Supply and Demand Analysis in NDLTD
Based on Patron Specialty and Contents
Statistics
The 9th International Symposium on
Electronic Theses and Dissertations
Quebec City, Quebec, Canada
June 7-10, 2005
Seonho Kim, Seungwon Yang, Edward A. Fox
Digital Library Research Laboratory, Virginia Tech
Blacksburg, VA 26061 USA
Overview
•
•
•
•
•
Purpose of Study
Data Set (ETDs, patrons, queries)
Our Approach
Data Analysis
Conclusions and Future Work
ETD 2006, Quebec, Canada
2
Purpose of Study
•
•
•
•
•
Analysis of ETD Subjects (supply)
Analysis of Users, Queries (demand)
Comparison of the Above Two
Users’ Years of Experience in Their Field
Distribution of Date Stamp of ETDs
ETD 2006, Quebec, Canada
3
Overview
•
•
•
•
•
Purpose of Study
Data Set (ETDs, patrons, queries)
Our Approach
Data Analysis
Conclusions and Future Work
ETD 2006, Quebec, Canada
4
Data Set - ETDs
• Up-to-date Union archive harvested from
Online Computer Library Center (OCLC)
• Using OAI/ODL Harvester [2] by Hussein Suleman
http://oai.dlib.vt.edu/odl/software/harvest/
• Total of 242,688 records
ETD 2006, Quebec, Canada
5
Example – ETD Metadata
•
<dc oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/"
dc="http://purl.org/dc/elements/1.1/"
xsi="http://www.w3.org/2001/XMLSchema-instance"
schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/
http://www.openarchives.org/OAI/2.0/oai_dc.xsd"><title>Composer-Centered
Computer-Aided Soundtrack Composition</title><creator>Vane, Roland
Edwin</creator><subject>Computer Science</subject><subject>human
computer interaction</subject><subject>music
composition</subject><subject>soundtracks</subject><subject>creativity</
subject><description>For as long as computers have been around, people
have looked for ways to involve them in music….
</description><publisher>University of
Waterloo</publisher><date>2006</date><type>Electronic Thesis or
Dissertation</type><format>application/pdf</format><identifier>http://etd.u
waterloo.ca/etd/revane2006.pdf</identifier><language>en</language><rights
>Copyright: 2006
ETD 2006, Quebec, Canada
6
Patrons, Queries
• User Profile Data (Oct. 2005 – May 2006)
• Online User Survey [3] as part of User Modeling study
• Total 1100 User Data that include
– Majors, specialties, years of experience, and demographic
information.
– Queries and detailed research interests
ETD 2006, Quebec, Canada
7
User Profile Form
ETD 2006, Quebec, Canada
8
Example – User Profile Data
• <user> <userID>shk</userID> <email>[email protected]</email>
<name><first>Sh</first> <last>King</last>
</name><major>CS</major><broadresearch>Digital Library
<specific>User interface</specific>
<experience>8,2</experience></broadresearch><group />
<query><item freq="79">digital library</item> <item
freq="33">computer science</item> <item freq="25">virginia
tech</item> <item freq="9">artificial intelligence</item> <item
freq="5">digital library.</item> </query><selected><item
freq="15">Digital Library</item> <item freq="6">Electronic Theses
and Dissertations</item> </selected><proposed><item
freq="80">Digital Library</item> <item freq="65">Data</item>
</proposed></user>
ETD 2006, Quebec, Canada
9
Overview
•
•
•
•
•
Purpose of Study
Data Set (ETDs, patrons, queries)
Our Approach
Data Analysis
Conclusions and Future Work
ETD 2006, Quebec, Canada
10
Categorization of Academic Subjects
• Built Our Own Classification Categories
• Based on Colleges / Faculties in
- Virginia Tech, University of Virginia, George Mason University,
VCU and Virginia State University
• Identified
- 7 categories and 77 subcategories
- Word patterns for each subcategories
ETD 2006, Quebec, Canada
11
Categorization of Academic Subjects
• 7 categories and selected 77 subcategories
7 Categories
Selected 77 Sub-categories
1
Architecture and Design
ArchitectureConstruction,
LandscapeArchitecture,…
2
Law
Law
3
Medicine, Nursing and
Veterinary Medicine
Dentistry, Medicine, Pharmacy, Nursing,…
4
Arts and Science
Agriculture, AnimalPoultry,Biology,...
