 So moving on from most of the geeks over here, but trying to see how exactly do we use location in a very, very general sense. I work for a company called Onze, and we have developed an application which I believe a lot of users, a lot of non-geeks would actually use it. So this is a journey with respect to figuring out how to translate what a lot of people actually know about location. I'm sure most of us are quite familiar with the maps, but when we go outside, how exactly do we, most of the people will not even have seen the map or if you show them a location on the map, or whether in the daily, let's say, how exactly do they interpret location. So that is the kind of journey that I would like to take. This is a brief introduction. I work at this company called Onze for two years. The company came up with a direction-based service. So we actually went ahead and created our own maps for Bangalore Engineering to create a driving direction, and as a prototype, we actually created this application over SMS. And over SMS itself, we've learned quite a bit of stuff where regular non-geeks would, how would they use the service and how would they provide information to them, which would be relevant without an aid of a map or a smartphone or even a GPS device to begin with. So let's start with the very basic question. There's not something that you ask after a heavy drink or something like that. Where am I right now? Or let's say you want to go to a place. What's the address of the place? How do I locate that place? Or let's say you're traveling to a unknown city or something, and you just want to know where exactly what's mine. If you have a map, you have a location, but mentally, how do you align yourself, which part of the world you are currently? Or rather, you're visiting a friend who is in a remote part of the town and he's trying to give you a direction or you're trying to tell him that I'm at the best bakery or something. He goes by now, how do I come to your place? So those are the kind of interaction where this kind of question often comes up and it's not always the case that you have a smartphone or a GPS device to give you an answer. But even if you have such a device, what would an answer be? In a pure GPS sense, it will just tell you latitude of, let's say this place, 12.98, 17.63. But without a map to actually put this thing in perspective or in context view, it wouldn't really make a lot of sense. The other thing is address. In India, address doesn't work that much in a lot of developing countries. Sunil was mentioning about Johannesburg and New York and Tokyo, where it works very well. But in India, you have to have the landmarks. The landmarks are, this thing has a water tank, not really a big one. So address is something, so this is the address that I just took it from the platform on the website, which once you get to have a reason to understand, okay, if I'm not being to this part of the town, probably I can come to, let's say, Dome Road Club and next to Dome Road Club, I may find this place. So I just stop. The other, obviously, if you have a smartphone or something, you could put it on this thing and see better perspective. Okay, right. I'm not here. I want to get to this place. Then probably this is the way to go. And finally, there's something called Location Persuade, which is nothing new, but a lot of us actually do use it. But this is in one sense what we have standardized over the years as we provide our service on SMS. That is to provide in-textual terms where exactly you are in a manner that will actually help you reach out to that place. So this particular tip, and this is all generated through our map data that we have, specifically in Bangalore and in Chilling. It will tell you the place. It will tell you the road on which it is. This is, by the way, a local named Shankar Nagrod. I'm sure most of the maps will not provide you, will probably do something, second process, something, but people around this place will be familiar with this road. It will provide you the locality in which this place is. It may provide you a couple of landmarks based on which you will be able to go through this place. And a main road, and this is from the main road from where this place is. And this is from a major junction. So once you've left. Sir, I'm sorry, Peter. How do you get this information? Do you get it on? Do you get this? Multiple things. In smartphone cases, yes, you give a latron, like the latrons that we saw previously, this kind of latron, and we have an API that will provide you this information. The second thing is we also have a search engine that you specify energy and research institute, and it finds out that energy and research institute is there in our map database. And so this is the location-specific product base. So it works both dubious as well as the search base. Now, location-specific as, you know, is very, very friendly in India, and this is something that we have erected over the time, and it actually works quite well in India. But why exactly location-specific? What's the problem with other things that I talked to you about? First and foremost, the latron. Obviously, without the context or without the map, it doesn't really make sense. And even once you have a map, you actually have to pinpoint and then depending on the granularity of the latitude and longitude, you can actually reach to that place. So without a context, pretty much useless. Poscillators, I mean, this particular place did have a pretty good poscillator, but consider, for example, this place. But another thing I don't know if you're familiar with, it's almost a 10-kilometer road. And in that 10-kilometer road, if you want to find out number five, it's probably a couple of days. So, I mean, address works quite well in some of the cases, but in most of the cases, especially in India, they don't quite work. Reading maps, definitely not trivial. Actually, I've broken an iOS 6.0 reading map. Sorry. And finally, I mean, the location-specific that we provide is, you know, human-friendly. This is a way, something that you tell your friend to come with, if I'm sure, I would have called up Kiran and asked him how exactly to come to this place. He would definitely give me a couple of landmarks around this place to get an idea. And it gives you a mental picture, okay, I know this place, so this is probably over here. And this is probably over here. And so, let's say, when 2005, when Google actually gave us the map, we didn't have a very good idea of how Bangalore looks like on a map. But we could find a way