Quality evaluation of free text information in the DSA Transparency Database Article 17 & 24(5) of Regulation (EU) 2022/2065

This opportunity is a planned negotiated procedure initiated by the European Commission's Directorate-General for Communications Networks, Content and Technology (DG CNECT). The objective is to develop a methodology and user-friendly tool for evaluating the quality of free-text data submissions to the DSA Transparency Database under Articles 17 and 24(5) of Regulation (EU) 2022/2065, known as the Digital Services Act (DSA). The contract will focus on assessing the clarity, specificity, and relevance of content moderation statements made by online platforms.

Eligible applicant types include any qualified entities, such as companies, research institutions, or consortia with expertise in machine learning and data quality assessment. The funding mechanism represents a procurement contract for services rather than a grant. The approximate budget for the contract is capped at €142,000, with a project duration of 6 months.

The tender does not explicitly require a consortium to apply, suggesting that single applicants can submit bids. However, current EU procurement rules generally allow for participation from entities based in EU member states, EEA countries, or associated nations, even though no specific geographic restrictions are mentioned. The target sector encompasses information and communication technology, focusing heavily on digital policy and online platform governance.

The project is categorized as being in the development and validation stage, requiring the establishment of an information quality assessment framework. This involves training machine learning models to automate the evaluation process and creating a graphical user interface to facilitate usability and result visualization.

The application process involves expression of interest and is not categorized as a traditional call for tenders; rather, it indicates an intent to move towards a negotiated procurement process. The deadline for expressing interest is July 31, 2025, with the expected launch date of the negotiated procedure set for September 8, 2025.

Overall, this procurement contract aims to support the European Commission's enforcement of the DSA by enhancing the quality of data submitted to the DSA Transparency Database. This includes generating better compliance monitoring tools for content moderation decisions across online platforms.

This is a publication announcing the European Commission's Directorate-General for Communications Networks, Content and Technology (DG CNECT) intention to publish a future negotiated low or middle value procedure, not a call for tenders. The procedure identifier is EC-CNECT/2025/MVP/0069-EXA.

The objective of the support contract is to assess the information quality of providers’ submissions to the free-text data fields of the DSA Transparency Database at scale, to improve the data quality of the submissions to this database and to support the Commission’s DSA compliance monitoring work. The goal is to develop a methodology for such an assessment and a user-friendly tool with which it can be performed in an automated manner, using a machine learning-based approach to support the systematic monitoring of the data quality of the DSA Transparency Database in multiple quality dimensions, including clarity, specificity, and relevance.

Article 17 of Regulation (EU) 2022/2065 (DSA) obliges providers of hosting services to inform users when they moderate their content and sets out the “clear and specific” information they need to share with users, in particular with regards to explaining how and why their content was moderated, in so-called statements of reasons. Article 24(5) obliges providers of online platforms to submit anonymised versions of such statements of reasons to a publicly accessibly database hosted by the Commission (The DSA Transparency Database), to establish a public track record of the content moderation decisions taken by online platforms for the monitoring of illegal and harmful content online. The data fields which providers of online platforms, including of Very Large Online Platforms (VLOPs) and Very Large Online Search Engines (VLOSEs) need to fill in to complete their submissions are publicly documented (API and Schema - DSA Transparency Database) and have been aligned with the content format required for the DSA’s transparency reports by Commission Implementing Regulation (EU) 2024/2835 as of 1 July 2025.

The tenderer will develop an information quality assessment framework, including by defining relevant indicators of information quality and completeness (e.g. text complexity, recurring patterns, clarity, reference to the relevant Terms of service section) against which the free text field submissions of the DSA Transparency Database are evaluated and scored. To detect recurring patterns and templated texts used by providers, pattern and part of speech recognition should also be applied. To apply this methodology at scale, the tenderer will train relevant Machine Learning models (including for example a classifier or regressor and a speech model to reconstruct the decision trees of statements of reasons) which can perform such an assessment at scale. To allow for user friendly interactions with the models and to communicate the assessments produced in an understandable and user-friendly manner, the tenderer should develop a graphical user interface which delivers the assessment in the form of user-friendly per-platform visualisations, e.g. in a dashboard interface or through the generation of comparative reports.

Since accurate and granular data from the DSA’s transparency reporting tools, including the DSA Transparency Database, serve as an input into and compliance indicator in many other DSA areas (e.g. systemic risks, protection of minors, online marketplace safety), such a methodology and tool can produce valuable inputs supporting the Commission’s DSA enforcement work also beyond its content moderation transparency provisions.

The duration of the contract is 6 months. The maximum amount is 142,000.00 EUR. The procedure type is a planned negotiated procedure for a middle/low value contract. The estimated total value is 142000 EUR. The lead contracting authority is the European Commission, DG CNECT - Communications Networks, Content and Technology. The main classification (CPV) is 73000000 - Research and development services and related consultancy services. The nature of the contract is services. The maximum contract duration is 6 months. There is no framework agreement.

