Peace Analytics

Our solution team is currently part of the Peace and Conflict Resolution Platform (www.peacerep.org), a research programme located at the University of Edinburgh, but involving a consortium of peacebuilding and research organisations, and teams of local researchers in conflict-affected countries. The solution team comprises the data engineers, visualisation researchers, developers of the Tracker interface, and those with peacebuilding or mediation experience who are responsible for technical development, engagement, and iteration.

We aim to solve the problem of an absence of 'peace analytics', that is - data analytics in the peacebuilding field that could support better outcomes for peace and transition processes.  

Specifically, we aim to support better 'adaptive management' of the implementation of peace and transition processes reached. 

Background    

Peace agreements were once viewed as a moment at which warring parties ended their conflict and agreed to peace. However, many agreements break down or remain only partially implemented even many years later.  For example, while it is over 25 years since the Belfast Agreement was signed in Northern Ireland, implementation took many additional agreements.  Indeed, until January of this year, power-sharing had broken down for several years, meaning that Northern Ireland’s political institutions were not in place. While in this case, peace largely held, breakdown of agreements in Yemen, and Sudan, among others, have seen catastrophic new phases of conflict.

Organisations such as the United Nations, regional organisations, and institutions created specifically to deal with implementation such as joint monitoring committees, or international contact groups, are often given an active role in ensuring implementation.  Implementation often involves not just identifying failures by the parties to an agreement to 'keep the promises' they made in the agreement, but also a failure to deliver the types of actual change in a lived experience of peace that people expected.  Both types of failure undermine public confidence in the peace process and affect whether any form of peace results.   

Research has revealed, that rather than a simple matter of 'enforcing' an agreement’s commitments, supporting implementation often needs 'adaptive management' of a set of complex new challenges. These challenges can include the long duration of peace and transition process, where changes in balances of power over time are seen due to new armed groups taking up arms, or new issues emerging to threaten the peace if not dealt with.  

It is therefore important to identify gaps between promises made and kept, and also gaps between what people expected of peace, for example, improvement of economic, human rights, or reconciliation and the realities they experience. Yet, it is often difficult to identify where gaps between the promises of often-lengthy agreements, and the reality of implementation, in a timely enough way to enable them to be effectively addressed. 

A mix of quantitative and qualitative data relevant to implementation exists across structured and unstructured sources.

BUT: putting the right data together in a way that enables it to be used to intelligently track progress over time, together involves a huge effort that without a solution, would need to be replicated across multiple settings through manual work. Most organisations do not possess the time, resources, or data expertise.

Our solution addresses both the problem of peace process implementation and the underlying challenges of using data better to support peacebuilding, through a mix of technologies.  It involves four components:

1. We provide a Peace and Transition Process Tracker, which brings together different qualitative and quantitative conflict and peace data from a variety of trusted sources, and uses interactive visual interfaces designed to facilitate novel insights on implementation, and support critical understanding of whether and how a peace process is progressing, and what is falling apart. 

2. We work with local data collectors to improve and extend the granularity of data relevant to each country’s specific context, such as 'peace perceptions' data from local communities in conflict. This data helps to bridge the gap between large country-level institutional datasets that do not take local nuances and dynamics into consideration, or need 'ground-truthed' through local perceptions. 

3. Taking a user-centered design approach, we aim to provide bespoke versions of the platform of our data-driven tools for parties who wish to see new forms of qualitative and quantitative data incorporated.  

4. We make our research infrastructure and innovative visualizations open source and able to be adapted, to support those working at the country level to re-purpose/reuse data, and visualization tools. 

We use the following technologies across a platform that attempts to provide 'different ways of knowing'. These technologies primarily include database design, large computing infrastructures, natural language processing, and creative interactive visualization.  