5
Engineering and Applied
Science
ComputerScience, Material, Electronics,…
6
Business and Commerce Buisiness, Economics, Management,…
7
Education
8
Others (unclassifiable)
Education
ETD 2006, Quebec, Canada
12
Categorization of Academic Subjects
• Each subcategory has a set of word patterns
- Matching table developed
• Process of word matching table development
1. Run our subject-matching classifier program
2. Count each unclassified subject & sort them.
3. If num. > 10, add the unclassified subject to
matching table
4. Repeat 1 – 3 until num. < 10 for all unclassified subjects
ETD 2006, Quebec, Canada
13
Categorization of Academic Subjects
• Matching Table
77 categories
Word Patterns
Education
/bildung/, /pedagog/, /fakul/, /educa/,
/teaching/,…
Geology
/geolog/, /geoscience/,…
LibraryScience
/librari/, /library/, /informatik/,…
…
…
ETD 2006, Quebec, Canada
14
Categorization of Academic Subjects
Unclassified ETD Subjects:
• Approx. 85 %  unique
• Approx. 10 %  only two occurrences
ETD 2006, Quebec, Canada
15
Measuring Supply – Demand
• ETD Supply (Num. of Resources)
- 242,688 ETDs classified into 7 categories and
counted
• Patron’s Demand (Num. of Queries)
- 4519 queries (in 1100 user data) classified into 7
categories
- “Sum of all queries” in each category calculated as
Demand of a Category 
 number
of queries
user  category
ETD 2006, Quebec, Canada
16
ETD Classification
• Based on the “first” subject field
•
<dc oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/"
dc="http://purl.org/dc/elements/1.1/"
xsi="http://www.w3.org/2001/XMLSchema-instance"
schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/
http://www.openarchives.org/OAI/2.0/oai_dc.xsd"><title>Composer-Centered
Computer-Aided Soundtrack Composition</title><creator>Vane, Roland
Edwin</creator><subject>Computer Science</subject><subject>human
computer interaction</subject><subject>music
composition</subject><subject>soundtracks</subject><subject>creativity</
subject><description>For as long as computers have been around, people
have looked for ways to involve them in music….
</description><publisher>University of
Waterloo</publisher><date>2006</date><type>Electronic Thesis or
Dissertation</type><format>application/pdf</format><identifier>http://etd.u
waterloo.ca/etd/revane2006.pdf</identifier><language>en</language><rights
>Copyright: 2006
ETD 2006, Quebec, Canada
17
User Classification
• Based on the “major”, ”broadresearch”, and “specific”
fields in each user profile
• <user> <userID>shk</userID> <email>[email protected]</email>
<name><first>Sh</first> <last>King</last>
</name><major>CS</major><broadresearch>Digital Library
<specific>User interface</specific>
<experience>8,2</experience></broadresearch><group />
<query><item freq="79">digital library</item> <item
freq="33">computer science</item> <item freq="25">virginia
tech</item> <item freq="9">artificial intelligence</item> <item
freq="5">digital library.</item> </query><selected><item
freq="15">Digital Library</item> <item freq="6">Electronic Theses
and Dissertations</item> </selected><proposed><item
freq="80">Digital Library</item> <item freq="65">Data</item>
</proposed></user>
ETD 2006, Quebec, Canada
18
Challenges
• Varieties in describing research subjects
 Solution: we built a subject matching table
77 categories
Decision patterns
Education
/bildung/, /pedagog/, /fakul/,
/educa/, /teaching/,…
Geology
/geolog/, /geoscience/,…
LibraryScience
/librari/, /library/, /informatik/,…
Arts
/music/, …
ETD 2006, Quebec, Canada
19
Challenges
• Interdisciplinary ETDs
– e.g., “Music Education”
– Solution: adjust matching order
• Unclassifiable ETDs
–
–
–
–
–
Null Entry (No subject field data)
Erroneous entries (e.g., “Ph.D”, “Georgia”,[email protected])
Typo (e.g. “edcuation”, “poluition”)
Too much detail (e.g., “pulsars”, “muon”, “cytochrome”)
Abbreviations (e.g., “MOCVD”, “OFDM”)
ETD 2006, Quebec, Canada
20
Overview
•
•
•
•
•
Purpose of Study
Data Set (ETDs, patrons, queries)
Our Approach
Data Analysis
Conclusions and Future Work
ETD 2006, Quebec, Canada
21
Resource Distribution
Resource Distribution in NDLTD
2
1
3
4
8
1
2
3
4
5
6
7
1
Architecture and
Design
2
Law
3
Medicine, Nursing
and Veterinary
Medicine
4
Arts and Science
5
Engineering and
Applied Science
6
Business and
Commerce
7
Education
8
Others.
