You are Home   »   News   »   View Article

Maana launches "Maana Q" to make it easier to build graphs

Friday, August 24, 2018

Knowledge technology company Maana is making it easier for companies to build the 'graphs' which the computer system uses to compile data together, with the launch of 'Maana Q'.

US digital knowledge technology company Maana has launched 'Maana Q', a system which makes it easier to build the 'graphs' which are used to map and understand company data.

The graph is the 'map' which is built to help the system understand how a company's data fits together - and so gather together data from multiple documents and databases to support decision making.

Because every company has different data, the graph needs to be built by every company.

Maana's customers include Fortune 500 companies such as Airbus, Chevron, GE, Maersk, and Shell. Its 'strategic' investors include Frost Data Capital, GE Ventures, Intel Capital, Chevron Technology Ventures, Saudi Aramco Energy Ventures, Shell Technology Ventures, Accenture Ventures, CICC, Eight Square Capital, and Sino Capital.

Explaining the graph

Maana describes the graph as a 'digital knowledge layer over the operational data.'

To explain it, consider the way that search engines can find a bundle of related data about something you search for. For example, if you search for a movie star, Google will return photographs, data from the Wikipedia page, quotes, movies, social media profiles, and related people. Google does this because it has a programmed 'graph' about information someone looking for might want to find, and where to find it on the internet.

For a business example, consider the different questions a company might answered about one of its vessels (ships).

If it wants to know the length of the vessel, it can be simply looked up from a database. If someone wants to know the weight of the vessel, that depends on the cargo, so it needs to find the unladen weight and the weight of the cargo and add them together.

If someone wants the current position, that needs to be taken from a live web service which receives vessel tracking data.

If someone wants to estimate a vessels' arrival time, that becomes a much more complex calculation, involving understanding the vessel's fuel consumption strategy, sea zone restrictions, sea routes (since vessels do not always go in straight lines), routes around peninsulas.

The idea is that the company would have a 'computational graph', which tracks the different elements of data needed to answer this question, where they are in the corporate systems, and how they should be retrieved and put together.

In this case, Maana calls it a 'computational knowledge graph', because it is able to perform computations from different pieces of data.

Another example is how the graph could be used to help drillers better understand a problem, for example, what kinds of problems are we seeing at something at 12,000 feet in an offshore block? The graph could be used to extract and store the problems (and depths) mentioned in the drillers comments in the daily drilling reports. These problems could be correlated to geological features depths such as sub-surface salt domes.

The graph could also be used to help put together data about the multiple of different activities being done at the same time in a complex organisation.

For example, an aircraft manufacturer has multiple staff working with different spare parts, and each spare part has data about incidents, mechanical history, design specifications, certifications and more. The graph is a way for the computer to put all the information together so it can be used as a whole.

Another analogy to the graph is a bubble chart, which people sometimes draw, to show how a system fits together, with different components, perhaps tasks, people, and equipment.

The graph software is built using GraphQL, a data query language developed originally by Facebook, which can be used to query multiple different types of databases to get exactly the data the graph wants, in whatever form the graph wants, in one go, thus saving a lot of to and fro.

Build your own graph

The critical component of Maana software turns out not to necessarily be the graph itself, but the capability to build a graph, and how much expertise is needed, and this is what Maana has been focussing on, with the launch of 'Maana Q'.

Bluntly, it enables people who are less adept with digital technology to build their own graphs.

The idea is that building a graph should be 'self-service'. Subject matter experts and business analysts can build their own tools which they can use to support their decision making, based on all kinds of data in the archive.

This means that the people who will ultimately use the results of the graph are also the people building it, so they can make sure they build exactly what they want.

The ultimate aim is that people have the best possible information from the corporate data systems available to them at the right time, leading to better decision making, says Donald Thompson, CTO of Maana.

Customers often say they are craving the capability to build graphs, but lack the resources in their companies to build sophisticated solutions.
'We're on a continuous mission to lower that barrier of entry,' Mr Thompson says.

With Maana Q, 'The core ideas haven't changed, how we are delivering them have drastically changed,' Mr Thompson says.

Maana has made the improvements from watching how people work with the software, and how the projects work.

With more people able to work on the graph, there is more likelihood that the graphs will be 'kept alive' and relevant to the needs of the technical experts. There is also less likelihood that a graph will end up siloed, only serving the needs of one department but inaccessible to other departments, Maana says.

The new version has been designed to be easier to deploy on the corporate IT networks, with all of the software tools placed in 'containers'.

There is no need to have a more technical person to assemble the bits and pieces, there are components pre-assembled.


Maana has also been improving the 'bots', software agents which continuously run through new data and try to understand it.

For example, if someone takes a photograph for an insurance purpose which is saved on the corporate network, a bot will immediately try to work out what kind of image it is - a photograph, scanned document or diagram. If it is classified as a photograph, a second bot looks at the photo to see if there is something useful in it. For example it might be a photo of equipment, and the computer can analyse the image and look for defects.

There are domain specific bots to try to understand the facts of a specific problem. For example if a driller's comment is about stuck pipe, the bot might try to identify what caused the stuck pipe (Hole Pack-Off, Differential Sticking, Wellbore Geometry) based on other text and data in the graph.

The image processing bot could be used for drillbit grading, automatically analysing a photo of a drillbit to generate a IADC Dull grade score for its quality. You could set up other bots to start whenever a drillbit with a certain grade is discovered, such as a suitability score of the bit to be re-run in hole based on how much hole is left to drill.

Bots can also be used to classify new documents, for example, given a well document is it a mud report, directional drilling report, incident report, or a wellbore schematic?

Running on the cloud

The original versions of Maana were designed to work on in-house corporate servers, because a couple of years ago, this is how most corporate data was stored.

Over the past two years there has been a big change, and much more of corporate data is now held on cloud servers. So it makes sense for Maana to run on the cloud too. Maana Q is designed as 'cloud native'.

By having it on cloud, you can take advantage of other 'higher level' cloud services, for example Microsoft Azure has tools available for machine learning, data extraction, transformation and loading (ETL). Maana Q can leverage these micro-services.

General use graphs

Maana is also developing general use graphs, which can be used by all of its customers.

It has built a 'whole geological domain model,' covering geological formations, locations, chemicals. Some companies have built their own statistical models on top of it, proprietary to them, to analyse their own exploration data.

It has also worked with PPDM, an industry standard data model, putting the PPDM model into a 'graph' environment, says Jeff Dalgliesh, director of oilfield digital transformation with Maana.

By doing this, a graph model can turn text based information into structured data - for example a description of problems encountered during drilling, which are only included within a PPDM model as a text based comment, in the Drilling Remarks table. Maana extracts these problems and stores then in the graph as problems that are related to the well.

The Maana software can be programmed to recognise different types of problems and understand their causes. For example, it could spot different geological conditions which tend to lead to stuck pipe, or different weather conditions which tend to be more likely to lead to delays.

Associated Companies
comments powered by Disqus


To attend our free events, receive our newsletter, and receive the free colour Digital Energy Journal.


Clustering Considerations in the Machine Learning Workflow – Examples with Exploration Data
Philip Lesslar
from Precision DM


Latest Edition Jun-Jul 2020
Jun 2020

Download latest and back issues


Learn more about supporting Digital Energy Journal