Our solution directly serves a wide range of local and international actors who wish to support the peace process to progress, and the peace agreement to be implemented.  These actors include:  

1.  National civic actors in conflict contexts who seek to influence armed conflict actors, other civic actors, and international donors, to ensure that the peace endures, by providing an evidence basis for where they need to put attention and effort. These groups often have limited capacity to access data and apply it to problem-solving. There is often 'too much' data, and understanding what it might mean, what its limitations are, and how to bring different types of data together in ways that could support monitoring peace agreement progress is not easy. Our Tracker has addressed these problems.

2. International organisations supporting conflict resolution. These organisations often have formal roles monitoring peace and transition processes, and they have access to a broad range of data, but again, they do not have the time or energy to develop custom-built visualization interfaces to portray information in novel ways, that readily let them see what they want/need to see in an engaging, easy-to-understand, and consumable manner.   

3.  A range of state, international, and philanthropic donors who support multi-donor trust funds, who often remain convinced of implementation challenges, and respond to evidence-based approaches.

We bring together a unique and diverse team of PeaceTech experts from various disciplines including data engineering, information visualization, peacebuilding, constitution-making, and peace mediation. Our distinctive added value lies in our ability to match data solutions to a strong grounded knowledge of 'what is useful' to people seeking to understand implementation through data-driven solutions.  

Our wider research team has created a unique peace agreement database, with all the world’s peace agreements from 1990 to date across all stages of a peace process (ceasefires to comprehensive and implementation agreements), coded for all the key topics. This database is the only database of this scope and has built-in interoperability with other datasets, such as deaths in conflict data. As such, it provides a unique framework to drive the implementation of the Tracker.  

Our team has also developed the Tracker pilot over two years, to bring together appropriately curated 'measurements' and other ways of understanding change over time, across different forms of data, in an automated way.  

Through this work, the team has also developed a set of digital capacities to serve a wider digital transformation of the peacebuilding field (PeaceTech).

We have made the first iteration of our Tracker publicly available in early 2024, having spent two years trialing proof-of-concept work of its various components and responding to feedback from actors in conflict countries and experts in the field.  

We have also created a unique peace analytics data infrastructure, with new forms of visualization and natural language processing, that is capable of supporting the service provision element of our plans.

We are now working to 'socialise' the Tracker with the above actors, using a strong network across international organisations in addition to local researchers and activists. These interactions and feedback opportunities will be continuously used to re-iterate the Tracker, so as to amend and develop it further.   

To bring this venture beyond the pilot stage, we need a new institutional and staffing model that enables us to build a self-sustaining way of working that would enable the Tracker to be sustained and scaled up over time. Solve seems to provide good training and support on exactly these aspects. 

We would also really appreciate being part of a supportive network of people involved in similar challenges and values to meet, collaborate, and learn from fellow travelers. 

We would really value the connection with MIT due to our academic grounding in the solution as academic researchers.

The primary barriers we hope to address through Solve include:

1. Being more strategic in our taking end-user engagement to scale in ways that enable good prioritisation of re-iteration of the initial pilot.  

2. The difficulty of developing a new business model that enables us to move from one major grant to a mixed funding model that is sustainable.   

3. The difficulty of scaling up and staffing to provide the 'service' element of our ambition in terms of building in local data collections or supporting local partners to modify/customize components of our platform, or reuse our interactive visualizations and PeaceTech tools, to scale up our peace analytics efforts to support those who peace processes. 

Our solution stems from a state-of-the-art peace agreement data and research project (www.peaceagreements.org), that has collected, curated, and digitized data and metadata of peace agreements and peace and transition processes. This data archives different types of agreements to end conflict over time, enumerating their stage, which actors were involved, and what issues were addressed. We store the agreement texts both as artifacts and in the form of plain text in a corpus that enables a wide variety of research methods on agreement content, that can be used to ground assessments of implementation.

From this baseline data collection, we can then monitor implementation of different agreement commitments. We have created a highly visual interface that enables very different types of data and different ways of assessing implementation to be triangulated.

The associated interactive interface enables progress to be viewed through:

- timelines of institutional change

- a curated approach to utilising widely used quantitative data, that acts as a proxy lens to understand different types of progress with relation to agreement commitments, that is based on underlying research as to data relevance, quality, and scope. 

