Skip to contents

Getting Started with scimapR

Source: vignettes/scimapR.Rmd
scimapR.Rmd

What is scimapR?

scimapR is a comprehensive toolkit for bibliometric and scientometric analysis in R. It is designed as a complement to the foundational bibliometrix package, adding:

  • Live corpus refresh with staleness tracking and locking
  • Research questions as first-class objects with LLM-grounded screening
  • Reproducible-by-construction corpus certificates for exact re-derivation
  • Author trajectory analysis with topic pivots and collaborator turnover
  • Equity and representation auditing with built-in epistemic humility
  • LLM-grounded corpus chat with retrieval-constrained citations

Quick start 

library(scimapR)

# Generate a synthetic example corpus
corpus <- sm_example_corpus(n_works = 100, seed = 42)
print(corpus)
#> 
#> ── <sm_corpus> ─────────────────────────────────────────────────────────────────
#> Works: 100 | Authors: 80 | Institutions: 0
#> Years: 2015 - 2024
#> Sources (journals): 10
#> Embeddings: 100 x 64
#> Provenance: synthetic (100)
#> Status: Unlocked (last refreshed: 2026-06-01 16:49:55)

Exploring the corpus 

# View works
head(corpus$works[, c("work_id", "title", "year", "cited_by_count")])
#> # A tibble: 6 × 4
#>   work_id    title                                           year cited_by_count
#>   <chr>      <chr>                                          <int>          <int>
#> 1 W000000001 Colorectal Cancer in colorectal cancer: a met…  2023             17
#> 2 W000000002 Drug Resistance in machine learning: a cohort…  2021             24
#> 3 W000000003 Spatial Transcriptomics in single-cell RNA-se…  2024             16
#> 4 W000000004 Single-Cell RNA-Seq in gene expression: a met…  2024             22
#> 5 W000000005 Immune Checkpoint in immune checkpoint: a pro…  2024              2
#> 6 W000000006 Biomarker Discovery in spatial transcriptomic…  2022             50
# View authors
head(corpus$authors[, c("author_id", "display_name")])
#> # A tibble: 6 × 2
#>   author_id  display_name  
#>   <chr>      <chr>         
#> 1 A000000001 Anna Garcia   
#> 2 A000000002 Maria Kumar   
#> 3 A000000003 Fatima Mueller
#> 4 A000000004 Yuki Kumar    
#> 5 A000000005 David Brown   
#> 6 A000000006 Hiroshi Sato

Visualisation

All plots use viridis colour palettes by default.

Annual production

Annual production 

sm_plot_top(corpus, level = "authors", n = 10)
Top authors

Top authors

Networks and large graphs

Network plots such as sm_plot_citation_network() and sm_plot_collab() return ggraph objects whose layout is computed lazily at print time. On large graphs that lazy layout (and the embedded graph) can crash a knitr/callr/ workflowr render subprocess. For documents, pass precompute = TRUE: the layout is computed eagerly and a self-contained plain ggplot is returned, which you can saveRDS() and print later in the document without recomputing anything — even for a graph with a couple of thousand nodes.

The max_nodes cap is opt-in (NULL by default), so existing renders are never silently downsampled; set it only when you deliberately want to bound a very large graph (a cli message reports when it engages).

p <- sm_plot_citation_network(corpus, precompute = TRUE)
# saveRDS(p, "fig_network.rds")  # print later in the document, cheaply
p

Filtering 

recent <- sm_filter_works(corpus, year_range = c(2020, 2024))
nrow(recent$works)
#> [1] 53

Next steps