Page 26: of Offshore Engineer Magazine (Oct/Nov 2014)

25 27

and chief technology offcer, at Arria. At the time, an early

VISUALIZATION TECHNOLOGY

NLG engine was developed to create weather reports based on meteorological data collected by students.

In 2008-9, Data2Text, a University of Aberdeen spin-out com- pany, led by Reiter, was launched. In 2012, Arria bought 20% on the frm, before taking it over completely in late 2013. Now, the Arria NLG engine is used to write 5000 weather reports a day across the UK for the Met Offce, where previously the company only created 60.

Narrating data “The fundamental goal of the technology is to take data and turn it into text, or voice,” Dale says. “It involves a two-step hen people talk about visualizing data it usually process. First, the data, such as raw sensor data, is turned into Software that can turn your oilfeld data into refers to how data is displayed visually, on screens, information (through reasoning), and then the information is readable reports is coming to an oilfeld near you.

W in infographics, and perhaps in 3D. turned into written text or narrative (communication). In the frst

The idea is to help an engineer see the data more clearly and step, the engine does analysis to identify patterns and trends and

Elaine Maslin found out how this technology could quickly, in order to carry out analysis, or make decisions. turn that into information. For example, if a piece of equipment help create articulate oilfelds.

As more and more data is generated from the oilfeld, from stops working, it will look at why that is happening and what electric subsea Xmas trees, pipeline or mooring integrity moni- other machines are around that, to determine what is happening. toring systems, rotating equipment monitoring, environmental The information is then turned into text to tell a story.” Both the data, downhole pressure and temperature gauges, hydrocarbons reasoning and communication require knowledge “as a fuel” to streams, and so on, the need to not only gather, but also collate, enable it to interpret and present the data and information. “What analyze and make decisions based on the data also increases. signifcance is a particular sensor sparking a certain alert going

Data mining companies are already helping to analyze this to have and at the same time as another sensor going off? This is data, looking for trends. But what if software could be used to the kind of knowledge, gained from subject collate, analyze, and then also present, in seconds, reports in matter experts that the software embodies.”

For the oil and gas industry, the frm has narrative format, tailored to a specifc audience, based on the started out providing its technology for dis- data and analysis (work, which would take a human hours)?

creet equipment areas, specifcally, an excep-

Technology to do this has been developed, over three tion-based alert system on rotating equipment decades, and is now being used by an operator in the US Gulf on a platform in the Gulf of Mexico. When of Mexico. Its origins is in natural language generation (NLG), a an alert indicates a temperature or movement subfeld of artifcial intelligence. Unlike natural language under- threshold has been breached, the NLG system standing (NLU), which takes language and turns it into data, kicks into action. It has 77.6 million sensor

NLG takes data and turns it into language. NLU, as a research points that could be relevant, which it assesses, analyzes and then area, started in the 1960s. NLG then developed in the 1980s. feeds into a 500 word report, describing what is happening, and why

Professor Ehud Reiter and Dr Robert Dale have been involved it has come to this summary, all in 60-90 seconds. “Normally, that from the start, from when they were both researching the feld could take the relevant expert 2-3 hours,” Dale says. at their respective universities, Harvard and Edinburgh as PhD

The processing power is based on a standard Intel desktop com- students in the 1980s, before joining forces in the 1990s.

puter. The engine knows how to analyze the relevant data, includ- “That is when we started looking at how to take machines ing associated machinery, and how to understand what informa- and produce language. There was very little interest in the tion is important and reportable. It knows how to put together a problem at that point,” says Dale, now chief strategy scientist

How the NLG engine works story to explain the data, emphasizing what is important. It knows how to package up information into sentences of the right size,

Analysis & Interpretation Information delivery and it knows the rules of grammar and the right terms to use.

DATA can be ingested NARRATIVE can be output

FACTS DOCUMENT RAW MESSAGES SENTENCE SURFACE from a wide variety of data in a variety of formats

Further applications are planned in the Gulf of Mexico

PLAN DATA PLANS TEXT sources, both structured (HTML, PDF, Word...), context and ultimately Arria sees a scenario when Arria NLG and unstructured combined with graphics would be used not just on particular pieces of equipment, but as appropriate, or across platforms as a whole, enabling any level of report to be h delivered as speec produced, from specifc equipment analysis, to a performance summary for the entire platform, each written for a specifc audience, at the touch of a button. “Anywhere where there is a lot of data and people are strug- gling to deal with that data is where this technology could be use-

MICRO- DATA DATA DOCUMENT SURFACE ful,” Dale says. “At the moment we are doing some work looking

PLANNING INTERPRETATIONPLANNING ANALYSIS REALIZATION at electrical submersible pumps, and drilling reports is another

DATA ANALYSIS processes the data area people seem interested in. We are starting with components,

SURFACE REALIZATION ensures that to extract the key facts that it contains the meanings expressed in the but you could imagine how you could aggregate that information,

DATA INTERPRETATION makes sense of the data, sentences are conveyed using correct then look at chains of equipment and then the entire platform, particularly from the point of view of what grammar, word choice, morphology information can be communicated and punctuation correlating and integrating that information for a complete report of the system, creating an articulate oil and gas feld.”

DOCUMENT PLANNING takes the messages derived from the data and

MICROPLANNING works out how to package the information

While it might sound relatively simple, the research to get the works out how to best structure the information they contain into a narrative into sentences to maximize ?uency and coherence engine to where it is has taken years, drawing on technologies

The Arria “engine” Image from Arria.

October 2014 | OE oedigital.com 28 028_1014_OE_Viz3.indd 28 9/23/14 11:47 AM