Datasets:
rcds
/

Nina Baumgartner commited on
Commit
21d3508
·
2 Parent(s): 275caca 5629529

Merge remote-tracking branch 'origin/main' into main

Browse files
Files changed (1) hide show
  1. README.md +24 -74
README.md CHANGED
@@ -52,7 +52,8 @@ task_ids: []
52
 
53
  ## Dataset Summary
54
 
55
- This dataset contains an implementation of occlusion for the SwissJudgmentPrediction task.
 
56
 
57
  ### Documents
58
  Occlusion-Swiss-Judgment-Prediction is a subset of the [Swiss-Judgment-Prediction](https://huggingface.co/datasets/swiss_judgment_prediction) dataset.
@@ -70,89 +71,38 @@ Switzerland has four official languages with 3 languages (German, French and Ita
70
  ### Data Instances
71
  #### Multilingual use of the dataset
72
 
73
- When the dataset is used in a multilingual setting selecting the the 'all_languages' flag:
74
-
75
- ```python
76
- from datasets import load_dataset
77
- dataset = load_dataset('occlusion_swiss_judgment_prediction', 'all_languages')
78
-
79
- ```
80
-
81
-
82
- For occlusion:
83
-
84
- ```json
85
- {
86
- "id": 57542,
87
- "year": 2017,
88
- "label": "dismissal",
89
- "language": "de",
90
- "region": "Espace_Mittelland",
91
- "canton": "BE",
92
- "legal_area": "social_law",
93
- "explainability_label": "Supports judgment",
94
- "occluded_text": "Mit Entscheid vom 4. Mai 2017 hiess das Verwaltungsgericht des Kantons Bern die Beschwerde teilweise gut, hob die angefochtene Verf\u00fcgung auf und wies die Sache an die IV-Stelle zur\u00fcck, damit sie, nach Vornahme der Abkl\u00e4rungen im Sinne der Erw\u00e4gungen, \u00fcber den Rentenanspruch neu verf\u00fcge",
95
- "text": "Sachverhalt: A. Die 1987 geborene, aus Bolivien stammende A.A._ leidet an einer pr\u00e4natalen Retinopathie und einer dadurch bedingten schweren Sehbehinderung beidseits. Im Dezember 2003 reiste sie in die Schweiz ein. Im Mai 2007 verheiratete sie sich mit B.A._. Am 22. Juni 2007 meldete sie sich bei der Invalidenversicherung zum Leistungsbezug an. Am 6. Februar 2008 lehnte die IV-Stelle Bern das Leistungsgesuch verf\u00fcgungsweise ab, weil die versicherungsm\u00e4ssigen Voraussetzungen nicht erf\u00fcllt seien. Am 22. Oktober 2009 stellte die Versicherte ein neues Leistungsgesuch, auf welches die IV-Stelle mit Verf\u00fcgung vom 16. Dezember 2009 nicht e**int**rat. Am 27. Juli 2010 meldete sich A.A._ erneut bei der Invalidenversicherung zum Leistungsbezug an, wobei sie eine Verschlechterung der Sehkraft geltend machte. Mit Verf\u00fcgung vom 10. November 2010 sprach ihr die IV-Stelle ab 1. August 2010 eine Entsch\u00e4digung f\u00fcr leichte Hilflosigkeit zu. Soweit das Gesuch den Invalidenrentenanspruch betraf, trat die IV-Stelle darauf nicht ein (Verf\u00fcgung vom 19. November 2010). Im Juli 2011 erteilte die IV-Stelle Kostengutsprache f\u00fcr berufliche Eingliederungsmassnahmen, welche die Versicherte vorzeitig abbrach. Mit Verf\u00fcgung vom 27. Januar 2016 er\u00f6ffnete die IV-Stelle der Versicherten, dass sie keinen Anspruch auf eine Invalidenrente habe. Der Versicherungsfall sei bereits vor der Einreise in die Schweiz eingetreten. B. A.A._ liess Beschwerde einreichen mit dem Rechtsbegehren, unter Aufhebung der Verf\u00fcgung vom 27. Januar 2016 sei ihr eine ausserordentliche Invalidenrente zuzusprechen. . C. Die IV-Stelle f\u00fchrt Beschwerde in \u00f6ffentlich-rechtlichen Angelegenheiten mit dem Antrag, der vorinstanzliche Entscheid sei aufzuheben. "
