Consider, for example, the problem of search in audio podcasts. One approach is for the creator of the podcast to associate metadata with it, for instance the name of the author, the title, key words, key phrases, and categories. This approach has two drawbacks: the creator must do extra work and users will not agree on the appropriate keywords and categories.
An alternate approach is to attempt to automatically recognize the audio words in the podcast, converting them to ASCII text, which can then be searched by a conventional search engine. Speech recognition is imperfect, particularly if one has varying recording quality and different speakers, but it may be good enough to be used in matching search terms. This approach is being taken by Podscope a search engine designed to find relevant audio material.
Podscope uses technology from a company called TVEyes which also extracts the closed caption material from video transmissions to use in search.
Search engines are also being developed to find relevant material in text blogs, for example:
Blogdigger is able to search for bloggers by geographic area as well as topic.There are also search engines for podcasts like Yahoo's and Podzinger.
Video search engines that use a variety of techniques are:
For searches that consider your current context, the Web page you are reading when you search, see Yahoo Y!Q.
In general, watch Yahoo and Google's beta test pages for new search tools.
We are all familiar with Google which searches for documents containing specified search terms. Web designers may add keywords to a page using the <meta> tag, but many do not.
Other services, like flickr, del.icio.us and Yahoo My Web, encourage users to add key word tags to the Web bookmarks and images they save.
Those tags are used for searches. For example, if you go to the URL:
http://del.icio.us/tag/ajax?setcount=100you will retrieve 100 bookmarks with the tag "ajax," along with the entries of everyone who has bookmarked that page.
This article discusses user-supplied tags for search.
Note we can search for people as well as data. One frequently contacts the author of a document retrieved using Google. Tag-based systems carry this idea further by allowing users to register a personal profile, and, optionally, including their contact information (typically email address or Web URL) as part of the profile. This encourages one to contact someone who has posted interesting material. It encourages the formation of communities of common interest.
Search and tags may be used in intranets as well as on the Internet. In that way they would be used to find information and people within the organization or project.