 Hi, good evening, my name is Arvind and I'm gonna talk about how big data analysis and social semantic web can bring you about a good web experience. So every single blog post, every single Twitter update, every single status update that you make on Facebook actually adds to the big data but it's already there available on the web. And what also happens is that there is a parallel talk going on about whatever status update or whatever tweet that you made but it's very hard to interlink those communications because it's not happening in say a particular Maka language. So say that there is, maybe take up for example that there is a Twitter update that you made and there is a Facebook, one of your friends is talking about your Twitter update on his Facebook timeline but you can't connect them essentially. So let us take again for another example that say you want to send across a link to your friend but what you see is that on the link there is a wrong number, a wrong phone number. So what the option that you would have right now is maybe to contact the administrator of the web page and tell him that could you please change this number to this number but then more often than not those kind of conversations don't happen with like immediate replies. You can't see that the number has been changed like immediately, maybe it would take a day. So what you would do is maybe send the number with the correct number, send the correct number on your Twitter and with the link you would send it across to him. So then when your friend actually sees that page, he would have to cross reference the number on your Twitter update and check it out and then he would have to dial that number but think of this web experience when he can actually see that number on the page rather than having to cross reference that particular correct number. So you could just, when he's on the page you could just see that the number is there, the correct number is there. So this kind of web experience is a very different one. Say for doing these kind of annotations maybe a browser bookmarklet or maybe a browser extension could help you doing these kind of annotations. So say there is a user profile maybe built into the browser which would say like these are my friends and only the annotations from these people I would accept. So you can see that there is a social networking aspect that comes into this. So you would say because he's my friend I would turn on the annotations from him but because X is not my friend I would block the annotations from him. So let us say maybe there's an element in page bar which has got the X bar. So all these annotations or whatever happens are basically linked with element to element. There is a particular element which you know the X bar for and there is another element in say another page. Say maybe a fire safety page and you have a blog about it. So you would cross link both of them saying like these two elements are related to each other in a particular way. And when your friend has turned on all these annotations from you he can see that all those comments or those annotations turn up because he has subscribed to you. So let us take up this page for an example. This blog is done by Jay's. He talks about outsource for source in this particular blog. At the end of the blog you can see that he says that he has blocked the comments because he can't handle spam over here. So you can see that but what he also does is he gives you a link so where you can go and follow the comments or the talks that are going on about this particular blog. But what you can think is that there's only one place that the talks are happening but they could also be all these talks could also be going on maybe Twitter, Facebook, blogs or anywhere else on any other some media which has not been linked over here. So let us go to this page and see. So you can see that there's a whole big commenting going on over here. So now what you would need to do is if you want to read the comments you need to come here and then again you need to go back to read the blog. Frankly speaking I go to the most of the blogs just to read the comments so that actually gives you a very different experience when you read what somebody else is saying or what somebody else's experience is of that particular blog. We can actually see that there is a debate going on whether GPLv3 is good or BSD is good. So those kind of comments are like totally good for Flame Wars. So say that maybe say that there is a somebody has say for example somebody has done a feed say somebody tweets about saying that there are comments. So say there are comments for this particular element here in Jace's page and then he says that there are comments for that for the Jace's page over here and you could say just maybe Twitter it. And when somebody comes to this page say Jace's page using so somebody could do say maybe do a feed like this he says there's a Jace's page and there is the expert for that particular element. So you could say like this particular unique element is related to that element in that page with say maybe with the tags with GPLv3 and stuff like that. So say there's a bookmarklet the user has it and you can see the annotations up over here. So say a user has this bookmarklet enabled and whenever he comes to this page he can actually see just this. So there is a recommendation engine which would say maybe like push up. They're saying because you are this person's friend you can have you can look up these annotations done by your friend. So maybe say I am interested in this. So you can actually see is now all the comments from that page are lowering up over here. So this kind of this kind of semantic linking it would give you a very different experience of the web all together. So yeah and then also like