Thursday, October 20, 2016

Watson Crowdsources Cloud Computing


Recently I've been doing quite a bit of analysis work using the IBM Watson cognitive business platform. The really exciting thing about this opportunity is the way data can seem to have a conversation with you.  This got me wondering if social media data could carry on a conversation as well.  Given my almost unhealthy interest in cloud computing, we ran a one week experiment to "crowdsource the internet" in order to see if it held any interesting cloud computing insights . To narrow the volume of documents down to a reasonable number, I limited providers to those on the most recent Gartner IaaS Magic Quadrant:
  • Microsoft
  • Amazon Web Services
  • Google
  • VMware
  • IBM
  • Rackspace
  • Verizon
  • CSC
  • Interoute
  • CenturyLink
  • Dimension Data
  • Fujitsu
  • Joyent
  • NTT Communications
  • Virtustream


Leveraging Watson, I gathered cloud computing related social media documents. According to Watson, in one 24-hr period, there were 46,869 documents that mentioned these Cloud Service Providers (CSP) a total of 57,997 times. Google was totally dominating the online conversation with 73% of all mentions. Microsoft was a poor second at 17%.

Figure 1- Social media cloud computing "Share of Voice"


At this this point I took a look at overall industry sentiment. From this vantage point, Interoute outshines all rivals for positive sentiment.  Of particular note, however, was that Dimension Data simultaneously held the crown for largest percentage of negative and lowest percentage of positive sentiment (which seems to be centered mostly around the dropout of a rider from its Tour de France team and a recent internal restructuring). The Dell/EMC cloud provider Virtustream doesn’t even seem to be present in social media conversations. 



Figure 2- Customer Sentiment Regarding Cloud Service Providers
Figure 3 - Cloud Service Model "Share of Voice"
Microsoft dominated that segment of the conversation that specifically addressed the three standard cloud computing service models (Infrastructure-as-a-Service [IaaS], Platform-as-a-Service [PaaS], Software-as-a-Service [SaaS]). Over 53% of the working set referenced Microsoft with second place AWS coming in at 13.5%. Software-as-a-service is the unsurprising overall service model leader but Microsoft seems to be edging out AWS for Infrastructure-as-a-Service mentions.  Platform-as-a-Service is a distant laggard with only three providers (Microsoft, AWS and VMware) represented in social media exchanges.



Figure 4- Industry Vertical Cloud Computing "Share of Voice



In order to glean some business value, the documents were binned across thirteen industry verticals and analyzed for share of voice and author sentiment. The initial industry bins were:

  • Construction
  • Manufacturing
  • Wholesale trade
  • Information technology
  • Retail trade
  • Utilities
  • Financial services
  • Educational services
  • Transportation and warehousing
  • Entertainment, accommodation, and food services
  • Healthcare and social services
  • Public administration


Across this set, entertainment, government, education and healthcare industries seem to be most interested in the cloud. Surprising to me is that the construction industry interest surpasses that of financial services. Google seems to be driving industry related social media conversations with Microsoft and IBM rounding out the top three.

Although I wouldn’t use this non-scientific experiment to make any big bets, it does demonstrate how actionable data can be gleaned from the social media stream.  It may also shed a little light on the power of cognitive computing in the business world.

One especially intriguing capability that I didn’t use in this experiment is the use of Watson Explorer technologies with Semantic Analytics.  This solution is currently being used by IBM GTS to deliver “built to purpose” cognitive systems for the information technology industry vertical.
Figure 5- Cloud Service Provider Industry "Share of Voice



A key differentiator of this approach is its ability to extract meaning from the fragmented sentences normally found in unstructured IT service ticket description fields. Due to the global nature of GTS Services, this unstructured text is typically in multiple languages. Additionally, due to the different language skill levels of the globally sourced pool of agents, the grammar quality varies. This solution is used by GTS to uncover patterns and trends in the identification of contributing incident causes in order to prescribe appropriate preventative actions.



The digital transformation couple with cognitive computing is accelerating almost every industry. In the IT world, at least, cognitive computing promises to deliver the ability to bridge the gap between unstructured language data and effective maintenance action by correlating social media chatter and customer sentiments with the root causes of operational IT issues.

This post was brought to you by IBM Global Technology Services. For more content like this, visit Point B and Beyond.




Cloud Musings
( Thank you. If you enjoyed this article, get free updates by email or RSS - © Copyright Kevin L. Jackson 2015)