Large volumes of data from the Norwegian Offshore Directorate ready for AI

"92,000 documents in Diskos have now been converted into structured files (JSON). This means that the content is searchable, and it can be analysed and used in various AI tools. This can save the industry time and money and provide considerably better analyses," says Louis Vos, Assistant Director for Data and Facts at the Norwegian Offshore Directorate.

First phase complete

The 92,000 documents are merely the first phase of the effort to make subsurface data available for AI, analysis tools and digital assistants. With structured information, the companies can work directly with data from documents including geological information, reservoir properties, drilling events, core analyses, liquid properties and test results.

"The initiative for this project originated from the industry collaboration forum FORCE. It is a good example of how cooperation between industry and government authorities can help unlock greater value from existing data," says Vos.

Diskos holds considerable amounts of raw data, particularly seismic data, production data, well data, etc., but PDF files are not a format that can be read by machines.

These files are a significant source of data, but it has been difficult for modern digital tools and automated analyses to access the content.

This process has standardised the data and converted unstructured information into an organised and useful format. This provides a good basis for future use of artificial intelligence, machine learning and advanced data analysis.

Opening up brand-new opportunities

Once all the work is complete, several hundred thousand PDF and TIFF files (Tag Image File Format) with reports, logs and other document materials will have been converted into JSON files.

Structured information makes it possible to establish connections between wells, fields, production licences, seismic surveys, reports, samples and interpretations.

"This means that Diskos will transition from storing documents and data and providing access to files, to providing knowledge, supporting decisions and enabling automation. In other words, Diskos will function even better as digital infrastructure on the NCS," Vos says.

He estimates that the effort to convert all the files into structured JSON files will be completed in 2027.