Print

Print


Some "recommender systems" history ...

Amazon's book recommendation technology was originally provided by
NetPerceptions, the commercial spinoff of U of Minnesota's GroupLens
project (collaborative filtering of UseNet, movies and more - papers
available from http://www.cs.umn.edu/Research/GroupLens ).  Today,
NetPerceptions appears to be a shadow of its former self and their web site
(www.netperceptions.com) no longer claims Amazon as a customer.

Back in 1999, Amazon acquired Alexa Internet (www.alexa.com), which
provides the "What's Related" feature in Netscape Navigator 4.0+, along
with a downloadable Toolbar and various "related sites" report
services.   Alexa was the subject of a class-action privacy lawsuit,
recently settled for [($1.9M / no. of claimants) or $40/claimant], plus a
$100K fund for $10K donations to university public-policy programs
(http://www.alexa.com/settlement/index.html ).

Alexa's founder, Brewster Kahle, founded the Internet Archive
(www.archive.org) in 1996 and invented WAIS in 1989.  Some data for the
Internet Archive has been provided by Alexa, while most funding has been
provided by Brewster (http://www.sciam.com/0397issue/0397kahle.html
).  IA's 40 terabyte dataset is available for research study
(http://www.archive.org/internet/index.html ).

In 1995, Shardanand & Maes "Social Information Filtering: Algorithms for
automating 'word of mouth'" was presented at CHI95
(http://citeseer.nj.nec.com/context/82360/0 ).  MIT's research produced a
commercial spinoff, Firefly Network Inc.  In 1997, FireFly began offering a
cross-website, profile management tool called "Passport Office", which
provided a "Firefly Passport"
(http://www.zdnet.com/products/stories/reviews/0,4161,58324,00.html ).  In
1998, Firefly Network was purchased by Microsoft.  In 1999, Firefly.com was
shut down and their technology was incorporated into Microsoft Passport.

In July 2001, AOL began rolling out "Magic Carpet", their cross-website,
profile management response to Microsoft Passport
(http://www.betanews.com/article.php3?sid=995975421 ).  AOL also made a
$100M investment in Amazon.com, with the fine print allowing for a possible
future buyout of Amazon by AOL.  A Business 2.0 article concludes
(http://www.business2.com/articles/web/0,1653,16645,FF.html ) that the
investment was driven by Amazon's expertise in " 'click-stream analysis' -
a method of determining consumer buying patterns to offer more personalized
services".

So today, we have:

   (U of Minnesota + NetPerceptions + Alexa + Amazon + AOL Magic Carpet)

     vs.

  (MIT + Firefly + MSN + Microsoft Passport)

Minor contenders in this competition are search engines (Google, Yahoo,
AltaVista), which perform cluster analysis on user queries and indexed
pages.  Of these, Yahoo has enough cross-application _profile_ visibility
to become a significant contender to AOL & MS.  That other large online
marketplace, eBay, has signed up for MS Passport.  Their "feedback" system
has propelled them to a market cap greater than the remaining 25
publicly-traded e-commerce companies *combined*.  Included in that list of
25? Amazon.com
(http://www.thestandard.com/article/0,1902,28310,00.html?printer_friendly= ).

A recent summary of "identity" tools:

   http://www.oreillynet.com/cs/weblog/view/wlg/481

---
Rich Persaud, CEO
Autometa Corporation
388 Market Street, Suite 400, San Francisco, CA 94111
voice/fax: 415.707.2003
www.autometa.com
Who. Where. When.  Why.


