Lilambd commited on
Commit
24ad2a4
·
verified ·
1 Parent(s): 7c1e43d

Card update: README.md (quick start, use cases, citation, metadata)

Browse files
Files changed (1) hide show
  1. README.md +59 -0
README.md CHANGED
@@ -11,8 +11,20 @@ tags:
11
  - research
12
  - countries
13
  - monthly
 
 
 
14
  size_categories:
15
  - 1M<n<10M
 
 
 
 
 
 
 
 
 
16
  configs:
17
  - config_name: papers_by_field_year
18
  data_files: papers_by_field_year.csv
@@ -87,6 +99,53 @@ rising = load_dataset("Lilambd/world-science", "rising_subfield_country_2025", s
87
 
88
  Rebuilt from each new OpenAlex snapshot (about monthly). The script that made it is `scripts/science_face.py` (it reads only the needed columns over HTTP range requests, about 2.4MB per 750MB file, 4.6GB for the whole snapshot, then builds the tables).
89
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
90
  ## Attribution
91
 
92
  Data from [OpenAlex](https://openalex.org) (Priem, Piwowar and Orr, 2022), released CC0. Grouping and tables by [Lilambd](https://huggingface.co/Lilambd), also CC0. Questions or a slice you need: hello@lilambd.com
 
11
  - research
12
  - countries
13
  - monthly
14
+ - scientometrics
15
+ - time-series
16
+ - open-data
17
  size_categories:
18
  - 1M<n<10M
19
+ task_categories:
20
+ - time-series-forecasting
21
+ - tabular-regression
22
+ annotations_creators:
23
+ - no-annotation
24
+ language_creators:
25
+ - found
26
+ source_datasets:
27
+ - original
28
  configs:
29
  - config_name: papers_by_field_year
30
  data_files: papers_by_field_year.csv
 
99
 
100
  Rebuilt from each new OpenAlex snapshot (about monthly). The script that made it is `scripts/science_face.py` (it reads only the needed columns over HTTP range requests, about 2.4MB per 750MB file, 4.6GB for the whole snapshot, then builds the tables).
101
 
102
+ ## Quick start: datasets, pandas, DuckDB
103
+
104
+ ```python
105
+ import pandas as pd # hf:// paths need `pip install huggingface_hub`
106
+ BASE = "hf://datasets/Lilambd/world-science/"
107
+ kw = dict(keep_default_na=False, na_values=[""]) # country_code "NA" is Namibia
108
+ pcy = pd.read_csv(BASE + "papers_by_country_year.csv", **kw)
109
+ wide = pcy[pcy["year"] >= 2000].pivot_table(index="year", columns="country_code", values="papers_fractional")
110
+ print(wide[["US", "CN", "JP", "IN", "DE"]].tail())
111
+ ```
112
+
113
+ ```sql
114
+ -- DuckDB 0.10+ reads hf:// paths directly
115
+ -- the fastest-rising subfield x country cells in 2025
116
+ SELECT subfield, country, papers_2025, mean_papers_2020_2024, index_2025_vs_base100
117
+ FROM 'hf://datasets/Lilambd/world-science/rising_subfield_country_2025.csv'
118
+ ORDER BY index_2025_vs_base100 DESC LIMIT 20;
119
+ ```
120
+
121
+ A runnable notebook is in [`notebooks/quickstart.ipynb`](https://huggingface.co/datasets/Lilambd/world-science/blob/main/notebooks/quickstart.ipynb).
122
+
123
+ ## Use cases
124
+
125
+ - Country and field research-output trends from the full OpenAlex snapshot (no sampling).
126
+ - Spot rising subfields per country (2025 vs the 2020-2024 mean).
127
+ - Global South vs North comparisons with fractional counting.
128
+
129
+ ## Update cadence
130
+
131
+ About monthly: rebuilt from each new OpenAlex snapshot (see Refresh).
132
+
133
+ ## Citation
134
+
135
+ ```bibtex
136
+ @misc{lilambd2026_world_science,
137
+ author = {Lilambd},
138
+ title = {World Science: papers by field x country x year, from OpenAlex},
139
+ year = {2026},
140
+ publisher = {Hugging Face},
141
+ howpublished = {\url{https://huggingface.co/datasets/Lilambd/world-science}},
142
+ url = {https://huggingface.co/datasets/Lilambd/world-science}
143
+ }
144
+ ```
145
+
146
+ Please also credit the original sources listed in this card.
147
+
148
+
149
  ## Attribution
150
 
151
  Data from [OpenAlex](https://openalex.org) (Priem, Piwowar and Orr, 2022), released CC0. Grouping and tables by [Lilambd](https://huggingface.co/Lilambd), also CC0. Questions or a slice you need: hello@lilambd.com