(unclassifiable)
8
5
7
6
ETD 2006, Quebec, Canada
22
User Distribution
User Distribution in NDLTD
1
1
Architecture and
Design
2
Law
3
Medicine, Nursing
and Veterinary
Medicine
4
Arts and Science
5
Engineering and
Applied Science
6
Business and
Commerce
7
Education
8
Others.
(unclassifiable)
2
3
8
4
7
1
2
3
4
5
6
7
8
5
6
ETD 2006, Quebec, Canada
23
Query Distribution
1 Architecture
and Design
Query Distribution in NDLTD
1
2 Law
2
3
4
8
5
1
2
3
4
5
6
7
8
3 Medicine,
Nursing and
Veterinary
Medicine
4 Arts and
Science
5 Engineering
and Applied
Science
6 Business and
Commerce
7 Education
7
6
ETD 2006, Quebec, Canada
8 Others.
(unclassifiable)
24
Supply-Demand Comparison
1 Architecture
and Design
ETD Resources and User Demands (Number of Queries) in NDLTD
50%
ETDs
2 Law
Demands
3 Medicine,
Nursing and
Veterinary
Medicine
45%
40%
35%
30%
4 Arts and
Science
25%
5 Engineering
and Applied
Science
20%
15%
10%
5%
6 Business and
Commerce
0%
7 Education
1
2
3
4
5
Academic Categories
6
ETD 2006, Quebec, Canada
7
8
8 Others.
(unclassifiable)
25
ti ngF
ETD supply
Ar t
Astr o
nom
y
B io c
B io lo
hem
i st ry
gi cal
Eng in
eeri n
g
B io lo
gy
B ot a
ny
B usi
ness
Chem
i cal
Chem
i stry
Co m
mun
i cat io
Co m
n
put e
Cro p
rSc ie
Soi lE
nc e
n vSc
ie nce
s
Dai ry
Scien
ce
Dent
istr y
Eco lo
gy
Eco n
o mic
s
Educ
at io n
El ec
Eng in
tr on ic
eeri n
s
g Scie
n ce
Eng li
sh
Ent o
mo l o
gy
Envi r
o nm
e nt
F am
F or e
i ly
ig nLa
ngu a
F oo d
g esL
i ter a
t ures
F or e
str y
Geo g
ra ph
Go ve
y
r nme
Geo l
nt Int
e rna
og y
ti on a
l Affa
ir
in anc
e
Aer o
spac
e
Ag ri c
ul tur
e
An im
a lPo
ul try
An th
r opo
lo gy
Appa
r elHo
using
Ar ch
Ar ch
i tect
a eol
o gy
ure C
o nst
r uct io
n
Acc o
un
Supply-Demand of 77 Subcategories
(1/2)
Supply/Demand 77 Subcategories (1/2)
12%
10%
User Demand
8%
6%
4%
2%
0%
ETD 2006, Quebec, Canada
26
ETD Supply
Law
Lib r ar
yScie
nce
Lin gu
i sti cs
Lit er a
tu re
Man a
g eme
nt
Mat er
i al s
Mech
a ni cs
Medic
i ne
Mete o
r ol o gy
Mat he
ma ti c
s
Min ing
Min er
al
Music
Nava l
Nucl e
ar
N
u
r
si ng
Oce a
nEng in
eeri ng
Pha rm
a cy
Phi lo s
op hy
Phy si
cs
Pl ant
Po li ti c
s
Pub li c
Psych
o l og y
Admi
nistr a
t io nPo
li cy
Pub li c
Affai r
So cio
lo g y
Sta ti s
tic s
Ur ba n
Pla nn
i ng
V eter in
ar y
Wil dli
fe
Woo d
Zoo lo
gy
Hist or
y
Ho rt ic
ul tur e
Ho sp
it al it yT
our ism
Huma
nDeve
Huma
l o pme
nNutr
nt
it io nE
xer cis
e
Indus
tri al
Infor m
at ic s
Int erd
isci pl in
Lan ds
ary
ca peA
rchi te
c tur e
Supply-Demand of 77 Subcategories
(2/2)
Supply/Demand 77 Subcategories (2/2)
12%
User Demand
10%
8%
6%
4%
2%
0%
ETD 2006, Quebec, Canada
27
User Expertise Years
Users' Expertise in Years
200
180
160
120
100
80
60
40
20
50
35
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
0
Users
140
Years
ETD 2006, Quebec, Canada
28
Expertise Years and Demand
Expertise Years and Demand
25%
Users
Demand
20%
15%
10%
5%
50
er
ro
r
39
40
30
35
27
28
25
26
23
24
21
22
19
20
17
18
16
14
15
12
13
10
11
9
8
7
6
5
4
3
2
1
0
0%
Years
ETD 2006, Quebec, Canada
29
17 xx
18 xx
19 0x
19 1x
19 2x
19 3x
19 4x
19 5x
19 6x
19 7x
19 8x
19 90
19 91
19 92
19 93
19 94
19 95
19 96
19 97
19 98
19 99
20 00
20 01
20 02
20 03
20 04
20 05
20 06
error
Date Stamp of ETD
60,000
50,000
40,000
30,000
20,000
10,000
0
Year
ETD 2006, Quebec, Canada
30
Date Stamp of ETD
• ETDs from seventeen hundreds ?
