How our origin informs our methodology and technology

First as Syrian Archive in 2014, the Mnemonic team was among the first groups within the international justice field to conduct investigations and documentation based on open source information utilising a clear, rigorous methodology. From the outset, Mnemonic cultivated robust technical expertise and developed tools to secure the content often shared by documenters at great personal risk. These tools and methods have allowed Mnemonic to comply with international standards for documenting human rights violations and international crimes, thereby honouring the efforts of those who document and bear witness. At the same time, through its regionally embedded teams and strong network of partners, Mnemonic ensures its archiving and investigative work meaningfully supports affected communities in their effort to use the archived records in the pursuit of justice.

Since 2014, this field – and Mnemonic with it – has evolved substantially. Social media documentation of human rights violations and international crimes has dramatically expanded. Shared understandings of how open source information might be used as credible evidence have significantly advanced. Actual use in legal and accountability processes has also progressed: open source information submitted by Syrian Archive contributed to the issuance of a historic arrest warrant against then-President of Syria Bashar Al Assad in connection to the 21 August 2013 sarin gas attack on the outskirts of Damascus and the landmark filing to the Human Rights Committee alleging Russian responsibility for a deliberate attack on Kafr Nabl Surgical Hospital in May 2019. The relevance and use of open source information in the human rights, transitional justice, and accountability sectors are now widely recognised.

Today, Mnemonic’s programmes secure and analyse open source information with the objective of advancing accountability for and memorialisation of human rights violations, international crimes, and other grave injustices.

Mnemonic hosts and manages the following research programmes: Syrian Archive, Yemeni Archive, Sudanese Archive, and Ukrainian Archive. Our Rapid Response programme has had a recent focus on Iran and Palestine, launching significant archival and investigative efforts. Mnemonic’s methodological and technical infrastructure empowers its programmes and enables partners to collect, preserve and organise open source information in such a way that safeguards its reliability and its evidentiary potential. Mnemonic’s records and research are used by international accountability mechanisms, domestic prosecution authorities, human rights litigation and advocacy organisations, investigative journalists, academic and memorial institutions, and policy and regulatory bodies.

All Mnemonic research is done in line with its Research Handbook: an internal resource created to document and communicate organisation-wide research guidance, policies and practices. Below we provide a general overview of Mnemonic’s methodological and technical infrastructure. Should you have any questions about our tools, methods, workflows or research policies: please contact us at info@mnemonic.org.

Our core research principles

All Mnemonic teams work in line with four core research principles: Do No Harm, Integrity, Humanity, and Inclusivity.

Do No Harm: Having chosen to work in high risk environments, we also choose to prioritise safety above all else. We commit to reflecting carefully on the possible impacts of our work and taking action to minimise harm so that whenever possible we “do no harm.”

Integrity: We recognise the duties and responsibilities inherent to our choice to take up this work. In undertaking Mnemonic’s mission, we demand from ourselves principled decisions, high ethical standards, and methodological rigour.

Humanity: We see and respect the humanity of all people contributing to, impacted by, or otherwise involved in our work. We commit to treating the subjects of our investigations with dignity and respect as human beings. In recognising the humanity of others, we also recognise the humanity in ourselves; in doing so, we commit to working with compassion and humility.

Inclusivity: We recognise the absolute imperative of diverse perspectives and voices in this work. We seek opportunities to challenge status quo practices that accept exclusion as a matter of convenience, habit, or prioritisation of selected characteristics. For every new door that we open, we commit to ensuring that it stays open for others to join in building more diverse and welcoming spaces.

Our roles and positionality

How the Archives relate to their working contexts differs team-by-team, reflecting the needs and dynamics of those contexts. However, the following characteristics generally apply to all:

Mnemonic programmes are flexible and tailored, taking lead from and working in solidarity with local actors.
Mnemonic maintains open source information archives with personal, cultural, community, and emotional value.
Mnemonic maintains open source information archives containing records that could potentially be used as evidence for justice and accountability.
Mnemonic engages with and advocates to decision-makers about open source research methods and evidence, including governments and tech companies.

Our in-house research tools

Mnemonic has developed two key tools that enable and enhance our archiving and research work on a daily basis: Diwan and Elfehrest.

Diwan (preservation software)

Mnemonic preserves online information using Diwan, a software developed and maintained by Mnemonic with international standards guiding its operations to ensure credibility and reliability for human rights investigations. The word “Diwan” or “ديوان” in Arabic means a collection of works that are connected by theme or style.

