Appropriate quality data is data that is suited to the level of functionality, security, and longevity required to fulfil the purpose for which it is used [1]. Thus, it's not just about having data; it's about having quality data that is defined, transparent, and measurable to build trust, assure users and facilitate continuous improvement [1]. Data preparation, including normalizing, filtering, imputing missing values, detecting outliers, and harmonizing, is often necessary to improve the quality of data [2].

Mechanisms - how to embed Quality

Planned Information Management Framework

When we talk about digital twins, data quality is at the core of effective implementation. The report ‘An integrated approach to information management’ describes the Information Management Framework (IMF) as a formal mechanism that allows the availability of the right information at the right time to the right people, and that the quality of this information is known and understood [3].

A key technical component of the IMF is the development of a common language, consisting of the Foundation Data Model (FDM) and the Reference Data Library (RDL). These will enable a top-down and bottom-up approach for data management, so that they converge in the middle for a comprehensive system [4]. How can information quality for decision support be facilitated?

• Thorough analysis of related activities and decisions is paramount, and there is a need for a method to translate this analysis into clear, actionable information requirements [3]. The Foundation Data Model (FDM) within the IMF is set to/should provide a clear and implementable set of information requirements that support data quality [3].

• The RDL will provide a common reference for parties wanting to exchange data, offering a meaningful basis on which to map and interpret each other’s data. The RDL is not just a published part by the information management commons, but an evolving ecosystem of libraries linked together. The resulting ecosystem will form a hierarchical and integrated set of reference data libraries, each building on existing definitions and adopting the same community-wide information quality management practices.

In summary, the FDM and RDL will embed the Gemini Principle of Quality by establishing unambiguous meanings for data and using common terms and definitions across multiple systems and organizations and specifying the architecture and protocols for consistent implementations, similar to the World Wide Web [4].

Additionally, Integration Architecture, the second key technical component of the IMF, facilitates data to meet a certain threshold of quality, with major issues tackled at the source and minor errors potentially corrected automatically, as the report ‘Integration Architecture Pattern and Principles’ [5] highlights. Key considerations for the Integration Architecture to boost quality are:

• Data Quality Measurement: The architecture must provide a standard way to measure and report on data quality, recognizing that data quality can drift over time [5].

• Data Quality Improvement: Continuous monitoring and improvement of data quality through root cause analysis and preventive action are suggested [5].

• Data Quality Attributes: Accuracy, timeliness, completeness, and provenance are identified as important attributes for architectural design [5].

In summary, data quality is a key requirement for any data represented in the digital twin's models and ontologies. A digital twin is only as effective as the quality of the data input into it.

Standards

Connected digital twins involve bringing together a diverse range of stakeholders, each one operating with their own IT infrastructure and at different stages of data maturity. To enable the successful integration of these varied components, the adoption of recognized standards is crucial. These standards serve as a guideline to harmonize the process and facilitate a uniform approach towards data quality, security, and compliance.

The joining rules must be balanced to provide a low enough barrier to entry so as not to inhibit take-up, but high enough that data quality, security and compliance are not detrimentally affected [5].

One such standard, amongst others, is ISO19650: Managing information with Building Information Modelling (BIM). However, it is important to note that this standard may need to be upgraded to fit the specific requirements of connected digital twins [5].

Skills and Competencies for Quality

To establish a strong skill set within the team, competency scorecards presented in the Skills & Competencies Framework are a valuable tool. These scorecards will help identify skill and competency gaps, build cross-functional teams, and develop a resource plan and pipeline of skills needed over a specific time frame [7].

Several business skills as well as digital skills are relevant to the Gemini Principle of Quality. For example, a commercial mindset, which involves the ability to advocate for better data management and quality both internally and externally. Communication skills are key in understanding data management challenges and making a compelling argument for enhanced information management and data quality. Analytical skills and a fundamental understanding of data is necessary, including knowing what good quality data looks like, articulating its purpose and value, and recognizing how to leverage it to generate value and make informed decisions. In terms of digital skills, relevant skills include data validation, setting information requirements and governance, quality analysis and improvement, and process modelling [7].

