slnader commited on
Commit
c6bec59
1 Parent(s): c2cfbc4

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +276 -12
README.md CHANGED
@@ -1,20 +1,92 @@
1
- ## Description
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
 
3
- Online comment floods during public consultations have posed unique governance challenges for regulatory bodies seeking relevant information on proposed regulations. How should regulatory bodies separate spam and fake comments from genuine submissions by the public, especially when fake comments are designed to imitate ordinary citizens? How can regulatory bodies achieve both breadth and depth in their citations to the comment corpus? What is the best way to select comments that represent the average submission and comments that supply highly specialized information?
4
 
5
- We present the comment corpus from the Federal Communications Commission's (FCC) 2017 "Restoring Internet Freedom" proceeding, augmented with metadata to assist in prototyping innovative search and discovery techniques. This data release contains structured metadata and the raw text of nearly 24 million comments submitted during the proceeding. The comment data were downloaded directly from the FCC's Electronic Comment Filing System (ECFS) between January and February of 2019, processed to be in a consistent format (machine-readable pdf or plain text), and augmented with information on which comments were cited in the FCC's final order.
 
6
 
7
- The release also includes query-term and document-term matrices to facilitate keyword searches on the comment corpus. An example of how these can be used with the bm25 algorithm can be found [here](https://github.com/slnader/fcc-comments/blob/main/process_comments/1_score_comments.py).
8
 
9
- Please cite any use of these data as follows:
 
 
 
 
 
 
10
 
11
- Handan-Nader, Cassandra. 2022. Do fake online comments pose a threat to regulatory policymaking? Evidence from Internet regulation in the United States. Policy & Internet. [https://doi.org/10.1002/poi3.327](https://doi.org/10.1002/poi3.327)
 
 
 
 
 
12
 
13
- ## Datasets
 
 
14
 
15
- FCC relational database (fcc.pgsql): The core components of the database include a table for submission metadata, a table for attachment metadata, a table for filer metadata, and a table that contains comment text if submitted in express format. In addition to these core tables, there are several derived tables specific to the analyses in the paper, including which submissions and attachments were cited in the final order, which submissions were associated with in-person meetings, and which submissions were associated with interest groups. Full documentation of the tables can be found in fcc_database.md.
16
 
17
- Attachments (attachments.tar.gz): Attachments to submissions that could be converted to text via OCR and saved in machine-readable pdf format. The filenames are formatted as [submission_id]_[document_id].pdf, where submission_id and document_id are keys in the relational database.
 
 
 
 
 
 
 
18
 
19
  Search datasets (search.tar.gz): Objects to facilitate prototyping of search algorithms on the comment corpus. Contains the following elements:
20
 
@@ -27,6 +99,198 @@ search_index_express.pickle | Pandas dataframe containing unique id and total te
27
  search_dtms.pickle | Document-term matrix for standard comment attachments (44655x3986) in sparse csr format (rows are comment pages, columns are bigram keyword counts). |
28
  search_index.pickle | Pandas dataframe containing unique id and total term length for standard comment attachments. |
29
 
30
- ---
31
- license: cc-by-nc-sa-4.0
32
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators:
3
+ - expert-generated
4
+ language:
5
+ - en
6
+ language_creators:
7
+ - found
8
+ license:
9
+ - cc-by-nc-sa-4.0
10
+ multilinguality:
11
+ - monolingual
12
+ pretty_name: fcc-comments
13
+ size_categories:
14
+ - 10M<n<100M
15
+ source_datasets:
16
+ - original
17
+ tags:
18
+ - notice and comment
19
+ - regulation
20
+ - government
21
+ task_categories:
22
+ - text-retrieval
23
+ task_ids:
24
+ - document-retrieval
25
+ ---
26
+
27
+ # Dataset Card for fcc-comments
28
+
29
+ ## Table of Contents
30
+ - [Table of Contents](#table-of-contents)
31
+ - [Dataset Description](#dataset-description)
32
+ - [Dataset Summary](#dataset-summary)
33
+ - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
34
+ - [Languages](#languages)
35
+ - [Dataset Structure](#dataset-structure)
36
+ - [Data Instances](#data-instances)
37
+ - [Data Fields](#data-fields)
38
+ - [Data Splits](#data-splits)
39
+ - [Dataset Creation](#dataset-creation)
40
+ - [Curation Rationale](#curation-rationale)
41
+ - [Source Data](#source-data)
42
+ - [Annotations](#annotations)
43
+ - [Personal and Sensitive Information](#personal-and-sensitive-information)
44
+ - [Considerations for Using the Data](#considerations-for-using-the-data)
45
+ - [Social Impact of Dataset](#social-impact-of-dataset)
46
+ - [Discussion of Biases](#discussion-of-biases)
47
+ - [Other Known Limitations](#other-known-limitations)
48
+ - [Additional Information](#additional-information)
49
+ - [Dataset Curators](#dataset-curators)
50
+ - [Licensing Information](#licensing-information)
51
+ - [Citation Information](#citation-information)
52
+ - [Contributions](#contributions)
53
 
