A recently sparked a fascinating email discussion at Tocci:
Pierce Reynoldson: Last night at an MIT event I met this guy who works with startup Narrative Science. They’ve created an algorithm that pulls from a specific pool of data (like sports) and spits out rough copy. He mentioned the article in the New York Times, In Case You Wondered, a Real Human Wrote This Column.
He emphasized that it is not intended to replace reporters, but automate the menial task of data collection and presentation.
Moises Berrun: The technology behind Narrative Science can be applied to building plans. An algorithm that can read the plan and identify the code (ADA, etc.) that governs decisions. So when changes are made, a checklist/report can be made that reminds the designer. For example:
Return fan is in a return duct that is fed by a courtroom. Duct does not require a duct detector (for detecting smoke in a duct, this is a fire alarm issue) because courtroom is returning less than 2000 cfm to the fan. Ducts get redesigned. Now duct returns air from two courtrooms and cfm now exceeds 2000. Duct detector is now required in duct. A report/warning is generated that notifies designer to include in the model/drawings a duct detector on the fire alarm drawings.
Something like this can [also] be applied to Specifications. Algorithm reads specs and produces a report that can become a construction QAQC checklist.
Jennifer Heikkinen: Is this example that Moises is referring to below similar to the Autocodes presentation that Target Stores was part of at DC BIM forum? [ and ]
MB: Probably similar. Though I don’t remember the Target presenters going that far.
John Tocci: Okay, let’s do it.