around, right? Now, interesting thing, this location-specific is something that we actually generate from our map database, right? So, let me get down to how exactly, what are the different components of it and how it is useful for you to make sense of where you are. The person-formers, it obviously tells you the name of the place. It may be local name as well, so that when you come to this place, it gives you what is it locally known as. The road that is, that it is on. So, that helps you, okay, the moment you get on Shankar Nagar road, it should be somewhere on the road. It gives you a reasonably well-known locality. So, if you're coming from a far off place, you don't really need to have directions from that particular place. You know, first you have to reach to don't lose a second straight, then you could figure it out. It gives you a couple of landmarks that get closer to this place. Now, these two information, the road, the main road and intersection, that actually helps you in your commute. So, you know that on the HL airport road, it is about 300 meters off from the main junction. There is a flyover, it is about half a kilometer off. So, these are the roughly kind of information. And I'm sure this is not exhaustive in this, and I'm sure you could find some places which this information doesn't really make sense, but definitely it works more than the post letters that work. So, that is the kind of survey that we carry out in building the map. So, we try to get a... So, we have a scale of one to five, saying that one is something that is totally a problem. Second is something that we think that probably in this particular road, people would know about. Three is something that people in this area would know about. Four is maybe a bigger locality. Five means the entire city knows about. So, that is the kind of guidance. So, whenever we pick a waypoint, we pick placemarks in a city by creating a database, we assign that kind of a level to it. So, that helps us once we get our location, when we see around what are the recently famous, recently known, and this is quite subjective image. So, at times it could go wrong. Most of the time it is... Can't you crowd-source that, assigning that one to five, you know? More number of four-square check-ins means a bigger score or something like that. Probably, but I'm sure with a lot of data that could help, but I don't think we have that kind of data, especially in Bangalore, to actually help make that kind of a decision. For example, I'm sure, I mean, we'd get a lot of check-ins for, let's say, forum or something, but to this place, the recently famous, you may not get that kind of thing. And the other alternative to, you know, location-specific, obviously, reverse view coding, and a lot of people who use Google's reverse view coding. So, just to give a perspective of LS versus... I mean, LS is the location-specific, as far as something which works with only our map. So, it's there, frankly, in Bangalore and Chennai. So, this is the look, and the Google reverse coding of this particular latitude, that Google would give you. There's the fourth main road, the Umru second stage, Bangalore. And there's a big part of, you know, our data collection, which I'm not touching over, just to give you a brief of how we organize, what constitutes our map data. And this is nothing visual. It is basically a list of points of which we have latitude and longitude. And in our web application and other applications, we actually use Google Maps to overlay. So, there's nothing pretty to look at in one sense. So, we have place marks, waypoints, points with accurate latitude, longitude, and a reasonable amount of subjective information over the popularity of that place. Then we have roads, which are essentially dotted lines. We have intersections. So, again, I mean, when we build this database, it was primarily to provide a dragging direction kind of a service of a lot of emphasis on roads and intersections. And finally, locality, which is basically a polygon, which captures not at an accurate level, but with roads dividing a particular area, it captures the whole locality. So, based on it, we have an application which is very simple, which is big hand one sense, which we call in locasi, locasi, is in Malay, it means location. So, very simple thing. The moment you launch this application, in your location specifications, it will tell you that this is something that we captured in our office in general. So, this is the information that it tells you. It shows you on the map, where exactly it is there. And then something interesting that you could do is send this location to your friends. With most of the other applications, let's say if it's a map-based application, the other person needs to have a smartphone or at least a map on which to look at this information. But what we do is, since we have all of the location, LS, whatever you could place, you could actually send this whole text as an SMS or as basically a complete text, or even the person who's receiving it without the const text of a smartphone or a map or something like that, he'll be able to get a reasonable idea of where exactly you are. I can't understand that. This was a complete game of this topic. It was a celebrating independence day. Let's reiterate. Basically, this application is very simple. It just tells you where you are in terms or in context, which would actually make a lot of sense to you without any extra position with the map. It helps you share your location to your friends who may not have an application or who may not have access to internet, or who may not have access to a smartphone or any such thing to know exactly where you are and probably help you guide through if you are visiting them. And a lot of questions are raised about the security of your location. Pretty secure because there's a short link or something like that that you generate and you only share it with your friends. So, you know that it actually goes out. So, you could probably get the application. So, it has a web interface also where the short link that you click on gives you a location of where you are on the map. Yeah, that's all. Nice. Just to give you a brief, we started with providing driving directions in Bangor and later on we moved on to providing a location-based service to a detailed establishment. So, now we have data across the country but the kind of granularity of the places that we have, the roles that we have, the accuracy, the popularity of these places is pretty much only in Bangor and generally. The other places, it's not that