54
+ ## Dataset Description
55
 
56
+ - **Repository: https://github.com/slnader/fcc-comments **
57
+ - **Paper: https://doi.org/10.1002/poi3.327 **
58
 
59
+ ### Dataset Summary
60
 
61
+ Online comment floods during public consultations have posed unique governance challenges for
62
+ regulatory bodies seeking relevant information on proposed regulations.
63
+ How should regulatory bodies separate spam and fake comments from genuine submissions by the public,
64
+ especially when fake comments are designed to imitate ordinary citizens? How can regulatory bodies
65
+ achieve both breadth and depth in their citations to the comment corpus? What is the best way to
66
+ select comments that represent the average submission and comments that supply highly specialized
67
+ information?
68
 
69
+ `fcc-comments` is an annotated version of the comment corpus from the Federal Communications Commission's
70
+ (FCC) 2017 "Restoring Internet Freedom" proceeding. The source data were downloaded directly from the FCC's Electronic
71
+ Comment Filing System (ECFS) between January and February of 2019 and include raw comment text and metadata on
72
+ comment submissions. The comment data were processed to be in a consistent format
73
+ (machine-readable pdf or plain text), and annotated with three types of information: whether the comment was cited in the
74
+ agency's final order, the type of commenter (individual, interest group, business group), and whether the comment was associated with an in-person meeting.
75
 
76
+ The release also includes query-term and document-term matrices to facilitate keyword searches on the comment corpus.
77
+ An example of how these can be used with the bm25 algorithm can be found
78
+ [here](https://github.com/slnader/fcc-comments/blob/main/process_comments/1_score_comments.py).
79
 
80
+ ## Dataset Structure
81
 
82
+ FCC relational database (fcc.pgsql): The core components of the database include a table for submission metadata,
83
+ a table for attachment metadata, a table for filer metadata, and a table that contains comment text if submitted in express format.
84
+ In addition to these core tables, there are several derived tables specific to the analyses in the paper,
85
+ including which submissions and attachments were cited in the final order, which submissions were associated with in-person meetings,
86
+ and which submissions were associated with interest groups. Full documentation of the tables can be found in fcc_database.md.
87
+
88
+ Attachments (attachments.tar.gz): Attachments to submissions that could be converted to text via OCR and saved in machine-readable pdf format.
89
+ The filenames are formatted as [submission_id]_[document_id].pdf, where submission_id and document_id are keys in the relational database.
90
 
91
  Search datasets (search.tar.gz): Objects to facilitate prototyping of search algorithms on the comment corpus. Contains the following elements:
92
 
 
99
  search_dtms.pickle | Document-term matrix for standard comment attachments (44655x3986) in sparse csr format (rows are comment pages, columns are bigram keyword counts). |
100
  search_index.pickle | Pandas dataframe containing unique id and total term length for standard comment attachments. |
101
 
