
RDF-based signature algorithms for computing differences of IFC models
Jyrki Oraskari
⁎
, Seppo Törmä
Department of Computer Science, Aalto University, School of Science, P.O. Box 15400, FI-00076 Aalto, Finland
abstractarticle info
Article history:
Received 1 November 2014
Received in revised form 29 May 2015
Accepted 30 May 2015
Available online 15 June 2015
Keywords:
BIM
RDF
IFC
Diff
The capability to accurately detect changes between successive versions of the IFC representation of a BIM
model would enable the development of generic change management functionalities for construction projects.
Unfortunately, IFC models consist mostly of anonymous objects without stable identities; therefore the
computat ion of differences between versions is complica ted. When IFC models are converted into an RDF
repr esentation, the uniform graph structure offers new algori thmic opportunities. We study how to assign
unique and stable identities to anonymous nodes based on signatures generated from their graph environment,
and present the Short Paths Crossings Algorithm (SPCA) that computes sets of paths with limited length from
anonymous nodes taking into account their crossings. Empirical tests show that SPCA produces significantly
smaller difference sets for IFC-derived graphs than previous algorithms for RDF change detection.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
Large building projects nowadays use building information modeling
(BIM) to produce models of different aspects of building: architecture,
structures, plumbing, heating, air-conditioning, and so on [1]. Multiple
interrelated models are produced in individual design tasks by different
parties with discipline-specific tools. However, the overall design
processes are characterized by frequent changes and the need to man-
age the impact of the changes on interrelated models. Tra ditional
change management practices require repeated comparisons of models
and manual work from designers. To enable more agile and automated
change management, information about detailed relations between
entities in different models would be useful [2].
Solutions for in terlinking entities across different models in a
standard-based manner using the Web of Data technologies are in
development [3]. BIM models can be exported in industry foundation
class (IFC) format [4] and converted into resource description frame-
work (RDF) [5]. Links can then be generated between the URIs of
entities across the models to support cross-model information access
and change management (Fig. 1).
An important enabler to efficiently manage changes between inter-
related models is the capability to accurately detect changes from one
version of a model to the next. These kinds of changes are typically
called diffs and they are represented as sets of deletions and additions
between versions. If correct di ffs can be computed, the objects that
have changed can be determined; if these are linked to objects in
other models, change requests or notifications can be produced [2],
possibly in the form of BIM Collaboration Format (BCF) messages. In
addition, diffs can be used to m anage the change history of a model,
for instance, to undo changes or to carry out analyses.
However, computing diffs between IFC model versions is not
straightforward. Only a small minority of instances are identified with
globally unique identifiers (GUID) [6,7] and most instances are
anonymous:theyareidentified only with line numbers but unfortunate-
ly in IFC files the line number of an instance is almost always different in
each version. Without stable identi fiers it is difficult to trace objects
between ve rsions, and consequently determine which objects have
changed.
The problem of anonymous nodes remains when an IFC model is
converted into an RDF graph, where the anonymou s objects are
represented as blank nodes, the RDF equivalent to anonymous objects.
However, the uniform graph structure of RDF creates new opportunities
to produce unique and stable identities for objects by collecting charac-
teristic information from their unique environment in the graph.
The goal of this research is to develop efficient algorithms for computing
diffs between successive versions of IFC-derived RDF graphs. The diffs have
to be correct (contain the whole change) and they should be close to
mini mal (contain only the change). Because diffs of RDF graphs are
trivial to compute in the absence of blank nodes , the approach is to
compute unique and stable identities for all blank nodes, i.e., to assign
the same identities to the same objects in different versions. There are
currently numerous BIM tools in widespread daily use; the presented
algorithms do not require any modification s to those tools but work
on the b asis of the IFC standard and W3C-specified Web of Data
technologies.
As the main result we present the Short Paths Crossings Algorithm
(SPCA) which determines the identity of each blank node by computing
a set of paths with limited lengths starting from the node. In addition
to utilizing the information contained in the paths themselves, the
Automation in Construction 57 (2015) 213–221
⁎ Corresponding author. Tel.: +358 40 77 44 99.
E-mail address: Jyrki.Oraskari@aalto.fi (J. Oraskari).
http://dx.doi.org/10.1016/j.autcon.2015.05.008
0926-5805/© 2015 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Automation in Construction
journal homepage: www.elsevier.com/locate/autcon