Knowledge Transfer Partnerships (KTPs) link forward thinking businesses with the UK's world class academics to deliver innovation projects led by inspired graduates. CSols Ltd has been embedding knowledge within their company from academics at the University of Liverpool for eight years through two successful KTPs, and together they are pushing the frontiers of ontological engineering and the use of data within software development and data science.
Overview
CSols Ltd, a software company that offers solutions to improve efficiency in analytical laboratories, has been working with academics from the Department of Computer Science through two KTPs stretching across eight years. These partnerships represent part of a long journey, following a multi-decade strategy, for Phil Goddard, CEO of CSols Ltd, playing an important and essential role addressing specific parts of a larger puzzle. As Mick Card, Knowledge Transfer Adviser from Innovate UK KTN, describes, “CSols have worked very shrewdly and well with the KTP program; the offering of the KTP program is to assist businesses to work with academia in order to advance their own business, and CSols have embraced that well.”
Challenge
CSols Ltd works with analytical laboratories (those that measure the chemical composition of samples), part of a huge global market. The principle output for these laboratories is data. The data is typically produced by analytical instruments working in conjunction with robots and data systems (together these are called laboratory resources). Most obviously, these data are simply the results of whatever they are analysing, but also includes a vast wealth of other information about how the results came to be produced by the laboratory resources and the people working in the lab.
The overall challenge that CSols Ltd is addressing through a multi-year research and development plan is how to produce more effective software applications (LabApps) in a compellingly cost-effectively manner. These LabApps could potentially address any problem that might prevent a laboratory from working effectively and efficiently. It has traditionally been easy to imagine many LabApps that (if they existed) would revolutionise laboratories. It has been harder, and sometimes impossible, to produce these LabApps at an effective price/performance point.
One significant issue to address is how the LabApps can interface and network these laboratory resources and access the ‘hidden’ knowledge within the data. It requires software than can collate, save, retrieve and integrate (potentially huge amounts of) information whilst allowing multiple pieces of technology within a laboratory to speak with each other. Their KTPs with the University of Liverpool are tackling specific gaps within this overall challenge.
In their first KTP, the team developed nDrites, software wrappers that allow analytical machines and laboratory resources to interact seamlessly with LabApps regardless of the type of machine or manufacturer. This creates many new opportunities for a laboratory, for example systems can be built that allow laboratory resources not to work in isolation. Rather, they can operate in a connected manner across the wider laboratory. This, along with the many other possibilities nDrites provide, has the potential for dramatic operational and financial benefits for companies.
The next challenge for the team is around the management of the large amounts of data produced by laboratory analytical processes. It is important to be able to save and retrieve the data in a usable format that can be queried in the future. Developing a solution for this was a key aim for the most recent KTP with the University of Liverpool.
Solutions
Working with Dr Valentina Tamma and Professor Frans Coenen from the Department of Computer Science, the most recent KTP was to find a model and a process for accessing the analytical data produced by the networked analytical instruments and providing a structure, interpretation and context for the data to make it usable. This model is referred to as an ontology, and is akin to 'explaining’ the data, defining categories and relationships between data to develop a framework and giving context for the information, which adds value to the data. By understanding the data that CSols provides for every audience, they can ensure they are gathering the correct data and can present it in the most appropriate way. Having this background knowledge available means software is correctly designed to provide the right interpretation of incoming data from laboratory resources. It also paves the way for the application of machine learning to the data so as to extract hidden, but useful, knowledge.
The team began by working on data handling within CSols Ltd to ensure clear understanding of the relationships between each bit of data, employing a dedicated KTP Associate to lead the project. Their aim was to ensure they fully understood their data and the constraints on its usage from the perspective of their software, but also for what their customers required.
The team designed an ‘ontology-first’ approach and assessed CSols Ltd’s data along with creating a data governance strategy to be used by the company moving forwards. Having a clear ontology outlines how data relates to the rest of the business in addition to understanding all aspects of the data. For software, it allows regular optimisation, ensuring systems are operating at their greatest potential. Ultimately, a clear ontology will mean data can be queried by users, enabling better understanding of systems and resources.
Phil Goddard, CEO of CSols Ltd, outlined why the ontology research has been key for their company. “The most important thing to come out of this work in the future will be the updating and optimisation of software through deep, systematic refactoring that is guided by the ontology in steps that are operationally manageable.”
The ‘ontology-first’ approach allows regular development and improvements to existing software, but also takes information about the data that the software can collect and adds it to the ontology framework. In addition, the team have created glossaries that describe what the data includes and their relationships and context in an understandable format for software developers and stakeholders to use.
Benefits
This approach has meant that CSols Ltd can give their customers a complete understanding and new interpretations of their data, which means data can begin to become an increasingly key focus and asset for the company. By completing the research around the ontology framework through a KTP and managing it in-house, the company has saved time and money on a high-level academic process which would otherwise be too expensive for a SME to undertake.
“This is the next step in ontology engineering,” described Dr Valentina Tamma, “because even larger companies don’t get the time and resources to invest in building ontologies from scratch as it is a complex process involving a lot of discussions and documentations for their specific applications.” Legacy data becomes tidy data and can be used efficiently, and everything has been documented so that all new employees can adhere to data management process, and new data produced is easily documented and usable by all stakeholders.
The university has also benefitted by the opportunities for various undergraduate and Masters projects as part of the KTP, along with support from CSols Ltd with research proposals, such as that which received funding from the Office of Students for 30 scholarships for minority students to study data analytics.
Dr Valentina Tamma added, “In general, there is a trend in ontology engineering towards this ontology-light approach, positioning us in-line with other groups.” The use of ontological knowledge to add value and represent the meaning of data will be further expanded in a collaboration between Dr Tamma and Unilever, which is funding a PhD studentship, together with the EPSRC on the topic of knowledge graphs and ontologies for autonomous formulation.
The ontology engineering light approach adopted by CSols Ltd was presented at the Eleventh International Conference on Knowledge Capture (K-Cap), that took place in December 2021. KCap is a forum that aims to unite researchers from different areas of AI who are interested in efficiently and precisely capturing knowledge from a variety of heterogeneous sources, and in creating representations that can be useful for automated reasoning, analysis, and other forms of machine processing, as well as to support users in knowledge-intensive collaborative tasks. The conference is sponsored by the Association for Computing Machinery's (ACM) Special Interest Group on Artificial Intelligence.
The paper was a runner up for the conference best paper award, and received a Best Paper Honourable Mention as a "great example of the use of semantic knowledge capture in practice."
Back to: Collaborate with us