albertvillanova HF staff commited on
Commit
ba5319c
1 Parent(s): f09a51f

Delete legacy dataset_infos.json

Browse files
Files changed (1) hide show
  1. dataset_infos.json +0 -194
dataset_infos.json DELETED
@@ -1,194 +0,0 @@
1
- {
2
- "full": {
3
- "description": "\nThe ASSIN (Avalia\u00e7\u00e3o de Similaridade Sem\u00e2ntica e INfer\u00eancia textual) corpus is a corpus annotated with pairs of sentences written in\nPortuguese that is suitable for the exploration of textual entailment and paraphrasing classifiers. The corpus contains pairs of sentences\nextracted from news articles written in European Portuguese (EP) and Brazilian Portuguese (BP), obtained from Google News Portugal\nand Brazil, respectively. To create the corpus, the authors started by collecting a set of news articles describing the\nsame event (one news article from Google News Portugal and another from Google News Brazil) from Google News.\nThen, they employed Latent Dirichlet Allocation (LDA) models to retrieve pairs of similar sentences between sets of news\narticles that were grouped together around the same topic. For that, two LDA models were trained (for EP and for BP)\non external and large-scale collections of unannotated news articles from Portuguese and Brazilian news providers, respectively.\nThen, the authors defined a lower and upper threshold for the sentence similarity score of the retrieved pairs of sentences,\ntaking into account that high similarity scores correspond to sentences that contain almost the same content (paraphrase candidates),\nand low similarity scores correspond to sentences that are very different in content from each other (no-relation candidates).\nFrom the collection of pairs of sentences obtained at this stage, the authors performed some manual grammatical corrections\nand discarded some of the pairs wrongly retrieved. Furthermore, from a preliminary analysis made to the retrieved sentence pairs\nthe authors noticed that the number of contradictions retrieved during the previous stage was very low. Additionally, they also\nnoticed that event though paraphrases are not very frequent, they occur with some frequency in news articles. Consequently,\nin contrast with the majority of the currently available corpora for other languages, which consider as labels \u201cneutral\u201d, \u201centailment\u201d\nand \u201ccontradiction\u201d for the task of RTE, the authors of the ASSIN corpus decided to use as labels \u201cnone\u201d, \u201centailment\u201d and \u201cparaphrase\u201d.\nFinally, the manual annotation of pairs of sentences was performed by human annotators. At least four annotators were randomly\nselected to annotate each pair of sentences, which is done in two steps: (i) assigning a semantic similarity label (a score between 1 and 5,\nfrom unrelated to very similar); and (ii) providing an entailment label (one sentence entails the other, sentences are paraphrases,\nor no relation). Sentence pairs where at least three annotators do not agree on the entailment label were considered controversial\nand thus discarded from the gold standard annotations. The full dataset has 10,000 sentence pairs, half of which in Brazilian Portuguese\nand half in European Portuguese. Either language variant has 2,500 pairs for training, 500 for validation and 2,000 for testing.\n",
4
- "citation": "\n@inproceedings{fonseca2016assin,\n title={ASSIN: Avaliacao de similaridade semantica e inferencia textual},\n author={Fonseca, E and Santos, L and Criscuolo, Marcelo and Aluisio, S},\n booktitle={Computational Processing of the Portuguese Language-12th International Conference, Tomar, Portugal},\n pages={13--15},\n year={2016}\n}\n",
5
- "homepage": "http://nilc.icmc.usp.br/assin/",
6
- "license": "",
7
- "features": {
8
- "sentence_pair_id": {
9
- "dtype": "int64",
10
- "_type": "Value"
11
- },
12
- "premise": {
13
- "dtype": "string",
14
- "_type": "Value"
15
- },
16
- "hypothesis": {
17
- "dtype": "string",
18
- "_type": "Value"
19
- },
20
- "relatedness_score": {
21
- "dtype": "float32",
22
- "_type": "Value"
23
- },
24
- "entailment_judgment": {
25
- "names": [
26
- "NONE",
27
- "ENTAILMENT",
28
- "PARAPHRASE"
29
- ],
30
- "_type": "ClassLabel"
31
- }
32
- },
33
- "builder_name": "parquet",
34
- "dataset_name": "assin",
35
- "config_name": "full",
36
- "version": {
37
- "version_str": "1.0.0",
38
- "major": 1,
39
- "minor": 0,
40
- "patch": 0
41
- },
42
- "splits": {
43
- "train": {
44
- "name": "train",
45
- "num_bytes": 986499,
46
- "num_examples": 5000,
47
- "dataset_name": null
48
- },
49
- "test": {
50
- "name": "test",
51
- "num_bytes": 767304,
52
- "num_examples": 4000,
53
- "dataset_name": null
54
- },
55
- "validation": {
56
- "name": "validation",
57
- "num_bytes": 196821,
58
- "num_examples": 1000,
59
- "dataset_name": null
