Tip:
Highlight text to annotate it
X
Hi My name is Yori I am cofounder at Local Sciences and I would like to start with a
question. Who remembers his or her first google search? Specifically Google. I actually do.
Back in 1998 I was an Alta Vista user. I was running the first of eventually three startups
in the natural language space when a coworker came to me said "hey there is a new search
engine called Google". I was intrigued and after a couple of trials spelled Google correctly
and typed my first search.And I remember it as vividly as if it was today. I typed IBM.
And the results came back in. And my jaw dropped. You see, right there at the top of the listIsaw
IBM.com and I did not expect it. Back in the Alta Vista days I would have expected to get
a long list of totally unrelated results. In other words what Google did that was pretty
interesting - they figured out that for me as a consumer certain web pages are more valuable
than others and they did it using network topology analysis and some otehr signals,
and it works great to these days other than it totally fails in the local search space.Local
search is stuck in the Alta Vista world. There is no network topology, no traffic, in fact
many of these businesses do not even have a website. in other words, scoring and ranking
businesses is a very different task than scoring and ranking web pages. and that is exactly
where Local Sciences comes in. We created a new kind of technology that takes local
businesses and ranks them and score them allows yellow pages and search engines local media
companies to present them based on their value to the consumer. And that all by itself disrupts
the local search space. Local Search when you search for something-somewhere or search
with local intent is very valuable. Because it carries a very strong very strong propensity
for action in the real world. and there is plenty of it. Plenty of it going around. In
fact over 40% of all searches are local and according to Google over 50% of mobile searches
carry some kind of local intent. and so if it so important how is it handled today? Quite
unbelievably but it is a manual task. sites such as Google Plus, Yelp, TripAdvisor source
from the crowd ratings and reviews about businesses and use this data to enhance local search
and local discovery and it is AWESOME it is working really well if you live in San Francisco
or New York and you search for a SPA or a restaurant. But outside these few keys geographies
and categories all of these players together combined have very thin coverage leaving over
70%of local search poorly served. So in other words, in the local search space we are all
Alta Vista users again. And this is me back as an Alta Vista user. So what have we done...
At Local Sciences we use big data techniques combined with analytical modeling to actually
score and rank every single business in teh United States without the human bottlenecks.We
rely on dozens of different signals not just reviews. Eitan, my cofounder and myself we
had good runs in both local media and analytics and we combined both to actualy create to
solve and resolve three key challenges: The first one is the semantic challenge.When people
talk about a business or when they write about it, what do they really mean. The second one
whicvh is a toughy is which business are tehy talking about when they talk about Domino's
which Domino's are they talking about. And lastly, because over the long tail there are
a lot of missing and sparse signals we have implemented a pretty clever imputation of
missing values in order to get meaningful results from our statistical machines. And
meaningful they are. In fact even all-mighty Google took notice they link to us as a trusted
review source from Google Plus pages They send a lot of high quality searchers our way
and these searchers interact with our business pages. They have very high engagement in fact
we show phenomenal results