In this guest post, Mehmet Yalcinkaya (PhD Candidate) and Vishal Singh (Assistant Professor - both at Aalto University, Finland) introduce their innovative approach to project information visualisation. Their background research is solid and the tool they've developed is very promising:
It is astounding that not many of us realise that we spend nearly one year of our life looking for misplaced objects. Yes, one year lost in looking for misplaced objects! No wonder, the same problem plagues us in our professional lives as well. On average, AEC-FM professionals and technicians spend 46% of their time looking for the information they need (Juhla and Suvanto 2015). That is, when we are seeking some input to base our actions on, nearly half of the time we can’t find the “right information”. Of course, this is also true of data and digitised information. So we know there’s a problem in locating information, but is there a solution to this?
In this short post, we will summarise our research covering key pain points - information management challenges that our industry is suffering from. We will also introduce a few basic principles that we’ve adopted, and a digital solution that we’ve developed, to address these pain points.
What is the problem?
The effective access, delivery and management of digital facility information has remained a major challenge over the past few decades. Despite the increasing digitalisation of the AEC-FM sector, and the proliferation of Information and Communication Technologies (ICT) and Internet of Things (IoT) solutions, challenges related to information management continue to expand and grow. Project-related information is vast, diverse, fragmented, and widely distributed across different sources, systems and actors. More often than not, we – the information managers - assume that those who need specific project information would be able to easily search diverse information sources and make sense of what they find. We also assume that everyone who needs information knows where information they seek is stored, and how to seamlessly navigate between vastly different data sources and data structures. We also assume that, in order to support data integration, we only need to create standards, enforce their adoption top-down, and then everyone will magically understand what to do and when to do it! In all these assumptions, we are actually overly reliant on the information seeker’s ability to find information, especially if we’ve not clarified where information is stored and how information retrieval need to be handled. So when these poor information seekers can’t find the information they need, many of us tend to either blame the information standards, the individual seeker, or the software systems used. Many of us may attempt to remedy this by creating new information standards or developing new information management tools. In doing so, we tend to forget that, without addressing the root causes, the problem is bound to persist and we’ll find ourselves facing the same challenge, over and over again.
How can we address this challenge?
The most important way to address the aforementioned challenge is to question the many underlying assumptions we’ve discussed above. In this post, we propose two key approaches: first, we need to focus on the information seekers, understand their decision making processes, and the cognitive aspect dictating how information is being parsed. Second, we need to acknowledge that current information standards and information structures have merit and do not need to be replaced or replicated. So rather than trying to create new standards, we need to focus on making them easier to understand and follow by non-experts.
Putting the approach to test: from concept to implementation.
To test our aforementioned approach, we’ve built a data aggregation and visual search platform - called VisuaLynk – using a network/graph-based data infrastructure. In building this platform, we avoided capturing information entities in a siloed manner as most current IT systems do. That is, most project information systems try to capture the whole lifecycle project information in files, folders and digital repositories. Such an approach cannot capture the relations between the information entities, cannot clarify the hidden interactions within the network, and thus these relations and interactions would remain misunderstood and underutilized.
To counter this, instead of capturing data in a static and siloed system, we’ve adopted a dynamic, network-based configuration to display relations between information entities (Figure 1).
To further improve the user’s experience, we introduced an interactive graph-based visualization and a visual search - a knowledge map of the linked facility data. This visual approach allows information seekers to more easily find the information they need and understand the relation between information entities (Figure 2). Also, the knowledge map simplifies the data structure that can correspond – in theory – to any information standard. For example, we’ve developed VisualCOBie, a graph visualization of the COBie standard (Construction Operations Building information Exchange – refer to Yalcinkaya & Singh 2016a) which makes COBie data structure visual and thus easier to understand by the non-expert user. Our preliminary research trials attracted much the positive feedback from industry partners in Finland. The key finding is that interactive graph visualizations of standardised data structures is a good way to clarify information standards to non-experts, and to increase their acceptance and adoption rates. Based on these positive results, it may be worth investigating to take trial this on more complex data structures and possibly develop a VisualIFC module or similar.
There is a limit to the amount of information that people can process. Efficient documentation and effective representation of data is critical to improve communication and collaboration between people. This has always been the best approach (to date) to deliver structured project information through CAD/BIM files, PDF documents, spreadsheets, emails, etc. However, despite the importance of documentation to exchange project/facility information, the knowledge and experience of individuals cannot be stored as documents. Even basic knowledge similar where information is stored, how to read different data formats, and how to build connections/links between stored data and documented information across different sources, remain stubbornly tacit (hidden) within the mind of the individual information seeker.
To address this, we need to develop a new kind of data aggregation platform which – to a degree – can act as a Transactive Memory System (Wegner 1995). That is, if the system can represent how information is connected, information seekers – a project team for example – can more easily locate the information they need, when they need it.
So, instead of categorizing project information into disconnected silos (files and folders), we’ve managed to create a visual layer that (a) uncovers existing data structures, and (b) clarifies many of the relations between these structures across the project’s whole lifecycle. The relations uncovered include (c) those between a single individual (project actor) with other individuals, and (d) those between individuals and information entities. By making these relations explicit, we have come closer to building a ‘transactive memory system’ that – not only represents the original hierarchical relations between information entities but also – shows many alternative paths between entities (Yalcinkaya & Singh, 2016b).
