search

alan-turing-institute / tric-dt

Open Project Pages for the TRIC-DT
Other
5 stars 0 forks source link

[Topic] Ontologies and Dynamic Knowledge Graphs for Digital Twins #23

Open ShenXiaoxue opened 7 months ago

ShenXiaoxue commented 7 months ago

Topic Ontologies and Dynamic Knowledge Graphs for Digital Twins

How is the topic relevant to the tric-dt themes? Consider at least two of the three themes health, natural environment and infrastructure

For natural environment, health, and infrastructure, the implementation of an integrated cross-sector digital twin platform is essential. Such a platform would serve as a critical link, bridging the existing gaps between these diverse sectors. Take, for instance, the realm of the natural environment, encompassing climate, flood, and geographic data. Currently, the segmentation of data and knowledge across these sectors impedes the ability to accurately predict system behaviors, due to a lack of consideration for the interplay among different sectors.

Structural dynamics in the context of enginnering infrastructure, e.g. buildings, bridges, transportation vehicles, typically involves numerical modeling, measurements, validation & verification, uncertainty quantification, model updating, etc., which assists in structural design, qualification, and control. Currently, these steps lack seamless integration due to the multi-user and multi-disciplinary nature of the process.

Introducing a cross-sector digital twin platform can bridge these gaps, thereby enhancing the overall efficiency and effectiveness of both predictive analysis and testing methodologies.. The foundation of this platform could be built upon ontologies and knowledge graphs. These elements are capable of articulating domain-specific knowledge and illustrating the intricate web of relationships and data interconnections. As such, they stand as pivotal research components in the development of Digital Twins.

Moreover, the Digital Twin platform should be engineered to facilitate comprehensive life-cycle management. A key research question in this domain involves the development of dynamic knowledge graphs. These graphs should possess the capability to autonomously identify and integrate the most current information within the system, ensuring continuous updates and relevance. This approach could revolutionise the way sectors interact and operate, leading to more informed decision-making and efficient resource utilization.

How does it relate to the wider topic of digital twinning?

Defining domain-specific knowledge and automating information updates are central challenges within the broader scope of digital twinning. Digital twins, essentially dynamic virtual representations of physical systems, rely heavily on accurately defined domain-specific knowledge to mirror the real-world environment effectively. This involves understanding the unique characteristics and requirements of the specific sector or system being modeled, whether it's in manufacturing, healthcare, urban planning, or any other field.

Furthermore, the power of a digital twin lies in its ability to stay synchronized with its physical counterpart. This synchronization depends on the system's capability to automatically update itself with the latest data. This ongoing update process is crucial for ensuring that the digital twin remains a relevant and reliable tool for decision-making, predictive analysis, and optimization.

In the wider context of digital twinning, addressing these questions isn't just about technical implementation; it's about ensuring that digital twins can provide accurate, real-time insights and forecasts that are applicable and valuable to their respective domains. The integration of technologies like IoT, machine learning, and advanced data analytics plays a vital role in this process, making the digital twin a powerful tool for innovation and efficiency across various sectors.

Suggested speakers or contributors

  1. Jethro Akroyd. Department of Chemical Engineering and Biotechnology, University of Cambridge. Topic on "Ontologies and Knowledge Graphs for Digital Twins"
  2. Jiaru Bai. Department of Chemical Engineering and Biotechnology, University of Cambridge. Topic on "Dynamic Knowledge Graphs"

Any resources you can recommend on this topic? Drop any links or references here that could help other people relate this to their own domains

  1. CReDo technical paper 1: Building a cross-sector digital twin. https://doi.org/10.17863/CAM.81779
  2. A derived information framework for a dynamic knowledge graph and its application to smart cities. https://www.sciencedirect.com/science/article/pii/S0167739X23003825

What format do you think would serve this topic best? e.g. seminar series, but in the future we might want to organize co-working or discussion sessions

Seminar series.

kallewesterling commented 7 months ago

Thank you for opening this issue @ShenXiaoxue! This looks really great, and I am looking forward to this seminar! :)

aranas commented 7 months ago

@ShenXiaoxue, @mhauru and I had a great little chat today about ongoing plans regarding knowledge graphs for DTs across our respective teams. Some Highlights:

During our discussions a couple of questions came up:

  1. Which query languages are appropriate when working with knowledge graphs and how do the different languages differ?
  2. How easily can knowledge graph DBs interface or be integrated in SQL DBs?
  3. Generally, what are the best practices for storing knowledge graph time data, what are possible backend solutions for this, what are pitfalls to avoid to make the code sustainable?

resources & tools mentioned: