undeleted commited on
Commit
01894a3
1 Parent(s): 3979779

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +13 -1
README.md CHANGED
@@ -69,7 +69,7 @@ This dataset is primarily intended for unsupervised training of text generation
69
 
70
  ### Data Splits
71
 
72
- No splitting of the data was performed. This dataset is intended for unsupervised training.
73
 
74
  ## Dataset Creation
75
 
@@ -83,9 +83,21 @@ Common Crawl, do not provide enough variety, or are simply outdated.
83
 
84
  #### Initial Data Collection and Normalization
85
 
 
 
86
  First, a list of active Fandom wikis was gathered into a text file. Active is defined as "having at least 250 images on the wiki."
87
  This list was gathered in early January 2023, despite the actual wiki content being more recent.
88
 
 
 
 
 
 
 
 
 
 
 
89
  #### Who are the source language producers?
90
 
91
  The contributors of each wiki.
 
69
 
70
  ### Data Splits
71
 
72
+ No splitting of the data was performed.
73
 
74
  ## Dataset Creation
75
 
 
83
 
84
  #### Initial Data Collection and Normalization
85
 
86
+ *More information about any referenced scripts, commands, or programs used may be found in the BigKnow2022 GitHub repository.*
87
+
88
  First, a list of active Fandom wikis was gathered into a text file. Active is defined as "having at least 250 images on the wiki."
89
  This list was gathered in early January 2023, despite the actual wiki content being more recent.
90
 
91
+ Second, the `scrape_fandom.py` script was used to generate and download an up to date dump for each of the wikis.
92
+
93
+ Third, `wikiextractor` was used to process these dumps into single XML files containing each article stripped of all formatting
94
+ besides links.
95
+
96
+ Fourth, `dump2jsonl` was used to convert the XML files into JSONL files with an article per line. Light markdown formatting was
97
+ applied, converting the HTML links to markdown-formatted links, and automatically making the article's title a header.
98
+
99
+ Finally, the JSONL files were concatenated into the Fandom23K dataset.
100
+
101
  #### Who are the source language producers?
102
 
103
  The contributors of each wiki.