Bibliometric analysis: Difference between revisions
Jump to navigation
Jump to search
Created page with "== Bibliometric analysis == Standard way of doing it: * Download database from WebofScience or Scopos * visualize with network viz tool 'Gephi' === Important tasks and chall..." |
|||
| Line 20: | Line 20: | ||
=== What to evaluate === | === What to evaluate === | ||
* Impact analysis | * Impact analysis | ||
* altmetrics - broad network impact | * altmetrics - broad network impact outside publications (news, social media ...) | ||
* Relevance over time, can it be normalized to some degree? Relation to GDP, number of employees per institution? | * Relevance over time, can it be normalized to some degree? Relation to GDP, number of employees per institution? | ||
* co-authorships | * co-authorships | ||
* citation networks (x-order) | * citation networks (x-order) | ||
* filter by internal citation in a community, not single high citations | * filter by internal citation in a community, not single high citations | ||
* Compare number of collective group citations vs. without self-citations | * Compare number of collective group citations vs. without self-citations | ||
=== Example tools === | === Example tools === | ||
Latest revision as of 13:43, 10 December 2018
Bibliometric analysis
Standard way of doing it:
- Download database from WebofScience or Scopos
- visualize with network viz tool 'Gephi'
Important tasks and challenges
- Cleaning up data (duplicate authors/publications) - heuristics to identify different/same persons built in? (80/90% of the issue)
- Multiple fields to analyze simultaneously
- Merging WebofScience and Scopos (does not automatically add article references)
- Orcid not yet default and standard (works for author to publications link but not the other way around). Is there a tool automatically checking this?
- Merging institutions
- Other networks where one can harvest bibliometric data? e.g. ResearchGate, Academia, Google scholar?
- Limit with download reference entries (WebofScience - 500 per go, Scopus easier)
- Clustering tasks - what are the different strategies and routines? (orthogonal, ...)
- Classification of journals into themes (can be changing} - how to find links/important authors/institutes beyond co-authorship networks?
- Topics classification by WebofScience - sometimes problematic, different from key word search
- Account for time factor
What to evaluate
- Impact analysis
- altmetrics - broad network impact outside publications (news, social media ...)
- Relevance over time, can it be normalized to some degree? Relation to GDP, number of employees per institution?
- co-authorships
- citation networks (x-order)
- filter by internal citation in a community, not single high citations
- Compare number of collective group citations vs. without self-citations
Example tools
- VOSviewer to construct and visualize bibliometric networks
- CitNetExplorer to visualize citation networks
- Gephi - visualize networks, even over time
- yEd - for layouting graphs