{"@context":"http://iiif.io/api/presentation/3/context.json","id":"https://amiastreaming.aviaryplatform.com/iiif/z02z31p054/manifest","type":"Manifest","label":{"en":["Tagasauris: The HyperTED Project"]},"logo":"https://d9jk7wjtjpu5g.cloudfront.net/organizations/logo_images/000/000/016/original/AMIA-Logo-17.jpg?1556650205","metadata":[{"label":{"en":["Type"]},"value":{"en":["Presentation"]}},{"label":{"en":["Coverage"]},"value":{"en":["Museum of Modern Art (Place of Recording)","New York, NY (USA) (Place of Recording)"]}},{"label":{"en":["Date"]},"value":{"en":["2015-05-08 (created)"]}},{"label":{"en":["Agent"]},"value":{"en":["Todd Carter, Founder/ CEO of Tagasauris Inc (Speaker)","Chris Lacinak (Programmer)"]}},{"label":{"en":["Publisher"]},"value":{"en":["Association of Moving Image Archivists"]}},{"label":{"en":["Description"]},"value":{"en":["\u003cp\u003eBuilding off of TED Talks, HyperTED was created to explore how ideas could be interlinked and consumed at the concept level, further pushing the idea of relationship aware hypervideo and allowing users the opportunity to explore over 1681 inspirational TED talks at the fragment level. This fine level of granularity  allows for the non-linear exploration and sharing of online media. Chapters and particular regions of interest are automatically created, sequentially organized and used to delimit important sections of the narrative flow within a given TED talk. HyperTED also utilizes “Hot Spots” in order to highlight important concepts and topics which can then serve as the basis for recommendations on other talks and interlinked concepts. Furthermore, HyperTED is enriched with background knowledge from DBpedia, Freebase and online educational resources from catalogs provided by the Open University and Open Courseware. Outcomes of the HyperTED project include:\u003c/p\u003e\r\n\r\n4.5X amplification beyond the Google Search index\r\n30% increase in total video views by suggesting relevant content based on the Hot Spot recommender system\r\n20% increase in inbound traffic from Social Network Services\r\n20% increase in time spent on the site per visit\r\n (general)"]}},{"label":{"en":["Language"]},"value":{"en":["English (Primary)"]}}],"summary":{"en":["\u003cp\u003eBuilding off of TED Talks, HyperTED was created to explore how ideas could be interlinked and consumed at the concept level, further pushing the idea of relationship aware hypervideo and allowing users the opportunity to explore over 1681 inspirational TED talks at the fragment level. This fine level of granularity\u0026nbsp; allows for the non-linear exploration and sharing of online media. Chapters and particular regions of interest are automatically created, sequentially organized and used to delimit important sections of the narrative flow within a given TED talk. HyperTED also utilizes \u0026ldquo;Hot Spots\u0026rdquo; in order to highlight important concepts and topics which can then serve as the basis for recommendations on other talks and interlinked concepts. Furthermore, HyperTED is enriched with background knowledge from DBpedia, Freebase and online educational resources from catalogs provided by the Open University and Open Courseware. Outcomes of the HyperTED project include:\u003c/p\u003e\r\n\u003cul\u003e\r\n\u003cli\u003e4.5X amplification beyond the Google Search index\u003c/li\u003e\r\n\u003cli\u003e30% increase in total video views by suggesting relevant content based on the Hot Spot recommender system\u003c/li\u003e\r\n\u003cli\u003e20% increase in inbound traffic from Social Network Services\u003c/li\u003e\r\n\u003cli\u003e20% increase in time spent on the site per visit\u003c/li\u003e\r\n\u003c/ul\u003e"]},"provider":[{"id":"https://amiastreaming.aviaryplatform.com/aboutus","type":"Agent","label":{"en":["AMIAstreaming"]},"homepage":[{"id":"https://amiastreaming.aviaryplatform.com/","type":"Text","label":{"en":["AMIAstreaming"]},"format":"text/html"}],"logo":[{"id":"https://d9jk7wjtjpu5g.cloudfront.net/organizations/logo_images/000/000/016/original/AMIA-Logo-17.jpg?1556650205","type":"Image"}]}],"thumbnail":[{"id":"https://d9jk7wjtjpu5g.cloudfront.net/collection_resource_files/thumbnails/000/035/691/small/Todd_Carter_Final.mp4_1556717272.jpg?1556717273","type":"Image","format":"image/jpeg"}],"items":[{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691","type":"Canvas","label":{"en":["Media File 1 of 1 - Todd_Carter_Final.mp4"]},"duration":607.27333,"width":640,"height":360,"thumbnail":[{"id":"https://d9jk7wjtjpu5g.cloudfront.net/collection_resource_files/thumbnails/000/035/691/small/Todd_Carter_Final.mp4_1556717272.jpg?1556717273","type":"Image","format":"image/jpeg"}],"items":[{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/content/1","type":"AnnotationPage","items":[{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/content/1/annotation/1","type":"Annotation","motivation":"painting","body":{"id":"https://aviary-p-amiastreaming.s3.wasabisys.com/collection_resource_files/resource_files/000/035/691/original/Todd_Carter_Final.mp4?1556717272","type":"Video","format":"video/mp4","duration":607.27333,"width":640,"height":360},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691","metadata":[]}]}],"annotations":[{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644","type":"AnnotationPage","label":{"en":["AUTO_Todd_Carter_Final.mp4 [Transcript]"]},"items":[{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/1","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"we think that video is the new culture role vernacular and that the technology is the key of the key capability needed to bridge the gap between people and the %HESITATION content that surrounds the things that they're interested in it and passionate about ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=9.38,25.49"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/2","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"so how do we do this %HESITATION and this is sort of background set up for the hyper Ted talk ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=28.61,34.28"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/3","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"so come ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=35.03,36.04"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/4","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"there's three parts of our system one human computation the idea that we can get humans and computers to work together in a cooperative manner to achieve outcomes that neither constituency could achieve in isolation so that is effectively machine learning meets crowdsourcing and the machine learning world it's computer vision and natural language processing and crowds humans humans AB this sort of interpretive layer this contextual layer that machines often are extremely poor at doing and so together we can do things like created genome for Disney television where we know character character character action relationships on a scene by scene or shot by shot basis because humans can fill in where the machine's up kind of balk at that task the second part of this is semantics right so we think of the world as a connected graph that your computer can inference across both forward and backward and so we use that triple store RDF %HESITATION that supports the inference seeing %HESITATION that triple store has models %HESITATION has vocabularies in schema %HESITATION and semantics so we model what it means to be a story line or to be a media object onto a week when we look at video we're decomposing video into a model and then persisting it in that's decompose state and then so finally this a I layer so that allows us basically a smart way to smart way to annotate smart way to store and and a process that allows us in this is key to this whole system this process allows us to capture consolidated machine expert our expert and end user perspectives for the continuous in Richmond improvement of the data so I'm ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=37.5,140.74"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/5","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"this leads us to hyper cat ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=141.83,143.07"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/6","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"so hyper Ted was a project in a way set the stage for a lot of this work that we're doing and we did with the Disney GM folks so it's a research collaboration between us between your calm in the south of France %HESITATION between media mixer between the link TV projects and the idea was we wanted to ingest the Ted corpus of talks and make that information more easily searchable more easily discoverable and more easily sharable %HESITATION ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=144.28,175.3"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/7","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"and so ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=176.2,177.31"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/8","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"here's the problem space right searching for the relevant portion of the of the relevant portion of the of of content in a particular video talk is almost impossible right because we currently treat video is a unitary object on the web we don't index it at the level of Chater scene granularity so Ted we said with hybrid which we're gonna break that mold we're gonna break this video down into regions of interest in this is a a screen shot of %HESITATION talk %HESITATION and I'll take you through that in short order what all the different sections are so there's a few core technologies that went into making this work I'm and again I mentioned the idea that we worked aggressively with European and web standards bodies we don't want to do this in a black box technology this is about accessibility this is about making web a first are making video a first class object on the web so just as text is a first class object so we use the media fragments specification this allows us to address spatial temporal regions of video of through a browser so Firefox safari chrome they know how to resolve those things %HESITATION media fragment creation the idea that you gonna play video and then use computer vision and technology to break it down automatically into shots and scenes and then media fragment description the idea that you're gonna take those decomposed lego parts right we're gonna take a video we're gonna turn it into a bunch of legos that we're gonna describe each of the lego parts %HESITATION and that's what media fragment description is ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=178.39,268.64"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/9","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"so there we mention automatic %HESITATION video shot scene %HESITATION and concept of Texas on a distant actually breaking it into shots and scenes and this is logically breaking or not physically chopping logically breaking it up we're also using computer vision to extract about two hundred and thirty two different concepts automatically we built a little media player that knows how to read and parks those fragments in HTML five and a no J. S. framework for %HESITATION serving that all up all very very efficiently ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=270.96,301.54"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/10","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"so here's the sort of story nineteen eighty four the the gig a tad was you pay them ten thousand dollars you get to watch the conference and they send you home with a D. V. D. that was the model and %HESITATION there was a time when it was thought to put they thought about putting Ted online in the folks from Ted were like Chris Anderson was like no one's gonna calm if you do have we won't get the DVD but they put it online in two thousand six in and obviously the rest is history it's become a very ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=302.52,330.45"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/11","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"famous a place to go for education and inspiration and so we said well in two thousand fourteen it's time to do it again what if we could take the videos and break it down into chapters and hot spots and extract entities and link to related data