose specific solutions, it serves as a starting point for deeper exploration.

Key Insights from the Analysis:

1. Over the past five years, there has been a decline in the number of contributors in most countries analyzed.

2. The volume and pace of mapped elements, including buildings, roads, amenities, and health facilities, has consistently increased in most countries analyzed.

3. The number of contributors does not show a significant correlation with the volume of elements mapped.

4. There are unexplained peaks in mapped elements may be attributed to mass imports or corporate mapping

1. There is an overall decline in number of contributors over the last five years

The analysis showed a general decrease in the number of contributors in most countries year over year since at least 2017, except for some countries in ESA. There was a significant decline in the number of contributors in AP and LATAM countries, while Caribbean countries had a mix of stable and unstable charts.

The WNA region had the most volatile charts, with significant increases and decreases over the observed years. Spain remained flat, while Czechia and the United States also experienced a decline in the number of contributors.

2. There is a consistent Increase in elements being mapped

Most countries experienced a consistent increase in the number of buildings, roads, amenities, and health facilities being mapped every year. This means that elements keep being added to the map at a growing pace.

3. Number of Contributors have no significant correlation on the volume of elements mapped

The number of contributors have no significant correlation on the volume of elements mapped. While the number of contributors is important, it may not be the only or most important factor in determining the volume of elements mapped.

4. Unexplained Peaks in Elements Mapped, Possibly Due to Mass Imports or Corporate Mapping

We observed unexplained peaks in the number of elements mapped, which could be attributed to mass imports or corporate mapping. The peaks coincide with Apple’s active mapping efforts using GPS, LiDAR, and cameras to collect data, and Kaart’s automated mapping technology.

These corporations have the resources to make mass edits to OpenStreetMap, resulting in the unexplained peaks. Further investigation is needed to determine if these peaks are indeed the result of mass imports or corporate mapping.

You can check the complete graphs for the countries we analyzed or explore the raw data.

Open questions and next steps

Based on the insights gained, we have identified at least three key questions that warrant further investigation:

1. Who is mapping the most?

Understanding the different levels of contributors can lead to better support for their specific needs.

• Which types of contributors are not well retained?

• How much of the total do the most active contributors’ edits represent?

• How is AI-assisted mapping affecting the contributors/contributions ratios?

2. Beyond remote mapping

How does the analysis change when we focus on changes requiring on-the-ground/local knowledge, excluding corporate mapping?

Tools like StreetComplete and the new Field Tasking Manager signal spaces where more research and analysis would provide a lot of valuable insights.

3. What can we learn from fruitful countries?

Such as Timor-Leste. Are there dynamics that can be applied to other countries?

Given our belief in the significance of local/on-the-ground knowledge for solving local challenges, we are currently conducting a follow-up analysis to explore potential differences with the initial research findings.

We would love to keep collaborating with HeiGIT and other organizations and individuals so OSM Data can bring value to everyone who is using and editing the map.

Increasing accessibility of OSM Contribution Data

Parallel to our analysis of contributor data, we also pursued the goal of improving the accessibility of OSM data, ensuring that individuals without technical or data analysis backgrounds can easily find answers to their questions.

If I’m part of the local Kenya community and I want to know how the number of contributors evolved in the past year, is there an easy way I can get a straight answer about this?

Inspired by the ChatGeoPT project, which employed Large Language Models (LLMs) and AI to query OSM data using natural language, we sought to replicate this approach using the ohsome-api, optimized for contributors and contributions data.

In a controlled three-week experiment, we aimed to address this challenge. However, we encountered significant obstacles when utilizing the commonly used LLM (gtp-3.5) to interact with the ohsome-api. This model lacked training on utilizing the API effectively, hindering our progress.

Nonetheless, we identified tremendous potential in pursuing this avenue in collaboration with other individuals and organizations interested in OSM data. We are actively exploring conversations with HeiGIT and others to refine and fine-tune LLM models, ideally open-source, to accomplish this task.

Get Involved: Let’s Collaborate!

We invite you to connect with us, whether you are an OSM community member, contributor, organization, or an individual passionate about OSM data. Together, we can unlock the value of OSM data, fostering collaboration and enabling meaningful insights for all.

If you are interested in exploring these opportunities, please reach out to me at ruben.martin at hotosm dot org. We look forward to engaging in fruitful discussions and driving impactful change through collaborative OSM data analysis.