- tracking implementation of agreement commitments through narrative reporting related to the agreement, using visualizations in the tool to, for example, check if ceasefires are being held.  This narrative information can be triangulated with other curated quantitative data, say relating to violent incidents over time. The Tracker is unique in this approach.  

This qualitative data on implementation uses natural language processing to provide fast access to relevant parts of agreement texts, and aggregate mentions of text relevant to these provisions in a range of other implementation texts/sources/reports (eg the OSCE reports into the Minsk agreements between Ukraine and Russia between 2014 and 2022).  

- Actor network mapping, to understand who is involved and how

Our solution will have an impact on the problem because it will provide an innovative data-driven way to understand how peace and transition processes are progressing. 

We therefore expect it to: 

(a) help a range of actors identify implementation gaps in a more timely way

(b) support informed adaptive responses to implementation failures

(c) provide a working example of how 'peace analytics' can be developed and used 

(d) provide a reusable set of peace analytics engineering and interactive visualizations, that enable 'peace analytics' not just to be scaled up by ourselves, but across organisations.  

For example, we expect the Tracker to provide civic actors in countries with data-driven, interactive, and engaging visual evidence to supplement/enrich their contextual political arguments for projects to address deficits in implementation or risks of new conflict outbreaks. Groups often need to provide forms of evidence to persuade donors and other influential actors to support them.  

We also expect the Tracker to help local groups to bring new data to the table, that can add to, and triangulate or contextualise our existing qualitative and quantitative data, in particular by providing better information on local experiences of implementation, and local expectations of the peace impacts they expect to see. 

More generally, we expect that our provision of a new open infrastructure will enable and empower groups to 'do this for themselves'. The drive for this is to help inspire a general reactive rise in 'peace analytics' that can diversify the field by shifting priorities from monitoring conflict and trying to provide 'early warning', to monitoring progress in ending conflict, by pointing to risk moments and providing insights into possible entry points and strategies to address these risks effectively.

Our goals are: 

·       To better develop and promote 'peace analytics' - defined as data analytics to inform research and decision-making in the peacebuilding field. 

·       To promote open source re-usable data frameworks and visualizations for 'peace analytics' to build easier use of data and supplement other ways of evaluating implementation to a range of end-users. 

·       To encourage greater interoperability among data providers relating to peace and conflict, by demonstrating how existing data can be better brought together to solve peacebuilding problems, without resource-intensive solutions that 're-invent the wheel'.  

We would measure success in: 

·      The number of times and variety of contexts the Tracker is used and quoted in reports by relevant parties.    

·       Case studies of the following (the number set per year according to a planned scaling up): 

-  Meaningful examples of the use of insights from the tracker by key local actors to effectively address implementation challenges

- Local actors working to collect data and have it incorporated into the platform 

- Local actors building their own capacity to use the underlying infrastructure to give them better organisational capacity to use and present their data to influence the implementation of peace  

As the Tracker uses mixed approaches and is a curated suite of different PeaceTech tools, with different technologies driving different components. Below is an overview of our core technologies: 

Below is an overview of our core technologies: 

· The tracker interface is hosted on our peace agreement database public website, driven by a Django back-end infrastructure on University of Edinburgh servers. We store our data and curated external data feeding the tracker in an underlying PostgreSQL database, utilising the world-class computing services available in Edinburgh. These external sources are automatically updated as often as possible to provide users with as close to real-time updates as we can.      

·       How we display and make this data accessible differs depending on the intended users of the tool and their purposes of use. This determines the data being displayed to ensure robustness and relevance. For some features, we utilize PowerBI as this is offered by many organisations we work with as part of the office suite – it is low code, relatively cheap, and can encourage our users to develop their own customized dashboards and interactive data reports by re-using our data model and approaches. Additionally, it allows for PDF exports of the data and visualizations, which is particularly useful in areas with limited internet connectivity who need to work offline/from print most of the time. We utilise other open source, free/low cost, or low/no-code solutions, such as Tableau Desktop, knight lab’s Timeline.js library, Graph Commons, and Kumu to encourage others in the field who may not have coding experience, to utilise these tools with a small learning curve. 

