Database Reference
In-Depth Information
Gutierrez:
Was there an aha! moment that “This is powerful”?
Lenaghan:
Our “this-is-powerful” moment came when we saw the predict-
ability of human behavior. Location histories tend to cluster very tightly, so it
was fascinating how, with a small amount of data, you could build interesting
profiles of devices. Most people are at home or at work most of the time. So
in that sense, it is not terribly difficult to infer high-level demographic informa-
tion and associate it with a device, even when you know nothing else about
that device.
Gutierrez:
How did you learn and get up to speed on the geospatial analyt-
ics industry?
Lenaghan:
When I joined the industry, it was still more or less a very new
field. On one hand, geography is a very old field. On the other hand, the scale
on which we were approaching it was fairly new, so there was not much lit-
erature to help with the learning process. A number of other companies are
working in the location space, although they are not doing exactly what we
do. So I started learning by looking at some of them, such as Factual—which
is a data provider for locations.
Research-wise, Google was publishing a lot of white papers around latitude
studies based on the experiments that they were running on people's loca-
tions. So the good and bad of it was that—aside from the Google white papers
and a few blog posts—it was completely wide open, so we had to figure things
out as we went along.
Gutierrez:
Has the literature and industry expanded?
Lenaghan:
Though the industry has expanded a great deal in terms of com-
petitors, the literature on location histories is still fairly sparse. There has been
some interesting statistical work Albert-László Barabási, who has done some
things on location and, in particular, predicting people's habits from mobile
data. The academic literature is focused on such questions as: What is the
minimum set of location histories that you could join with another data set to
infer information about that device?
Another research area is looking at the power law distributions of people's
aggregate behaviors. In terms of academic research that is specific to location
and applicable to what we do, there is not a great deal being produced. That
being said, we have a lot of competitors now, who, in the past year, are claiming
to do the things that we do. This is both good and bad.
Gutierrez:
Generally, it is viewed as good to have competition.
Lenaghan:
Exactly. Having no competitors is bad.
