Patterns surround us in our everyday lives.

Subconsciously, we navigate our way around the world by

recognizing patterns, and taking appropriate actions based on our

experiences of what to expect when we encounter that same (or a

similar) set of patterns. Just as in life, the data that flows

through our organizations is chock-a-block with patterns, and

just as in life, if we can recognize those or similar patterns,

our organization can take the appropriate actions.

Patterns in Data

There are two levels of patterns in and about data that

we need to be aware of. The first is the patterns hidden within

the data being processed and managed by our traditional systems.

Let's call them type 1 patterns. As humans looking at the data,

we might, for instance, conclude that multiple customer records

from different systems are actually about the same customer. Of

course, it would be easy if each record had exactly the same

content; in real life, of course they often don't. There are

numerous differences, inconsistencies and errors we need to deal

with. Some of these are illustrated below:

href="http://www.thetibcoblog.com/wp-content/uploads/2012/06/Untitled.png"> class="aligncenter size-full wp-image-3326" title="Untitled"

src="http://www.thetibcoblog.com/wp-content/uploads/2012/06/Untitled.png"

alt="" width="617" height="106" />

Humans take these types of

differences and errors in stride and, with enough similarity in

enough of the attributes, can easily reach an accurate conclusion

about what the patterns in the data are telling them; in this

case, that these records are actually about people in the same

household.

Pattern recognition of this type by

humans includes a great deal of fuzziness. This enables us to not

ignore, but take into account differences and errors; for

instance, in names and addresses, misfielding and sparse data.

The great part is that we do this without really thinking about

it. Our systems, on the other hand, have a terrible time when

faced with this sort of data. There is no fuzziness in the way

current systems try and determine similarity. To a large degree,

the content of attributes are the same - or are not the same;

this, for instance, leads to the situation where systems can't

tell that the sample data is about people in the same

household.

It seems obvious that we need to

provide our systems with the same type of fuzziness humans employ

when faced with the types of differences and errors that are rife

in our data. This would enable these systems to become much

smarter about how they deal with this type of problem. And, of

course, this in turn enables higher degrees of automation in

business processes where today only humans are able to deal with

the data. For an illustration of capabilities against several

different types of data in several languages.

To see powerful capabilities at work, href="http://patterns.tibco.com/">check out real, live

demos against a wide variety of types of data in a

variety of languages and href="http://www.tibco.com/products/business-optimization/pattern-matching/default.jsp">check

here for more detailed information, including customer

success stories, recorded webinars, whitepapers etc.

Stay tuned for more about the power of

patterns!

No related posts.