60
- }
61
- },
62
- "download_size": 1335013,
63
- "dataset_size": 1950624,
64
- "size_in_bytes": 3285637
65
- },
66
- "ptpt": {
67
- "description": "\nThe ASSIN (Avalia\u00e7\u00e3o de Similaridade Sem\u00e2ntica e INfer\u00eancia textual) corpus is a corpus annotated with pairs of sentences written in\nPortuguese that is suitable for the exploration of textual entailment and paraphrasing classifiers. The corpus contains pairs of sentences\nextracted from news articles written in European Portuguese (EP) and Brazilian Portuguese (BP), obtained from Google News Portugal\nand Brazil, respectively. To create the corpus, the authors started by collecting a set of news articles describing the\nsame event (one news article from Google News Portugal and another from Google News Brazil) from Google News.\nThen, they employed Latent Dirichlet Allocation (LDA) models to retrieve pairs of similar sentences between sets of news\narticles that were grouped together around the same topic. For that, two LDA models were trained (for EP and for BP)\non external and large-scale collections of unannotated news articles from Portuguese and Brazilian news providers, respectively.\nThen, the authors defined a lower and upper threshold for the sentence similarity score of the retrieved pairs of sentences,\ntaking into account that high similarity scores correspond to sentences that contain almost the same content (paraphrase candidates),\nand low similarity scores correspond to sentences that are very different in content from each other (no-relation candidates).\nFrom the collection of pairs of sentences obtained at this stage, the authors performed some manual grammatical corrections\nand discarded some of the pairs wrongly retrieved. Furthermore, from a preliminary analysis made to the retrieved sentence pairs\nthe authors noticed that the number of contradictions retrieved during the previous stage was very low. Additionally, they also\nnoticed that event though paraphrases are not very frequent, they occur with some frequency in news articles. Consequently,\nin contrast with the majority of the currently available corpora for other languages, which consider as labels \u201cneutral\u201d, \u201centailment\u201d\nand \u201ccontradiction\u201d for the task of RTE, the authors of the ASSIN corpus decided to use as labels \u201cnone\u201d, \u201centailment\u201d and \u201cparaphrase\u201d.\nFinally, the manual annotation of pairs of sentences was performed by human annotators. At least four annotators were randomly\nselected to annotate each pair of sentences, which is done in two steps: (i) assigning a semantic similarity label (a score between 1 and 5,\nfrom unrelated to very similar); and (ii) providing an entailment label (one sentence entails the other, sentences are paraphrases,\nor no relation). Sentence pairs where at least three annotators do not agree on the entailment label were considered controversial\nand thus discarded from the gold standard annotations. The full dataset has 10,000 sentence pairs, half of which in Brazilian Portuguese\nand half in European Portuguese. Either language variant has 2,500 pairs for training, 500 for validation and 2,000 for testing.\n",
68
- "citation": "\n@inproceedings{fonseca2016assin,\n title={ASSIN: Avaliacao de similaridade semantica e inferencia textual},\n author={Fonseca, E and Santos, L and Criscuolo, Marcelo and Aluisio, S},\n booktitle={Computational Processing of the Portuguese Language-12th International Conference, Tomar, Portugal},\n pages={13--15},\n year={2016}\n}\n",
69
- "homepage": "http://nilc.icmc.usp.br/assin/",
70
- "license": "",
71
- "features": {
72
- "sentence_pair_id": {
73
- "dtype": "int64",
74
- "_type": "Value"
75
- },
76
- "premise": {
77
- "dtype": "string",
78
- "_type": "Value"
79
- },
80
- "hypothesis": {
81
- "dtype": "string",
82
- "_type": "Value"
83
- },
84
- "relatedness_score": {
85
- "dtype": "float32",
86
- "_type": "Value"
87
- },
88
- "entailment_judgment": {
89
- "names": [
90
- "NONE",
91
- "ENTAILMENT",
92
- "PARAPHRASE"
93
- ],
94
- "_type": "ClassLabel"
95
- }
96
- },
97
- "builder_name": "parquet",
98
- "dataset_name": "assin",
99
- "config_name": "ptpt",
100
- "version": {
101
- "version_str": "1.0.0",
102
- "major": 1,
103
- "minor": 0,
104
- "patch": 0
105
- },
106
- "splits": {
107
- "train": {
108
- "name": "train",
109
- "num_bytes": 522994,
110
- "num_examples": 2500,
111
- "dataset_name": null
112
- },
113
- "test": {
114
- "name": "test",
115
- "num_bytes": 392880,
116
- "num_examples": 2000,
117
- "dataset_name": null
118
- },
119
- "validation": {
120
- "name": "validation",
121
- "num_bytes": 105618,
122
- "num_examples": 500,
123
- "dataset_name": null
124
- }
125
- },
126
- "download_size": 706661,
127
- "dataset_size": 1021492,