maybe say imagine threading in with the use of timestamp. So we can see that if people could say mark for this particular comment there is a comment somewhere else on Twitter and it could be threaded according to a particular use case. Can I ask a question? I see that two connected documents got interlinked and you can use them in one place. But what is semantic about all of this? So the semantic thing is say these are these are related by somebody. So you are defining your own semantics to see when you say semantic where there is there's a particular maybe an ontology which defines the semantics of the semantic web. So you define your ontology. So say imagine it for a very small custom web of yours. You say like I'm interested in these pages and these are related to these. So you define your own ontology on your and that is how it's a semantic web in itself. I mean it's semantic in a sense everything is interconnected. That's the web right. Right. Yeah. It's the web but then what you can see is you can't you don't see the interconnectedness on a page given page. So say I just go to Jason's page and I can't see the comments are there. I need to either have says maybe a bookmark like this or maybe I need to keep jumping through the through the loops. So I need to go say even here for this matter say you have a link over here. So you have a link on this particular page. I mean this is not coming showing up over here. So you can see that think of annotations. I mean think of this in a multiple level. So say maybe when you when you can come up here because you're interested in this there could be other recommendations by your friend with say maybe like your third friend did this. He says like this comment is related to something else and that could come up over here. So that kind of experience of the web is a very different. Right. It's not about it's not about semantic tools. It's about semantic web how how you can define your own. How you can have a custom web of your own. Right. Right. But shouldn't the semantic layer be about meaning. It see you define the meaning. Where's the meaning piece in this. I don't understand. I mean I understand connected documents but that's what the web was always. It was always bringing it on. You're bringing it on one page and understand that. But then where's the meaning exchange. That's what semantic web is about. It's about communicating more meaning than just plain text. Right. So I mean this is just text maybe what so think of something else over here. I mean not just text but maybe it could be any documentary or whatever. I mean don't think of it just as plain text. It's just an example over here that I'm taking. So say maybe think of I mean in the future slide in the next slide I'll show you how this could make sense. So let it go forward. Right. I mean so see a book marking in a sense so you can't do this in as of now. I mean how we do this you would need to write. So then what happens is there is content with them. So say even even take up for I mean for example take up this course so you can do a commenting engine when discuss you could just embed in your site. But then again what happens is your content is with them. It's not with you. It's more about the content is distributed. For example I mean where I took up this one. This could be coming from your own blog. It need not be somewhere else. So you could say that I mean like I blocked up all this. So I mean look look this up over here. So when you do maybe a feed over here. Yeah. So when you want to do the feed you would say like I mean I'm the author and I say like these two elements are related here. Say you take up your blog and say Jason's page has got comments has got comments in my particular blog. So that would come up. The idea of like doing this this kind of feed passing. I mean so basically what so these kind of feeds could actually be be passed by a service. Maybe somebody runs a service and picks them up and does it and does it for everybody. So maybe say a browser could could read these feeds and show it to the people. So whenever they come up so you say let us say that there's a browser profile of yourself and then the browser could recommend could recommend you all these annotations or renaissance or comments basically looking up your profile. So what we did is we implemented this particular idea of feed passing into into one of the platforms that we have been developing. It's a renaissance web web framework. Yeah. It's called LAP dot us. So what LAP LAP is about doing enabling renaissance on the web. So you could basically renaissance a maybe a paragraph of text to another to another language or the same language or maybe annotate it with say audio audio content of your in your own voice. It's not like machine machine reading or anything. It's your own voice which are doing it or maybe say putting up an image in place of in place of an in place of the text. So let us take for example this particular. So this page is about the minimum wages by the government of India which we took up to do narrations so that it could reach a lot of people because most of the government class and I mean almost all the content of the web is in English. We wanted to reach people who can't read English who can maybe who can't even even read their own mother tongue. So we took the help of a few of the NGOs and and also a law forum and then re-narrated it in all these languages. So let us check the Hindi one maybe. So maybe say again think of this as a browser profile. So your browser would know that you know only Hindi and it could suggest you all these things. So so this kind of I mean an experience when you when you come to this page it's somebody else has done this for you. Maybe a friend of you maybe a friend of yours has done this for you. And