StepUpLaw commited on
Commit
adc2d9e
·
verified ·
1 Parent(s): 1eec608

Dataset, metadata and card from legal-empirics

Browse files
.gitattributes CHANGED
@@ -58,3 +58,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
 
 
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
61
+ disputed-terms-national.csv filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,90 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-4.0
3
+ language:
4
+ - en
5
+ pretty_name: Disputed Contract and Instrument Terms in US State Appellate Courts
6
+ size_categories:
7
+ - 100K<n<1M
8
+ tags:
9
+ - legal
10
+ - law
11
+ - empirical-legal-studies
12
+ - contract-interpretation
13
+ - ambiguity
14
+ - corpus-linguistics
15
+ - legal-drafting
16
+ - empirical-legal-studies
17
+ - state-courts
18
+ configs:
19
+ - config_name: default
20
+ data_files:
21
+ - split: train
22
+ path: disputed-terms-national.csv
23
+ ---
24
+
25
+ # Disputed Contract and Instrument Terms in US State Appellate Courts
26
+
27
+ One row per (term, decision) pair: a word or phrase a court quoted as the disputed language in a decision containing an ambiguity holding, across the state appellate courts of all 50 states and DC. Terms are extracted from the whole opinion by cue-anchored quotation ('the term "X"', '"X" as used in'), classified into functional drafting categories, and linked to the holding by three graded levels of evidence.
28
+
29
+ **286,846 rows.** Licence CC BY 4.0. Not legal advice.
30
+
31
+ ## Where this comes from
32
+
33
+ | | |
34
+ |---|---|
35
+ | Canonical record | https://doi.org/10.5281/zenodo.22247377 |
36
+ | Code and methodology | https://github.com/stepuplaw/legal-empirics |
37
+ | Research page | https://stepuplaw.com/research/ |
38
+ | Author | Kevin D. Klagge, [ORCID 0009-0002-1385-8498](https://orcid.org/0009-0002-1385-8498) |
39
+ | Source corpus | CourtListener bulk export, snapshot 2026-06-30 |
40
+
41
+ The DOI above identifies the **code**, which is a different object from this
42
+ dataset. Cite the code when you are describing the method and cite this dataset
43
+ when you are using the numbers.
44
+
45
+ ## Columns
46
+
47
+ | Column | Type | Meaning |
48
+ |---|---|---|
49
+ | `term` | string | the disputed language, lowercased and normalised, as the court quoted it |
50
+ | `category` | string | functional drafting class: nexus, degree, temporal, scope, modal, role, succession, property, event, conduct, condition, mental, uncategorised |
51
+ | `source` | string | the instrument the words sit in: testamentary, deed, insurance, contract, statute, constitution, uncertain |
52
+ | `posture` | string | found | rejected | alleged | uncertain -- whether the court held the language ambiguous |
53
+ | `link` | string | direct | proximate | inferred -- strength of the link between this term and the holding; inferred rows must not be pooled with the other two |
54
+ | `oid` | integer | CourtListener opinion id |
55
+ | `cid` | integer | CourtListener cluster id |
56
+ | `court` | string | CourtListener court id |
57
+ | `state` | string | two-letter state postal code |
58
+ | `year` | integer | year the decision was filed |
59
+ | `statement` | string | the row written as one self-contained English sentence, so it can be retrieved, quoted and checked on its own |
60
+
61
+ Every row carries a `statement` column, which is the row written as one
62
+ self-contained English sentence. A row of codes can be downloaded but not
63
+ retrieved or quoted, and the sentence is what makes each row usable on its own.
64
+
65
+ ## How it was built
66
+
67
+ Retrieval and extraction are deterministic code over a local corpus of 10.8M US
68
+ judicial opinions. Classification uses rules written against a hand-coded sample
69
+ that ship with their measured accuracy, so the error rate is reported rather
70
+ than assumed. Every study states its exclusion funnel with counts, because
71
+ silent filtering is the commonest defect in research on opinions and it is
72
+ invisible in the result.
73
+
74
+ `datapackage.json` carries the Frictionless schema, `croissant.json` the
75
+ MLCommons Croissant description, and `dataset.jsonld` the schema.org form.
76
+
77
+ ## Limits
78
+
79
+ **This is exploratory.** The coded samples behind it were coded once, so it
80
+ supports a described pattern rather than a measurement. Inter-annotator
81
+ reliability has not been established.
82
+
83
+ **Published appellate opinions are not disputes.** Most disputes settle, most
84
+ settlements are unpublished, and appellate coverage varies by court and decade.
