Wednesday, November 5, 2014

Twitter, IBM Bluemix, Watson & Interact

IBM introduced couple of weeks ago IBM Watson Developer Cloud with a variety of services for building cognitive apps. One of them is very relevant in Marketing activities: Watson User Modeling.

The IBM Watson User Modeling service uses linguistic analytics to extract cognitive and social characteristics, including Big Five, Values, and Needs, from communications that the user makes available, such as email, text messages, tweets, forum posts, and more. By deriving cognitive and social preferences, the service helps users to understand, connect to, and communicate with other people on a more personalized level.

I have prototyped a solution using Watson User Modeling service analysing 200 customer's tweets from Twitter and using the cognitive and social characteristics coming from Watson User Modeling to push the most relevant offer to that customer.

High-level architecture of the solution:

Examples of scenario:
1/ A customer is on the WebSite  Acme Inc., a travel agency, looking for his future travel destination. As a known customer, Acme Inc. has already built his profile based on demographics and his previous navigation history. But, Acme Inc can implement cognitive segmentation to push some special offers based on the personality of the customer like "Adventurousness".
2/ A customer is on his mobile retail application, looking for a product to buy. Instead of just using Product recommendation or previous purchase approaches, the retailer can have implement Watson capabilities in his offer recommendation engine.

Products Used:

- IBM Interact: With IBM® Interact, marketers can personalize, in real time, the experience of customers interacting with websites, call centers and other inbound marketing channels. 
- IBM Bluemix: an open-standards, cloud-based platform for building, managing, and running apps of all types, such as web, mobile, big data, and smart devices
- IBM Watson Developer Cloud: a bold new partnership between people and computers, providing you the tools and resources to create cognitive apps 
- Twitter: as source of customer non-structured information helping to retrieve cognitive and social characteristics, in my case "Openness to change". But, other informations are available like "Artistic interests", "Imagination", "Self-discipline", "Excitement-seeking"...

Flow of the process:

When the user accesses to one personalized channel of communication with Acme Inc (Mobile, Web...), Interact is requested to return the best offer or a list of offers. In the Interactive flowchart, a "CallOut" is done to retrieve the score of a specific social or cognitive characteristic. A process is running on the Bluemix platform. Based on the request containing the twitter account of the customer, the process will retrieve the last 200 tweets and send that content to Watson User modeling process to obtain the requested score for that customer.
Finally, via that score, we can determine if that customer is eligible for that special offer.

IBM Bluemix Interface: 

I have one Liberty Profile and one service for Watson user Modeling running on Bluemix.

IBM Interact Interface:

Flowchart with a Select box
Select Box Configuration
My ExternalCallout (Watson_Mood) method return "Open" with the score provided by Watson user Modeling is greater than 50 and "Not_Open" if lower.

That new technology opens new possibilities to provide advanced personalization in real-time or in batch for Marketers. We are just at the beginning of a beautiful journey...

More to come...