Periodic Reporting for period 1 - CLARUS (Building clarity and preventing bias in digital forensic examination, interorganisational communication and interaction)
Periodo di rendicontazione: 2023-11-01 al 2025-04-30
In the digital realm, forensic procedures walk a tightrope between justice and potential bias. Judgement bias and errors often seep into investigations through communication gaps, organisational cultures and linguistic disparities. Recognising this perilous landscape, the EU-funded CLARUS project brings together a consortium of six academic research organisations and four forensic and three police agencies from six European nations. The project aims to untangle the threads of bias in digital evidence examination. With a mission to chart ambiguous information flows, pinpoint sources of misunderstanding, establish a common lexicon and craft a bias-checking tool, CLARUS promises to fortify objectivity, neutrality and fairness in the pursuit of justice.
Clarus’ vision is: by detecting the organisational context for bias and errors in methods of informal and formal inter-agency communication, as well as the implementation of a common lexicon and bias checking tool, Clarus will improve objectivity, neutrality and fairness in the pursuit of safer justice outcomes in terrorism and serious crime exploiting digital technologies. This will enhance communication and improve efficiency and increase trust between citizens and the professional services in policing, forensic science and the courts. This will enable Clarus to contribute to the destination impact: "Improved forensics and lawful evidence collection, increasing the capabilities to apprehend criminals and terrorists and bring them to the court".
Project Objectives
• Mapping the systematic properties and practitioners’ perceptions of forensic contexts including organisational and professional structures, processes, relations, and cultures that may contribute to misunderstandings, biases, and errors and potentially undermine the impartiality and objectivity of forensic and investigative work.
• Analysing the patterns of interaction and communication across organisational, professional, and national borders through a hypothetical case study that will allow an in-depth examination of the sources of misunderstanding, errors and biases that may be evident in a digital forensic environment.
• Develop a common multilingual lexicon, verified by police and forensic science professionals, through the analysis of oral and written communication about digital evidence.
• Develop bias-checking tools/prototype for digital evidence and investigation through the assessment of the case study.
• Test and refine the common multilingual lexicon of digital evidence and investigation as a method of cross-border communication at all stages of the process, from citizens to the judiciary.
• Test and refine the bias-checking prototype/tool.
• Develop and disseminate project outcomes and training to identify sources of bias in organisational systems in digital evidence and investigation, to use the tools to identify bias and to use the common lexicon.
Clarus’ vision is: by detecting the organisational context for bias and errors in methods of informal and formal inter-agency communication, as well as the implementation of a common lexicon and bias checking tool, Clarus will improve objectivity, neutrality and fairness in the pursuit of safer justice outcomes in terrorism and serious crime exploiting digital technologies. This will enhance communication and improve efficiency and increase trust between citizens and the professional services in policing, forensic science and the courts. This will enable Clarus to contribute to the destination impact: "Improved forensics and lawful evidence collection, increasing the capabilities to apprehend criminals and terrorists and bring them to the court".
Project Objectives
• Mapping the systematic properties and practitioners’ perceptions of forensic contexts including organisational and professional structures, processes, relations, and cultures that may contribute to misunderstandings, biases, and errors and potentially undermine the impartiality and objectivity of forensic and investigative work.
• Analysing the patterns of interaction and communication across organisational, professional, and national borders through a hypothetical case study that will allow an in-depth examination of the sources of misunderstanding, errors and biases that may be evident in a digital forensic environment.
• Develop a common multilingual lexicon, verified by police and forensic science professionals, through the analysis of oral and written communication about digital evidence.
• Develop bias-checking tools/prototype for digital evidence and investigation through the assessment of the case study.
• Test and refine the common multilingual lexicon of digital evidence and investigation as a method of cross-border communication at all stages of the process, from citizens to the judiciary.
• Test and refine the bias-checking prototype/tool.
• Develop and disseminate project outcomes and training to identify sources of bias in organisational systems in digital evidence and investigation, to use the tools to identify bias and to use the common lexicon.
During the first 18 months, the CLARUS project made substantial progress across its technical and scientific objectives aimed at improving objectivity, neutrality, and fairness in digital forensic and investigative communication. Overall, the CLARUS project has delivered key analytical frameworks, validated data collection tools, and foundational technical components (including the lexicon and bias detection system) that will be refined and tested in the next phase. In specific, key achievements include:
* Completion of harmonized methodological guidelines for field research
* Development of system maps analyzing over 100 documents and highlighting potential sources of bias and miscommunication
* Generation of mental maps, draft finalized) based on 29 research diaries and interviews, reflecting practitioners’ perceptions of systemic bias
* Assessment of the impact on professional awareness via the Report on Professional Practice
* Full development and piloting of the case study protocol
* Completion of 19 focus groups and 75 surveys across five countries, now transcribed and translated
* Compilation of written and spoken corpora from WP1 and WP2 sources
* Extraction of high-frequency and domain-specific terminology using corpus linguistics techniques
* Engagement with practitioners to validate linguistic relevance and capture cross-border variability
* Establishment of a practitioner-validated typology of linguistic bias indicators
* Initial implementation of a neuro-symbolic model combining AI and rule-based systems for bias detection
* Construction of a synthetic annotated dataset using LLMs to support tool training and validation
* Completion of harmonized methodological guidelines for field research
* Development of system maps analyzing over 100 documents and highlighting potential sources of bias and miscommunication
* Generation of mental maps, draft finalized) based on 29 research diaries and interviews, reflecting practitioners’ perceptions of systemic bias
* Assessment of the impact on professional awareness via the Report on Professional Practice
* Full development and piloting of the case study protocol
* Completion of 19 focus groups and 75 surveys across five countries, now transcribed and translated
* Compilation of written and spoken corpora from WP1 and WP2 sources
* Extraction of high-frequency and domain-specific terminology using corpus linguistics techniques
* Engagement with practitioners to validate linguistic relevance and capture cross-border variability
* Establishment of a practitioner-validated typology of linguistic bias indicators
* Initial implementation of a neuro-symbolic model combining AI and rule-based systems for bias detection
* Construction of a synthetic annotated dataset using LLMs to support tool training and validation