About CRI Disinformation Monitoring
Why We Do It:
In an era where information can significantly shape public debate, misleading or manipulative narratives may weaken trust, intensify social divisions, and affect democratic participation. At CRI, we seek to better understand these information dynamics and support the integrity of the public information environment.
Our Disinformation Monitoring project examines narratives and influence patterns relevant to the Baltic states, including content that may reflect or advance the perspectives of foreign state or state-aligned actors, particularly those associated with China and Russia. By documenting and contextualizing potentially misleading or divisive narratives, we aim to contribute to informed public discussion and strengthen societal resilience.
Why It's Important:
Disinformation is generally understood as false or misleading information distributed deliberately to achieve a particular objective. However, intent is often difficult to establish conclusively. For that reason, CRI’s analysis distinguishes, where possible, between demonstrably inaccurate claims, misleading framing, unverified information, coordinated amplification, and legitimate opinion or political advocacy.
Historical, linguistic, and geopolitical factors may make parts of the Baltic information environment particularly exposed to external influence. Our work seeks to identify and analyze relevant narratives and to help the public and policymakers understand their possible origins, characteristics, reach, and effects. These assessments can inform proportionate and evidence-based responses while respecting freedom of expression and legitimate differences of opinion.
Our Approach:
Our methodology, described below, outlines how we monitor and analyze publicly available content. Our approach is grounded in transparency, contextual analysis, and recognition of uncertainty.
CRI’s assessments represent the organization’s analytical judgments based on the information available at the time of review. They should not be interpreted as judicial findings or definitive determinations regarding the motives, affiliations, or conduct of any individual or organization.
Our strategy combines monitoring, quantitative indicators, qualitative analysis, and engagement with relevant stakeholders. Automated tools assist researchers in identifying material for review, but they do not independently determine whether particular content or actors constitute disinformation. Material findings are subject to human review and contextual assessment.
Consequences of Inaction:
Without systematic monitoring and analysis, potentially misleading or manipulative narratives may go undocumented, making their development and possible effects more difficult to understand. Such narratives may contribute to declining trust, social polarization, or confusion in public debate.
CRI’s initiative seeks to support a media environment in which significant claims can be examined critically and public discussion can be informed by accessible evidence. Our role is analytical and educational: we aim to identify relevant patterns, explain our reasoning, and acknowledge uncertainty where the available evidence does not support firm conclusions.
Looking Forward:
The information environment is continuously evolving. As new platforms, narratives, and methods of amplification emerge, CRI will continue to refine its monitoring practices and analytical criteria.
Our objective is not to police opinion or establish a single authorized version of events. Rather, we seek to help audiences assess the reliability, context, and possible effects of information for themselves. We remain committed to transparent, proportionate, and evidence-based analysis of developments affecting the Baltic information environment.
Methodology of Disinformation Monitoring in the Baltic States
Introduction:
The Civic Resilience Initiative (CRI) works to strengthen societal resilience in Lithuania and the neighboring Baltic states. Its monitoring program studies potentially false, misleading, manipulative, or divisive narratives and examines indications of foreign state or state-aligned influence in the Baltic information environment, including influence potentially associated with actors in China and Russia.
References to alignment, influence, or amplification are analytical assessments based on defined indicators and available evidence. They do not necessarily establish formal direction, coordination, funding, or control unless CRI expressly identifies reliable evidence supporting such a conclusion.
A specialized team oversees monitoring across the three Baltic states. The team uses a combination of subject-matter expertise, public-source research, and technical tools to identify relevant content and patterns. The project may also examine the broader information ecosystem, including prominent actors, networks, amplification patterns, and publicly observable relationships.
Legal-readiness assessments, practical recommendations, and outreach activities support the project’s wider objectives. CRI aims to conduct this work consistently with applicable law, freedom of expression, privacy, and responsible research practices.
Our Methodology:
1. Adaptive Monitoring Approach:
The changing nature of the media environment requires an adaptive approach. Monitoring priorities may therefore be adjusted to reflect recent developments and emerging narratives in the Baltic states.
Purpose:
The primary goal is to follow evolving stories and narratives that may materially affect the information environment. Selection for monitoring does not, by itself, indicate that content is false, unlawful, coordinated, or attributable to a particular actor.
Implementation:
Researchers periodically review monitored topics, sources, keywords, and events. Focus areas may be adjusted when narratives emerge, evolve, or decline in relevance. These decisions are based on factors such as public significance, apparent reach, repetition, cross-platform activity, and relevance to the project’s research questions.
2. Two-Part Analysis:
CRI combines quantitative and qualitative analysis to provide both breadth and context. Neither component should be interpreted in isolation, and numerical indicators do not independently establish intent, coordination, or impact.
Quantitative Analysis:
Focus:
The analysis may include outlets that CRI assesses, under stated criteria, as regularly reflecting or amplifying Kremlin-aligned perspectives. The sample may also include independent media, social-media content, and other relevant sources to provide context and points of comparison.
The description of an outlet as “Kremlin-aligned” is an analytical classification, not necessarily a claim of formal affiliation, direction, funding, or control. Where relevant, reports should explain the indicators supporting the classification.
