Web-based video search engines allow users to enter keywords into a search box, just as one would into a Yahoo! or Google search box. However, rather than get back Web pages, users are provided related video clips from across the Web. While traditional search engines are skilled at indexing, understanding and finding text-based content, they are inadequate for finding video content results. They focus only on textual or metadata within web pages rather than looking at actual video files themselves. Video search engines have emerged to compensate for the weakness of such straight HTML-focused search engines.
Today, the field of online video search is rapidly-evolving—an overview of the evolution of video search (from first to second generation) follows.
First generation video search
First generation video search solutions depended entirely on metadata. Including examples are SingingFish, Alta vista Video (now used at Yahoo!). These engines are extremely similar to regular web search engines. Just as with a standard web search engine, the spider propagates across the Internet, recording and looking for content to index. Unlike a standard web search engine, text documents and pages are ignored and the spider focuses instead only on video (and sometimes audio) content. Once such content types are discovered they are examined for relevant metadata. Metadata is the textual data that is applied to a piece of multimedia content in order to describe it and can include user-provided tags, an editorially written title or summary, a transcript of the speech in the video or even information stored in the video file itself pertaining to its resolution, frame-rate and creation date.
Still part of the first generation, but much improved, display-oriented spidering has been used to great effect in video search. First developed for the closely related problem of image and photo search, display-oriented spidering looks at the web page text that lies near a video. Using a specialized algorithm, display-spidering evaluates the physical attributes of the way the page is designed and rendered to decide which portions of it are closely related or linked to the video. It then extracts the text within these areas and applies them, as further metadata, to the videos being indexed. As many web pages contain commentary or description that is related to the video but may not be contained in the official metadata, this approach can provide more detail on the meaning of the video being spidered. The best example of display-oriented spidering for video search today is that found at AOL’s SearchVideo.com.
However, whether augmented with display-oriented analysis or not, the methodology of first-generation, metadata spidering is still flawed because the engines still rely heavily upon the quality of the metadata that has been provided. As the metadata is often provided as an afterthought, it may be incomplete or lacking in detail and, as it is provided by the owner or publisher of the video, may even be false or misleading. First generation video search is a reasonable solution that borrows on existing web search technology to simplify the video search problem. By doing so, however, it limits itself to never actually understand an actual video, but rather focusing only on pieces of text that may be related to the video but are, fundamentally, of second order to it.
Second generation video search
Second generation video search engines emerged as a reaction to the faults of the first generation. As well as spidering textual meta data, second generation video search aims to understand and extract meaning from the video itself.
Second generation video search engines use methods such as speech recognition, visual analysis and recognition and video optical character recognition to allow software to listen to, watch and read the text appearing on the video content itself. As well as providing more information, this approach provides objective information—if a video contains speech on a particular topic, it really is about that topic, whereas if a video has been tagged as pertaining to a certain topic, it may, actually be about something entirely different.
Second generation video search is still primarily used in government and enterprise settings, but blinkx. com and Pod zinger exist as examples of technologies that have been applied to general, consumer Web video search. Pod zinger, as the name suggests, focuses more of audio and video podcasts, while blinkx indexes all audio and video content on the Web, whether amateur or professional.
Regardless of the technology involved, both first and second generation video search engines exist and are popular today. For the purposes of a successful video SEO campaign, it is important to be included in both types of engine.
Direct to India is one of the seo offshore organization providing a video search engine optimization and having a big team of search engine and video search engine optimization for getting more detain on SEO service and for free SEO analysis pls. contact us
0 comments
Post a Comment