To achieve such a visual memory system, we developed a knowledge graph/network using a linked data approach. The integration between different data sources and the aggregation platform is established via API (application programming interface) connections. In addition, the integration of BIM/IFC files is established via the semantic web technology by conversion of IFC file to resource description framework (RDF) format (Hoang & Seppo 2015). Therefore in our approach, instead of trying to collect all data in a single repository, we’ve opted to aggregate and semantically link existing data from across various sources. Such an approach has also been adopted by BuildingSmart and the Linked Building Data Community Group, which are currently conducting much research activity around BIM and Web of Data.
Since we are focusing on the individual user, we go a step beyond the back-end linked data capability to provide a user-friendly and interactive interface that also explicitly shows the relationships between the linked data. The User Interface presents the data structure, data connections and spatial organization in a visually-appealing, space-saving node-link diagram.
Such visualization facilitates navigation between information entities using pre-attentive figures (symbols) assigned to information entities (nodes) which are directly related (links) to individuals and/or other information entities (Figure 3).
The node-link visualization is also able to connect bi-directionally with interactive 2D floor plans and 3D BIM models thus allowing information seekers to interrogate specific spaces, equipment and components. This ability to concurrently navigate data structures using varied representations (2D, 3D, tree-structure, node-link diagram) enhances user experience.
Figure 3. Representation of a graph-based data model with node-link diagram
In summary, our research has addressed the ‘information management’ challenge in two ways: focusing on the individual information seeker, and enhancing the usability of data aggregated from different sources. This has been accomplished by applying the following key principles:
- Developing a transactive memory system (TMS) which allows project team members to locate information in multiple formats;
- Developing a semantic web of data residing in different databases;
- Developing a Visual management and communication system for enhanced user experience and reduced cognitive load;
- Showing explicit relations between information entities/sources; and
This post reflects our early research findings with more empirical evidence currently being collected through pilot projects. However, these findings are very promising as they suggest that visual communication, explicit representation of relations between information entities, and server-side linked-data approach can markedly (a) improve user experience, and (b) increase adoption of information systems!
Hoang, Nam Vu, and Seppo Törmä. "Implementation and Experiments with an IFC-to-Linked Data Converter." https://itc.scix.net/data/works/att/w78-2015-paper-029.pdf
Tuuli Jylhä , Maila Elina Suvanto , (2015) "Impacts of poor quality of information in the facility management field", Facilities, Vol. 33 Iss: 5/6, pp.302 - 319
Wegner, Daniel M. "A computer network model of human transactive memory." Social cognition 13.3 (1995): 319.
Yalcinkaya M. & Singh V., (2016a). Evaluating the Usability Aspects of Construction Operation Building Information Exchange (COBie) Standard. CIB World Building Congress, At Tampere, Finland https://www.researchgate.net/publication/303811016_Evaluating_the_Usability_Aspects_of_Construction_Operation_Building_Information_Exchange_COBie_Standard
Yalcinkaya M. & Singh V., (2016b). A Visual Transactive Memory System Approach Towards Project Information Management, [Manuscript in review]
This study is a part of an ongoing research at Aalto BIM Collaboration led by Dr. Vishal Singh (@SinghV_Aalto), Assistant Professor at the Department of Civil Engineering, Aalto University, Finland. To read peer-reviewed paper related to this post, please see the references above. Additional and upcoming articles related to this blog can be found through the author’s public research profiles.
This is currently an ongoing research and development project, requiring improvements and revision of both on theoretical and practical aspects. For any inaccuracies, inquiry for further discussions and/or comments/suggestions, please contact the researcher directly.
We would like to thank Dr. Bilal Succar for this opportunity to publish this research through BIM ThinkSpace. We would also like to thank Senior Researcher Seppo Törma and PhD Candidate Nam Vu Hoang for their collaboration and intensive support to this research.
PhD Candidate at Aalto University, Finland
Mehmet is a PhD Candidate and Researcher at Aalto BIM Collaboration where he is currently investigating a number of topics including: Facility Management (FM) practices, computational and usability aspects of BIM-based FM solutions, and how standardization processes affect industrial and technological knowhow. His PhD study resulted in the development of a data aggregation and visual search platform called VisuaLynk. Mehmet has previously worked on the DRUMBEAT research project and is currently the project manager for the DigiBuild research project. He has industrial experience in large-scale residential and infrastructure construction projects in both Turkey and the USA.
Assistant Professor at Aalto University, Finland
Vishal Singh is the professor of Computer Integrated Construction (Building Information Modeling) at the Department of Civil Engineering, Aalto University, Finland. He leads the BIM research program and works closely with both industry and national/international research partners. Vishal’s research focuses on the interaction of products, processes and people - aiming to understand the decision making processes, and developing tools and methods to support this decision making. Vishal’s research combines his expertise in the areas of design thinking and computational thinking with a focus on BIM, multidisciplinary collaboration, and digitally-enabled innovation across the AEC-FM industry.