Citation Analysis for the Free, Online Literature

1
Citation Analysis for the Free, Online Literature

• Tim Brody
• Intelligence, Agents, Multimedia Group
• University of Southampton

2
Content

• Current services for Open Access Literature
• Institutional Archives Registry
• Metadata Harvesting through Celestial
• Citebase Search
• Citation Linking
• Search and Navigation Service
• Web Impact as a predictor of Citation Impact

3
Institutional Archives Registry
4
(No Transcript)
5
Sites in the IAR

• Things we want to know
• GNU EPrints sites
• Other research collections (Other Archives, Open
  Journals)
• BOAI 1. vs BOAI 2.
• A submission form consisting of
• URL, Name, OAI URL, Country, type, full-text,
  software
• Cant (yet) track full-texts
• (Create a master-list so archives only
  register-once?)

6
Celestial

• Designed to
• Be an abstraction over OAI-PMH versions
• Caching OAI metadata records
• Technological questions
• How big can the OAI-PMH go (ok for 5 million
  records so far)
• How reliable are OAI-PMH implementations
• Feeds Citebase, IAR, some external users

7
(No Transcript)
8
(No Transcript)
9
Services for Open Access Literature
OAIster
Scirus
Google
Search Engines Navigation Tools Analysis
Assessment
BMC
Citeseer
Citebase
Citation Analysis/Linking Services (Citebase /
Citeseer / OpenURL / DOI) Version Linking Services
OAI-PMH Transport
arXiv.org
Self-Archived Full-texts (Pre/Post-prints) Open
Access Publishing
n.b. Scirus/OAIster arent citation-analysis
aware yet, Google indexes Citeseer. Not an
exhaustive list
10
Citation Analysis Linking

• A citation is a reference from one work to
  another as a hyperlink a citation link
• Citation analysis uses citation relationships to
  analyse patterns in research
• As a graph a work (paper, book etc.) is a vertex
  and a citation an edge
• Bibliometrics
• (study of patterns in literature)

11
Digitometric/Infometric Analysis

• Bibliometrics for the online age
• Couple citation analysis with Web analysis
• (how many times has x been accessed?)
• Similar to readership studies, but easier to
  survey and more comprehensive
• (though subject to the same problems of copies
  being re-distributed, multiple accesses etc.)

12
Citebase Search
13
Citation Linking

• Retrieve and cache full-texts
• LaTeX, PDF, XML
• Extract reference list
• Extract individual references
• Parse references into components
• Author, year, title, journal, volume, pagination
• Store in structured database

14
Citebase Search
15
(No Transcript)
16
Citebase SearchNavigation by Citation Links
Article withreference list
Future
Referencelink
Related
Current Article
Co-cited
Past
17
(No Transcript)
18
Predicting Citation Impact

• The Web gives us access to new metrics
• Download/access frequency
• Can early-day download frequency give an
  indication of longer-term citation frequency?
• (Web logs from the UK arXiv.org mirror, Citation
  data from Citebase Search)
• Pearson correlation after 6 months of web logs
  0.42 for the High Energy Physics sub-arXiv

19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Assessing Research(ers)

• Citation Impact
• By-Paper, Author, Journal, Institution
• Web Impact
• Predictor of citation-impact, combine with
  citation-impact
• Search Engines
• More detailed research assessment

25
Comparing Online/Offline Impact

• Using ISI CD-ROM data
• Use Web crawlers to find online articles
• Compare citation impact of online and offline
  articles
• By discipline, by journal, by author?
• Initial results for Physics show 2-3x increase
• arXiv.org
• Southampton, U. Quebec, Oldenburg (de)

26
Relevant Web Pages

• EPrints http//www.eprints.org/
• IAR http//archives.eprints.org/
• Citebase Search
• http//citebase.eprints.org/
• Celestial
• http//celestial.eprints.org/
• Correlation Generator
• http//citebase.eprints.org/analysis/correlation.p
  hp
• Tim Brody lttdb01r_at_ecs.soton.ac.ukgt