Premises, Challenges and Suggestions for Modelling Building Knowledge using the Configuration Paradigm. Bart Deschoolmeester1,*,† , Elise Vareilles2,3,† 1 Lammekensstraat 25 - 2140 Borgerhout, Belgium 2 ISAE SUPAERO, University of Toulouse, France 3 IMT Mines Albi, Toulouse University, Albi, France Abstract This problem instance paper addresses the need for an industry wide modelling paradigm and language that allows the formalisation and representation of building knowledge by domain experts (architects, engineers). Herein, the special nature of the construction industry (e.g. its openness and semantics) in comparison to other industries and the complexity that arises from this, is recognised. The research needed covers a computation independent meta-model and accompanying modelling language and the added value of the knowledge-based configuration paradigm therein. The research outcome might spark renewed interest in an all-round universal knowledge representation language in the field of building information modelling (BIM) and even prove valuable for other ‘less complex’ industries. Keywords Knowledge Modelling, Building Sector, Configuration, Universal Language 1. Introduction the ’open’ nature specific to the building industry is pre- sented. In Section 3, the need to call some basic premises A modelling environment for the design, construction, of previous efforts into question is addressed. Section 4 in- operation and end-of-life of buildings, in which it is im- troduces the knowledge configuration paradigm and out- possible for the end user to make modelling mistakes lines the work of examining the possible benefits and chal- because of the integration of personal, company, stan- lenges of its application for building knowledge. Lastly, dardised and regulatory knowledge, has been envisioned possible further extension of the research is outlined in since at least 1999 [1]. In addition, the introduction of en- Section 5. vironmental, social, cost, organisational, etc. objectives would further automate the modelling process through optimisation. 2. Building Industry as an ‘Open’ While some attempts have been made in the field Industry of building information modelling, also named BIM, [2][3][4], the quest for a universal knowledge represen- The need for a universal knowledge representation lan- tation language has also been met with scepticism [1][5]: guage (or at least a common meta-model) and the re- claiming that immediate practical needs should be priori- search challenges this provides, arise from the fact that tised or even that this is not (yet) feasible. It can even be the building industry is possibly the most open industry argued that the field has adopted a pragmatic approach [1]: by focusing on information (as opposed to knowledge) [6], its translation from one environment to another [7], • Many parties are involved in a project and parties and constraint verification only after modelling [8]. Our change with every project. proposed research returns to an idealistic view, but finds • Vast numbers of manufacturers and products for it promising if based on revised conceptual foundations any building part (from traditional to innovative), and the knowledge-based configuration paradigm. on any scale (up to the building itself) are avail- The rest of the paper is as follows. First, in Section 2, able on the market. • Both a product directly and an onsite composition ConfWS’24: 26th International Workshop on Configuration, Sep 2–3, from products might provide a solution for a re- 2024, Girona, Spain quired part (e.g. a wall as prefabricated masonry * Corresponding author. or on site masonry). † These authors contributed equally. • Project specifications often don’t prescribe spe- $ bartdeschoolmeester@hotmail.com (B. Deschoolmeester); elise.vareilles@imines-albi.fr (E. Vareilles) cific products. € https://pagespro.isae-supaero.fr/elise-vareilles/ (E. Vareilles) • Product delivery might not include some parts but  0000-0001-6269-8609 (E. Vareilles) only list its requirements (called ‘open systems’ © 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). CEUR ceur-ws.org Workshop ISSN 1613-0073 Proceedings in this text, as opposed to proprietary, ‘closed’ physical item (called positions in this text): a win- systems’). dow can be open or closed, supports for raised office floors having an adjustable height or a ven- This openness is reenforced at a European level tilation unit with different flow rates. Therefore, through regulation (Construction Product Regulation[9], at least conceptually, properties must be thought public procurement[10]) and standardisation (CEN - Eu- of as potentially having different domains over ropean Committee for Standardization). This openness its parts, items, variants and positions. entails that most knowledge is generic and generally • Any level of abstraction should be allowed from available in ample building regulations and standards. the obvious generic concept ’door’, over ’parti- Designers, contractors and manufacturers refer to these tion’ (covering window, door, wall, floor, etc. ) up documents and generally only complement them with to a ‘building object’ concept. their specific requirements. • Innovative products exist for any building part The need for a common language for all the stake- and therefore must be expected: a generic concept holders is even more acute because of the challenges fac- should not be confined to its traditional meaning ing the construction industry: climate and environment, but allow almost unlimited heterogeneity. robotics, artificial intelligence, digital twins, etc. and this • The semantics of the aforementioned ‘position’ while facing a shrinking workforce (both engineers and can be further developed to also hold changes workers). like the onsite length adjustment of a beam, the removal, addition or replacement of a part (e.g. 3. Work Part 1: Basic Premises a filter change), or the different installation or use options of a product. With the addition of In light of the unsuccessful attempts to develop a uni- a ‘location‘ and ‘time’ property an item could versal knowledge representation language for the con- be tracked in space and time, with each change