128
- "size_in_bytes": 1728153
129
- },
130
- "ptbr": {
131
- "description": "\nThe ASSIN (Avalia\u00e7\u00e3o de Similaridade Sem\u00e2ntica e INfer\u00eancia textual) corpus is a corpus annotated with pairs of sentences written in\nPortuguese that is suitable for the exploration of textual entailment and paraphrasing classifiers. The corpus contains pairs of sentences\nextracted from news articles written in European Portuguese (EP) and Brazilian Portuguese (BP), obtained from Google News Portugal\nand Brazil, respectively. To create the corpus, the authors started by collecting a set of news articles describing the\nsame event (one news article from Google News Portugal and another from Google News Brazil) from Google News.\nThen, they employed Latent Dirichlet Allocation (LDA) models to retrieve pairs of similar sentences between sets of news\narticles that were grouped together around the same topic. For that, two LDA models were trained (for EP and for BP)\non external and large-scale collections of unannotated news articles from Portuguese and Brazilian news providers, respectively.\nThen, the authors defined a lower and upper threshold for the sentence similarity score of the retrieved pairs of sentences,\ntaking into account that high similarity scores correspond to sentences that contain almost the same content (paraphrase candidates),\nand low similarity scores correspond to sentences that are very different in content from each other (no-relation candidates).\nFrom the collection of pairs of sentences obtained at this stage, the authors performed some manual grammatical corrections\nand discarded some of the pairs wrongly retrieved. Furthermore, from a preliminary analysis made to the retrieved sentence pairs\nthe authors noticed that the number of contradictions retrieved during the previous stage was very low. Additionally, they also\nnoticed that event though paraphrases are not very frequent, they occur with some frequency in news articles. Consequently,\nin contrast with the majority of the currently available corpora for other languages, which consider as labels \u201cneutral\u201d, \u201centailment\u201d\nand \u201ccontradiction\u201d for the task of RTE, the authors of the ASSIN corpus decided to use as labels \u201cnone\u201d, \u201centailment\u201d and \u201cparaphrase\u201d.\nFinally, the manual annotation of pairs of sentences was performed by human annotators. At least four annotators were randomly\nselected to annotate each pair of sentences, which is done in two steps: (i) assigning a semantic similarity label (a score between 1 and 5,\nfrom unrelated to very similar); and (ii) providing an entailment label (one sentence entails the other, sentences are paraphrases,\nor no relation). Sentence pairs where at least three annotators do not agree on the entailment label were considered controversial\nand thus discarded from the gold standard annotations. The full dataset has 10,000 sentence pairs, half of which in Brazilian Portuguese\nand half in European Portuguese. Either language variant has 2,500 pairs for training, 500 for validation and 2,000 for testing.\n",
132
- "citation": "\n@inproceedings{fonseca2016assin,\n title={ASSIN: Avaliacao de similaridade semantica e inferencia textual},\n author={Fonseca, E and Santos, L and Criscuolo, Marcelo and Aluisio, S},\n booktitle={Computational Processing of the Portuguese Language-12th International Conference, Tomar, Portugal},\n pages={13--15},\n year={2016}\n}\n",
133
- "homepage": "http://nilc.icmc.usp.br/assin/",
134
- "license": "",
135
- "features": {
136
- "sentence_pair_id": {
137
- "dtype": "int64",
138
- "_type": "Value"
139
- },
140
- "premise": {
141
- "dtype": "string",
142
- "_type": "Value"
143
- },
144
- "hypothesis": {
145
- "dtype": "string",
146
- "_type": "Value"
147
- },
148
- "relatedness_score": {
149
- "dtype": "float32",
150
- "_type": "Value"
151
- },
152
- "entailment_judgment": {
153
- "names": [
154
- "NONE",
155
- "ENTAILMENT",
156
- "PARAPHRASE"
157
- ],
158
- "_type": "ClassLabel"
159
- }
160
- },
161
- "builder_name": "parquet",
162
- "dataset_name": "assin",
163
- "config_name": "ptbr",
164
- "version": {
165
- "version_str": "1.0.0",
166
- "major": 1,
167
- "minor": 0,
168
- "patch": 0
169
- },
170
- "splits": {
171
- "train": {
172
- "name": "train",
173
- "num_bytes": 463505,
174
- "num_examples": 2500,
175
- "dataset_name": null
176
- },
177
- "test": {
178
- "name": "test",
179
- "num_bytes": 374424,
180
- "num_examples": 2000,
181
- "dataset_name": null
182
- },
183
- "validation": {
184
- "name": "validation",
185
- "num_bytes": 91203,
186
- "num_examples": 500,
187
- "dataset_name": null
188
- }
189
- },
190
- "download_size": 639490,
191
- "dataset_size": 929132,
192
- "size_in_bytes": 1568622
193
- }
194
- }