85
+ Any rate here is a rate among decisions that reached an appellate court and were
86
+ published, which is not the same population a drafter cares about.
87
+
88
+ **Read the study's own limitations section** in the repository before quoting a
89
+ number. Each one names the specific threats to its own validity, including the
90
+ ones that are unflattering.
croissant.json ADDED
@@ -0,0 +1,239 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "@context": {
3
+ "@language": "en",
4
+ "@vocab": "https://schema.org/",
5
+ "cr": "http://mlcommons.org/croissant/",
6
+ "sc": "https://schema.org/",
7
+ "dct": "http://purl.org/dc/terms/",
8
+ "data": {
9
+ "@id": "cr:data",
10
+ "@type": "@json"
11
+ },
12
+ "dataType": {
13
+ "@id": "cr:dataType",
14
+ "@type": "@vocab"
15
+ },
16
+ "extract": "cr:extract",
17
+ "field": "cr:field",
18
+ "fileObject": "cr:fileObject",
19
+ "fileProperty": "cr:fileProperty",
20
+ "column": "cr:column",
21
+ "recordSet": "cr:recordSet",
22
+ "source": "cr:source",
23
+ "subField": "cr:subField",
24
+ "references": "cr:references",
25
+ "repeated": "cr:repeated",
26
+ "examples": {
27
+ "@id": "cr:examples",
28
+ "@type": "@json"
29
+ },
30
+ "conformsTo": "dct:conformsTo"
31
+ },
32
+ "@type": "sc:Dataset",
33
+ "conformsTo": "http://mlcommons.org/croissant/1.0",
34
+ "citeAs": "Klagge, Kevin D. legal-empirics. https://doi.org/10.5281/zenodo.22247377",
35
+ "name": "disputed_terms_national",
36
+ "description": "One row per (term, decision) pair: a word or phrase a court quoted as the disputed language in a decision containing an ambiguity holding, across the state appellate courts of all 50 states and DC. Terms are extracted from the whole opinion by cue-anchored quotation ('the term \"X\"', '\"X\" as used in'), classified into functional drafting categories, and linked to the holding by three graded levels of evidence.",
37
+ "url": "https://stepuplaw.com/research/disputed-terms-national/",
38
+ "license": "https://creativecommons.org/licenses/by/4.0/",
39
+ "version": "2026-09-02",
40
+ "keywords": [
41
+ "contract interpretation",
42
+ "ambiguity",
43
+ "corpus linguistics",
44
+ "legal drafting",
45
+ "empirical legal studies",
46
+ "state courts"
47
+ ],
48
+ "creator": {
49
+ "@type": "sc:Person",
50
+ "name": "Kevin D. Klagge",
51
+ "sameAs": "https://orcid.org/0009-0002-1385-8498"
52
+ },
53
+ "distribution": [
54
+ {
55
+ "@type": "cr:FileObject",
56
+ "@id": "disputed-terms-national.csv",
57
+ "name": "disputed-terms-national.csv",
58
+ "description": "The dataset as a single UTF-8 CSV with a header row.",
59
+ "contentUrl": "https://stepuplaw.com/research/disputed-terms-national/disputed-terms-national.csv",
60
+ "encodingFormat": "text/csv",
61
+ "sha256": "41794102c0d5c052b40fda0c7779415be95d43a61d08bbbe5462fb66269b11bd"
62
+ }
63
+ ],
64
+ "recordSet": [
65
+ {
66
+ "@type": "cr:RecordSet",
67
+ "@id": "rows",
68
+ "name": "rows",
69
+ "description": "Disputed Contract and Instrument Terms in US State Appellate Courts",
70
+ "field": [
71
+ {
72
+ "@type": "cr:Field",
73
+ "@id": "rows/term",
74
+ "name": "term",
75
+ "description": "the disputed language, lowercased and normalised, as the court quoted it",
76
+ "dataType": "sc:Text",
77
+ "source": {
78
+ "fileObject": {
79
+ "@id": "disputed-terms-national.csv"
80
+ },
81
+ "extract": {
82
+ "column": "term"
83
+ }
84
+ }
85
+ },
86
+ {
87
+ "@type": "cr:Field",
88
+ "@id": "rows/category",
89
+ "name": "category",
90
+ "description": "functional drafting class: nexus, degree, temporal, scope, modal, role, succession, property, event, conduct, condition, mental, uncategorised",
91
+ "dataType": "sc:Text",
92
+ "source": {
93
+ "fileObject": {
94
+ "@id": "disputed-terms-national.csv"
95
+ },
96
+ "extract": {
97
