Page 83 - Building Big Data Applications

78 Building Big Data Applications

Internet is open 24x7x365 for business

Vendors are selling in a continuum
Prospects and Customers are there and looking for deals, and there is a social media buzz on deals and prices

Each vendor needs to be able to trace the steps of each prospect as they browse for items
Every millisecond of stay increases an opportunity to sell, there needs to be a just in time marketing that
needs to happen
The prospect can be attracted by other vendors also communicating with cookies and offers, while they are
on your site
The ability to convert from look to buy is the most interesting conversion and the point of victory for any
vendor
The crowd that strolls, looks and converts to buy can form a community that provides the word of mouth
marketing for a vendor
The events discussed here are the eCommerce happenings, which can be experienced today if you need to
see this in action
FIGURE 3.4 eCommerce Interactions.

learning where the machine was taught to compute in the parallel domains how the
behavior of the prospect was during the entire browse period. We developed a series of
algorithms for the family of behavior which includes recommendation algorithms,
sentiment analysis algorithms, influencer analytics algorithms, location behavior algo-
rithms, and socio-economic algorithms.
The compute logic in the world of internet evolved from the first minute of its birth.
We realized that using a database and a storage area network was good but cannot scale
in terms of the volume of data, the velocity of its production and the variety of formats.
We needed to have an ecosystem where the compute shall move to where the data lives;
and this was proven by Google, Netscape, and America OnLine first and repeated by
Yahoo, Amazon, Facebook, Twitter, and others soon. The new compute mechanism
introduced distributed data processing as the foundational family of architecture and on
this design and patterns wereborn a series of platforms including Apache Hadoop,
Google Dremel, Google File System, Amazon DynamoDB, Apache Cassandra, Apache
HBase, and other evolving platforms. A newer series of programming languages has also
evolved in terms of Javascript, Python, Go, Scala, and others.
The ecommerce model has evolved into a mcommerce model with mobile platforms
being a provider of information to any prospect at any time anywhere. This means we
have now learned multiple lessons and strategies in terms of information management
and analytics, these include the following:
Data collection expands from being just about what we had browsed to what is be-
ing searched. This means collection of clicks, links, pages, products, comparison,
time, active connection and presence, and potential conversion opportunities. All of
this data needs to be collected automatically and analyzed for next steps.