when you come to this page you can see that all these annotations come up right there for you for you to consume. Right. So in translation on Google.com it's a machine translation here. Somebody has typed it to somebody. It's a mix human based but something like Wikipedia. So what you do is you take up a document because you want to send it to your friend. You add the context to it. Generally what happens in I mean all the translators use the grammar and the context to know not really. So it is not about the document. I mean right now it's only about the web pages on on the web. It's not about no you could see the the emphasis over here is about how you could make your friend understand what the content is. It's not about translating. It is not about being in the language that I mean that you essentially want it to be. But think of it say maybe you could annotate this image and put up a different image over here. So think of this page if you say like you want to change the image over here to something else you wanted to be a table. And you take up a picture of a table and put it up over here and they send it to him. Or maybe you could remove all the text in the page and just put images or just put one audio link. One audio recording of yours and it could be the summary of the page. That can be done by the translators. No there's I mean right now there's no plug-in. What I'm saying is they could be I mean in the future they could be an extension or the browser could browser would have these capabilities. It would only be case so all this. So I'll just show you this. So all this content is coming from somebody else's blog. So as I show you over here. See there are re-narrations. Somebody has done a re-narration and the content is sitting somewhere else. It's a distributed model. It's not coming from the server. It's only case at the server. So you can see that there are. So yeah this is the X part and then there is another. This is the X part of the blog. And you can also see that there is. So here is the content that is coming from. How do you stop spam? So okay I mean another way. So you could see you could add filters and you say like because he's my friend. I would only read his annotations or his particular content. And the others I would blog. How is his different. So here see here there's a blog. So you would know like so you could say maybe maybe you could like redefine the whole filter as an XML based one and you would say like these are my friends. And as I told like think of it as a browser extension where the browser knows your profile and you would say like these are my friends. So it would have an FOF and it would say okay because these are your friends I would only push of these. Yeah we wanted to extend this particular idea to do a digital heritage war. So a digital heritage is basically when you experience the heritage on a web. So you can walk around you can walk around a particular area which has got edited and you can experience all of it on your web browser. So we took up this mural from Le Paxi which is say this one in question is particular 16th century mural. And think of it say maybe like one of your friends who could tweet about this and say this particular face over here. It looks very similar to the one found in Egyptian murals. So yeah coming back to what you asked so how I mean this this would make a lot of sense. So say your friend says like this particular face has got a very much resemblance to an Egyptian mural. So when you're walking say you're walking through the through the mural and there's a mural on your left side. And when you're looking at it so this could pop up. I mean all the annotations done for this particular mural come up and so yeah I mean something like this. So all the annotations would come up because you are looking at and you have turned on your friend's annotation. But see I mean this is index because right now we are the only one who are running this service. So you could also build up your own indexer and you could also put it up like where you don't want it anyone else to see it. Right it is. So if you go here you can find the source code. I mean we don't transform the user does it. So yeah so maybe like let us take. So let us maybe so say a user comes up to this page and he has this bar. So maybe so you could just go I mean click on this one. Say you know it and say I want to change this image. So I would get like say replace image. You could just paste and URL over here or maybe say like anything and that would be that would be how it would transform. Or also say you want to add audio to a to a particular text piece of text. So you could add audio and then just say the changes and it would be published to your blog. So you don't have to I mean store it with somebody else. It would be on on your particular blog and if you want to change the content just just ask for a reindex. Right now it's I mean it's from this URL. So if you if you use this but yeah I mean if you let us say you have a browser extension which could do that. Yeah you could see it on the place. So you said you could change that image over there. Yeah. Yeah. Now where is this edited HTML being stored. So yeah as I showed you there is so what happens is the crawler there would be a crawler which would run on and pick up pick up attributes. So it is everything is on your blog. It is not stored anywhere. So see I mean as I told you see here is a element type of audio. So here here is the blog link. So if you go to this blog link you can see a blog post which is done with this particular audio content and whenever I mean a base is requested it's only index. So we can see that the content is here. It's not being stored in the central server somewhere. Yeah. So the content is coming from here. We have not done that.