Despite the overwhelming scientific evidence supporting the impact of humans on the environment, climate misinformation remains pervasive. This persistent spread of falsehoods is often achieved through the misrepresentation of scientific evidence and the promotion of pseudoscientific narratives that hinder efective climate action. To combat this issue, we introduce ClimaFactsKG, a knowledge graph that links common climate change denial narratives with scientific corrections. ClimaFactsKG currently consists of 252 common climate myths and the corresponding scientific counter-arguments. A key feature of ClimaFactsKG is its strategic integration with CimpleKG, one of the largest existing misinformation knowledge graphs. This connection allows the interlinking of scientific corrections with over 611 misinforming climate claims found in CimpleKG and significantly enhances the utility of ClimaFactsKG. By providing a structured and interlinked repository of climate change myths and their scientific rebuttals, ClimaFactsKG ofers a valuable resource for researchers studying climate misinformation, fact-checkers seeking reliable counter-evidence, and educators aiming to improve climate literacy. This work represents a crucial step towards developing more robust, data-driven approaches to identify, analyse, and mitigate the efects of climate misinformation.
Although the scientific consensus about the role of human activity in altering the climate is overwhelming,1 climate misinformation remains pervasive. The persistent spread of climate misinformation both online and ofline cultivates scientific misunderstanding that hinders the understanding of climate change by the general public and the development of efective responses to the climate crisis.
In response to the rise of climate misinformation, several key strategies have emerged to combat its spread. These include fact-checking initiatives that directly debunk false claims, the development of climate scientific resources to organise and provide access to accurate information, and research-led communication strategies and tools that help identify and classify misinforming claims (see Section 2).
To deal with the continuously expanding amount of climate misinformation, fact-checking and research initiatives require easy access to scientific resources and historical climate misinformation. Although many websites ofer scientific resources on climate misinformation, their human-centric design makes the content dificult for machines to access.
In this paper, we introduce ClimaFactsKG, a climate Knowledge Graph (KG) that connects climate
misinformation myths with scientific corrective evidence. ClimaFactsKG relies on data extracted from
the SkS website and connects it with the CARDS taxonomy 2 [
At the time of writing,3 ClimaFactsKG contains 252 common climate myths with their scientific 5th International Workshop on Scientific Knowledge: Representation, Discovery, and Assessment, Nov 2025, Nara, Japan $ gregoire.burel@open.ac.uk (G. Burel); harith.alani@open.ac.uk (H. Alani) 0000-0003-0029-5219 (G. Burel); 0000-0003-2784-349X (H. Alani)
© 2025 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). 1NASA scientific consensus, https://science.nasa.gov/climate-change/scientific-consensus/. 2CARDS, https://cardsclimate.com.
3The statistics provided in the article are based on data collected on the 8th of July 2025. correction. These myths are linked with 611 fact-checks from CimpleKG.
Although there are some information sources designed to provide scientific knowledge related to
climate misinformation, these resources tend to be available in a format that is primarily designed for
human readability (i.e., website instead of an API or downloadable database) and come mostly from
the fact-checking community. As a result, these resources cannot be easily integrated into automated
fact-checking workflows and analysed easily. Skeptical Science 4 (SkS) is one of the most established
sources of scientific knowledge for fighting common climate misinformation and myths. SkS is backed
by multiple climate scientist.5 and the website main purpose is to "debunk misinformation that is
harming our species’ ability to deal with climate change caused by excessive anthropogenic greenhouse
gas emissions".6 This purpose is achieved by presenting peer-reviewed science in a concise form and
explaining the techniques used for climate science denial. This takes the form of articles around specific
climate change myths. Each article contains two boxes containing a summary of the misinforming
claim ("Climate Myths...") and the scientific correction ("What the science says..."). Detailed scientific
explanations, including links to peer-reviewed research, can be found in the main content of each
article, and some articles are provided in diferent languages with various levels of detail and complexity.
The Climate Disinformation Database7 (DeSmog) is also a relevant source of climate misinformation,
but contrary to SkS, it provides information about individuals and organisations spreading climate
misinformation rather than knowledge about common climate myths. A similar focus on individuals
and organisations is also followed by Hot Air.8 The CLIMATE-FEVER dataset [
Even though climate fact-checking data has been produced by fact-checking organisations (e.g.,
Climate Feedback10, Euro Climate Check11), fact-checkers tend to only cover specific misinformation
instance and do not provide scientific knowledge that can be easily reused outside the fact-check
context. Although fact-checking KGs such as ClaimKG [
Classification tools and eforts for categorising and identifying climate misinformation have been
created by researchers [
The Climate Facts Knowledge Graph (ClimaFactsKG) is a KG that integrates data from the SkS website, the CARDS taxonomy and the CimpleKG fact-checks. The following sections detail the methodology and steps involved in constructing ClimaFactsKG.