Diwan enables the automated and manual large-scale collection and preservation of digital open source material from social media platforms and websites. Diwan is designed to ensure four core digital forensics principles: Immutability, Integrity, Authenticity, and Reliability.

Immutability: Diwan treats every piece of information like a permanent note. Once data is archived, it is not possible for users to erase, or overwrite it. It is only possible to add new details alongside it. This way, Diwan has a clear, unchangeable record of everything that’s ever been collected and archived.

Integrity: Diwan double-checks every piece of scraped data against a simple template (a “schema”). If a required field is missing or collected badly, it logs a warning to re-archive the affected record and ensure that all expected files and metadata are collected and archived properly.

Reliability: All processes that Diwan uses in handling open source information are logged, auditable and repeatable. The results of applying such processes are reproducible. Diwan logs all new updates to the archived record, What happened, when it happened, and from which device, operating system, and IP address.

Authenticity: Diwan has the ability to demonstrate that a digital item remains unchanged from when it was collected. Diwan generates multiple hash values to the collected metadata and files of the archived record.

These core principles ensure verification of an archived record’s authenticity, preservation of its integrity, and full traceability of its chain of custody.

In line with recommended evidence collection practices, Diwan archives:

The metadata associated with the URL (the fields may vary depending on the platform);
A screenshot of the URL for platforms from which the metadata cannot be directly extracted via a publicly available API;
The embedded media content (videos, images and audio); and
The HTML, WARC, or the JSON file of the archived material.

Elfehrest (archive interface and annotation platform)

Information archived through Diwan is available for viewing and analysis on Elfehrest, the web-based platform that Mnemonic’s team developed and is using to interact with archived data that has been collected from online open sources. The word “Elfehrest” or “الفهرست” in Arabic means the index.

Elfehrest is a platform designed to enable Mnemonic’s team and its partners to securely access, search, and analyse millions of archived records for use in public investigation reports, reports supporting litigation, responses to requests for information and assistance from accountability bodies, and more.

Elfehrest allows users to submit URLs for Diwan archiving; view, tag, and annotate archived records; create and manage tagging schemas; cluster records into groups or incidents and then group those into broader datasets; link records, incidents, and events to actors or potential suspects; define command structures and map deployments of state and non-state actors; publish data to programme websites; and export archived records to external partners. Elfehrest also employs machine learning modules to prioritise records and improve their discoverability.

Our research policies

Mnemonic’s Research Handbook details quite a few policies that were drafted and designed to ensure our work meets professional standards, reflects ethical practice, and adheres to legal requirements. Here are some examples of policies covered by the Research Handbook:

Research Code of Conduct;
Secure data sharing;
Virtual identities and research profiles;
Graphic and violent content;
Intermediary consultations; and
Maintaining research notes and records.

Our core research tasks

While specific practices vary among teams, all Mnemonic programmes endeavor to accomplish three core research tasks: 1) securing relevant open source information by ensuring successful and reliable collection and preservation of materials; 2) tagging and organisation of the preserved materials – the Archives – so that they are as useful and usable as possible; and 3) producing reliable, research-based projects with additional analysis of the preserved materials.

Collection and preservation

Researchers identify and collect online material by monitoring platforms and websites, and searching for material following a specific request by a partner or to advance their team’s own strategic research projects. At least one person per programme is responsible for maintaining an updated, written record of specific sites that are being monitored and any other specific activities undertaken during monitoring. Upon identification of potentially relevant information, a researcher conducts a preliminary assessment of its relevance and any risks associated with the preservation of the online material. Individual links are then submitted for preservation.

Additionally, each programme maintains a vetted list of sources that produce online material consistently relevant to their working context and objectives. These sources can include legal practitioners, human rights activists, journalists, governments of all levels, bloggers, religious organisations, scholars, propagandists, members of security forces, and alleged violators of human rights. Sources on this vetted list may be considered impartial or may have a history of biased statements: a source may produce consistently relevant material even if not impartial. Generally, consistently relevant sources are notable contributors to the digital information landscape of the working context. These consistently relevant sources are regularly archived as part of automated preservation activities.

Mnemonic programmes may also collect and preserve material as part of research projects carried out in partnership with or at the request of external groups or individuals.

Most preservation of online material is done through Diwan’s automated process, which by design directly captures all the relevant information and metadata associated with the material. For cases where Diwan is not able to capture all the necessary data, Mnemonic has a manual archiving workflow, which includes documenting chain of custody information alongside capturing the relevant online material.

