About PubStat.org

About the Metrics

Display Options

Display Option Definition
Rank The rank of the author in a certain topic area order by the corresponding metric.
RankPer Rank percentage is dividing the rank value by the total number of authors in this topic area, and then *100%.
Percentile Inside a certain topic area, the percentile of an author is the cummulative value of all the authors who have higher ranks for the certain metric.
Letter (RankPer) By combining authors into 5 groups, we assign each group one of the letter "A", "B", "C", "D", "E" according to authors' rank percentile.
Letter (Contri) By combining authors into 5 groups, we assign each group one of the letter "A", "B", "C", "D", "E" according to authors' contribution percentile.
Author Types
  • (Include All) - No filtering applied during the calculation.
  • (Active YYYY - now) - Only include authors active YYYY-now.
Domain Types
  • 24 sub-domain - Following the categoraization provided by Microsoft academic research - Computer Science Top Domain
  • CS overall (Type A) - Consider all the papers (including those that do not belong to any of the 24 sub-domains) forming one single domain.
  • CS overall (Type B) - Merge the 24 sub-domains into one single domain, i.e., those papers belong to CS top domain, but not belong to any of the 24-subdomains are excluded.

Author Metrics

Metric Name Definition
Influence (Inf.) Influence is computed by distributing each author's influence across her publications, then onto those authors citing those publications. This computation extends the page-rank fixed point computation used in Google Search service. In this case, the solution gives each author an influence value that is in turn used to rank all the authors in a particular field.
Connections (Conn.) Connections is computed as the fixed point of the co-author matrix. An author ranked high by this metric co-authors with many other people, and/or people highly connected. Note, in this metric, no citation information is involved in this calculation.
Exposure (Expo.) The value of exposure is computed based on both an author’s publications as well as from the venue of those publications. By publishing more in venues of high influence, an author can accumulate more exposure; in this sense, exposure can be considered a predictor of future influence.
# of citation counts (#Cit Cnt) Total number of citations received by the papers written by this author
Citation rank (#Cit) In this case, authors are ranked according to his/her total citation counts
H-index Standard h-index (the largest integer h, that the author has h papers, each of which receives no less than h citations) for an author in a certain domain
H-in In this case, authors are ranked according to his/her h-index value
H-index
G-index Standard g-index (the largest integer g, that the author has g papers, the total #citations received by them is no less than g*g) for an author in a certain domain
G-in In this case, authors are ranked according to his/her g-index value
Avg#Cit Average number of citations received by each paper.
#B-Cit In this case, authors are ranked according to his/her balanced citation counts. The the value of balanced citations an author gets by one coauthored paper is equal to the #citations devided by the #coauthors of this paper.
Fol Follower is computed as the fixed point of the author-follower matrix. An author ranked high by this metric has been followed by many other people, and/or people largely followed. Note, in this metric, we do not consider any weighting factors according to the times an author has followed another. Each author will evenly distribute values to its following authors.
#Fol A follower of an author X is defined as an author who has cited a paper by X. This metric ranks authors by number of unique followers.
Cred Credit is computed in a similar way as "influence", but the focus is how to attribute the credit of each publication to its authors [RFMV2009]. The calculation of author scientific credits is based on the credit re-distribution matrix. In each matrix iteration, authors re-distribute their current credits to other authors according to the paper citation links. The re-distribution weight is affected by the number of authors of both citing and cited papers.
#Coa Number of distinct collaborators of this author (coauthored at least one paper)
Pval Paper value (Pval) is to first calculate each paper's influence value by applying the page-rank like algorithm on the paper citation network, then evenly distributed these values to all the coauthors.

Author-based Organisation Ranking

Metric Name Definition
Metrics selected We selected five metrics, Inf, #Cit, Expo, H-index and G-index. Two letter assignment appraoch (contribution percentile and rank percentile)
Total Score and Rank For each metric, we calculate the total score according to authors' letters, "A" = 1, "B" = 0.5, "C" = 0.25, and "D"="E" = 0
#As Score and Rank For each metric, we only count the #of "A" authors

Venue metrics

Since topologically, venues are symmetric to authors, the definition of the venue metrics are the same as the author metrics. The only differerence is that venues do not have co-author related metrics.