Process:
Researchers review selected articles and posts to identify the frequency and distribution of defined narrative categories, including content that may be divisive, polarizing, misleading, or likely to undermine trust.
Coding decisions are based on documented criteria. Where classification is uncertain or reasonable analysts could disagree, the content may be labeled as inconclusive, mixed, or requiring further review.
Outcome:
The analysis produces descriptive statistics about the monitored sample, including narrative frequency, repetition, and apparent patterns of amplification. Results describe the material collected during a defined period and should not automatically be generalized to the entire media environment or population.
Quantitative findings indicate observable patterns; they do not, without additional evidence, establish causation, audience persuasion, coordination, or the intent of the people or organizations involved.
Qualitative Analysis:
Focus:
Qualitative analysis considers prominent or relevant stories appearing across monitored media outlets and social platforms during the reporting period. Selection may reflect reach, repetition, public significance, novelty, or relevance to the project’s research questions.
Process:
Researchers assess each selected story in its linguistic, political, historical, and social context. The analysis may consider framing, sourcing, factual support, audience engagement, amplification patterns, and the groups apparently addressed by the content.
Researchers may also assess whether a story could contribute to polarization or reinforce previously observed narratives. Such conclusions are expressed as analytical judgments and are qualified according to the strength of the available evidence. CRI avoids attributing intent, coordination, or institutional affiliation unless supported by sufficiently reliable evidence.
Outcome:
The analysis results in expert reports explaining the observed narratives, available evidence, possible interpretations, and potential implications. Where appropriate, reports distinguish between established facts, third-party claims, researchers’ interpretations, and unresolved questions.
Assessments of potential impact do not necessarily demonstrate that the content caused a measurable change in public opinion or behavior. Where direct impact data is unavailable, reports describe impact as possible, apparent, or estimated rather than established.
3. Data Collection & Reporting:
Data collection and reporting are designed to provide a structured view of the monitored information environment. The findings remain subject to the scope, availability, and limitations of the underlying data.
Frequency:
Weekly collection helps the team follow recent developments, while monthly reporting provides a more consolidated analytical view. Reporting periods and relevant cutoff dates are stated where appropriate.
Scope:
Monitoring may cover selected state-aligned or state-sympathetic outlets, independent media, publicly accessible comment sections, and other sources relevant to the research questions.
Social-media monitoring may include publicly available content from Facebook, YouTube, X, and other platforms considered relevant during a particular period. Access restrictions, deleted content, platform policies, algorithmic changes, language limitations, and technical constraints may affect the completeness of the dataset.
Monitoring a person, organization, outlet, or account does not imply wrongdoing or establish that the monitored entity knowingly participated in a disinformation activity.
Report Structure:
Experts from each Baltic country prepare monthly reports summarizing principal findings, relevant context, quantitative indicators, qualitative assessments, and results generated with the assistance of monitoring tools such as Awario.
Reports should, where practicable:
- identify the period, sources, and selection criteria used;
- distinguish factual findings from analytical interpretations;
- cite or preserve supporting material;
- disclose significant methodological limitations;
- use calibrated confidence language;
- avoid inferring intent or coordination from similarity alone;
- record material corrections or updates; and
- provide an appropriate channel through which affected parties may submit clarifying information.
4. Tool Integration & Enhancement:
Technical tools help researchers collect, organize, and examine large volumes of publicly available information. Tool-generated outputs are treated as indicators for further analysis rather than conclusive findings.
Web Scraping
Where legally and technically permitted, Oxylabs’ web-scraping tools may assist in collecting publicly accessible material relevant to the defined research scope. Collection may be incomplete because of access restrictions, technical errors, content deletion, changes to websites, or other factors.
The inclusion of content in a collected dataset does not mean CRI has determined that the content is false, misleading, coordinated, or unlawful.
Sentiment Analysis:
The sentiment analysis process, based on Awario’s media monitoring tools undergoes continuous refinement. This enhancement ensures a nuanced understanding of public reactions to various narratives, providing a clearer picture of their impact. Awario and related media-monitoring tools may assist in identifying patterns in language, engagement, and apparent sentiment. Automated sentiment analysis has recognized limitations, particularly in relation to irony, sarcasm, coded language, multilingual content, and culturally specific expression.
Sentiment outputs are therefore treated as approximate indicators. Material conclusions should be reviewed by researchers familiar with the relevant language and local context.
Data Analysis:
AI-assisted tools may be used to accelerate tasks such as categorization, summarization, translation, clustering, or the identification of potentially relevant patterns. These tools may produce errors, omit context, or reflect biases present in their models or input data.
AI-generated outputs do not independently determine whether content constitutes disinformation or whether an actor is responsible for an influence activity. Researchers review material findings, compare them with available evidence, and revise or reject automated outputs where appropriate.
Quality Assurance:
Before publication, material findings should undergo an appropriate level of human review. Stronger allegations-particularly those concerning intent, coordination, funding, control, or attribution-require correspondingly stronger and clearly documented evidence.
Where evidence remains incomplete, CRI uses qualified language and states the level of confidence or uncertainty. If significant new information becomes available, CRI may clarify, correct, or update its published assessment.
The content on this website represents solely the analytical opinions, published in good faith for informational purposes only, with no intent to cause harm, offense, or defamation to any party.