Emulating Amazon: what lessons can we learn what features can we implement EndUser 2006 - PowerPoint PPT Presentation

1 / 79
About This Presentation
Title:

Emulating Amazon: what lessons can we learn what features can we implement EndUser 2006

Description:

Emulating Amazon: what lessons can we learn what features can we implement EndUser 2006 – PowerPoint PPT presentation

Number of Views:55
Avg rating:3.0/5.0
Slides: 80
Provided by: jimrob6
Category:

less

Transcript and Presenter's Notes

Title: Emulating Amazon: what lessons can we learn what features can we implement EndUser 2006


1
Emulating Amazon what lessons can we learn
what features can we implement EndUser 2006
  • Jim Robertson
  • Assistant University Librarian
  • NJIT (New Jersey Institute of Technology)
  • www.library.njit.edu/staff-folders/robertson/pr
    esentations/

2
Abstract
  • Amazon has raised the bar for libraries with
    their innovations in presenting rich added-value
    information, fostering community, and presenting
    actionable options for users.
  • What lessons can we learn from a thoughtful
    analysis of Amazon? How can we apply those
    lessons to our libraries?
  • Jim Robertson will analyze the Amazon website
    with an eye towards bringing its principles into
    our libraries. Jim will demonstrate how the NJIT
    library has enabled some of these and will
    present a new Web Services paradigm for our
    catalogs.

3
Success of Amazon
4
Success of Amazon
5
Amazon the 300 lb gorilla
6
Amazon
7
Amazon
Remembers who you are (personalization)
For you (personalization)
Ubiquitous search box
Book covers (graphical content)
Aggregate rating
Add to YOUR content (customization)
Suggestions (linked content)
8
Amazon
9
Amazon
Suggestions for additional content based on usage
(purchasing)
Tell a friend (share this)
Reviews (floated to the top because of quality
decision)
10
Amazon
11
Amazon
Popularity (inference of quality) (usage)
Suggestion for more content based on usage
(searching behavior)
12
Amazon
13
Amazon
Tagging / sharing (wisdom of the crowds,
distributed cataloging)
Reviews by customers (filtered to the top based
on quality indicators of the reviewer, signed
reviews) (content built by the users)
14
Amazon
15
Amazon
Other customer reviews (pushed down by quality
inference)
Feedback loop (how are we doing?)
16
Amazon
17
Amazon
Chat about the book share discuss learn
Comments and more users help build the content
Customer-authored bibliographies participation
18
Amazon
19
Amazon
Put theory into action move book learning into
experience customer-authored
Multiple categories to search and filter upon
Multiple categories to search and filter upon
20
Amazon
21
Amazon
Feedback
What are the status of your current processes,
requests, orders ?
Books viewed
Terms searched
Suggestions for more content (based upon your
history)
22
Amazon
Some books have excerpts
23
Amazon
View pages
Statistical text analysis of books to summarize
content (automatic cataloging)
Search inside the book
View pages
Text cloud to graphically summarize content
24
Amazon
  • What do nearly all of these things have in
    common?
  • Foster community
  • I am here / Others are here
  • I did this / Others did that
  • I think this / Others think that
  • Interact with each other (albeit asynchronously
    and mostly anonymously)
  • Share / participate / peer-to-peer teaching and
    learning
  • Provide added-value to the book record
  • More ways to search, find, discover
  • More ways to learn
  • Present the user options and actionable
    choices
  • Participate / contribute
  • Give feedback
  • Build content for self or others
  • Buy / read
  • Customize

25
Our physical libraries
  • Places to discover and explore
  • Places to communicate and exchange ideas
  • Where and how do our digital libraries enable
    communication and exchange of ideas?

26
Our catalogs
27
Our catalogs
28
Our catalogs
  • What are we missing?
  • No sign of any community or activity or
    footprints
  • Little added-value information in the book
    record
  • Few to none user options or actionable choices
  • Aside from putting a request in or writing down
    the call number to go to the stacks, our catalogs
    are dead ends.
  • Neither good ways to link out of them nor to link
    in to them.