The deadline for expression of interest is 31/07/2025 04:59 Europe/Brussels. The indicative date of launch of the negotiated procedure is 2025-09-08 Europe/Brussels. Submissions must be sent exclusively at the address for submission given below. The method of expression of interest is electronic. The address for expression of interest is "Express interest".

A question in the Public Q&A section asks: Where can I find the procurement documents for calls for tenders with ‘ExA' in the reference? The answer is: References that feature an ‘ExA’ are not calls for tenders. They are a publication announcing the contracting authority’s intent to launch in the future a low or middle negotiated procedure. This was published on 03/10/2024 16:32.

In summary, this is an announcement of a forthcoming opportunity for a contract to develop a methodology and tool for assessing the quality of free-text information submitted to the DSA Transparency Database. The project involves creating an information quality assessment framework, training machine learning models, and developing a user-friendly graphical interface. The goal is to improve the accuracy and reliability of data related to content moderation decisions by online platforms, supporting the European Commission's efforts to enforce the Digital Services Act (DSA). The project has a duration of 6 months and a maximum budget of 142,000 EUR. The deadline for expressing interest is July 31, 2025, and the negotiated procedure is expected to launch on September 8, 2025.

Eligible Applicant Types: The text refers to "tenderer" which suggests that the eligible applicant types could include companies, research institutions, or other organizations capable of providing the required services. The tenderer will develop an information quality assessment framework, including by defining relevant indicators of information quality and completeness (e.g. text complexity, recurring patterns, clarity, reference to the relevant Terms of service section) against which the free text field submissions of the DSA Transparency Database are evaluated and scored. To detect recurring patterns and templated texts used by providers, pattern and part of speech recognition should also be applied. To apply this methodology at scale, the tenderer will train relevant Machine Learning models (including for example a classifier or regressor and a speech model to reconstruct the decision trees of statements of reasons) which can perform such an assessment at scale. To allow for user friendly interactions with the models and to communicate the assessments produced in an understandable and user-friendly manner, the tenderer should develop a graphical user interface which delivers the assessment in the form of user-friendly per-platform visualisations, e.g. in a dashboard interface or through the generation of comparative reports.

Funding Type: Procurement, specifically a service contract. The text mentions "support contract" and "planned negotiated procedure for middle/low value contract".

Consortium Requirement: The text refers to "the tenderer" in the singular, but it does not explicitly exclude consortia. Therefore, it is possible that either a single applicant or a consortium could apply.

Beneficiary Scope (Geographic Eligibility): The opportunity is published by the European Commission, suggesting that eligible applicants should be based in the EU or countries associated with EU programs. The text does not explicitly define geographic eligibility.

Target Sector: The program targets the ICT sector, specifically focusing on data quality, content moderation, transparency, and the application of machine learning in the context of the Digital Services Act (DSA).

Mentioned Countries: No specific countries are mentioned, but the context is the European Union due to the reference to the European Commission and the Digital Services Act (DSA).

Project Stage: The project stage is development and implementation. The goal is to develop a methodology and a user-friendly tool for assessing the information quality of submissions to the DSA Transparency Database. This involves developing an information quality assessment framework, training machine learning models, and creating a graphical user interface.

Funding Amount: The maximum amount is 142,000.00 EUR. The procedure is described as a "planned negotiated procedure for middle/low value contract" with an estimated total value of 142,000 EUR.

Application Type: Planned negotiated procedure. The text indicates that this is not a call for tenders but a publication announcing the contracting authority’s intention to publish a future negotiated low or middle value procedure. An expression of interest is required.

Nature of Support: The beneficiaries will receive money in exchange for services. The opportunity is a service contract where the tenderer will be paid to develop a methodology and tool.

Application Stages: There are at least two stages: expression of interest and a negotiated procedure. The text mentions a deadline for expression of interest and an indicative date for the launch of the negotiated procedure.

Success Rates: The success rates are not mentioned in the text. As this is a negotiated procedure, the success rate will depend on the number of candidates invited to negotiate and the quality of their offers.

Co-funding Requirement: The text does not mention any co-funding requirement.

Summary:

This opportunity is a planned negotiated procurement procedure initiated by the European Commission, DG CNECT, related to the Digital Services Act (DSA). The goal is to improve the data quality of the DSA Transparency Database, which collects anonymized statements of reasons for content moderation decisions from online platforms. The Commission intends to contract a tenderer to develop a methodology and a user-friendly tool for assessing the quality of free-text data submitted to the database. This involves creating an information quality assessment framework, training machine learning models to automate the assessment, and developing a graphical user interface for visualizing the results. The maximum budget for this 6-month contract is 142,000 EUR. Interested parties must first express their interest electronically. This is not a call for tenders but an announcement of a future negotiated procedure for a middle/low value contract. The deadline for expressing interest is 31/07/2025, and the indicative date for launching the negotiated procedure is 2025-09-08. The project aims to support the Commission's DSA compliance monitoring work and improve the accuracy and granularity of data used in various DSA areas, including systemic risks, protection of minors, and online marketplace safety.