+ "column": "category"
98
+ }
99
+ }
100
+ },
101
+ {
102
+ "@type": "cr:Field",
103
+ "@id": "rows/source",
104
+ "name": "source",
105
+ "description": "the instrument the words sit in: testamentary, deed, insurance, contract, statute, constitution, uncertain",
106
+ "dataType": "sc:Text",
107
+ "source": {
108
+ "fileObject": {
109
+ "@id": "disputed-terms-national.csv"
110
+ },
111
+ "extract": {
112
+ "column": "source"
113
+ }
114
+ }
115
+ },
116
+ {
117
+ "@type": "cr:Field",
118
+ "@id": "rows/posture",
119
+ "name": "posture",
120
+ "description": "found | rejected | alleged | uncertain -- whether the court held the language ambiguous",
121
+ "dataType": "sc:Text",
122
+ "source": {
123
+ "fileObject": {
124
+ "@id": "disputed-terms-national.csv"
125
+ },
126
+ "extract": {
127
+ "column": "posture"
128
+ }
129
+ }
130
+ },
131
+ {
132
+ "@type": "cr:Field",
133
+ "@id": "rows/link",
134
+ "name": "link",
135
+ "description": "direct | proximate | inferred -- strength of the link between this term and the holding; inferred rows must not be pooled with the other two",
136
+ "dataType": "sc:Text",
137
+ "source": {
138
+ "fileObject": {
139
+ "@id": "disputed-terms-national.csv"
140
+ },
141
+ "extract": {
142
+ "column": "link"
143
+ }
144
+ }
145
+ },
146
+ {
147
+ "@type": "cr:Field",
148
+ "@id": "rows/oid",
149
+ "name": "oid",
150
+ "description": "CourtListener opinion id",
151
+ "dataType": "sc:Integer",
152
+ "source": {
153
+ "fileObject": {
154
+ "@id": "disputed-terms-national.csv"
155
+ },
156
+ "extract": {
157
+ "column": "oid"
158
+ }
159
+ }
160
+ },
161
+ {
162
+ "@type": "cr:Field",
163
+ "@id": "rows/cid",
164
+ "name": "cid",
165
+ "description": "CourtListener cluster id",
166
+ "dataType": "sc:Integer",
167
+ "source": {
168
+ "fileObject": {
169
+ "@id": "disputed-terms-national.csv"
170
+ },
171
+ "extract": {
172
+ "column": "cid"
173
+ }
174
+ }
175
+ },
176
+ {
177
+ "@type": "cr:Field",
178
+ "@id": "rows/court",
179
+ "name": "court",
180
+ "description": "CourtListener court id",
181
+ "dataType": "sc:Text",
182
+ "source": {
183
+ "fileObject": {
184
+ "@id": "disputed-terms-national.csv"
185
+ },
186
+ "extract": {
187
+ "column": "court"
188
+ }
189
+ }
190
+ },
191
+ {
192
+ "@type": "cr:Field",
193
+ "@id": "rows/state",
194
+ "name": "state",
195
+ "description": "two-letter state postal code",
196
+ "dataType": "sc:Text",
197
+ "source": {
198
+ "fileObject": {
199
+ "@id": "disputed-terms-national.csv"
200
+ },
201
+ "extract": {
202
+ "column": "state"
203
+ }
204
+ }
205
+ },
206
+ {
207
+ "@type": "cr:Field",
208
+ "@id": "rows/year",
209
+ "name": "year",
210
+ "description": "year the decision was filed",
211
+ "dataType": "sc:Integer",
212
+ "source": {
213
+ "fileObject": {
214
+ "@id": "disputed-terms-national.csv"
215
+ },
216
+ "extract": {
217
+ "column": "year"
218
+ }
219
+ }
220
+ },
221
+ {
222
+ "@type": "cr:Field",
223
+ "@id": "rows/statement",
224
+ "name": "statement",
225
+ "description": "the row written as one self-contained English sentence, so it can be retrieved, quoted and checked on its own",
226
+ "dataType": "sc:Text",
227
+ "source": {
228
+ "fileObject": {
229
+ "@id": "disputed-terms-national.csv"
230
+ },
231
+ "extract": {
232
+ "column": "statement"
233
+ }
234
+ }
235
+ }
236
+ ]
237
+ }
238
+ ]
239
+ }
datapackage.json ADDED
@@ -0,0 +1,110 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "disputed-terms-national",
3
+ "title": "Disputed Contract and Instrument Terms in US State Appellate Courts",
4
+ "description": "One row per (term, decision) pair: a word or phrase a court quoted as the disputed language in a decision containing an ambiguity holding, across the state appellate courts of all 50 states and DC. Terms are extracted from the whole opinion by cue-anchored quotation ('the term \"X\"', '\"X\" as used in'), classified into functional drafting categories, and linked to the holding by three graded levels of evidence.",
