Autoplay: ON Autoplay: OFF
CNET News Video
Gravity takes advantage of the 'interest graph'Gravity CEO Amit Kapur explains how combining what you like with who you know can make the ultimate filter for the things that matter to you.
-If you look at sort of a standard, any kind of information media as it exist today, so take the internet, early on in its life cycle, you tend to have less content but a lot of very popular content that appeals to a lot of people so it has a high signal-to-noise. As time goes on and as more and more-- you have a larger and larger volume of information, you tend to follow a power-law distribution where there's a significant amount of noise in that system, in the absence of filters, to help us find the good stuff. So, what technology companies do is they come in and say, at periodic points in time, "Let's create a better filter in order to compensate for that loss in signal and actually as more information comes on, we'll help you find better stuff." So, the first set of companies that did that were companies like AOL and Yahoo and Google who said, "We're gonna actually curate this content early on, we're gonna create links and curate them in a directory so that we can help you find really good stuff." Google came in and actually took another level and said, "We're gonna apply science within this environment and within these links that exist to say, based on having sites only linking back and forth, we're gonna rank them and then when you search for something, we're gonna show you really good stuff." What ended up happening eventually, and this is around 2003, is user generated content started to come online in mass so this is when MySpace really got going, YouTube really got going, and it started to, at a certain point, overwhelm those filters a bit. With companies like, you know, Yahoo and Google, they're amazing companies, right? They changed the way that we use the web but I think even a lot of them will admit that they could do better with respect to filtering social data and real time data and even when you hear a lot of quotes that are coming out of sites like Google is we need to figure out how we get better at, you know, integrating real time content and social content. So, what ended up happening around that time, around that 2003 timeframe is a new set of companies came in and they said, "Well, we're gonna create a different type of filter within all that user-generated and social content," and that's when companies like MySpace and Facebook and Twitter entered and said, "Well, let's actually let you look at this content through the lens of your friends. Let's help you actually curate it based on who I'm friends with and people that I actually trust and so today, as we look at our environment, that's one of the primary ways that we discover new content is via people sharing it on Twitter, via people sharing it on Facebook. What we believe we're in a time right now is just another inflection point where we're starting to once again get inundated with information and these sites, these products, they're amazing, these social companies, but if you're an avid Twitter user or an avid Facebook user, YouTube user, you're seeing too much content coming into the screen for you to actually effectively parse it and find the stuff that interests you. Even myself, I'll log in to Twitter 5, 6 times a day and I know that I'm missing things because I'm not constantly, I'm never finding the nuggets that are really important to me so we believe that what is once again happening is the filters are starting to fail a bit within that kind of an environment. So, what we're doing at Gravity and the reason that we started this is we're trying to create a new type of filter that lets you find great content within this larger volume of information and the way that we do that is we believe that we're uniquely positioned to look at all of the social data within this environment, understand what you're interested in and then be able to take content, identify that within our anthology of interests and say, "This is the very best stuff to show you right now, based on your interests." If all of this, in terms of phases and how this all relates is early on, in the web, as it was with music and radio, that was their web, that was their radio. This is editors and professionals curating the content for us. The second phase, a lot of people call "Web 2.0" was our web, it was the social web where our friends and people that we trust and follow are curating content for us. This third phase is what we like to call "Your Web" or in music's case, your radio, which is where we believe we're taking this kind of transformative step to understand who you are, what you're interested in, take any information, any people, any advertising, and target them to you based on those interests. That doesn't mean that these other things go away, it just means that this is a new web. We still have radio and editors and professionals curating content for us 'cause they know that stuff, they're good at it. We still wanna see content and music through our friends because we want that shared experience, we want something to talk about with them, but there's a better way to do things, too, in terms of filtering a lot of the rest of this noise and that's by understanding you. So, the next time I pick up, like go to Wallstreetjournal.com or ESPN.com or CNET.com, I should be able to see more content that interests me because it should be personalized to me and that's the kind of internet that we're trying to create with this technology. So, I'm gonna show TWINterest. It's very easy. The first thing you do is connect with Twitter, allow it to connect. It authorizes and then what happens in the background as soon as it connects is we start to analyze my tweets so what we do here is we use sophisticated natural language processing to process the 140-character tweets to identify what are things that I'm interested in, what are concepts that I'm interested in and then kind of create a list here for me to view. Eventually, I'll be able to X these things out. Also, what this doesn't show is weighted detachment which we have on the back and it's how interested am I on these things and also the time basis for your attachment to those interests so is it something that I'm constantly talking about or is it something that I recently talked about or talked about a long time ago. What I'm doing is I've got my friends here on the right that have connected. My friend Max has 1200 interests. He tweets all the time and so what I wanna see is what do Max and I have in common, so right now Max is kind of my interesting friend, let's play this, but I can see all of Max' interests here and then I can click on the overlap to see, okay, Max and I have video games in common and karaoke, and computers, obviously, education, certain movies, and it works on the internet, works with Gravity actually, and so we obviously have a lot of things in common on the internet. So, here's a basic picture of TWINterest.