David Watson, CET, CSP
President, Digicon Information Inc.
For the past 20+ years, buildingSMART International (bSI) has been developing open standards to enable software programs to exchange building information. While the initial focus was on digitizing “geometry,” Building Information Modelling (BIM) has evolved as a means to store information about objects within the modelled asset. For the purposes of this article, “information” refers to any data other than geometry. But first, let’s start with identifying the problem that bSDD is designed to solve.
The more information that users had access to in a building information model (BIM), and the more applications that used this information, the more exchanges began to occur between users. It was here in these exchanges that users noticed a critical element was missing. The meaning of the data being exchanged relies heavily on the terminology used by the BIM author. As we all know, even when speaking a single language, terminology can vary across regions and between disciplines.
In the specifications world, harmonization of terms between drawings and specifications is the panacea. Unless they were developed simultaneously (never in this author’s lifetime), the terms used in each environment were created by different people with different depths of knowledge. Often, multiple terms for the same thing were used across documents.
Over time, information companies sprang up that began to offer cost data, product information, and other useful information relating to building objects. Their challenge was that to “enable” access to their information by clients, they first needed to recognize which relevant building elements existed in the BIM. In a model containing thousands of objects, finding the correct one demanded that a human, not a computer, make the match.
Europeans who regularly trade within the European Union noticed very quickly that a BIM developed in one language could not be used effectively by someone speaking another language.
We needed a method for computers to solve these problems—removing the human component and minimizing human error.
The Proprietary Approach
One of the few benefits of operating in a proprietary domain is that the software vendor ensures that connections between software applications work as intended. However, experience has shown that every application has limitations; no single software works best for every application. An application designed to ‘sketch’ does not usually work well to produce a highly detailed manufactured model.
There will always be tools that are missing or inadequate, and therefore team members must compromise in the interest of maintaining compatibility with the other applications.
This approach is normally identified in most BIM presentations as a “silo” approach (it works great, as long as you operate within the limits of the silo).
The Classification Approach
Today, the most common solution used to identify BIM objects and access information about them is to embed a classification code. Classifications have been most useful in a paper-based society where getting “close” was good enough. Classification allows us to conveniently group a large collection of data into predefined groups, in order to assist humans to find the things they seek. This aspect means that classifications can help group or sort a bit of data, but does not uniquely identify it.
In a digital world, classification still serves a useful purpose to organize like objects into groups that humans understand for convenience. Various classification facets can be used to organize the types of objects in a BIM into smaller groups that are more familiar to the user.
The problem with classifications is that they are by definition “groups” of things (grouped by some shared characteristic). This means they can never uniquely identify individual object types that are more specific than the general class associated with them. Further, the level of detail associated with a class can vary widely across classifications. Classifications also tend to be “national” in usage, which means they are only useful in the geographic locations where they are adopted.
The bSDD Solution
bSDD aims to globally standardize the terms used in construction, independent of language and classification. The definition of each term is not only described using narrative text (as in a dictionary) but also through semantics (explicit relationships with other terms). For example, a semantic dictionary may establish a “typeof” relationship between a door and a fire door, or a property relationship between a fire door and its fire rating.
bSDD doesn’t exist as a publication per se, but as an open source database, which is accessed through a service operated by bSI. Its primary purpose is to enable users to assign unique identifiers (not unlike UPC codes) to BIM object terms, and to store definitions as well as semantics about those terms.
To be useful, a BIM object sent from an author to a receiver should be properly defined, so that the receiving party can perform their service whatever that might be. The recognition of the object by the receiver is not dependent on the term used by the creator, nor the classification code that may be associated with it. Instead, the sender could include the bSDD code, which the receiver can then use to extract the following using the bSDD service:
- Explicitly identify the incoming object type as uniquely distinct from other objects of similar type
- Associate any external data to this unique object type (such as specification clauses or product data)
- Gain access to translated names if they need to communicate with a user who speaks a different language, or from a user who speaks the same language but a different domain (e.g., the English distinction between trunk and bonnet)
- Gain access to other classifications that exist within bSDD, for reporting purposes or to access data that is organized using a different classification
Today, development of the bSDD service infrastructure is evolving. Use of bSDD is free, but it requires real hardware and services to operate. New access protocols are being developed by bSI to attracting funding from developers who profit from accessing bSDD services. Fortunately, bSDD offers more than a simple repository of identifiers and terms; it can provide some valuable additional services.
Additional bSDD Services
bSDD is more than a dictionary; it is a web service based on infrastructure capable of storing multiple ontologies. As such, bSDD is not typically visible to end users, but instead is a tool that can be used by software developers to enable highly useful functionality.
bSDD can be used to do the following:
- Store multiple classifications, enabling the service to map between them
- Once stored, bSDD can enable a user to automatically re-classify objects in a BIM project using any supported classification
- Deliver sets of properties related to an object (not values, but terms for the properties)
- Deliver lists of objects that relate in a specific manner, such as subtype objects or component objects
- Translate model terms into supported languages, enabling a user in Japan to open and read a model created in Quebec
One of the tasks recently completed by bSI was the addition of every IFC class and IFC property set into the bSDD service (IFC is treated as a language in bSDD, with unique and distinctive terms). That means that software developers need only query the bSDD to acquire the proper IFC class names for their model items, thus enabling importing or exporting of IFC model files in their applications.
bSDD is an open-source semantic dictionary owned and managed by bSI. It has the powerful potential to enable users to exchange model data and minimize misinterpretation of the data regardless of language or classification. It also has the potential to serve as the nexus to connect any type of external data to model data in a reliably automated or semi-automated way.
To find out more about bSDD, browse to the website at http://bsdd.buildingsmart.org/.