5
+ "licenses": [
6
+ {
7
+ "name": "CC BY 4.0",
8
+ "url": "https://creativecommons.org/licenses/by/4.0/"
9
+ }
10
+ ],
11
+ "homepage": "https://stepuplaw.com/research/disputed-terms-national/",
12
+ "version": "2026-09-02",
13
+ "created": "2026-09-02",
14
+ "sources": [
15
+ {
16
+ "title": "CourtListener bulk export, topped up nightly from the courts",
17
+ "path": "https://www.courtlistener.com/help/api/bulk-data/"
18
+ }
19
+ ],
20
+ "resources": [
21
+ {
22
+ "name": "disputed-terms-national",
23
+ "path": "disputed-terms-national.csv",
24
+ "format": "csv",
25
+ "mediatype": "text/csv",
26
+ "encoding": "utf-8",
27
+ "bytes": 80549261,
28
+ "rows": 286846,
29
+ "hash": "sha256:41794102c0d5c052b40fda0c7779415be95d43a61d08bbbe5462fb66269b11bd",
30
+ "schema": {
31
+ "fields": [
32
+ {
33
+ "name": "term",
34
+ "type": "string",
35
+ "description": "the disputed language, lowercased and normalised, as the court quoted it"
36
+ },
37
+ {
38
+ "name": "category",
39
+ "type": "string",
40
+ "description": "functional drafting class: nexus, degree, temporal, scope, modal, role, succession, property, event, conduct, condition, mental, uncategorised"
41
+ },
42
+ {
43
+ "name": "source",
44
+ "type": "string",
45
+ "description": "the instrument the words sit in: testamentary, deed, insurance, contract, statute, constitution, uncertain"
46
+ },
47
+ {
48
+ "name": "posture",
49
+ "type": "string",
50
+ "description": "found | rejected | alleged | uncertain -- whether the court held the language ambiguous"
51
+ },
52
+ {
53
+ "name": "link",
54
+ "type": "string",
55
+ "description": "direct | proximate | inferred -- strength of the link between this term and the holding; inferred rows must not be pooled with the other two"
56
+ },
57
+ {
58
+ "name": "oid",
59
+ "type": "integer",
60
+ "description": "CourtListener opinion id"
61
+ },
62
+ {
63
+ "name": "cid",
64
+ "type": "integer",
65
+ "description": "CourtListener cluster id"
66
+ },
67
+ {
68
+ "name": "court",
69
+ "type": "string",
70
+ "description": "CourtListener court id"
71
+ },
72
+ {
73
+ "name": "state",
74
+ "type": "string",
75
+ "description": "two-letter state postal code"
76
+ },
77
+ {
78
+ "name": "year",
79
+ "type": "integer",
80
+ "description": "year the decision was filed"
81
+ },
82
+ {
83
+ "name": "statement",
84
+ "type": "string",
85
+ "description": "the row written as one self-contained English sentence, so it can be retrieved, quoted and checked on its own"
86
+ }
87
+ ]
88
+ }
89
+ }
90
+ ],
91
+ "provenance": {
92
+ "repository": "https://github.com/stepuplaw/legal-empirics",
93
+ "commit": "a8774f7d8d99fd5c49e459ba6e55e12dee7971f1",
94
+ "run": {
95
+ "query": "(ambiguous OR ambiguity OR ambiguities)",
96
+ "scope": "national state appellate (S+SA)",
97
+ "corpus": "CourtListener bulk snapshot 2026-06-30",
98
+ "workers": 8,
99
+ "runtime_minutes": 119.7,
100
+ "funnel": {
101
+ "decisions_queued": 236561,
102
+ "noterm": 118750,
103
+ "ok": 117804,
104
+ "noamb": 7
105
+ },
106
+ "term_hits": 286846,
107
+ "distinct_terms": 100679
108
+ }
109
+ }
110
+ }
dataset-metadata.json ADDED
@@ -0,0 +1,84 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "title": "Disputed Contract Terms, 51 US Jurisdictions",
3
+ "subtitle": "Words courts were asked to call ambiguous, and whether the challenge won",
4
+ "id": "stepuplaw/disputed-terms-national",
5
+ "licenses": [
6
+ {
7
+ "name": "CC-BY-4.0"
8
+ }
9
+ ],
10
+ "keywords": [
11
+ "contract interpretation",
12
+ "ambiguity",
13
+ "corpus linguistics",
14
+ "legal drafting",
15