Tagging and organisation

Mnemonic’s Archives safeguard an immense amount of data pertaining to the documentation of human rights violations and other crimes. Using manual tags and automatically collected metadata the Archives become searchable resources, and research teams have the ability to conduct data analysis, potentially uncovering larger trends and patterns in the collected information.

When developing manual tags, programmes create general or dataset-specific tagging schemas. General schemas are applied Archive-wide and help ensure that key information, such as eyewitness or munitions remnant documentation, is more easily found. Dataset projects involve systematically tagging material for key information, according to a thematic and project-specific tagging guide.

A well-tagged Archive or dataset can support data analysis that speaks to:

Victim identification and mapping
Record-making, memorialisation and other truth-seeking or truth-telling efforts
Perpetrator identification, mapping, strategies and command structures, or mode of liability analysis (e.g., incitement, business complicity)
Incident or base crime patterns, patterns illustrative of criminal intent, indications of proportionality, and indicia of less-documented harms/crimes (e.g., SGBV)
And more

Additional analysis

Mnemonic researchers approach verification as a process of “reality-testing” the collected information and materials in order to determine what about them is reliable. Consistent with the international standards articulated by the Berkeley Protocol on Digital Open Source Investigations, Mnemonic teams generally consider three categories of analysis as part of their verification work: source, content and technical analysis. Researchers verify online material holistically – considering all types of analysis and information together – to get an impression of the reliability of the information collected.

Investigative analysis then involves piecing together the reliable information to reach reasonable conclusions or findings of fact. Generally, researchers seek to identify: locations (Where?); timing (When?); events (What happened?); and implicated people (Who? And how?).

For any given project, proper verification and investigative analysis necessarily involves a creative combination of multiple analytical techniques tailored to that particular set of facts and materials.

Commonly used analytical techniques include:

Visually comparing content to mapping imagery (e.g., satellite imagery, street view imagery, or 3D and terrain maps) to identify – “geolocate” – a depicted filming location;
Conducting shadow analysis on near-aftermath or eyewitness footage to estimate – “chronolocate” – the time of day that an event occurred; and
Reverse-searching images through online image libraries, to help track a visual’s online provenance and lifespan (including back to the earliest findable post).

Our research outcomes

To see how Mnemonic’s research methodology and technology are put into practice, explore the projects and impacts shared on our programmes’ websites: Syrian Archive, Yemeni Archive, Sudanese Archive, and Ukrainian Archive.

At the centre of this work is Archive-building: the systematic preservation, verification, and organisation of digital records documenting human rights violations. Mnemonic’s teams carry out four main types of research projects that shape how the archives are used and shared. These four activities – public investigations, dataset creation, strategic legal accountability work, and responses to information requests – form the backbone of our research and engagement.

Public investigations are online, open source, fact-finding investigations into a team’s working context. Carried out independently or jointly with partners, these investigations are strategic and public-facing interventions, usually designed to shed light on under-reported phenomena or incidents of harm.

Mnemonic programmes produce meticulously tagged and organised Datasets as large-scale investigations into particularly prominent phenomena or well-documented events in their working context, such as attacks on medical facilities or media workers. Each dataset project includes hundreds of meticulously tagged materials and is designed to enable systematic analysis, evidence-based reporting, and accountability processes. These projects often serve as a resource for journalists, investigators, and CSOs.

Together with our partners, Mnemonic programmes participate in Strategic litigation projects, advancing criminal case-building and other litigation efforts. This includes sharing evidentiary materials, contextual research, and technical expertise, always in alignment with international legal standards.

Requests for information are how our archives field and respond to external groups’ needs for access to our preserved records and research support from our teams. Through these requests, our teams are able to safely facilitate external access to the archives. Teams receive requests from a wide range of stakeholders, including: major news media outlets, family and victims associations, other civil society groups, investigating offices at international accountability mechanisms, documentary filmmakers, and national-level prosecutors and war crimes units. Fielding and responding to requests for information require highly confidential and often time-sensitive research and constitute a significant portion of our teams’ work.

If you would like to send a request to one of Mnemonic’s teams, please contact us at info@mnemonic.org.

.css-68jq5d{padding-bottom:1.45rem;color:#ffffff;-webkit-transition:0.25s;transition:0.25s;}.css-68jq5d:hover{-webkit-text-decoration:underline;text-decoration:underline;}ABOUT

Methods

Table Of Contents

ABOUT