Page 20 - newDATAmagazine | 04>08>2021

Another simple case that most cities, even documents, each with its presentation,
smaller ones, can surely think about is traffic,
and more broadly, our driving behavior.
I don't mean prying into each driver's own
quirks and habits. That's personal and it should
remain personal. But we could certainly benefit
from knowing some of them, anonymously, so
we could provide more solutions for everyone,
preventing, rather than giving you a ticket,
whenever some “bad” habit takes the wheel.
Surely, drivers everywhere would welcome the
change.
Each car today is a set of fine-tuned
computers interacting with each other (and of
course, the mechanical parts). And though
many of them with proprietary hardware and
software, they're still available to share data
with other systems, if the destination is a
trusted one.
How is this going to help the drivers? Well, it
might give the traffic department some insight
into what really happens with the driver and the
car. Build up real-life stories, data-based
stories (another pun intended).
This is something the manufacturers won't
easily share with a private contractor, but that
they usually share with authorities and would
probably be more interested in doing so, if used
not to harass the driver, but to provide him with
more comfort, in a consented way.
A third and last case I would like to present
to you is document analysis.
One of the most boring, but also most
needed use case is to go through loads of
20 newDATAmagazine.com
structure, images, shorter or longer texts
(preferably shorter, of course) to figure out
what to do with them.
Many have certainly dreamt of just put them
into the shredder but they couldn't. Others,
more conscious ones, have dreamt of a xerox-
like machine, where you would put them and
they would “magically” be separated into
stacks, with labels like “unpaid invoices”,
“Thank you letters” (yeah, sure, keep
dreaming), “Complaints” (let's get real), “Bills”,
etc.
Well, some dreams can come true. And
document classification is one of them. It's
already done, actually. You can put documents
into a xerox (though the mechanics of different
formats is still hard), digitalize it, read it,
separate the images, try to figure out what
those images are (Logos? Warnings?), find
keywords, group them as they appear together
(frequency, order, etc), and even process them
if we can get enough information.
Do you have to do all the heavy lifting
yourself?
Of course not. There are many vendors that
can help you put these pieces together.
Reditus is surely one of them, but you also have
many local companies across the globe
specialized in working with natural language,
you can power your system with open source
and/or proprietary technologies, from
Microsoft, IBM, Google, to name only some of
the most well-known companies. And CMS and
DMS are catching up too.