29
We have options
30
We have options
Jims law 2 A passionate, diverse, creative,
and oft-times desperate user base will always
out-power and out-pace vendors in pushing
innovations.
Jims law 1 If it can be hacked, it will be
hacked.
31
NJIT catalog
32
NJIT catalog
Top half of the page much like most catalogs
nothing new
33
NJIT catalog
The bottom half has extensive customization
34
NJIT catalog
Some book reviews, with a link to more
Book covers
Persistent links
Tag, store, and share
See how this was used evidence of others
Link to local libraries, OCLC, Google, etc.
35
NJIT catalog book reviews
Link opens window directly to Amazon.com page
with all reviews on it
36
NJIT catalog book reviews
Invitation to participate, to build to the
content opens window to the book on Amazon
37
NJIT catalog tagging
Opens del.icio.us account and sends custom link
to del.icio.us includes persistent shortcut
link title / author (book in NJIT library
phrase), ISBN
Upon save, returns user back to book in Voyager
38
NJIT catalog tagging
Book now tagged in your del.icio.us account
This works because of our NJIT permalink using
the ISBN
39
Shortcut links (PURLs)
Code that includes the image invokes a ColdFusion
book image resolver page. (More on this later.)
40
Shortcut links (PURLs)
Why do we need? Because OPAC does not have a
link-to-this-item feature and a
copy-and-paste action on the URL in the browsers
address bar will fail in the future due to timing
out. Plus, it is not human-readable
Allows users to code shortcut searches into
Voyager without having to know all the
cgi-bin/Pwebrecon.cgi syntax.
www.library.njit.edu/catalog/shortcut.cfm?isbn030
0093837
41
Shortcut links (PURLs)
  • Put the complexity on the back end! Dont
    torture the user.
  • Each bit of discrete intellectual content
    should be permanently addressable.

42
Reuse, share
43
Tags in Amazon
Amazon is starting to experiment with tagging
this may be an option to pursue, too
www.amazon.com/gp/tagging/glance/ajax/
44
NJIT catalog - usage statistics
Lists each copys location and circ type. Also
date added to collection.
Lists each loan, day borrowed and returned, days
out, renewals, borrower class.
Summarizes days available vs. days borrowed.
45
NJIT catalog - usage statistics
  • Potential to be used for
  •      Purchase and acquisition of additional or
    replacement copies
  •      Weeding
  •      Migrating on or off reserve
  •      Changing lending terms
  •      Adjusting book budgets
  • Calculating popularity for search results
    display
  • Also, this is evidence of community these are
    digital footprints in the catalog.

46
NJIT catalog local libraries
Opens new browser window to check local libraries
for copies of the book by ISBN
47
NJIT catalog local libraries
Checks Google, too
48
NJIT catalog live help
Opens chat session with NJIT libraries who are
automatically able to see the users search
history and screens done by mining
OPAC_SEARCH_LOG tables see my other EndUser
presentation for more details
Actionable choice brings together community of
users and librarians
49
NJIT catalog
50
Book covers in NJIT library catalog
NJIT has a local database of book cover images
51
Virtual new books shelf
Images pulled from our book cover image repository
Metadata grabbed on-the-fly from our catalog
52
Library homepage
Display on library web page, also. Random book
each time page is loaded.
Re-use, re-purpose, re-context
53
Journals in the catalog
Live tables of contents for current issue via RSS
feed.
Dont catalog! Start fetching.
54
NJIT catalog
  • Evidence of users and usage (usage data,
    reviews, Live Help)
  • Participation via reviews and tagging
  • Value-added information via reviews
  • Actionable choices such as tag in del.icio.us,
    check Rutgers Univ., check Google, etc.
  • Less dead ends better ways to move in and
    out of the catalog in meaningful ways

