A Case for Cognitive Computing in the Oil & Gas Industry - Renato Cerqueira - Keynote@Ontobras 2017

A Case for Cognitive Computing in the Oil & Gas Industry 
Renato Cerqueira

IBM Research Lab has been recently open in Brazil. The research of this lab focuses on natural resources. His focus today is in the Oil and Gas industry.

Cognitive Computing is the use of computational learning systems to augment human cognitive capabilities and accelerate, enhance and scale human expertise to solve real world problems. Systems in this field:

• Generate and evaluate hypotheses (learn and build knowledge)
• Understand natural language (hear, see)
• Interact more naturally with humans (reason, explore)

Informed Decision Making
Search x expert Q&A systems in Watson:

• Understanding questions
• produces possible answer and evidence
• Analyses the evidence
• Deliver the user: the answer, the evidence and the analysis result.

To calculate the evidence of a possible answer:

They apply different reasoning mechanisms to calculate the evidence and then use a formula to combine this, in a way a final score is calculated for each answer.

Watson Q&A Pipeline (interesting slide!)

see Watson videos on Youtube

Watson Oncology: this was one of the first applications.

• Knowledge Representation and Reasoning: this system used the Snomed ontology 
• The system analyses images (thus, computer vision techniques have now given Watson the ability to see).

Disruptive trends in the Oil and Gas industry - Work with Petrobras

Data is recognized as key to drive values. new methods required for new unconventional exploration, seismic technologies and data overload. 

• Drilling companies report collecting 2T bytes per day of data (only 5% gets to the shore)
• 80.000 sensors in a single modern platform
• 300.000 publications per year, according to the American Petroleum Institute.

Cognitive computing is set to re-define workflows across the entire O&G value chain.

Selected Use Cases:

- UC1: Basin/Reservoir Knowledge Base Builder

There is usually an initial resistance from the company to give away their knowledge, but IBM explains that the KB will not be taken away, it belongs to the company.

- UC2: Basin/Reservoir Similarity Adviser

- UC3: Seismic Interpretation Adviser

- UC4: Capital Project Management Adviser
The challenge here was to combine different kinds of information from different sources.

Scientific committee - gathered people from different divisions of the company to agree on how information could be integrated 

- UC5: UC5: Drilling Adviser

- UC6: IoT-driven Reservoir Digital Twin

Problems for interoperability:

1. uncertain or missing information (for some processes, the information is a must, for example, you must know the pressure in the pipes before opening valves); 
2. need for evolving the applied conceptual models (including ontologies);

The KB is composed of the ontology + instances + (standard) inferences and rules

Ontology = class + relation + constraints

They face many challenges to go from text to KBs (he gave some ex. in a slide)