+ "empirical legal studies",
16
+ "state courts"
17
+ ],
18
+ "description": "One row per (term, decision) pair: a word or phrase a court quoted as the disputed language in a decision containing an ambiguity holding, across the state appellate courts of all 50 states and DC. Terms are extracted from the whole opinion by cue-anchored quotation ('the term \"X\"', '\"X\" as used in'), classified into functional drafting categories, and linked to the holding by three graded levels of evidence.\n\nCanonical record and methodology: https://github.com/stepuplaw/legal-empirics\nResearch page: https://stepuplaw.com/research/\nAuthor: Kevin D. Klagge, ORCID https://orcid.org/0009-0002-1385-8498\n\nEvery row carries a `statement` column, which is the row written as one English sentence so it can be read and quoted on its own. Column definitions are in datapackage.json; the same data is also described in MLCommons Croissant.",
19
+ "resources": [
20
+ {
21
+ "path": "disputed-terms-national.csv",
22
+ "description": "Disputed Contract and Instrument Terms in US State Appellate Courts",
23
+ "schema": {
24
+ "fields": [
25
+ {
26
+ "name": "term",
27
+ "description": "the disputed language, lowercased and normalised, as the court quoted it",
28
+ "type": "string"
29
+ },
30
+ {
31
+ "name": "category",
32
+ "description": "functional drafting class: nexus, degree, temporal, scope, modal, role, succession, property, event, conduct, condition, mental, uncategorised",
33
+ "type": "string"
34
+ },
35
+ {
36
+ "name": "source",
37
+ "description": "the instrument the words sit in: testamentary, deed, insurance, contract, statute, constitution, uncertain",
38
+ "type": "string"
39
+ },
40
+ {
41
+ "name": "posture",
42
+ "description": "found | rejected | alleged | uncertain -- whether the court held the language ambiguous",
43
+ "type": "string"
44
+ },
45
+ {
46
+ "name": "link",
47
+ "description": "direct | proximate | inferred -- strength of the link between this term and the holding; inferred rows must not be pooled with the other two",
48
+ "type": "string"
49
+ },
50
+ {
51
+ "name": "oid",
52
+ "description": "CourtListener opinion id",
53
+ "type": "integer"
54
+ },
55
+ {
56
+ "name": "cid",
57
+ "description": "CourtListener cluster id",
58
+ "type": "integer"
59
+ },
60
+ {
61
+ "name": "court",
62
+ "description": "CourtListener court id",
63
+ "type": "string"
64
+ },
65
+ {
66
+ "name": "state",
67
+ "description": "two-letter state postal code",
68
+ "type": "string"
69
+ },
70
+ {
71
+ "name": "year",
72
+ "description": "year the decision was filed",
73
+ "type": "integer"
74
+ },
75
+ {
76
+ "name": "statement",
77
+ "description": "the row written as one self-contained English sentence, so it can be retrieved, quoted and checked on its own",
78
+ "type": "string"
79
+ }
80
+ ]
81
+ }
82
+ }
83
+ ]
84
+ }
dataset.jsonld ADDED
@@ -0,0 +1,116 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "@context": "https://schema.org",
3
+ "@type": "Dataset",
4
+ "name": "Disputed Contract and Instrument Terms in US State Appellate Courts",
5
+ "description": "One row per (term, decision) pair: a word or phrase a court quoted as the disputed language in a decision containing an ambiguity holding, across the state appellate courts of all 50 states and DC. Terms are extracted from the whole opinion by cue-anchored quotation ('the term \"X\"', '\"X\" as used in'), classified into functional drafting categories, and linked to the holding by three graded levels of evidence.",
6
+ "url": "https://stepuplaw.com/research/disputed-terms-national/",
7
+ "identifier": [],
8
+ "keywords": [
9
+ "contract interpretation",
10
+ "ambiguity",
11
+ "corpus linguistics",
12
+ "legal drafting",
13
+ "empirical legal studies",
14
+ "state courts"