Finding a good source of information for scientific climate knowledge that can be easily associated with
common climate misinformation is key to empowering climate fact-checks and research on climate
misinformation. For constructing ClimaFactKG, we decided to rely on the data found on the SkS website
as it provides common climate misinforming claims with short countering scientific corrections and
CimpleKG [
A good categorisation of climate myths and corrections is needed for the easy identification of
potential corrections of climate claims and the thematic analysis of climate misinformation. Although
SkS provides a taxonomy of myths,15 its structure is not always synchronised with the claims found on
the website. As previously highlighted in Section 2, the CARDS project [
In this paper, we rely on the Schema.org vocabulary16 (denoted with the sc prefix in the rest of this document) and the sc:Claim and sc:ClaimReview concepts for representing the claims and scientific corrections found in SkS since they align well with the structure of the SkS website data. Another reason for using the Schema.org ontology is that CimpleKG is also based sc:ClaimReview. This makes the integration between ClimaFactsKG and CimpleKG simpler as both KG share the same data model. sc:ClaimReview is also the standard used by fact-checking organisations to represent and publish their fact-checks. This data is generally embedded in fact-checking articles as JSON-LD.17 By following the sc:ClaimReview format, ClimaFactsKG makes it possible to easily relate fact-check and claims to scientific evidence from the SkS.
The claims (sc:Claim) and corrections (sc:ClaimReview) found in the SkS pages are linked through the sc:itemReview relation. As SkS only provides corrections to misinforming claims, the rating of the claims is fixed to 0 via the sc:ratingValue property of the sc:Rating. The sc:Rating are then associated with the sc:ClaimReview using the sc:reviewRating relation. To ensure that these claims are never perceived as trustworthy, the sc:worstRating property is set to 0 and the sc:bestRating property is set to 1.
When multiple languages are available, language attributes are added to the textual attributes. If more than one version of the SkS article is found (i.e. complexity/explanation level), only the simplest 15Climate Myths sorted by taxonomy, https://skepticalscience.com/argument.php?f=taxonomy. 16Schema.org, https://schema.org. 17JSON-LD, https://json-ld.org. available version of the article is made available as sc:ClaimReview. In the future, we will look at making the other versions available in ClimaFactsKG.
The content and metadata of the article associated with the climate myths (e.g., keywords, publication date, authorship, license and related articles) are linked to the sc:ClaimReview using relevant properties and relationships (e.g., sc:keywords, sc:description, sc:abstract) from Schema.org. The main mappings between the SkS content on sc:ClaimReview can be found in Table 1.
The connection to CimpleKG is achieved via the CARDS taxonomy, so the fact-checks found in
CimpleKG can be easily linked to SkS concepts and scientific corrections. The CARDS taxonomy is mapped
using the Simple Knowledge Organization System (SKOS) [
The ClimaFactsKG data and code are made available on GitHub18 and the ClimFactsKG namespace is: https://purl.net/climatesense/climafactskg/ns#.
As previously discussed, ClimaFactsKG is built on top of the climate myths articles found on the SkS website. For collecting the data, we created a web crawler that extracts the content of the SkS articles in a structured JSON format and then converts it to the sc:ClaimReview format used by ClimaFactsKG (Section 3.2). Although we collect all the available complexity levels for each article, we currently only map the lowest available level of each climate myth (typically this is the Basic level, but some articles have only the Intermediate complexity level). Although the crawler is currently manually executed, it can be updated to periodically update the scientific climate claims data found on SkS when new 18ClimaFactsKG code repository, https://github.com/climatesense-project/climafacts-kg. claims are added to the website. The code of the crawler can be found on the GitHub repository of ClimaFactsKG.
After collecting the data, we use the Augmented CARDS classifier [
To connect the SkS myths with existing fact-checks, we first identify the sc:Claim from CimpleKG that are related to the climate using the Augmented CARDS classifier. The CimpleKG sc:ClaimReview to the corresponding sc:ClaimReview in ClimaFactsKG are then connected through the CARDS taxonomy. To create these links, we query the CimpleKG Sparql endpoint using the query described in Listing 1. We only focus on the English sc:ClaimReview due to the language limitation of the Augmented CARDS classifier. Since sometimes the text of sc:Claim in CimpleKG is not in English despite the sc:inLanguage property, we use the langdetect library19 to make sure that we only link to English claims.
