1. Author Name Disambiguation in Medline
Vetle I. Torvik and
Neil R. Smalheiser
August 31, 2006

4. A Statistically Based Model:
Hypothesis: an individual tends to publish papers with similar attributes, sufficiently so that these attributes suffice to disambiguate the authors.
The case of Dr. Tom Jobe?
Large, automatically generated training sets of pairs of articles matching (Last Name, First Initial) written by the same person vs. by different individuals.

5. Name Attributes Suffix if present (III, Jr.) Middle initial match
Original model got very good performance without using first names at all!
First name (if available in Medline, or if it can be scraped from online papers)
Name spelling variants
Name frequency

6. Article Attributes Journal name Number of co-author names in common
Affiliation words in common
may not be given for all co-authors
Name correlations with affiliations (e.g. Ito and Japan are correlated)
Language of the article
Title words in common
addresses, if given
assign to right author
MeSH headings in common

7. A Monotone Model
Each pair of papers creates a vector of 10 dimensions, each of which has a matching score.
Assume monotonicity [more attributes in common, more likely written by the same person]
Allows for nonlinear and interactive effects across dimensions

8. Estimate Pairwise Frequencies
For a given pair of articles, compute the match vector, then look up its frequencies in pos vs. neg training sets: ratio = R value
For a given name, estimate the a priori probability P that any two papers will be written by the same person
This is a whole story in itself….
1/[1 + (1-P)/PR] = probability of a match
The Author-ity Site at (

9. Beyond Pairwise Comparisons
A and B share titles, journals
B and C share co-authors, affiliations
But A and C share nothing!
Yet p(AC) must be > (p(AB) + p(BC) -1)
Triangle inequality using probabilities, detect and correct anomalies
due to missing data or higher order correlations
Catch un-characteristic papers by an author
Another long story to optimize the methods!

10. Clustering “all” papers in Medline according to author-individuals
First we compute all pairwise probabilities for each (last name, first initial) modified with triplet correction
Then we form clusters
at p = 0.95 (high precision)
and at p = 0.5 (high recall)
i.e. the chance is greater than 0.5 that it belongs to some cluster, or it stays as a singleton

11. First-Pass Disambiguation is Complete!
Except for several hundred names having more than ~3000 papers each,
reach memory limit,
will assess if the model is reliable for the biggest names
For now, proceed for papers giving first names.
Monitoring for over-clustering and under-clustering
Summarizing global statistics

12. Immediate Next Steps Evaluate the clustering performance
Old vs. new papers
Importance of missing data
Very frequent names
Singletons, least confident assignments
Update the web interface

16. Upcoming Grant Renewal Aim 1: Special Cases
name reversal, hyphenated names, spelling errors, Gerald vs. Jerry, Rick vs. A. Rick
Use co-author assignment to help disambiguate another co-author
Compute confidence level of assignment for each paper, identify least confident assignments

17. Upcoming Grant Renewal Aim 2: Update the Model
Original model covers 1966-present, but new papers have different information, MeSH, s, online information
Modify training sets with recent papers.
Journal name partial match
Abstract words match?
Affiliations matched to each authors in PMC, online papers
References Cited information taken from PMC

18. Upcoming Grant Renewal Aim 3: Web Interface
Update the pairwise interface (given name, a particular paper, list all others in order of match probability)
Show clusters – given a name, show all clusters of author-individuals, link to Community of Science, searchable by attributes, can summarize and explore further (Anne O’Tate tool)
Author profile/collaboration finder tools
Data made available for bibliometrics and collaboration network research

19. Upcoming Grant Renewal Aim 4: Curation
Curator to identify errors and least-confident assignments
manually
machine methods (e.g. wobble in clustering)
change the database and alter the model as needed
Wiki Authors – will monitor postings to Wiki and change the database as verified and warranted (e.g. maiden name to married name)