Gulf Air

Big data and open-source cloud technology help Gulf Air pin down customer sentiment

Big data engine that performs sentiment analysis on both Arabic and English textual data helps airline to understand and react rapidly to customer preferences

This article can also be found in the Premium Editorial Download: CW Middle East: CW Middle East: July-September 2016

Summer is a particularly crucial time for airlines operating in the extremely competitive Middle Eastern market. By now, any carrier worth its salt is already gearing up to capitalise on the annual mass migration from the scorching heat of the Gulf and other parts of the region to more temperate climes.

“The airline industry is very price-sensitive, and the majority of travellers look for the cheapest fares available,” Jassim Haji, director of IT at Bahrain’s national airline, Gulf Air, told Computer Weekly. “During peak travel times such as summer, quick action is crucial, because any delay in responding to the competition could mean the loss of passengers and revenue.”

This summer, Gulf Air has in its enterprise IT arsenal what the carrier hopes will give it a competitive advantage over its regional rivals – something its IT director claims is the first of its kind in the world: a big data engine that performs sentiment analysis on both Arabic and English textual data.

Essentially, this enables Gulf Air to find out what customers think about the airline (and its competitors) at any given moment.

Sentiment analysis, or opinion mining, harnesses textual analysis, natural language processing and computational linguistics to track and dissect conversations and statements to decipher public attitudes and emotional responses to an experience or event.

Social media has proved a rich seam that corporations can mine for details of customers’ experiences, views or opinions on the goods, services or brands they have engaged with – using powerful big data engines to monitor social media conversations and by collecting and analysing the associated data.

The insights derived from this are used to inform corporate strategy, from customer relationship management capability through to the next marketing campaign or competitor analyses.

Given the breadth of its operations across the Middle East, Gulf Air sees itself as “more of a regional carrier than a national one”, said Haji.

Arabic Sentiment Analysis

The airline’s Arabic Sentiment Analysis (ASA) system was developed to make sense of the social media output generated by travellers that span the 18 Arab-speaking destinations that the airline services – along with the complexity of multiple dialects and variations of modern standard, classic and colloquial Arabic. 

The system was built and is hosted on the airline’s private cloud and an open-source software platform from Red Hat Technologies.

Red Hat’s JBoss enterprise application platform is a Java-based application server that provides the distributed environment over which multiple applications can be run. In Gulf Air’s case, it spans 200 servers, running more than 100 enterprise applications across the company’s enterprise IT operations.

The private cloud is a sub-section of Gulf Air’s hybrid environment through which it connects to industry and technology partners, including flight booking system supplier Sabre, airport management systems firm SITA and Microsoft, for the cloud solutions tools that run the airline’s corporate and administrative systems.

ASA, developed in-house over nine months, is part of a mass enterprise computing overhaul at Gulf Air since Haji became IT director in 2011. Company-wide restructuring in 2013 gave IT a central role in strategic business decision-making, rather than remaining at the fringes as a driver for cutting costs. 

ASA complements an earlier online learning management system for training pilots, as well as ongoing projects to introduce mobile technology and the internet of things to passenger services and flight operations.

Read more about Middle East enterprise IT

Gulf Air’s association with the cloud and open-source platforms, as well as its enterprise partnership with Red Hat, goes back to 2012, when the airline migrated its flavour of Unix to Red Hat Enterprise Linux.

The airline revamped its datacentre last year to accommodate the private cloud’s increasingly sophisticated demands. Red Hat Storage, a software-defined storage manager, handles ASA’s colossal demands, enabling Gulf Air’s private cloud to hold 50 terabytes of data.

The ASA itself sits on Cloudera’s version of the Hadoop analytics platform, whose open-source distributed file system (HDFS) facilitates the collection and processing of large datasets – in this case, social media output originating from a substantial segment of the Arab-speaking world, where customers do not necessarily tweet in English – across multiple node servers in parallel.   

Simple economics

The decision to use open-source tools and platforms for this big data project came down to simple economics. Looking outside of Red Hat to Java application server options such as Oracle’s WebLogic or IBM’s WebSphere would have gone against Gulf Air’s current drive to cut costs.

“In deploying Arabic Sentiment Analysis, no hardware investment was required,” said Haji. “The infrastructure where it runs is based on our existing open-source-based cloud environment and virtual servers, which meant no special licence fees to pay either.”

A further incentive was widespread support from a universal open-source community, regular updates and new releases. 

In terms of concrete benefits of the big data system, Haji points to various potential business opportunities the platform should be able to highlight through social media analysis.  

“Everything from preferred destinations to favourite meals will all come in under a business intelligence dashboard, feeding into market forecasts, marketing campaigns, trend analysis, promotions and offers,” he said. 

Read more on Big data analytics

Data Center
Data Management