In ML-BAU-DOK, the methodological foundations were developed to extract and process relevant information from analogue or digital documents for two special application areas (energy efficiency, life cycle analysis). For this purpose, documents provided by two practice partners (SKE, Architrave) were evaluated. The primary sources had already largely been digitised and comprised >600,000 document pages. The project describes document classes that are prioritised for the two application areas. This allows the building documentation to be evaluated in terms of its suitability for the application areas. Criteria of data quality and machine readability were used as a prerequisite for machine learning, which can also be transferred to other application areas.

A cooperation for research and knowledge transfer in the area of digitalization of the real estate industry has existed with the proptech company Architrave since 2016.

Architrave GmbH, based in Berlin/Frankfurt, was founded in 2012 by Maurice Grassau and Dr. Benjamin Staude and develops document management solutions for the real estate industry with over 100 employees.

As a non-profit foundation under public law, the Eberbach Monastery Foundation owns and operates the Eberbach Monastery in the Rheingau region, which was profaned in 1803.

ML-BAU-DOK builds on the following already completed research projects of the Unit of Real Estate Studies:

• Broad application of artificial intelligence for document classification, information extraction and predictive analytics in real estate
• Von der analogen zur digitalen lebenszyklusübergreifenden Gebäudedokumentation
• Fundamentals for automating due diligence processes in property transactions

Project Term: 04/2021 - 01/2023

The Full Research Report (in German) can be downloaded from the website of Zukunft Bau (BBSR): Link

The follow-up project ML-BAU-DOK+ continues the research: Link