Tutorial: Building Metadata Normalization with Plaster


Building metadata is one of the foundations of deploying applications and conducting research on buildings. Over the last four years at BuildSys, we have presented a series of effort in tackling the building metadata challenge --- from Brick, a unified metadata schema, to various algorithms (Scrabble, Zodiac, and BuildingAdapter) for normalizing building metadata. While these works have promised to facilitate the conversion of metadata in existing buildings into Brick, actual users with not much programming experience, such as building managers, are still lacking a practical tool that provides these algorithms in their work life. To this end, we have designed and developed Plaster and Plaster Web Service. In particular, Plaster is a Python library including recent metadata normalization algorithms and provides a unified interface to them, which enables benchmarking and integration of different algorithms as well as development of new ones with regard to the standardized programming model. Using Plaster’s Python interfaces, Plaster Web Service provides a user-friendly web interface for non-programmer users to easily explore the state-of-art metadata normalization algorithms. In this tutorial, we present an overview of the metadata normalization workflow and a couple of algorithms in Plaster. We will also go over Plaster Web Service to help users understand the underlying mechanisms and workflow. Plaster WebService is freely available at https://plaster.ucsd.edu.

Proceedings of the 5th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation, 2019