Page 27: of Offshore Engineer Magazine (Oct/Nov 2014)

26 28

and chief technology of? cer, at Arria. At the time, an early developed in arti? cial intelligence, data analytics, and natural

NLG engine was developed to create weather reports based on language processing, and has involved a number of blind alleys. meteorological data collected by students. For example, in an early explorations of the technology, it was

In 2008-9, Data2Text, a University of Aberdeen spin-out com- thought that a template could be used into which the data is pany, led by Reiter, was launched. In 2012, Arria bought 20% inserted to create the written report. But, when perhaps not all on the ? rm, before taking it over completely in late 2013. Now, the data expected was available, the report would be left with the Arria NLG engine is used to write 5000 weather reports gaps. The commercial version has systems that detect what infor- a day across the UK for the Met Of? ce, where previously the mation is available, and also what is the most relevant informa- company only created 60. tion that needs to be presented, and then produces the report, “The fundamental goal of the technology is to take data and organizing the presentation of the material appropriately. turn it into text, or voice,” Dale says. “It involves a two-step “The holy grail of this space is being able to use machine process. First, the data, such as raw sensor data, is turned into learning to automatically lean how to tell a story based on information (through reasoning), and then the information is data,” says Dale, using data, reports, and statistical techniques turned into written text or narrative (communication). In the ? rst to look at correlations between stored data and textual content. step, the engine does analysis to identify patterns and trends and But that is 10-15 years away, he says. An element of machine turn that into information. For example, if a piece of equipment learning, is used by Arria, but the basis of the technology is on stops working, it will look at why that is happening and what telling the system how to interpret the data it is given, to turn it other machines are around that, to determine what is happening. into information and then from information into text.

The information is then turned into text to tell a story.” Both the Gaining and incorporating the knowledge from the subject reasoning and communication require knowledge “as a fuel” to matter experts also sounds like a lengthy process, but, Dale enable it to interpret and present the data and information. “What says, using corpus analysis, a type of linguistics methodology, signi? cance is a particular sensor sparking a certain alert going existing, human-authored reports, can be scanned and “reverse to have and at the same time as another sensor going off? This is engineered” to aid the process. In fact, this process can reveal the kind of knowledge, gained from subject tacit knowledge the subject matter does matter experts that the software embodies.” not think to reveal, perhaps because they

For the oil and gas industry, the ? rm has think it is “obvious,” making it a valu- started out providing its technology for dis- able part of the process. The application creet equipment areas, speci? cally, an excep- already has general knowledge embed- tion-based alert system on rotating equipment ded about language – it just requires on a platform in the Gulf of Mexico. When any speci? c language, pertinent to the an alert indicates a temperature or movement application, adding any speci? c termi- threshold has been breached, the NLG system nology or linguistics required to suit the

Dr. Robert Dale

Professor Ehud Reiter kicks into action. It has 77.6 million sensor application.

So what are the safety safeguards? Dale is keen to point out points that could be relevant, which it assesses, analyzes and then that the human is still a crucial element in such a system, when it feeds into a 500 word report, describing what is happening, and why comes to mission critical applications. The report makes a recom- it has come to this summary, all in 60-90 seconds. “Normally, that mendation about what an action should be. The human still needs could take the relevant expert 2-3 hours,” Dale says.

The processing power is based on a standard Intel desktop com- to safeguard the right action is taken. “If it is a mission critical puter. The engine knows how to analyze the relevant data, includ- situation it is important to have a human in the loop,” he says. ing associated machinery, and how to understand what informa- “The reports are produced for the human to decide what to do.” “Another question that comes up is ‘isn’t it better to have tion is important and reportable. It knows how to put together a just graphs and charts.’ To some extent it is horses for courses, story to explain the data, emphasizing what is important. It knows but graphs can become unwieldy and to someone not used to how to package up information into sentences of the right size, those graphs and charts will just see a collection of graphs and and it knows the rules of grammar and the right terms to use.

Further applications are planned in the Gulf of Mexico charts,” he says.

While the technology has been in development for 20-30 years, context and ultimately Arria sees a scenario when Arria NLG it is only now, as data is becoming ever vaster, that the NLG engine would be used not just on particular pieces of equipment, but will come into its own, Dale suggests. “At the time (we produced across platforms as a whole, enabling any level of report to be the ? rst weather forecasting engine) there simply wasn’t a lot of produced, from speci? c equipment analysis, to a performance data around and it wasn’t economically viable to automate report summary for the entire platform, each written for a speci? c production. Fast forward and now the situation has completely audience, at the touch of a button. “Anywhere where there is a lot of data and people are strug- changed and the technology has commercial bene? t. The amount gling to deal with that data is where this technology could be use- of data you have to deal with has a bearing on what is achievable. ful,” Dale says. “At the moment we are doing some work looking

More and more data, at ever-? ner granularity, is emerging every at electrical submersible pumps, and drilling reports is another day. The more ? nely grained data gets, the more operations have area people seem interested in. We are starting with components, to be performed to get from data to information. And there is no but you could imagine how you could aggregate that information, sight of that changing. The challenge for us is to scale our tech- then look at chains of equipment and then the entire platform, niques to deal with this.” correlating and integrating that information for a complete report

Scan this page!

of the system, creating an articulate oil and gas ? eld.”

Scan this page with your smart phone or tablet to see

While it might sound relatively simple, the research to get the

Dr Robert Dale discuss Arria NLG at SPE in Dubai.

engine to where it is has taken years, drawing on technologies

The Arria “engine” Image from Arria. oedigital.com October 2014 | OE 29 028_1014_OE_Viz3.indd 29 9/23/14 11:47 AM