55
How?
56
How to put this into the catalog?
  • Customize or hack the catalog within the
    parameters of the product
  • Export all the data outside the catalog and run
    it within another product (e.g., NCSUs Endeca
    and Casey Bissons WPopac)
  • Proxy server between the user and the catalog
  • Client-side customizations (e.g., Javascript or
    GreaseMonkey in Firefox)

57
How?
58
How to fetch remote info?
Screen scraping (primitive and brittle, but
sometimes the only way) Z39.50 (outdated and
limited to specific record types) SRW/SRU
(ZING) XML gateways (vendor-enabled, powerful,
lightweight) Web services model
(vendor-enabled, powerful, lightweight) APIs
(vendor-enabled, powerful, lightweight) SQL to
database (must recreate the business logic of
the vendor, may require advanced programming, may
be proprietary, tightly-coupled) OpenURL
resolver (mixed support, evolving, built on a
article model, rather than for other
information) DOI resolver (limited to known
items)
59
Amazon Hacks andAmazon web services (AWS)
60
What are web services?
The friction-free linking enabled by web
services. Loosely coupled services, even if they
use incompatible system technologies, can be
joined together on demand to create composite
services, or disassembled just as easily into
their functional components. Participants must
establish a shared semantic framework to ensure
messages retain a consistent meaning across
participating services. www.looselycoupled.com

61
What are web services?
  • PC to PC communication
  • Pass structured query in URL via HTTP protocol
  • Your server asks the Amazon server give me 3
    book reviews
  • http//amazon.com/aws/isbn87423211xtaskreviews
    limit3
  • Information is given back in XML format
  • ltreviewsgt
  • ltreview_1gt
  • ltreview_textgtThis book is great highly
    recommendedlt/review_textgt
  • ltstar_ratinggt5lt/star_ratinggt
  • lt/review_1gt
  • ltreview_2gt
  • ltreview_textgtThis book is a waste of my
    timelt/review_textgt
  • ltstar_ratinggt1lt/star_ratinggt
  • lt/review_2gt
  • Your server can now parse and process and euse
    that info locally and present it to your users
    and use it in your interfaces and applications
    and websites

62
xISBN
www.oclc.org/research/projects/xisbn OCLCs
xISBN project
http//labs.oclc.org/xisbn/0441172717 lt?xml
version"1.0" encoding"UTF-8" ?gt - ltidlistgt
ltisbngt0441172717lt/isbngt ltisbngt0801950775lt/isbngt
ltisbngt0399128964lt/isbngt
63
del.icio.us APIs
Our concept is to harvest NJIT tags back out of
del.icio.us at future point to display back into
the catalog
We dont want to reinvent the wheel. Enter
locally store remotely retrieve for local use.
64
NJITs service architecture
Client-side solution
65
NJITs service architecture
HTML035altscript language"JavaScript"
src"http//www.library.njit.edu/apps/voyagerdb/re
solver-return-javascript.cfm?oclca"
type"text/javascript"gtlt/scriptgt HTML035altscr
ipt language"JavaScript" src"http//www.library.
njit.edu/apps/voyagerdb/voyager-amazon-reviews-ret
urn-javascript.cfm?oclca" type"text/javascript
"gtlt/scriptgt
66
How can we link our catalog with other bits of
information?
  • Hard links
  • Link records 123ABC --gt XYZ789
  • Hard coded
  • Controlled or authored purposefully (typically by
    hand, but could be batched or automated)
  • Brittle (link can break if destination link moves
    or is deleted)
  • MARC cataloging
  • This is the old model
  • Difficult to maintain and scale

67
How can we link our catalog with other bits of
information?
  • Semi-soft links
  • If ISSN variable available, then, link to
    XYZ/?issnISSN
  • If keywords from the entered search available,
    then link to EFG/?kwkeywordkeyword
  • Loosely coupled
  • Semi-controlled (strategy or rule is set up
    invoked when conditions are met)
  • Yes/No or If/Then logic
  • Destination may not resolve in all cases
  • Requires or assumes some kind of resolver
    application with error catching
  • Easier to maintain and scale
  • Results less predictable some control given up