struction sector (see Section 1), it is necessary to first list being a new position. Thus covering the complete these experiments, examine their potential shortcomings life-cycle. and generate new ideas and approaches. Based on this • The semantics of the hierarchical relations be- work, it will then be possible to define the premises of a tween a concept and its parts and items respec- meta-model and its accompanying modelling language. tively, should not be confined to their traditional A preliminary examination already allows some un- definitions. A concept is primarily a generali- derpinnings of previous efforts to be called into question. sation of its items but this relation can have a Firstly, are existing attempts sufficiently intuitive? The part-like meaning through emergent properties sheer volume of available building expertise will necessi- like cardinality, overall cost, energy loss etc. Like- tate the creation, verification and maintenance of knowl- wise, a concept might have properties that are a edge models as a collaborative endeavour to be done by generalisation of the part properties: for example, domain experts (e.g. architects and engineers) directly a masonry wall concept enforces the same colour without a need for intermediaries like knowledge engi- domain for mortar and bricks. neers. • The ontology should be polyhierarchical (a single Secondly, are these efforts ontological sufficiently concept occurs in more than in one place) [11]: sound? Some examples of overlooked building ontology: for example, products exist that act as roof boards and roof insulation or the window grille is simul- • A building concept can play different semantic taneously part of the window and the ventilation roles: it can simultaneous be a conceptual ‘con- system. tainer’ of parts, items, variants and positions. For • Within the partonomy there is also a need for the instance, a window is composed of parts for its idea of ‘breakdowns’: different ways of breaking operation: generally, a frame, glazing(s) and hard- down a concept into parts. These ways can be ware. Yet, in a project, the concept might also disjunct (variants): for example, the choices for represent more than one window, for example, a the building structure might be frame-like (e.g. generalisation of the 4 physical windows (items) wood or steel) or mass-like (e.g. prefab concrete of the front facade. The concept might also ex- or masonry). Breakdowns can also be conjunct press the variants allowed in the specification (within a single variant): a building can be sub- (e.g. the designer allows freedom in the choice divided into its structure and total air volume or of hardware to the contractor) or offered by the into floors (with each floor incorporating part of product (a window available in different heights). the structure and air volume). Each breakdown Lastly, variability can also exist within a single (and its parts) can be needed for the representa- tion of knowledge or user requirements. Lastly, what is the universe of discourse of the at- Product Generic concept tempts? In any industry, knowledge is interconnected, User Specifi- Products but in the construction industry, due to its open nature, cation this is scaled to the entire industry. It might therefore be impossible to effectively isolate a particular aspect in a model while striving for its universal use. The work should therefore outline the contours of what constitutes Figure 1: Left: relationship between product and as building knowledge. user solution space in a traditional configuration task. Right: relationships between generic concept, project specification and products solutions space in a con- 4. Work Part 2: Applying figuration task for a building project part Knowledge-based Configuration challenges to the paradigm can already be identified. The knowledge-based configuration paradigm defines a Can configuration cover the needs resulting from the configuration model as a set of variables with their do- work of Section 3: the ontology, the domain of discourse mains and with product and user constraints limiting and will it be enough to allow domain experts to take on the possible combinations of variable values, and a solu- the role of knowledge engineers? A task resembling the tion (a configuration) as an assignment of single values work of [14]. to all variables consistent with the constraints (e.g. a Will the configuration paradigm be able to fully absorb valid configuration), as in Chapter 6 of [12]. Knowledge- the open character discussed in Section 2? based configuration is a matured and successful area of • The knowledge base will be incomplete. This artificial intelligence, used and integrated across many because of the amount of standards, products, industries for more than 40 years, as presented in Chapter etc. , the gradual nature of the design process or 1 of [12]. The configuration paradigm will feel intuitive confidentiality (e.g. pricing information). Also, and familiar for most building professionals: a (product tacit knowledge is prevalent with construction independent) specification as a solution space; a building parties. as a configuration; design choices as constraints; con- figurable products like drywall systems, roof systems, • As it is impossible for any product knowledge insulation systems. An intensional, declarative represen- base to contain all building products available on tation through domains and constraints might therefore the market, the user requirements (the project prove to be a good fit for construction knowledge specification) do no operate ‘within’ or on a sin- Another appealing aspect is the possibility of a repre- gle product knowledge base, cf. Chapter 6 of [12]. sentation that is non-causal, meaning that in a particular It is rather that both constraints defining multi- constraint which variables are input and which are out- ple products and user requirements operate in put need not be defined. Though the building modelling the knowledge base of the generic concept (e.g. a process is largely experienced as procedural, directional, generic window, door, wall, etc.) and it is the in- top-down, where decisions thought of as the most im- tersection of