102
+ ### Data Fields
103
+
104
+ The following tables are available in fcc.pgsql:
105
+
106
+ ### comments
107
+ plain text comments associated with submissions
108
+
109
+ | column | type | description |
110
+ | ----------- | ----------- | ----------- |
111
+ | comment_id | character varying(64) | unique id for plain text comment |
112
+ comment_text | text | raw text of plain text comment
113
+ row_id | integer | row sequence for plain text comments
114
+
115
+ ### submissions
116
+ metadata for submissions
117
+
118
+ | column | type | description |
119
+ | ----------- | ----------- | ----------- |
120
+ submission_id | character varying(20) | unique id for submission
121
+ submission_type | character varying(100) | type of submission (e.g., comment, reply, statement)
122
+ express_comment | numeric | 1 if express comment
123
+ date_received | date | date submission was received
124
+ contact_email | character varying(255) | submitter email address
125
+ city | character varying(255) | submitter city
126
+ address_line_1 | character varying(255) | submitter address line 1
127
+ address_line_2 | character varying(255) | submitter address line 2
128
+ state | character varying(255) | submitter state
129
+ zip_code | character varying(50) | submitter zip
130
+ comment_id | character varying(64) | unique id for plain text comment
131
+
132
+ ### filers
133
+ names of filers associated with submissions
134
+
135
+ | column | type | description |
136
+ | ----------- | ----------- | ----------- |
137
+ submission_id | character varying(20) | unique id for submission
138
+ filer_name | character varying(250) | name of filer associated with submission
139
+
140
+ ### documents
141
+ attachments associated with submissions
142
+
143
+ | column | type | description |
144
+ | ----------- | ----------- | ----------- |
145
+ submission_id | character varying(20) | unique id for submission
146
+ document_name | text | filename of attachment
147
+ download_status | numeric | status of attachment download
148
+ document_id | character varying(64) | unique id for attachment
149
+ file_extension | character varying(4) | file extension for attachment
150
+
151
+ ### filers_cited
152
+ citations from final order
153
+
154
+ | column | type | description |
155
+ | ----------- | ----------- | ----------- |
156
+ point | numeric | paragraph number in final order
157
+ filer_name | character varying(250) | name of cited filer
158
+ submission_type | character varying(12) | type of submission as indicated in final order
159
+ page_numbers | text[] | cited page numbers
160
+ cite_id | integer | unique id for citation
161
+ filer_id | character varying(250) | id for cited filer
162
+
163
+ ### docs_cited
164
+ attachments associated with cited submissions
165
+
166
+ | column | type | description |
167
+ | ----------- | ----------- | ----------- |
168
+ cite_id | numeric | unique id for citation
169
+ submission_id | character varying(20) | unique id for submission
170
+ document_id | character varying(64) | unique id for attachment
171
+
172
+
173
+ ### near_duplicates
174
+ lookup table for comment near-duplicates
175
+
176
+ | column | type | description |
177
+ | ----------- | ----------- | ----------- |
178
+ target_document_id | unique id for target document
179
+ duplicate_document_id | unique id for duplicate of target document
180
+
181
+ ### exact_duplicates
182
+ lookup table for comment exact duplicates
183
+
184
+ | column | type | description |
185
+ | ----------- | ----------- | ----------- |
186
+ target_document_id | character varying(100) | unique id for target document
187
+ duplicate_document_id | character varying(100) | unique id for duplicate of target document
188
+
189
+ ### in_person_exparte
190
+ submissions associated with ex parte meeting
191
+
192
+ | column | type | description |
193
+ | ----------- | ----------- | ----------- |
194
+ submission_id | character varying(20) | unique id for submission
195
+
196
+ ### interest_groups
197
+ submissions associated with interest groups
198
+
199
+ | column | type | description |
200
+ | ----------- | ----------- | ----------- |
201
+ submission_id | character varying(20) | unique id for submission
202
+ business | numeric | 1 if business group, 0 otherwise
203
+
204
+
205
+ ## Dataset Creation
206
+
207
+ ### Curation Rationale
208
+
209
+ The data were curated to perform information retrieval and summarization tasks as documented in https://doi.org/10.1002/poi3.327.
210
+
211
+ ### Source Data
212
+
213
+ #### Initial Data Collection and Normalization
214
+
215
+ The data for this study come from the FCC's Electronic Comment Filing System (ECFS) system, accessed between January and February of 2019.
216
+ I converted the API responses into a normalized, relational database containing information on 23,951,967 submissions.
217
+ 23,938,686 "express" submissions contained a single plain text comment submitted directly through the comment form.
218
+ 13,821 "standard" submissions contained one or more comment documents submitted as attachments in various file formats.
219
+ While the FCC permitted any file format for attachments, I only consider documents attached in pdf, plain text, rich text,
220
+ and Microsoft Word file formats, and I drop submitted documents that were simply copies of the FCC’s official documents (e.g., the NPRM itself).