68
How can we link our catalog with other bits of
information?
  • Soft links
  • For the information at hand, what else can be
    inferred?
  • Quality filtering / collaborative filtering
  • Personalization
  • Fuzzy logic
  • Bayesian network
  • Book at hand recent search history of user
    borrowing history of user borrowing history of
    other users of the book at hand profile of user
    LCSH analysis unique recommendation for
    further books

69
Web services modelfor catalog enhancement
  • We have enough in the MARC records. Dont add
    more to MARC. Just use what we have smarter.
  • What can we use?
  • ISBN
  • ISSN
  • Author name
  • Title
  • LCSH
  • OCLC number
  • Call number
  • Etc.

70
Web services modelfor catalog enhancement
  • What can we get with this info?
  • Book reviews
  • Discussion of books in blogs and virtual book
    clubs
  • Usage information
  • Author biographies
  • Author blogs
  • Map standard classification terms to new
    folksonomies / tagging
  • Graphical representations of information
  • Etc.

71
Future for NJITs catalog
  • Tag NJIT-provided reviews going into Amazon so
    we can harvest them later for our users
  • Harvest NJIT-provided del.icio.us tags in the
    same way (PennTags)
  • Tease and lead approach
  • Not check Rutgers library, but this book is
    available in the Rutgers library or this book
    is due on Apr 21 in the Rutgers library
    Then, Click for more info
  • This book last read on Feb 21, 2006 click for
    more info.
  • This book borrowed 15 time last year 88th most
    popular book in the library see more detail.
  • Recommendation engine (people who borrowed this
    also borrowed these others )
  • Full text for this book exists at (Google
    Print) or (Open Content Alliance)
  • OpenWorldCat ?

72
My 2005 vision closerto reality in 2006
73
A manifesto
  • If it can be hacked, it will be hacked. Even
    if is users have to go to extraordinary lengths.
    The desire to create an improvement or fix a
    problem is a strong one. ILS vendors should not
    resist it. They should embrace it plan for it
    facilitate it.
  • A passionate, diverse, creative, and oft-times
    desperate user base above a critical mass will
    always out-power and out-pace vendor-based
    developers in pushing innovations. Recall
    Margaret Mead Never doubt that a small group
    of thoughtful people can change the world.
    Indeed, it's the only thing that ever has.)
    Thus, vendors should use the leverage of this
    user base. Exploit the gift economy.
  • Innovation thrives in open environments.
    Indeed, innovation may even require openness.
    Vendors need to open up. And they will reap the
    rewards many times back.
  • Vendors should not focus on creating new
    features. Instead, focus on creating new
    capabilities, possibilities, and potentials for
    their users to exploit.
  • Vendors should not create solutions to
    problems. They should create solution systems.
  • Bigger, more detailed, do-everything systems
    take longer to develop. Vendors should focus on
    small pieces, loosely joined that allow
    librarians to link them in infinite, unforeseen
    ways. (Metcalfs Law says the value of the whole
    network will logarithmically increase with each
    node added to the network).
  • Vendors need to create an ILS platform that
    allows for unintended consequences. We want
    unintended consequences. They want unintended
    consequences. Unintended consequences are good.
  • Vendors need to create an ILS platform that
    expects disruptive technologies. We want them,
    too.

74
Lipstick on a pig
75
Before
76
Goals Amazon lessons
  • Expose and foster community and activity
    (digital footprints)
  • Added valued information in the book record
  • Increase user options and actionable choices
  • Decrease catalogs are dead ends outbound
  • Increase access points inbound

77
After
78
Side-by-side
79
Emulating Amazon what lessons can we learn
what features can we implement EndUser 2006
  • Jim Robertson
  • Assistant University Librarian
  • NJIT (New Jersey Institute of Technology)
  • www.library.njit.edu/staff-folders/robertson/pr
    esentations/
Write a Comment
User Comments (0)
About PowerShow.com