15
+ ],
16
+ "license": "https://creativecommons.org/licenses/by/4.0/",
17
+ "isAccessibleForFree": true,
18
+ "version": "2026-09-02",
19
+ "dateModified": "2026-09-02",
20
+ "creator": {
21
+ "@type": "Person",
22
+ "name": "Kevin D. Klagge",
23
+ "url": "https://stepuplaw.com",
24
+ "identifier": "https://orcid.org/0009-0002-1385-8498",
25
+ "sameAs": "https://orcid.org/0009-0002-1385-8498"
26
+ },
27
+ "publisher": {
28
+ "@type": "Person",
29
+ "name": "Kevin D. Klagge",
30
+ "url": "https://stepuplaw.com",
31
+ "sameAs": "https://orcid.org/0009-0002-1385-8498"
32
+ },
33
+ "sameAs": [
34
+ "https://github.com/stepuplaw/legal-empirics",
35
+ "https://huggingface.co/datasets/stepuplaw/disputed-terms-national",
36
+ "https://doi.org/10.5281/zenodo.22247377"
37
+ ],
38
+ "isBasedOn": {
39
+ "@type": "SoftwareSourceCode",
40
+ "identifier": "https://doi.org/10.5281/zenodo.22247377",
41
+ "codeRepository": "https://github.com/stepuplaw/legal-empirics"
42
+ },
43
+ "citation": "(ambiguous OR ambiguity OR ambiguities)",
44
+ "measurementTechnique": "Full-text retrieval over a local CourtListener corpus, followed by deterministic sentence-level classification. Retrieval is wide and filtering is explicit; the exclusion funnel is reported with counts.",
45
+ "distribution": [
46
+ {
47
+ "@type": "DataDownload",
48
+ "encodingFormat": "text/csv",
49
+ "contentUrl": "https://stepuplaw.com/research/disputed-terms-national/disputed-terms-national.csv",
50
+ "contentSize": "80549261",
51
+ "sha256": "41794102c0d5c052b40fda0c7779415be95d43a61d08bbbe5462fb66269b11bd"
52
+ },
53
+ {
54
+ "@type": "DataDownload",
55
+ "encodingFormat": "application/json",
56
+ "contentUrl": "https://stepuplaw.com/research/disputed-terms-national/datapackage.json"
57
+ }
58
+ ],
59
+ "variableMeasured": [
60
+ {
61
+ "@type": "PropertyValue",
62
+ "name": "term",
63
+ "description": "the disputed language, lowercased and normalised, as the court quoted it"
64
+ },
65
+ {
66
+ "@type": "PropertyValue",
67
+ "name": "category",
68
+ "description": "functional drafting class: nexus, degree, temporal, scope, modal, role, succession, property, event, conduct, condition, mental, uncategorised"
69
+ },
70
+ {
71
+ "@type": "PropertyValue",
72
+ "name": "source",
73
+ "description": "the instrument the words sit in: testamentary, deed, insurance, contract, statute, constitution, uncertain"
74
+ },
75
+ {
76
+ "@type": "PropertyValue",
77
+ "name": "posture",
78
+ "description": "found | rejected | alleged | uncertain -- whether the court held the language ambiguous"
79
+ },
80
+ {
81
+ "@type": "PropertyValue",
82
+ "name": "link",
83
+ "description": "direct | proximate | inferred -- strength of the link between this term and the holding; inferred rows must not be pooled with the other two"
84
+ },
85
+ {
86
+ "@type": "PropertyValue",
87
+ "name": "oid",
88
+ "description": "CourtListener opinion id"
89
+ },
90
+ {
91
+ "@type": "PropertyValue",
92
+ "name": "cid",
93
+ "description": "CourtListener cluster id"
94
+ },
95
+ {
96
+ "@type": "PropertyValue",
97
+ "name": "court",
98
+ "description": "CourtListener court id"
99
+ },
100
+ {
101
+ "@type": "PropertyValue",
102
+ "name": "state",
103
+ "description": "two-letter state postal code"
104
+ },
105
+ {
106
+ "@type": "PropertyValue",
107
+ "name": "year",
108
+ "description": "year the decision was filed"
109
+ },
110
+ {
111
+ "@type": "PropertyValue",
112
+ "name": "statement",
113
+ "description": "the row written as one self-contained English sentence, so it can be retrieved, quoted and checked on its own"
114
+ }
115
+ ]
116
+ }
disputed-terms-national.csv ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41794102c0d5c052b40fda0c7779415be95d43a61d08bbbe5462fb66269b11bd
3
+ size 80549261