Distributed, RealTime Computation of Community Preferences

1
Distributed, Real-Time Computation of Community
Preferences

• Thomas Lutkenhouse, Michael L. Nelson, Johan
  Bollen
  
• Old Dominion UniversityComputer Science
  DepartmentNorfolk, VA 23529 USA
  
• lutken,mln,jbollen_at_cs.odu.edu
• HT 2005 - Sixteenth ACM Conference on Hypertext
  and Hypermedia
  
• 6.-9.Sept. 2005, Salzburg Austria

2
Distributed, Real-Time Computation of Community
Preferences
3
Outline

• Review of technologies
• buckets
• Hebbian learning
• previous results
• Experiment design
• Results
• Lessons learned
• Conclusions

4
Non-evolution of DL Objects
. . .
5
Buckets

• Premise repositories come and go, but the
  objects should endure
  
• Began as part of NASA DL research
• focus on digital preservation
• implementation of the Smart Objects, Dumb
  Archives (SODA) model for digital libraries
  
• CACM 2001, doi.acm.org/10.1145/374308.374342
• D-Lib, dx.doi.org/10.1045/february2001-nelson

6
Smart Objects

• Responsibilities generally associated with the
  repository are pushed down into the stored
  object
  
• TC, maintenance, logging, pagination display,
  etc
  
• Aggregate
• metadata
• data
• methods to operate on the metadata/data
• API examples
• http//www.cs.odu.edu/mln/teaching/cs595-f03/?met
  hodgetMetadatatypeall
  
• http//www.cs.odu.edu/mln/teaching/cs595-f03/?met
  hodlistMethods
  
• http//www.cs.odu.edu/mln/teaching/cs595-f03/?met
  hodlistPreference
  
• (cheat) http//www.cs.odu.edu/mln/teaching/cs595-
  f03/bucket/bucket.xml
  

7
Internal Structure
jaga.cs.odu.edu/home/mln/public_html/teaching/cs6
95-f03 ls bucket/ CVS/ index.cgi jaga.cs.od
u.edu/home/mln/public_html/teaching/cs695-f03
ls bucket/ bucket.xml content/ CVS/ lib/ log
s/ methods/ jaga.cs.odu.edu/home/mln/public_htm
l/teaching/cs695-f03 ls bucket/content/
syllabus.txt week1readings.html
week5readings.html week10readings.html week
1week-01.ppt week6readings.html
week11readings.html week2readings.html
week7readings.html week12readings.html week
2week-02.ppt week8readings.html
week13readings.html week3assignment1.ppt
week9readings.html week14readings.html week
3readings.html week15readings.html week3wee
k-03.ppt jaga.cs.odu.edu/home/mln/public_html/te
aching/cs695-f03 ls bucket/lib
CVS/ EZXML.pm mime.e style.css
jaga.cs.odu.edu/home/mln/public_html/teaching/cs6
95-f03 ls bucket/logs/ access.log CVS/ mylog.
log jaga.cs.odu.edu/home/mln/public_html/teachin
g/cs695-f03 ls bucket/methods/
addElement.pl getElement.pl
listMethods.pl setPreference.pl
CVS/ get_log.pl
listPreference.pl deleteElement.pl getlog.pl
log.pl display.pl getMetadata.pl
setMetadata.pl jaga.cs.odu.edu/home/mln/public
_html/teaching/cs695-f03
8
Examples

• 1.6.X bucket
• http//ntrs.nasa.gov/
• http//www.cs.odu.edu/mln/phd/
• 2.0 buckets
• http//www.cs.odu.edu/mln/teaching/cs595-f03/
• http//www.cs.odu.edu/lutken/bucket/
• 3.0 buckets (under development)
• http//beaufort.cs.odu.edu8080/
• uses MPEG-21 DIDLs
• cf. http//www.dlib.org/dlib/november03/bekaert/11
  bekaert.html
  

9
Hebbian Learning
Implementation issues - gather log files - p
roblematic when spread across servers/domains
- determine a ?T for session reconstruction
- typically 5 min - compute links weights
- update the network periodically
- typically monthly
10
Previous, Log-Based Recommendation Implementations

