Search Engine on TV…Tuesday, August 26, 2008 13:23
This is an idea which is doing rounds in my mind for quite a while. I tried promoting it but people could not see it realizing. So, I am just posting it out here. If any VC is reading and likes it, please mail me.. and if any developer likes it and gets it through, atleast drop me a word of appreciation if nothing else is possible.
Okay, After reading this post, you might say that whatz so new about this? this already happens on STB using the metadata. but then, I have left some hint towards the end of the post what doesn;t exist and I want to achieve. If you read carefully & get the feel of the actual stuff, you might like to take it forward. Ofcourse, I have taken out many important pieces in this post for obvious reasons.
Search Engine on TV is one thing that looks similar to search on internet and it is. Till now, we used to have search based on program name, time, actor, channel etc. and now, is the time to move one step forward to search the content of the program and choose the content among different channels. This paper deals with a concept of searching the whole content at the broadcaster’s end. The user just needs to key in the keywords he wants to search and he gets the search results on TV Screen. The user, then, can play the item selecting from the search. Of course, the catch is that the content can contain only the results from the programs or channels permitted by the broadcaster and user may need to pay for the item to be seen. This paper is attempted to create a search engine of the internet world in TV domain.
One day I was watching TV and I was looking for particular news I wanted to know about but there are too many news channels to look for particular news. And after many tries also, I failed to locate for particular news or its series that made me log on to the internet and search for it. And it took me a while on the internet to get the complete story. Then, also, I did miss the videos of the news.
Of course, this paper makes some assumptions before going into the details of the implementation. The major assumption made is that it is targeted for the IP-TV world or the STB’s with a high bandwidth return path.
Now, the whole concept of search is based on the broadcaster’s support where the broadcaster can have a new business model itself. The broadcaster will maintain a database of the broadcasted programs and store all the information like program names, synopsis, descriptions, closed-caption text, subtitles etc. in the indexed format like all the other search engines.
As a user, if I wish to look for any particular program which may have mentioned a word or a phrase. I can just input that in a search box on EPG UI and get back the search result containing all the programs listed which contained the reference to my search string. Now, I should be able to play the program and view the content.
The business model can be that the broadcaster returns the search result and let the user have the preview of the program and pop-up a buy message if the user continues to watch it. And couple of other business models can be drawn around this search engine.
Let’s consider a simple example of a database where we’ll maintain an index of all the keywords from the metadata. Now, this database can be searched for the search string given by the user and with the help of metadata, the actual item can be found and displayed as a search result.
Now, this database will contain all the information about the broadcasts and will be maintained by the broadcaster.
How this database can be populated? It can be done with the help of either the bots searching the program, schedule, catalogue, closed captioning text, subtitles, and any other information available for the program in the textual format and will be indexed in the database. These bots will run on the periodic basis from time to time and will keep on updating the database. These bots are nothing but automated programs running on the broadcaster’s side to ensure that every single byte of data is captured and stored so that the search can be more relevant.
Now, as the user keys in the search string, it goes back to the broadcaster over IP or the return path and it is being searched in the database for relevant results. And then, the results are returned to the user in displayable format.
The database has to be indexed regularly so that search is fast and appropriate. It will be just like any other search engine on the internet. The only difference it will make is that it this search engine will be based on real time information.
One more addition can be done to this concept is of speech recognition. If the information is not available in text format, then the bots can make use of speech recognition programs to identify the text spoken in the programs and store them in the database. But as speech recognition as high limitations, this can be used as a back up by the bots in case of unavailability of the textual information.
.Quote of the Day:"If there are any questions, direct them to that brick wall over there." -- Network President