the specification and products solu- pactful, like the overall shape of the building, are taken tion spaces that represents the configurations that first and then gradually more detailed decisions are taken, provides a solution and this only for the known it is argued that this must not be imposed by the mod- products (see Fig. 1). elling environment. Light requirements might determine • The user should be presented only with valid op- the number and shape of windows instead of the other tions at any one moment in the modelling process. way around [13], or standard sizes of plywood sheets It is therefore not enough to solve for one valid determine the size of a construction to avoid waste [13]. solution but continuously for the complete valid In light of circularity, products available for reuse might solution space. This is especially necessary in a even become requirements instead of solutions. The up- multi-user environment, where parties operate coming practice of early involvement of all stakeholders in each other’s solution space. entails the registering of big and small requirements be- • Building industry knowledge is distributed. Not fore designing is started. only for product knowledge (different manufac- The knowledge-based configuration paradigm might turers) but also generic knowledge (building reg- even make the typical iterative design process obsolete, ulations and standards) is generated by differ- creating substantial savings. Though the knowledge- ent institutions at different geographical levels based configuration paradigm seems promising, some (municipality, country, EU level, etc.). Expect- ing all of them to formalise their knowledge on one location seems unrealistic. The product and [4] J.-K. Lee, C. M. Eastman, Y. C. Lee, Implemen- generic knowledge base will be distributed and tation of a BIM domain-specific language for the maybe also the project requirements base. Consis- building environment rule and analysis, Journal tency, verification and maintenance of distributed of Intelligent & Robotic Systems 79 (2015) 507–522. generic knowledge might seem especially chal- doi:10.1007/s10846-014-0117-7. lenging. [5] R. S. Weygant, BIM Content Development. Stan- • A solution is not always a product variant (a sin- dards, Strategies, and Best Practices, 1st. ed., Hobo- gle product item). A product item position (a spec- ken, NJ, 2011. ification might require a specific height for a sup- [6] F. Liu, A. K. Jallow, C. Anumba, Building knowledge port, yet a support adjustable in height might be modeling: integrating knowledge in BIM, in: Pro- acceptable), a product item part (order the whole ceedings of the CIB W78 2013: 30th International product to use only one of its parts) or product Conference – Beijing, w78, 2013, pp. 199–208. items combined (concrete from different suppli- [7] ISO 16739 Industry Foundation Classes (IFC) for ers for one single structure or products combined data sharing in the construction and facility man- as parts to make up the specified whole) might agement industries - Part 1: Data scheme, 2nd. prove to be equally valid solutions. ed., International Organization for Standardization, • In open systems, as defined in Section 2, the con- Geneva, Switzerland, 2024. straints for the not included parts of a (supply [8] A. Koutamanis, Planning regulations and modelled side) product might in effect be a product inde- constraints in BIM: A dutch case study, Buildings pendent (demand side) specification. Making it 44 (2024). doi:10.3390/buildings14040939. necessary to solve the product knowledge base [9] Regulation (eu) no 305/2011 of the european par- first. liament and of the council of 9 march 2011 laying down harmonised conditions for the marketing of construction products and repealing council direc- 5. Further Expansion of Research tive 89/106/eec, in: Official Journal of the European Union, 1st. ed., European Union, 2011, pp. 4–43. Once the conceptual foundation and configuration as a [10] Directive 2014/24/eu of the european parliament solution established, the research could be extended: and of the council of 26 february 2014 on public • As touched up in the introduction, a need for procurement and repealing directive 2004/18/ec, in: optimisation might arise. Official Journal of the European Union, volume 57, • New solving methodologies: computationally 1st. ed., European Union, 2014, pp. 65–242. doi:10. more efficient surrogate models might proof to 3000/19770677.L_2014.094. be more practical or the use of generative design [11] ISO 5127 Information and documentation — Foun- where the solution space is explored in an itera- dation and vocabulary, 2nd. ed., International Orga- tive process through single exemplary solutions. nization for Standardization, Geneva, Switzerland, • Propositions for domain expert and end user in- 2017. terface might result from the work. [12] A. Felfernig, L. Hotz, C. Bagley, J. Tiihonen, • New ways of knowledge acquisition like through Knowledge-based Configuration: From Research to voluntary open collaboration of domain experts Business Cases, 1 ed., Morgan Kaufmann Publishers or the use of artificial intelligence (large language Inc., San Francisco, CA, USA, 2014. models, natural language processing) to extract [13] K. M. Kensek, D. E. Noble, Building Information knowledge. Modeling. BIM in Current and Future Practice, 1st. ed., Hoboken, NJ, 2014. References [14] M. Mohammad-Amini, T. Coudert, É. Vareilles, M. Aldanondo, System Configuration Mod- [1] C. M. Eastman, Building product models : computer els: Towards a Specialization Approach, in: environments supporting design and construction, 10th IFAC Conference on Manufacturing Mod- 1st. ed., CRC Press LLC, Boca Raton, Florida, 1999. elling, Management and Control, volume 55, [2] B.-C. Björk, Requirements and information struc- Nantes, France, 2022, pp. 1189 – 1194. URL: https: tures for building product data models, Ph.D. thesis, //imt-mines-albi.hal.science/hal-03833708. doi:10. VTT Building Technology, Espoo, Finland, 1995. 1016/j.ifacol.2022.09.551. [3] B.-C. Björk, Ratas, a longitudinal case study of an early construction it roadmap project, Journal of Information Technology in Construction 14 (2009) 385–399.