Sunday, August 30, 2009

Real-time unified search

It is surprising that a unified search application across different types of web content does not yet exist.

According to OneRiot, 40% of web searches at present are for real-time content such as that from Twitter, FaceBook, PeopleBrowsr, digg, bookmarking sites, blogging and microblogging sites (friendfeed, etc.).

One vision of a unified search 'megafeed' app would be a customizable html page with search across many types of web content, automatically updated and delivered together or organized into categories such as events, articles, comments, people, etc. The types of web content to search would be:

  • Traditional web search: Google, Bing, etc., which could be more richly granularized with content-tagging per a variety of parameters such as information type (news, blog, video, book, event, etc.), time (time added to web, time of occurrence), original vs. subsequent post and other distinctions.
  • Real-time web search: The emerging real-time content search engines should be extended and unified into one digital social interaction feed for Twitter, FaceBook, LinkedIn, bookmarking, email, blogging, microblogging and possibly IM/SMS notification AND response. User-permissioned credentials can be browser-stored for such a unified action platform. In addition to usability, the fast-growing real-time web search companies are also focused on monetization, reinventing generating AdSense-like models.
  • Local search: New restaurant and retail notifications, events, craigslist and other commercial postings of interest, friends traveling to the area (links to feeds from GeckoGo, Dopplr and other travel social networking and public calendaring websites).
  • Academic search: notification of new papers, articles or news. Federated PubMed, ArXiv-like journal portals are needed for all academic fields, including economics and liberal arts.
  • Multimedia search: notification of non-text postings of photo, music, podcast and video content.

Content demand mechanisms: ambience to supersede keywords
Content could be searched by the usual user-entered keywords or a deeper variety of content demand-interaction mechanisms could be developed, for example, permissioning-in by users such that ambient profiles from hard-drive content and previous web interactions automatically form and evolve (a precursor to pre-AI web interactions).

blog comments powered by Disqus