·       Other, more novel approaches to making this data accessible include custom visualizations of the data using D3.js, Leaflet.js, MapboxGL.js, and other JavaScript libraries (e.g., Vue.js or React.js), in addition to Python-powered models on demonstrator interfaces. Additionally, our project acts as a pilot case study that focuses on utilizing a new library developed internally at the University called NetPanorama. As part of this study, we develop network graph visualizations using our data to highlight the innovative and modern tools in use in the Tracker that allow for interrogation of processes.  

·       For more advanced network analysis aimed at exploration of our vast data collection of signatories to agreements and those involved in dialogue and mediation processes, we use matrix representations of bipartite graphs in C++ and Python to power the visualizations and platforms that allow users to query the data effectively and robustly.  

·       New natural language processing tools utilise a variety of techniques and libraries that we have incorporated into our pipelines, including Named Entity Recognition and Semantic Similarity models for topic modeling, clustering, diffusion, and summarisation.

Six 

A Chief Tracker Data Engineer (full time staff )

A visualization researcher (full time staff)

A developer (full time staff)

A natural language processing expert in semantic similarity models (contract staff)

A peace and conflict data expert (full time staff, but working part-time on the Tracker components, currently on maternity leave until fall of 2024)

A globally recognised peace process academic and mediation expert (full time but part-time on this project, but currently on a leave of absence from 2023-24, working as a United Nations Senior Mediation Advisor in the Department of Political and Peacekeeping Affairs). 

We have been working on the Tracker and Peace Analytics Framework for two years, to produce the pilot.

However, we have been working on the underlying peace agreements database for in total 10 years, and over time this has developed new forms of data interoperability on which the Tracker relies and also provided research on reliability of data and data quality.

We comply with a robust anti-discrimination framework covering all status equality categories, and a strong set of institutional policies on equality, diversity, and inclusion for all staff, and this is reflected in the make up of our small team.  We also have a pro-active team commitment to a diverse and inclusive staff and inclusion has been at the heart of our work on conflict and peace mediation. We have a proactive commitment to working with staff from the global south. However, current institutional constraints, together with visa constraints limit exactly how we do this. Despite this barrier, unique research partnerships with country researchers also facilitate the Tracker development, and provide a strong basis for iterative design. These partnerships also extend racial and national diversity.

The team is currently part of a larger consortium of researchers drawn from peacebuilding non-governmental organisations, and Universities, led by the University of Edinburgh (https://peacerep.org/about/people/). However, to develop a commercial offer, we would want to establish the Tracker team in the form of a non-profit company or similar, that can both retain a close relationship with the wider research team, while becoming a self-sustaining unit underpinned by a clearer business model that could include service provision.  We would seek advice on the best way to achieve an appropriate institutionalization.

We provide value that is likely to garner academic grants, tech-for-good start-up grants, donor support for peacebuilding and mediation support, and earned income for providing bespoke support.

Our plan to be financially sustainable is to use this end period of our current grant to move the Tracker into a self-sustaining project, funded by a mix of grants, 'earned' income, and philanthropic support.  

As regards evidence for this being possible, we have so far been successful in obtaining grant support sufficient to develop first the underlying data and then the pilot.  Our initial consultations on the pilot indicate the unique offer and a strong appetite for it, meaning that we think grant funding will be further forthcoming.  

We have also had a number of requests from international organisations, and local initiatives, to deliver a service by using our technology and components of the Tracker and our unique mediation experience.  This has included a request from an international organization to provide training, and then digital tools, drawing on the tracker, for ceasefire monitors; and a request to support confidential archiving and making accessible, witness testimony in transitional justice mechanisms. This type of request has been fairly persistent without us advertising services and leads us to believe that a business model would be useful. At the point of requests, we struggled with how to scale up staffing to enable this work to be undertaken alongside the further development of the pilot.