• LANL Journal Recommendations
• collection analysis based on journal readership
  patterns
  
• D-Lib Magazine, dx.doi.org/10.1045/june2002-bollen
  
  
• NASA Technical Report Server
• compared recommendations with those generated by
  VSM
  
• WIDM 2004, doi.org.acm/1031453.1031480
• Open Video Project
• generated recommendations for videos (little
  descriptive metadata)
  
• JCDL 2005, doi.acm.org/1065385.1065472

11
Hebbian Learning with Bucket Methods
http//b?methoddisplay refererhttp//b redir
ecthttp//a?methoddisplay 26redirecthttp//c?
methoddisplay
26refererhttp//b
http//a?methoddisplay refererhttp//a redir
ecthttp//b?methoddisplay
26refererhttp//a
12
Experiment

• Spin Magazines Top 50 Rock Bands of All Time
• something other than reports, journals, etc.
• harvest allmusic.com for metadata for all LPs by
  the 50 bands (total 800 LPs)
  
• Maintain hierarchical arrangement
• 1 artist ? N albums
• Initialize the network of 800 LPs with each LP
  randomly linked to 5 other LPs
  
• Send out email invitations to browse the network
• have them explore, and then examine the resulting
  network
  
• users not informed about the workings of the
  network

13
Display of LPs
14
Hierarchical, Weighted Links
- - w.cs.odu.edu/lutken/bucket/121/"
- -   Terrapin S
tation, Capital Centre, Landover, MD,
3/15/90     ative /     - ment wt"0.5" id"http//www.cs.odu.edu/lutken/b
ucket/11/" - -   leJealousy/Progress  
     
- .odu.edu/lutken/bucket/434/"
- -   Nevermind/title    
    - "0.5" id"http//www.cs.odu.edu/lutken/bucket/13
0/" - -   Tech
nical Ecstasy     inistrative /    
.
weights - initial 0.5 - frequency 1.0
- symmetry 0.5
- transitivity 0.3
15
Respondents

• August 2004 - October 2004
• 160 respondents
• self-identify at the beginning exit survey at
  the end
  
• 1200 bucket-to-bucket traversals (7.5 average
  traversals per session)

16
How to Evaluate the Resulting Network?

• Compute network analysis metrics
• PageRank
• Degree Centrality
• Weighted Degree Centrality
• Compare the results to
• Other expert lists (VH1, DigitalDreamDoor,
  original Spin Magazine list)
  
• Artist / LP best seller according to RIAA
• Artist / LP Amazon sales rank

17
Expert Rankings

• No correlation with
• VH1 artist list
• DigitalDreamDoor list
• original Spin Magazine list (!)
• (critics dont agree with each other, or the
  record buying public)

18
RIAA Results

• RIAA had only
• only 51/800 LPs
• only 14/50 artists
• (critics dont buy records!)

RIAA sales caveat
Figure 6. Probability of albums being
best-sellers.

Figure 7. Probability of artists being
best-sellers.
19
Amazon Sales Rank

• No correlation with individual LP sales rank
• but correlated with mean artist sales rank
• similar to RIAA data
• interpretation popular artists often have
  obscure LPs

20
Relatedness(?)
21
Relatedness(?)
22
Lessons Learned

• While the subject matter was interesting, it was
  oriented for music geeks
  
• i.e., no actual music was delivered to the users
  (intellectual property considerations)
  
• more traversals needed
• Random initial starting points were difficult to
  overcome
  
• cold start problem - pre-seed the links
  according to some criteria?
  
• weights did not decay over time/traversals
• Choosing only artists from Spin Magazine may have
  pre-filtered the response
  
• choose artists from Down Beat (Jazz), Vibe
  (Urban), Music City News (Country), etc.

23
Conclusions

• Can build a network of smart objects featuring
  adaptive, hierarchical links constructed in
  real-time without central state
  
• network is created without latency and with
  computations amortized over individual accesses
  
• Experimental testbed with popular music LP
  metadata shown to approach sales rank of artists,
  not LPs