96
- }
97
-
98
- ```
99
-
100
- #### Monolingual use of the dataset
101
-
102
- When the dataset is used in a monolingual setting selecting the ISO language code for one of the 3 supported languages. For example:
103
-
104
- ```python
105
- from datasets import load_dataset
106
- dataset = load_dataset('occlusion_swiss_judgment_prediction', 'de_1')
107
- ```
108
- For occlusion:
109
-
110
- ```json
111
- {
112
- "id": 57542,
113
- "year": 2017,
114
- "label": "dismissal",
115
- "language": "de",
116
- "region": "Espace_Mittelland",
117
- "canton": "BE",
118
- "legal_area": "social_law",
119
- "explainability_label": "Supports judgment",
120
- "occluded_text": "Mit Entscheid vom 4. Mai 2017 hiess das Verwaltungsgericht des Kantons Bern die Beschwerde teilweise gut, hob die angefochtene Verf\u00fcgung auf und wies die Sache an die IV-Stelle zur\u00fcck, damit sie, nach Vornahme der Abkl\u00e4rungen im Sinne der Erw\u00e4gungen, \u00fcber den Rentenanspruch neu verf\u00fcge",
121
- "text": "Sachverhalt: A. Die 1987 geborene, aus Bolivien stammende A.A._ leidet an einer pr\u00e4natalen Retinopathie und einer dadurch bedingten schweren Sehbehinderung beidseits. Im Dezember 2003 reiste sie in die Schweiz ein. Im Mai 2007 verheiratete sie sich mit B.A._. Am 22. Juni 2007 meldete sie sich bei der Invalidenversicherung zum Leistungsbezug an. Am 6. Februar 2008 lehnte die IV-Stelle Bern das Leistungsgesuch verf\u00fcgungsweise ab, weil die versicherungsm\u00e4ssigen Voraussetzungen nicht erf\u00fcllt seien. Am 22. Oktober 2009 stellte die Versicherte ein neues Leistungsgesuch, auf welches die IV-Stelle mit Verf\u00fcgung vom 16. Dezember 2009 nicht e**int**rat. Am 27. Juli 2010 meldete sich A.A._ erneut bei der Invalidenversicherung zum Leistungsbezug an, wobei sie eine Verschlechterung der Sehkraft geltend machte. Mit Verf\u00fcgung vom 10. November 2010 sprach ihr die IV-Stelle ab 1. August 2010 eine Entsch\u00e4digung f\u00fcr leichte Hilflosigkeit zu. Soweit das Gesuch den Invalidenrentenanspruch betraf, trat die IV-Stelle darauf nicht ein (Verf\u00fcgung vom 19. November 2010). Im Juli 2011 erteilte die IV-Stelle Kostengutsprache f\u00fcr berufliche Eingliederungsmassnahmen, welche die Versicherte vorzeitig abbrach. Mit Verf\u00fcgung vom 27. Januar 2016 er\u00f6ffnete die IV-Stelle der Versicherten, dass sie keinen Anspruch auf eine Invalidenrente habe. Der Versicherungsfall sei bereits vor der Einreise in die Schweiz eingetreten. B. A.A._ liess Beschwerde einreichen mit dem Rechtsbegehren, unter Aufhebung der Verf\u00fcgung vom 27. Januar 2016 sei ihr eine ausserordentliche Invalidenrente zuzusprechen. . C. Die IV-Stelle f\u00fchrt Beschwerde in \u00f6ffentlich-rechtlichen Angelegenheiten mit dem Antrag, der vorinstanzliche Entscheid sei aufzuheben. "
122
- }
123
-
124
- ```
125
 