Some roles which are central to the implementation and maintenance of the Gemini Principle of Quality are: Data Regulator, Data Consumer, Data Producer, Data Quality Analyst, Data Architect, Data Custodian, Data Steward, Data Leader, Process Modeller, Benefits Manager [7].

Ethical considerations

Quality, when it comes to the data that forms the foundation for digital twins, is not just about accuracy. It includes the value, transparency, and appropriateness of the data, as well as its use [2]. As the report on Digital Twins: Ethics and the Gemini Principles highlights, these characteristics of Quality can differ depending on the context [2]. This variance means that there is a need for further agreement on how the Gemini Principles should be applied in different scenarios [2], making the definition and assessment of Quality a critical aspect of this principle.

Data preparation, including normalizing, filtering, imputing missing values, detecting outliers, and harmonizing, is often necessary to improve the quality of data [2]. Without these, there is a risk of uninformed results, potentially leading to safety issues. However, the decisions made during data preparation can impact the outcome of the digital twin in ways that may raise ethical concerns.

Understanding how data is generated is also vital to Quality. There can be numerous sources of imprecision and bias related to data capture. Bias in data collection can present itself as an underrepresentation or overrepresentation of specific groups in the dataset, which risks creating issues of fairness and discrimination with a digital twin [2]. Therefore, selected data should be representative, relevant, accurate, and used to form generalizable datasets. Moreover, acknowledging and exploring the limits of the parameters of algorithms used in a digital twin for potential biases is essential [2].

By promoting the quality and ethical integrity of our data, we can make sure our digital twins serve their purpose effectively and fairly.

Examples

Case Studies

The case study outlined below demonstrate the practical applicability of digital twins in relevance to the Gemini Principle of Quality.

• National Underground Asset Register (NUAR) [6]: In the context of this case study, mechanisms such as data quality thresholds, measurement, and continuous improvement align with the Gemini Principle of Quality by making sure that the data is accurate, timely, complete, and has clear provenance.

Please see the DT Hub case study register (Case Studies - DT Hub Community (digitaltwinhub.co.uk) for further evidence of successful outcomes with digital twins.

References

[1] The Gemini Principles. Available at: https://digitaltwinhub.co.uk/files/file/12-gemini-principles/. Accessed March 18, 2024.

[2] Digital Twins, Ethics and the Gemini Principles. Available at: Digital_Twins_Ethics_and_the_Gemini_Principles.pdf (utwente.nl) Accessed March 18, 2024.

[3] An integrated approach to information management. Available at: An integrated approach to information management: Identifying decisions and the information required for them using activity and process models - Public Resources - DT Hub Community (digitaltwinhub.co.uk) Accessed March 18, 2024.

[4] The pathway towards an Information Management Framework - A ‘Commons’ for Digital Built Britain. Available at: The pathway towards an Information Management Framework - A ‘Commons’ for Digital Built Britain (cam.ac.uk) Accessed March 18, 2024.

[5] National Digital Twin: Integration Architecture Pattern and Principles. Available at: Integration Architecture Pattern and Principles - Public Resources - DT Hub Community (digitaltwinhub.co.uk) Accessed March 18, 2024.

[6] Gemini Papers: How to Enable at Ecosystem of Connected Digital Twins? Available at: The Gemini Papers - DT Hub Community (digitaltwinhub.co.uk). Accessed March 18, 2024.

[7] Skills and Competency Framework. Available at: Skills & Competency Framework - Public Resources - DT Hub Community (digitaltwinhub.co.uk) Accessed March 18, 2024.

Further Reading

• The Gemini Principles

• Our Vision for the built environment

• Data for the Public Good

• Flourishing systems

• Skills and Competency framework

• Pathway towards and IMF

• Digital Twins, Ethics and the Gemini Principles

• Digital Twin Navigator

• The Cyber-Physical Infrastructure Vision

• Our Shared Understanding

• Trust, Purpose, Value and Connected Digital Twins

• Gemini Papers: What are connected digital twins?

• Gemini Papers: Why connected digital twins?

• Gemini Papers: How to enable an ecosystem of connected digital twins?