scilist: Make reading lists

In MaximeRivest/scimeetr: Scientometrics Meet R

Description

scilist ranks papers according to different metrics. The metric is determined by the argument 'reading_list'.

Usage

scilist(scimeetr_data, reading_list = "core_papers", k = 5, m = 3)

Arguments

scimeetr_data	An object of class 'scimeetr'
reading_list	A vector of length one. Equal to either: 'core_papers', 'core_yr', 'core_residual', 'by_expert_LC', 'by_expert_TC', 'group_of_experts_TC', 'group_of_experts_LC', 'cite_most_others', 'direct_cite_eigen', 'betweeness', 'closeness', 'connectness', 'link_strength', 'page_rank', 'journal_dis', 'journal_unique_combn'
k	Length of list per community
m	Not always used. m is used when there is a two step filter. An example of a two step filter is the list generated when reading_list = 'by_expert'. First k will determine how many experts should be listed, then m will determine how many papers per experts should be listed.

Details

With the parameter reading_list, we can get any of the following 12 reading lists that fits into three categories:

• Core

  • core_papers

  • core_yr

  • core_residual

• Experts

  • by_expert_LC

  • by_expert_TC

  • group_of_experts_TC

  • group_of_experts_LC

• Centrality

  • cite_most_others

  • betweeness

  • closeness

  • connectness

  • page_rank

Using scilist with reading_list = "core_papers" will list the most cited paper. The parameter k controls the number of paper to list.

Using scilist with reading_list = "core_yr" will list the most cited paper for each year from three years before present to ten years before present. The parameter k controls the number of paper per year to list.

Using scilist with reading_list = "core_residual" will list the papers that diverge most from the expected number of citation for this particular paper.

Using scilist with reading_list = "by_expert_LC" we will get a list of recent papers by one or a few experts in the community. For the option by_expert_LC, authors are ranked based on their harmonic local H-index. The H-index is a measure of an other productivity and impact. An author with an H-index of 10 means that he has published at least 10 papers with 10 or more citation each. A local H-index means that only citations from other papers in the community are counted. A harmonic local H-index means that authors do not get the full credit for each citation their paper received. It is corrected depending on the authos position in the authors list. First authors gets most of the credit, then the last author gets the second most, and the authors gets credit as a proportion of their position. Once the authors harmonic-local-H-index is found they are ranked and the m most recent publication of the k most 'expert' authors are listed as a reading list.

Using scilist with reading_list = "by_expert_TC" instead of reading_list = "by_expert_LC", notice the _TC instead of the _LC will based the ranking calculation on total citation of it's publications instead of only the local citations.

Using scilist with reading_list = "group_of_experts_LC" we will get a list of papers for which many authors are experts in the community. For this option, authors are assigned a harmonic local H-index like described in the previous section. But this time, a weighted sum of the harmonic-local-H-index of each authors of a paper is calculated.

Betweeness measures the importance of a paper in connecting two clusters of papers. Papers with a high betweeness would therefore be a paper that tend to be more interdisciplinary.

Closeness measures the average number of link between a paper and all other papers. Papers with a high closeness would therefore be a paper that tend to have a large and wide list of citations.

Connectness measures the number of links a paper has. Papers with a high connectness would therefore be a paper that tend to have cited what most other studies cited.

With the option cite_most_others, the papers that cite most other papers of the community can be found. This is not a centrality measure but it is also based on papers connection to each other. It should tend to find litterature review and recent papers that have an especially good grasp on the community.

Value

A data frame.

MaximeRivest/scimeetr documentation built on May 8, 2019, 9:51 a.m.