and allow people to kind of traverse that whole thing %HESITATION automatically ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=332.32,352.29"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/12","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"so I'm ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=353.49,354.67"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/13","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"that's what we do so the project takes a a ten piece of Ted video and its associated transcript %HESITATION we use that those Ted talks to to break it into sort of this human made subdivision of titles the transcript itself we reflect that that pattern in the chapters which we create automatically we run the tax to through ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=355.53,378.65"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/14","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"at your com they've developed a tool called nerd it it's an ensemble are for entity extraction it allows you to run multiple text extractors against ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=379.49,390.57"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/15","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"a piece of text and in doing so increases the precision and recall %HESITATION so we run it through an ensemble you can see we're extracting words %HESITATION down with the arrows pointing down to named entities the words are color coded ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=392.28,406.13"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/16","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"because in fact we're not just extracting them as tax but we're we're we're typing them so this is a person this is the location this is an organization and then we're using tools like tax raiser for topic modeling so not only are we extracting entities but we're also modeling the topics ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=406.97,421.68"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/17","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"some fancy math the bottom line is what we're gonna do is automatically compute regions of interest in the video from the underlying tax so this idea of hot spots is the is a big deal %HESITATION we didn't just want the hot spots to be sort of arbitrary words that were time coded but we want them to be you know places where there were true interest events so maybe it's Tim Berners Lee giving his talk on linked open data at Ted I'm that part where he talks about link open data we want that to be a hot spot this allows us to do that ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=424.28,457.11"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/18","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"we hyperlink all that together %HESITATION we build a recommender systems of the cool thing about the recommender system is it now isn't working off the video is a unitary object it's working off the actual fragments off I have a pic a piece of video a ten second window where Tim early is talking about %HESITATION is talking about link open down and I fire the recommender system I get back ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=460.8,482.34"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/19","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"%HESITATION other clips that are related to that specific clip so we've we've now started to make video extremely granular and this is where I'm just describing this idea of almost like lego parts where you take video you decomposed into a series of sort of standardized building blocks that you can then reassemble for linear play out but also for non linear play out so if you want to have bite size content ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=483.69,508.25"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/20","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"or have your favorite up funny Patrick's you could sort of literally roll your own from the endless a content that uploaded to YouTube every day and then as a final step you know it is called linked open data so the cool thing about using it as a ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=509.09,527.54"},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/21","type":"Annotation","motivation":"transcribing","body":{"type":"TextualBody","value":"as the technology is the idea of linking to things so we thought that we should take these videos and make them connect to other important places on the web where %HESITATION online course is %HESITATION you know other kinds of destinations and so there's an an ability to do that and so in this way you know through this research project we we really outlined I think really the full complement of technologies that are required to be converged in the cloud some kind of cloud based offering that allow you to begin to think about processing audio visual information at scale I think the thing it's exciting for us about the hypertext project is completely automated this is done in a complete automated pass and then you could let your in that human intelligence later on so it gives you a way to think about processing real time video very very efficiently and then layering in the interpretive the uniquely human aspects the interpretive component as a as a back in process ","format":"text/plain"},"target":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691#t=528.6,595.41"}]},{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644","type":"AnnotationPage","label":{"en":["English [Transcript]"]},"items":[{"id":"https://amiastreaming.aviaryplatform.com/collections/48/collection_resources/4798/file/35691/transcript/3644/annotation/22","type":"Annotation","motivation":"subtitling","body":{"type":"TextualBody","value":"https://d9jk7wjtjpu5g.cloudfront.net/file_transcripts/associated_files/000/003/644/original/transcript_1558512387.vtt20190522-6031-1vqxm3x?1558512387","format":"text/vtt","language":"en"},"target":"https://d9jk7wjtjpu5g.cloudfront.net/file_transcripts/associated_files/000/003/644/original/transcript_1558512387.vtt20190522-6031-1vqxm3x?1558512387"}]}]}]}