much. Do you have a question about the distance from the junction? Yeah. How accurate is that? It's a road accurate. So, it's not as the profile kind of distance but actually the distance that would take you if you would travel on the road. So, GPS for this, where you calculate it? No. It pays on the road that you would take to reach to a place from the junction. How do you take the geometry of a road in that case? We have, our data has the road. So, we know where the junction is. We know the places. So, we know how exactly to reach to that place to do adopting and then we know the distance. Okay. In that case, why don't you provide this left side, right side, that kind of information? Yeah. We could provide. Did you use OpenStreetMap as your base initial data or did you build all this data yourself? When we started, OpenStreetMap was not that popular and especially in Bangor, we did not have that kind of information. So, we started building this data by ourselves. Having said that, over the last one year or so, we have kind of evaluated and explained the OpenStreetMap data. The thing that the data is there, it's very ad hoc in one sense and we need to have some subjective information over between to extract it. And the kind of demarbition that we do with, let's say, the roads, the areas that we have. We've seen in OpenStreetMap, we've met people draw, let's say, this building, right? And they use something called a road as a metatime, just because it looks pretty. So, if we try to extract this, the roads from OSM, we would get the structure of this institute. So, that has been a very big determinant. But again, we're trying to see how exactly, in one sense, this is more like a noise. So, when we extract data from OSM, how do we curtail this noise and this together? That would be actually a big help if we could get that. Sorry, just a follow-up question. What did you use then to generate the data? Can you give us an idea of just the kind of tools to... Most of the tools were generated by us. So, in a very, very raw sense, we just took a couple of these devices, just drove around the road, captured in audio of this thing, as we were trying to go back to the office in the transcription. Very, very basic. How long have you been doing this special map? So, this map, we have not updated in the last two years or so. But before that, we took close over six to nine months to build the Bangalow map. Chennai, we did faster because we had at least at least a possible X-ray of once we did Bangalow. So, Chennai, we did the working three months. And my question is going to be just in terms of keeping up-to-date as geography changes. Yeah, that is a challenge. What your maintenance plans are for how you maintain that data. Right. So, we have not really done that before. Yeah. How expensive was it to acquire all this data yourself? As I told you, Bangalow took about six to nine months approximately. So, you can make an estimate. But actually, it's basically... And it's just for speculation, depending on how many people you put in and what kind of... How many people did you put in? You put out about ten people, one car, a couple of laptops, a couple of heaters, devices, multiplied by nine months. Yeah. It's not an easy thing, definitely. And what people are doing with OSMS, definitely, if I come into it, the way they're building it up. Yeah. So, in the last two years, also, we have kind of moved away from providing driving directions. So, we have a service on SMS quite well, even now. But in the last two years, we have moved on to providing a more like a location-based platform to retail businesses. So, right now, what we do is, you have a retail chain of 10 to 15 stores. And you want to provide an application or SMS-based website, web plugin or something where you can help your customers locate the nearest store. So, that's the kind of location-based service you provide to retail businesses. Use a base of... Sorry? Use a base of... Is anyone using it? Use a base of... Use a base of our app. This is just really about three months or so. And this is not... Sorry? Yeah. So, Android app is available. Use a base of... I don't know, 100 people or something. But on the SMS side, we have a reasonable update. So, in SMS, we provide driving directions in Bangalore. You can send SMS from TV to... It's a generic to 980, 980, and it will provide your direction. So, we have a pretty big update. Yeah. So, do you deal with misspellings, like disambiguation? It's like if someone misspells how to deal with this? Yeah, we do. Is this at the database level or... Do you know what you use to deal with disambiguation? Yeah. We do deal with disambiguation, specifically because we provide this service primarily on SMS. And on SMS, if let's say you type Majestic as MAG, ISTI, SK, and we would need to resolve it. So, we do a good amount of spelling tolerance, which we have built up in our system. Yes. Yes. How is the popularity for you? We draw them. Yeah. What about the conventions that are being followed in the local areas? But for example, both the fuel joints are an option. That's what many people call it. But it's actually known as urban domain. So, how do you deal with this? A lot of location, a lot of local, or what you call as tribal knowledge, is actually built in. So, when we created this database, we did actually take into account of what exactly this road is more popularly known as and not the official name of the road. But again, it's very subjective. It depends on people to people. But we try to capture it as much as possible. So, it gives me both, if I say, full of local law, if I say, if I say, full of local law, if I say, full of local law, if I say, full of local law, if I use this in smaller cities, but information isn't as consistent or is that just still a big problem? Yeah. There are 10 subs that we could do. So, this could be probably one of them. But right now, the focus, timely is on the retail business platform. It's a non-technical question, but how do you plan to monetize your application and in your monitoring incentive? Right. So, we were running this driving variation on SMS for close to three years now We will not make any money on that part, so most of our focus is now on the retail platform. We provide this as a platform to retail businesses and we have more text from retail businesses. I mean not directly from the individuals also. So the SMS based service or this particular application, I don't think, it doesn't make us money and I don't think it will make money very quickly. Okay. Thank you man. Thanks a lot.