PREFIX schema: <http://schema.org/> PREFIX cimple: <http://data.cimple.eu/ontology#> SELECT DISTINCT ?claimrev ?date_published ?text WHERE { ?claimrev a schema:ClaimReview; schema:inLanguage "English"; schema:datePublished ?date_published ; schema:itemReviewed ?claim .
?claim schema:text ?text . } ORDER BY DESC(?date_published) Listing 1: SPARQL Query used for identifying CimpleKG claims that may be connected to ClimaFactsKG.
After collecting the SkS data and linking it to the CimpleKG sc:ClaimReview using the CARDS taxonomy, we identify 611 climate claims in CimpleKG that can be linked to the ClimaFactsKG. Currently, ClimaFactsKG consists of 252 base climate claims and scientific corrections that are available in up to 26 languages (not all 252 base claims are available in all the 26 languages).
To make it easy to use and integrate ClimaFactsKG into various research and fact-checking workflows, we decided to base the design of the KG on the Schema.org ontology and sc:ClaimReview. Although this simplifies the integration of the data with existing KGs such as CimpleKG, we identified some limitations of Schema.org in representing more complex content. For example, even though we collect all the versions of each SkS climate myth, we currently only represent the simplest available version of each claim, as sc:ClaimReview does not provide mechanisms for representing complexity variations of the same content. In the future, we plan to expand the Schema.org ontology so that the various versions of the content can be accessed through ClimaFactsKG.
Although new SkS myths are not added often, currently, the crawling SkS is not fully automated. Similarly, the mappings between CimpleKG and ClimaFactsKG are not yet automated. This means 19Langdetect library, https://github.com/Mimino666/langdetect. that new fact-checks from CimpleKG are not mapped automatically as a new version of CimpleKG is released. We are currently working on automating such a task so ClimaFactsKG remains synchronised with CimpleKG.
Two other areas of improvement are the validation of the claim classification using the CARDS taxonomy and the extraction of scientific publications from the content of the SkS articles. Future work should formally evaluate the quality of the CARDS classification on the SkS data and the CimpleKG claims, as the number of retrieved CimpleKG claims appears to be low compared to the amount of claims available in the KG. Additional matching methods should also be considered, such as identifying similar claims between ClimaFactsKG and CimpleKG using embedding similarity or the creation of a new CARDS classifier based on new manual annotations. The future version of ClimaFactsKG will include the links to the scientific publications associated to the myth correction. This will provide additional means for querying the data via scholarly KGs and identifying additional relevant scientific resources that can be used for fighting and understanding climate misinformation.
Concerning the maintainability and long-term availability of ClimaFactsKG, the code and KG snapshots are publicly available on the KG repository. We also created a persistent URL for the data releases so that the data will always be accessible even if the SkS repository changes. Although we currently do not host a SPARQL endpoint, an endpoint can be easily created using the code found in the repository. Additional information about ClimaFactsKG availability can be found in the Resource Availability Statement Section of this article.
We introduced the ClimaFactsKG semantic resource, a KG that connects scientific climate evidence
extracted from the SkS website with fact-checks from CimpleKG [
This work was supported by the European CHIST-ERA program within the ClimateSense project (Grant ID ANR-24-CHR4-0002, EPSRC EP/Z003504/1).
In addition to the URLs previously mentioned in the paper, ClimaFactsKG can be resolved using the following two persistent URLs: 1) the source code and the releases are available at: https://purl .net/climatesense/climaf actskg; 2) the KG is available using the following URL and namespace: https://purl.net/climatesense/climafactskg/ns#. The data collected to create ClimaFactsKG is licensed under a Creative Commons Attribution 4.0 International License (CC-BY 4.0).20 This license is the most recent version of the license used by the Skeptical Science website (Creative Commons Attribution 3.0 International (CC-BY 3.0)). 20CC-BY 4.0, https://creativecommons.org/licenses/by/4.0/.
During the preparation of this work, the authors used Google Gemini in order to: Grammar, spelling check and rephrase sentences. After using these tools, the authors reviewed and edited the content as needed and take full responsibility for the publication’s content.