- Some of scanned copies from European
universities
- Oldest ETDs are from British universities
- Some of the older dates are typos - you'd have
to check each one to know for sure
ETD 2006, Quebec, Canada
31
Overview
•
•
•
•
•
Purpose of Study
Data Set (ETDs, patrons, queries)
Our Approach
Data Analysis
Conclusions and Future Work
ETD 2006, Quebec, Canada
32
Conclusions
•
•
•
•
•
Analysis of ETD Subjects (supply)
Analysis of User Queries (demand)
Comparison of the Above Two
Users’ years of experience in their field
Date Stamp of ETDs
Learned the future directions…
ETD 2006, Quebec, Canada
33
Future Work
• Use of widely-used classification system
- e.g., Dewey Decimal Classification 22
• More detailed classification of ETDs
- Include title, abstract and other subject field data
- Utilize “discipline” in ETD_MS format records
 But only approx. 7000 records
• Use of user behavior data
- e.g., Clicking of query results in NDLTD
ETD 2006, Quebec, Canada
34
References
[1] NDLTD, Networked Digital Library of Theses and Dissertations,
available at http://www.ndltd.org, 2006
[2] Hussein Suleman, “OAI/ODL Harvester”, available at
http://oai.dlib.vt.edu/odl/software/harvest/
[3] Seonho Kim, Uma Murthy, Kapil Ahuja, Sandi Vasile, Edward A.
Fox, “Effectiveness of Implicit Rating Data on Characterizing
Users in Complex Information Systems”, Springer-Verlag
LNCS3652, 9th European Conference on Research and
Advanced Technology for Digital Libraries (ECDL 2005), 2005,
186-194
[4] Search Interface Embedded User Tracking System, available at
http://boris.dlib.vt.edu:8080/controller/index.jsp, 2006
ETD 2006, Quebec, Canada
35
Appendix
• 7 Categories and 77 Subcategories
7 categories
77 subcategories
1
Architecture and Design
ArchitectureConstruction, LandscapeArchitecture
2
Law
Law
3
Medicine, Nursing and
Veterinary Medicine
Dentistry, Medicine, Nursing, Pharmacy, Veterinary
4
Arts and Science
Agriculture, AnimalPoultry, Anthropology, ApparelHousing, Archaeology, Art,
Astronomy, Biochemistry, Biology, Botany, Chemistry, Communication,
CropSoilEnvSciences, DairyScience, Ecology, EngineeringScience, English,
Entomology, Family, Food, ForeignLanguageLiterature, Forestry, Geography,
Geology, GovernmentInternationalAffair, History, Horticulture,
HospitalityTourism, HumanDevelopment, HumanNutritionExercise, Informatics,
Interdisciplinary, LibraryScience, Linguistics, Literature, Meteorology,
Mathematics, Music Naval, Philosophy, Physics, Plant, Politics, Psychology,
PublicAdministrationPolicy, PublicAffair, Sociology, Statistics, UrbanPlanning,
Wildlife, Wood, Zoology
5
Engineering and Applied
Science
Aerospace, BiologicalEnginerring, Chemical, ComputerScience, Electronics,
Environment, Industrial, Materials, Mechanics, MiningMineral, Nuclear,
OceanEngineering
6
Business and Commerce
AccountingFinance, Business, Economics, Management
7
Education
Education
8
Others. (unclassifiable)
(Unclassifiable)
ETD 2006, Quebec, Canada
36
Thank You
ETD 2006, Quebec, Canada
37
Questions or Comments?
ETD 2006, Quebec, Canada
38
User Data - Fields
• <query> : entered by the user
• <proposed> : ETD results clustered and displayed
• <selected> : cluster labels clicked by the user
ETD 2006, Quebec, Canada
39
Descargar

EPortolio Presentation