If you went to the MIT Sloan CIO Symposium in the spring and want to review some of the discussions – or maybe you didn’t have a chance to attend and want to know what you missed about the future role of CIOs – there’s a cool way to catch up.

Head over to the MIT Sloan CIO Symposium On Demand site, and you’ll have access to videos of panel discussions at the conference, and transcripts of these events, as well. But what’s interesting is that it’s powered by technology from 3PlayMedia, a company founded by MIT grad students and drawing on research from the MIT Spoken Language Systems Group, that makes those transcripts intelligent and interactive.

Click on a word in the transcript and you are taken right to that spot in the video, so that you can get to the part you want to see in a snap. Additionally, while watching a video you can see each word in the text highlighted as it’s spoken, which Symposium chair and Fellow at the MIT Sloan School of Management Graham G. Rong says can be particularly helpful for viewers whose mother tongue isn’t English. You also can highlight a section of text and from there automatically get the corresponding video snippet and a unique URL of it for social sharing.

Click on a word in the transcript and you are taken right to that spot in the video, so that you can get to the part you want to see in a snap. Additionally, while watching a video you can see each word in the text highlighted as it’s spoken, which Symposium chair and Fellow at the MIT Sloan School of Management Graham G. Rong says can be particularly helpful for viewers whose mother tongue isn’t English. You also can highlight a section of text and from there automatically get the corresponding video snippet and a unique URL of it for social sharing.

“The key here is that when they generate the text transcript they also embed a time reference,” says Rong.  “In this case, the introduction of the time reference synchronizes the video or audio with the text.” Automated speech recognition generates the draft text transcript and then supervised adaptive learning enters the picture where humans review every word and, only when necessary, make adjustments. “In artificial intelligence there are two types of learning, unsupervised and supervised,” Rong says. The former leans to estimating if an evaluation of a result is or isn’t accurate, while supervised adaptive learning lead to more exact results.

Of course, people also can perform searches for a specific word or phrase on the transcript text. And as Rong sees it, semantic web technology potentially has a role to play to making the search of video transcriptions, regardless of vendor, even more powerful and accurate with the application of domain ontologies. “In that case, when you search something you can search a concept instead of just a term,” he says.

On top of that, they could be applied, along with adaptive learning, to making the automated speech-to-text process even more accurate in the first round. This could speed the human editing process by correctly applying ‘skull’ rather than ‘scull,’ for instance, when transcribing a video about the medical sector, and vice verse when discussing the boating industry. “There’s a lot of room for domain specific ontologies to apply,” he says.

 

 

 

SemTechBiz is Less Than 3 Weeks Away

The Semantic Tech & Business Conference (SemTechBiz) is coming to San Francisco on June 3-7! Join us for case studies, innovative panels, tutorials, and keynotes that will provide you with practical advice, hands-on guidance, and breakthrough approaches to solving business problems with semantic technology. Passes go up $200 at the door. Sign up now and save !