221
+ Using standard OCR software, I attempted to convert all attachments into plain text and saved them as machine-readable pdfs.
222
+
223
+ #### Who are the source language producers?
224
+
225
+ All submitters of public comments during the public comment period (but see note on fake comments in considerations).
226
+
227
+ ### Annotations
228
+
229
+ #### Annotation process
230
+
231
+ - Citations: I consider citations from the main text of the FCC's final rule. I did not include citations to
232
+ supporting documents not available through ECFS (e.g., court decisions), nor did I include citations
233
+ to submissions from prior FCC proceedings. The direct citations to filed submissions are included
234
+ in a series of 1,186 footnotes. The FCC’s citation format typically followed a relatively standard
235
+ pattern: the name of the filer (e.g., Verizon), a description of the document (e.g., Comment), and
236
+ at times a page number. I extracted citations from the text using regular expressions. Based on a
237
+ random sample of paragraphs from the final order, the regular expressions identified 98% of eligible citations,
238
+ while successfully excluding all non-citation text. In total, this produced 1,886 unique citations.
239
+ I then identified which of the comments were cited. First, I identified all documents from the cited filer
240
+ that had enough pages to contain the page number cited (if provided), and, where applicable, whose filename
241
+ contained the moniker from the FCC’s citation (e.g., "Reply"). The majority of citations matched to only one
242
+ possible comment submitted, and I identified the re- maining cited comments through manual review of the citations.
243
+ In this way, I was able to tag documents associated with all but three citations. When the same cited document was
244
+ submitted under multiple separate submissions, I tagged all versions of the document as being cited.
245
+
246
+ - Commenter type: Comments are labeled as mass comments if 10 or more duplicate or near-duplicate copies were
247
+ submitted by individual commenters. Near-duplicates were defined as comments with non-zero identical information scores.
248
+ To identify the type of commenter for non-mass comments, I take advantage of the fact that the vast majority of organized
249
+ groups preferred standard submissions over express submissions. Any non-mass comment submitted as an express comment was
250
+ coded as coming from an individual. To distinguish between individuals and organizations that used standard submissions,
251
+ I use a first name and surname database from the names dataset Python package to characterize filer names as belonging to
252
+ individuals or organizations. I also use the domain of the submitter’s email address to re-categorize comments as coming
253
+ from organizations if they were submitted on behalf of organizations by an individual. Government officials were identified by
254
+ their .gov email addresses. I manually review this procedure for mischaracterizations. After obtaining a list of organization
255
+ names, I manually code each one as belonging to a business group or a non-business group. Government officials writing in
256
+ their official capacity were categorized as a non-business group.
257
+
258
+ - In-person meetings: To identify which commenters held in-person meetings with the agency, I collect all comments labeled
259
+ as an ex-parte submission in the EFCS. I manually review these submissions for mention of an in-person meeting. I label
260
+ a commenter as having held an in-person meeting if they submitted at least one ex-parte document that mentioned an in-person meeting.
261
+
262
+ #### Who are the annotators?
263
+
264
+ Annotations are a combination of automated and manual review done by the author.
265
+
266
+ ### Personal and Sensitive Information
267
+
268
+ This dataset may contain personal and sensitive information, as there were no restrictions on what commenters could submit to
269
+ the agency. This dataset also contains numerous examples of profanity and spam. These comments represent what the FCC decided was
270
+ appropriate to share publicly on their own website.
271
+
272
+ ## Considerations for Using the Data
273
+
274
+ ### Discussion of Biases
275
+
276
+ This proceeding was famous for the large number of "fake" comments (comments impersonating ordinary citizens) submitted to the
277
+ agency (see [this report](https://ag.ny.gov/sites/default/files/oag-fakecommentsreport.pdf) by the NY AG for more information).
278
+ As such, this comment corpus contains a mix of computer-generated and natural language, and there is currently no way to reliably separate
279
+ mass comments submitted with the approval of the commenter and those submitted on behalf of the commenter without their knowledge.
280
+
281
+ ## Additional Information
282
+
283
+ ### Licensing Information
284
+
285
+ CreativeCommons Attribution-NonCommercial-ShareAlike 4.0 International.
286
+
287
+ ### Citation Information
288
+
289
+ ```
290
+ @article{handan2022,
291
+ title={Do fake online comments pose a threat to regulatory policymaking? Evidence from Internet regulation in the United States},
292
+ author={Handan-Nader, Cassandra},
293
+ journal={Policy \& Internet},
294
+ year={2022}
295
+ }
296
+ ```