Machine Learning: Definition of Elements and Concepts Dedicated to Bridges

Introduction

The purpose of a dictionary is to speak the same language, to avoid confusion.

A specialized dictionary includes words in specialist field, rather than a complete range of words in the language. Lexical items that describe concepts in specific fields are usually called terms instead of words.

The aim of our contribution is to create a multilingual data dictionary in the field of bridges, in order to learn to a computer how to support engineers to design Bridges. This specialized data dictionary will be a library of terms or objects dedicated to bridges, and their attributes, which maps relationships between objects as well as their property definitions.

This data dictionary enables to:

• •

  Ensure the sustainability of information over time (using standardized terms derived from a common frame of reference), considering the life cycle of an infrastructure lasting decades;

• •

  Facilitate information exchange between the actors of the same project (naming conventions, concept definition, work units, etc.)

• •

  Ensure interoperability between the software packages (authoring and simulation tools) and in particular to better define systems composed of a collection of objects.

The BIM environment (Building Information Modeling) is already adopted in the building sector, and there is a real need in the civil engineering sector. For our purpose, the data dictionary is a document related only to the Bridge concept definition. This data dictionary gathers bridge elements with their English and French translation, their descriptions, their hierarchical links and their characteristics. The aim of that Data Dictionary is to define, in a first step, bridge elements to describe typical bridges, and, in a second step, exceptional bridges (suspension bridges, cable-stayed bridges, moveable bridges, etc.)

This initiative uses several existing documents to provide the Bridge Data Dictionary. These existing documents are mainly terms oriented, often arranged alphabetically. Our purpose is to gather these terms in systems (load-bearing system, carrying system, crossed system, etc.) in order to avoid any lacks, to identify all the expected object attributes, and to allow machines to understand the associated concepts. Furthermore, it could be considered more as a classification than a dictionary, because of its system organization.

The challenge is to make this Data Dictionary precise, exhaustive and comprehensive. Its comprehensiveness will be validated by an international expert panel. This dictionary will be integrated in the Data Dictionary of BuildingSMART.

Our workgroup on that project is composed of engineers from civil engineering and construction companies such as Bouygues Travaux Publics and EGIS, software companies such as Vianova, or firms of engineering consultants specializing in civil engineering such as Servicad, and public scientific centers such as the CEREMA and the CSTB. Thus, this diversity of skills and experiences is the guarantee of the quality of the work.

Resource Documents

We selected some existing documents in order to have a solid basis to create the Bridge Data Dictionary. To do so, we had first to define a structured framework with the precise properties of each object that need to be completed. That is the reason why we have chosen a standard that defines an harmonized reference. Then, to fill the Data Dictionary, we started from a grouping of bridge objects classified hierarchically (in systems). In a second step, we used the classification system of buildingSMART Data Dictionary. Indeed, our aim is to transfer automatically all our bridge objects to that Data Dictionary.

In these domains, the following documents are references.