At 15:32 8/1/01, Gavin Hougham wrote:
>Linton,
>
>Good point, but that begs the same question, albeit with a twist, "Would
>they grant access in order to audit the book referral programming
>algorithms they use, THEN, would they allow data access for SNA
>purposes.... maybe mount some software to ride piggyback on top of their
>system to collect SNA data directly as the referrals are being made?"  Or
>download historical data that would have little proprietary or market
>value? Actually, now that I think about it, it seems like an amazing amount
>of human decision making time would have to be spent in order to push the
>"right" books if those referrals are NOT, in fact, real.  I.e, they'd need
>to know a LOT about each of the millions of books they sell in order to
>artificially generate a tie -- time that I doubt would be offset by any
>increased sales as a result of such "point of sale" referrals.  Could be
>wrong...
>
>These questions also bring us to those exploring "administrative databases"
>for research purposes...
>
>Gavin
>
>
>At 02:34 PM 8/1/01 -0700, Linton Freeman wrote:
> >No, that isn't the question.  The data are, as Vladis said, publicly
> >available.  I think the real question has to do with validity.  Obviously
> >Amazon has an economic interest in pushing big profit or slow moving
> >books, so one wonders if the co-buying information they provide has
> >anything to do with co-buying, or is it simply embodying a sales strategy?
> >I, at least, would be wary of believing Amazon without a lot more
> >information.
> >
> >Lin
> >
> >
> >
> >
> >On Wed, 1 Aug 2001, Gavin Hougham wrote:
> >
> >> Valdis,
> >>
> >> This begs the question, would (has?) Amazon let an academic access their
> >> databases directly to answer these kinds of questions?  You never know...
> >>
> >> Gavin Hougham
> >>
> >>
> >>
> >> At 04:52 PM 8/1/01 -0400, Valdis wrote:
> >> >If you are looking for network data try amazon.com.  They have great
> >> >network data for books -- those interested in two mode nets can look at
> >> >books and authors.  For those interested in how networks change over
> >> >time mine the data on amazon's site weekly, monthly, etc. and you will
> >> >see different connections and different nodes emerge.  Amazon has data
> >> >available for both ego networks and larger group networks.
> >> >
> >> >The key is in Amazon's 'people who bought this book also bought...'
> >> >data.   Put enough of these 'also bought' lists together and you have a
> >> >nice network.  For an example of book ego network see:
> >> >http://www.orgnet.com/buzznet.html
> >> >
> >> >For those that have been wondering...
> >> >
> >> >* Do Ron Burt's books span structural holes?
> >> >    - Start your data gathering here:
> >>
> >http://www.amazon.com/exec/obidos/tg/stores/detail/-/books/0674843711/simil
> >> arities/
> >> >
> >> >* Does Wayne Baker's book on Social Capital *have* social capital?
> >> >    - Start your data gathering here:
> >>
> >http://www.amazon.com/exec/obidos/tg/stores/detail/-/books/0787953091/simil
> >> arities/
> >> >
> >> >* Are Barry Wellman's books embedded in communities?
> >> >    - Start your data gathering here:
> >>
> >http://www.amazon.com/exec/obidos/tg/stores/detail/-/books/0813368219/simil
> >> arities/
> >> >
> >> >* Is Tom Valente's book in a position of diffusion?
> >> >    - Start your data gathering here:
> >>
> >http://www.amazon.com/exec/obidos/tg/stores/detail/-/books/1881303225/simil
> >> arities/
> >> >
> >> >Other 'interesting' questions that can be answered...
> >> >Which is the most central social network book?
> >> >Which SNA books are structurally equivalent?
> >> >Which SNA books form a clique?
> >> >What SNA emergent clusters have formed?
> >> >Which other 'communities of interest' is the SNA community connected to?
> >> >
> >> >If anyone pursues these questions, please share your results with the
> >> >rest of us.
> >> >
> >> >Enjoy!
> >> >
> >> >Valdis
> >> >
> >> >[log in to unmask]
> >> >http://www.orgnet.com
> >> >
> >> >
> >>
> >
> >--------------------------------------------------------------------------
> >Linton C. Freeman                         Phone: Office: 949 824-6698
> >Department of Sociology                   Sec.:          949 824-6336
> >School of Social Sciences  SSPA 2143      Home:          949 494-6139
> >University of California                  FAX:           949 824-3733
> >Irvine, CA 92697-5100                     E-mail: [log in to unmask]
> >http://eclectic.ss.uci.edu/~lin/lin.html
> >---------------------------------------------------------------------------
> >
> >
>-------------------------------------------------
>Gavin Hougham, M.A., Ph.D. (cand)
>The University of Chicago
>Project Director, Department of Medicine/Section of Geriatrics, and
>Adjunct Instructor, Graduate School of Social Service Administration
>5841 S. Maryland Ave. (MC 6098)
>Chicago, IL 60637
>
>(773) 702-6735 office
>(773) 702-3538 fax
>[log in to unmask]
>http://www.geocities.com/ghougham
>-------------------------------------------------