The 38th ISARC Symposium was held in Dubai from November 2-4. ISARC is the leading organization dedicated to promoting automation and robotics in the construction industry. Xuling Ye, Phillip Schönfelder, Daniel Napps, Dennis Pawlowski and Markus König attended digitally and presented their papers. 

The paper titled "Deep Learning-Based Entity Recognition in Construction Regulatory Documents" by Phillip Schönfelder and Markus König deals with automated rule checking of construction designs. For this purpose, design rules from regulatory documents have to be converted into a machine-readable format. A first step in this process is the recognition of key terms in these texts. Therefore a deep-learning transformer model is trained, which can then extract relevant terms from German-language regulatory documents. Within the scope of the paper, a training dataset was created, which contains about 2500 sentences from building regulation documents and is annotated with named entity tags.
Link to the paper 

The paper "From the graphical representation to the smart contract language: a use case in the construction industry" by Xuling Ye and Markus König deals with graphical construction process in the construction industry using smart contracts by proposing a method from graphical representations to smart contract languages. Smart contracts self-execute through "if-then" rules, which are the program for automated process execution. The use of blockchain-enabled smart contracts is not yet widespread in the construction industry. One reason for this is the research assumption that users such as contractors can program, which is generally not the case. Prefabricated smart contracts usually do not meet the different and specific requirements of the actors, therefore an interface solution is proposed in the paper. Graphical workflow notations (e.g., BPMN, YAWL) could be translated into Smart Contracts programming languages (e.g., Slidity, Vyper) to simplify the creation of a Smart Contract and make it possible for non-programmers. This could also improve the dissemination of this method.
Link to the paper

The paper by Daniel Napps, Dennis Pawlowski and Markus König is entitled: "BIM-based variant retrieval of building designs using case-based reasoning and pattern matching" and deals with the outline of a solution for BIM-based variant retrieval of building designs in early design phases.
The use of 2D plans is still widely used today, even though they are, for example, very difficult to read for people outside the field and comparability between plans can sometimes be very difficult. The use of BIM models, with the help of the IFC data structure, is a good approach, for example, to compare similar floor plans of existing buildings and to learn from old mistakes. Currently, information such as the respective experience of the architects is not available and is also very difficult to transfer across projects. One solution for finding variants of building designs can be Case Based Reasoning (CBR) and Pattern Matching (PM), which are based on the graph structure of BIM models. This can enable the selection of suitable variants for buildings in early design phases from a large pool of variants, thus providing architects with important inspiration and impulses that can also be integrated into various (building) analyses.
Link to the paper
 

Phillip Schönfelder, M. Sc. (more...)

Yuling Ye, M. Sc. (more...)

Daniel Napps, M. Sc. (more...)

Dennis Pawlowski, M. Sc. (more...)

Markus König, Prof. Dr.-Ing. (more...)

Link to the conference