IL05: Fueling Engines for the Future

DeWitt Clinton, Software Development Engineer, A9.com

David Mandelbrot, Vice President of Search Content, Yahoo!

Peter Norvig, Director of Search Quality, Google

A9.com

  • About
    • Subsidiary of Amazon.com
    • Founded in October 2003
    • Based in Palo Alto, CA
    • Powers amazon product search
  • Google supplies both Web and image search results
  • Growing a9
    • Started with web, images, books
    • Added Wikipedia, ref, yellow pages, moves, & more
    • Most search engine APIs are similar albeit proprietary
    • Which is why we introduced…
  • OpenSearch
    • It’s simple search syndication
  • Behind OpenSearch
    • Propose a common format for queries and results
    • Identified the minimal subset of data necessary
    • Reuse existing and familiar standards such as RSS
  • A9.com – Click “choose more columns”
    • Currently 298 choices
    • Flickr photo search [M: !!!]
    • PubMed OpenSearch
  • OpenSearch response in XML/RSS format
  • OpenSearchLaunch
    • March 2005
    • More than one/day added
    • Creative commons licensed
  • OpenSearch 1.1
    • Built into IE7
    • Flexible syndication formats including Atom support
    • Extensibility (can work with SRW/U)
  • http://opensearch.a9.com/
  • Seattle PL is doing this

Google: Research Search Innovations

  • GoogleAnswers
    • Type in factual question, get answer and source
  • What you might be looking for
    • javascript not
    • javascript not operator
  • Statistical Machine Translation
    • Translation on the fly of results
    • In research right now
    • Underlined words = not sure
    • Effects of more data
      • More words in data, better translations
  • Google Mobile
    • Local search on phone
  • Google Maps
    • Uses Ajax
    • Satellite results
    • Moon (no directions yet)
    • Integrating additional data
      • Katrina
      • Seattle911.com
      • Urinal dot net
      • New York in the Movies
      • Brewster Jennings Projects America
      • PlaceOpedia (Wikipedia place aritcles)

Yahoo!

· Innovation acceleration

· FUSE

o Enable people to Find, Use, Share and Expand all human knowledge

· Challenge

o Attempts to find “all human knowledge” didn’t include for-pay content sources – couldn’t find everything

o Search Subscriptions

§ Searching popular for-pay content

§ Personalization allows users to always get for-pay content

§ Feed from partner web sites ensure…

· Challenge: once you find content, how can you use it?

o Find pic, can I post it on my page?

§ Search for Creative Commons

§ Licencing system

§ Yahoo created interface to all users to search CC content

§ Then users know they can use it

§ Feature allows search based on type of use

· Challenge: enable users to share knowledge with their community to create a better search experience

o Limit to number of useful relevant results

o How can I share what I’ve learned from my searches

o My Web 2.0 (“social search”)

§ Save results

§ Tag results

§ Share results

· Challenge: expanding the amount of content made openly available online while not upsetting the ecosystem

o Open Content Alliance

§ Joint effort

§ Approval of copyright holders

§ Multimedia & text

§ Full text rather and snippets

§ Freely crawlable

§ International effort

§ Uses common formats

Leave a Reply

Your email address will not be published. Required fields are marked *