126
  ### Data Fields
127
 
128
  The following data fields are provided for documents (train, validation):
129
 
130
- id: (**int**) a unique identifier of the for the document
131
- year: (**int**) the publication year
132
- text: (**str**) the facts of the case
133
- label: (class label) the judgment outcome: 0 (dismissal) or 1 (approval)
134
- language: (**str**) one of (de, fr, it)
135
- region: (**str**) the region of the lower court
136
- canton: (**str**) the canton of the lower court
137
- legal area: (**str**) the legal area of the case
138
-
139
- The following data fields are provided for documents (occ_test):
140
-
141
- id: (**int**) a unique identifier of the for the document
142
- year: (**int**) the publication year
143
- label: (**str**) the judgment outcome: dismissal or approval
144
- language: (**str**) one of (de, fr, it)
145
- region: (**str**) the region of the lower court
146
- canton: (**str**) the canton of the lower court
147
- legal area: (**str**) the legal area of the case
148
- explainability_label (**str**): the explainability label assigned to the occluded text: Supports judgment, Opposes judgment, Neutral, Baseline
149
- occluded_text (**str**): the occluded text
150
- text: (**str**) the facts of the case w/o the occluded text except for cases w/ explainability label "Baseline" (contain entire facts)
151
 
152
  Note that Baseline cases are only contained in version 1 of the occlusion test set, since they do not change from experiment to experiment.
153
 
154
  ### Data Splits
155
  Language | Subset | Number of Documents (Training/Validation/Test)
 
156
  German| de_1| 35'452 / 4'705 / 427
157
  German| de_2| 35'452 / 4'705 / 1366
158
  German| de_3| 35'452 / 4'705 / 3567
 
52
 
53
  ## Dataset Summary
54
 
55
+ This dataset contains an implementation of occlusion for the SwissJudgmentPrediction task.
56
+ Note that this dataset set only provides a test set and should be used in comination with the [Swiss-Judgment-Prediction](https://huggingface.co/datasets/swiss_judgment_prediction) dataset.
57
 
58
  ### Documents
59
  Occlusion-Swiss-Judgment-Prediction is a subset of the [Swiss-Judgment-Prediction](https://huggingface.co/datasets/swiss_judgment_prediction) dataset.
 
71
  ### Data Instances
72
  #### Multilingual use of the dataset
73
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
74
 
75
  ### Data Fields
76
 
77
  The following data fields are provided for documents (train, validation):
78
 
79
+ id: (**int**) a unique identifier of the for the document<br/>
80
+ year: (**int**) the publication year<br/>
81
+ text: (**str**) the facts of the case<br/>
82
+ label: (class label) the judgment outcome: 0 (dismissal) or 1 (approval)<br/>
83
+ language: (**str**) one of (de, fr, it)<br/>
84
+ region: (**str**) the region of the lower court<br/>
85
+ canton: (**str**) the canton of the lower court<br/>
86
+ legal area: (**str**) the legal area of the case<br/>
87
+
88
+ The following data fields are provided for documents (test):
89
+
90
+ id: (**int**) a unique identifier of the for the document <br/>
91
+ year: (**int**) the publication year<br/>
92
+ label: (**str**) the judgment outcome: dismissal or approval<br/>
93
+ language: (**str**) one of (de, fr, it)<br/>
94
+ region: (**str**) the region of the lower court<br/>
95
+ canton: (**str**) the canton of the lower court<br/>
96
+ legal area: (**str**) the legal area of the case<br/>
97
+ explainability_label (**str**): the explainability label assigned to the occluded text: Supports judgment, Opposes judgment, Neutral, Baseline<br/>
98
+ occluded_text (**str**): the occluded text<br/>
99
+ text: (**str**) the facts of the case w/o the occluded text except for cases w/ explainability label "Baseline" (contain entire facts)<br/>
100
 
101
  Note that Baseline cases are only contained in version 1 of the occlusion test set, since they do not change from experiment to experiment.
102
 
103
  ### Data Splits
104
  Language | Subset | Number of Documents (Training/Validation/Test)
105
+ | ----------- | ----------- | ----------- |
106
  German| de_1| 35'452 / 4'705 / 427
107
  German| de_2| 35'452 / 4'705 / 1366
108
  German| de_3| 35'452 / 4'705 / 3567