added emotion

README.md

@@ -113,49 +113,48 @@ The data fields are the same among all splits.
 - `text_start`: a `int` feature.
 - `text_end`: a `int` feature.
 
+#### tweet_emotion
+- `text`: a `string` feature.
+- `gold_label_list`: a list of  `string` feature.
+
+
 ### Data Splits
 
 | task             | description                        | number of instances   |
 |:-----------------|:-----------------------------------|:----------------------|
-| tweet_topic      | multi-label classification         | 4585 / 573 / 1679     |
-| tweet_ner7       | sequence labeling                  | 4616 / 576 / 2807     |
-| tweet_qa         | generation                         | 9489 / 1086 / 1203    |
-| tweet_intimacy   | regression on a single text        | 1191 / 396 / 396      |
+| tweet_topic      | multi-label classification         | 4,585 / 573 / 1,679     |
+| tweet_ner7       | sequence labeling                  | 4,616 / 576 / 2,807     |
+| tweet_qa         | generation                         | 9,489 / 1,086 / 1,203    |
+| tweet_intimacy   | regression on a single text        | 1,191 / 396 / 396      |
 | tweet_similarity | regression on two texts            | 450 / 100 / 450       |
-| tempo_wic        | binary classification on two texts | 1427 / 395 / 1472     |
-| tweet_hate       | multi-class classification         | 5019 / 716 / 1433     |
+| tempo_wic        | binary classification on two texts | 1,427 / 395 / 1,472     |
+| tweet_hate       | multi-class classification         | 5,019 / 716 / 1,433     |
 | tweet_emoji      | multi-class classification         | 50,000 / 5,000 / 50,000 |
-| tweet_sentiment  | ABSA on a five-pointscale          | 26632 / 4000 / 12379  |
+| tweet_sentiment  | ABSA on a five-pointscale          | 26,632 / 4,000 / 12,379  |
 | tweet_nerd       | binary classification              | * / 407 / *           |
+| tweet_emotion       | multi-label classification      | 6,838 / 886 / 3,259      |
 
 
 ## Evaluation Metrics
-#### tweet_topic
-```macro-F1```
+- __tweet_topic:__ ```macro-F1```
 
-#### tweet_ner7
-``` ```
+- __tweet_ner7:__ ```TBA```
 
-#### tweet_qa
-``` ```
+- __tweet_qa:__ ```TBA```
 
+- __tweet_intimacy:__ ```TBA```
 
-#### tweet_intimacy
+- __tweet_similarity:__ ```TBA```
 
-#### tweet_similarity
+- __tempo_wic:__ ```Accuracy ```
 
-#### tempo_wic
-```Accuracy ```
+- __tweet_hate:__ ```TBA```
 
-#### tweet_hate
+- __tweet_emoji:__ ``` Accuracy at top 5 ```
 
-#### tweet_emoji
-``` Accuracy at top 5 ```
+- __tweet_sentiment:__ ```TBA```
 
-#### tweet_sentiment
-```   ```
-
-#### tweet_nerd
+- __tweet_nerd:__ ```TBA```
 
 
 
@@ -303,4 +302,24 @@ TBA
   journal={arXiv preprint arXiv:2210.08129},
   year={2022}
 }
+```
+
+- TweetEmotion
+```
+@inproceedings{mohammad-etal-2018-semeval,
+    title = "{S}em{E}val-2018 Task 1: Affect in Tweets",
+    author = "Mohammad, Saif  and
+      Bravo-Marquez, Felipe  and
+      Salameh, Mohammad  and
+      Kiritchenko, Svetlana",
+    booktitle = "Proceedings of the 12th International Workshop on Semantic Evaluation",
+    month = jun,
+    year = "2018",
+    address = "New Orleans, Louisiana",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/S18-1001",
+    doi = "10.18653/v1/S18-1001",
+    pages = "1--17",
+    abstract = "We present the SemEval-2018 Task 1: Affect in Tweets, which includes an array of subtasks on inferring the affectual state of a person from their tweet. For each task, we created labeled data from English, Arabic, and Spanish tweets. The individual tasks are: 1. emotion intensity regression, 2. emotion intensity ordinal classification, 3. valence (sentiment) regression, 4. valence ordinal classification, and 5. emotion classification. Seventy-five teams (about 200 team members) participated in the shared task. We summarize the methods, resources, and tools used by the participating teams, with a focus on the techniques and resources that are particularly useful. We also analyze systems for consistent bias towards a particular race or gender. The data is made freely available to further improve our understanding of how people convey emotions through language.",
+}
 ```

data/tweet_emotion/map.txt

@@ -0,0 +1,11 @@
+anger,0
+anticipation,1
+disgust,2
+fear,3
+joy,4
+love,5
+optimism,6
+pessimism,7
+sadness,8
+surprise,9
+trust,10

process/tweet_emotion.py

@@ -0,0 +1,52 @@
+import pandas as pd
+import urllib
+
+
+# format text
+def clean_text(text):
+    text = text.replace('\n', ' ').replace('\r', ' ').replace('\t', ' ')
+
+    new_text = []
+    for t in text.split():
+        # MAKE SURE to check lowercase
+        t = '@user' if t.startswith('@') and len(t) > 1 and t.replace(
+            '@', '').lower() not in verified_users else t
+        t = '{URL}' if t.startswith('http') else t
+        new_text.append(t)
+
+    return ' '.join(new_text)
+
+
+train = pd.read_csv('./emotion/2018-E-c-En-train.txt', sep='\t')
+validation = pd.read_csv('./emotion/2018-E-c-En-dev.txt', sep='\t')
+test = pd.read_csv('./emotion/2018-E-c-En-test-gold.txt', sep='\t')
+
+sem_emotions = train.columns.difference(['ID', 'Tweet', 'split', 'dataset'])
+
+# keep class mapping
+with open('../data/tweet_emotion/map.txt', 'w') as f:
+    for idx, em in enumerate(sem_emotions):
+        f.write(f'{em},{idx}\n')
+
+
+cols_to_keep = ['text', 'gold_label_list']
+# get list of verified users
+verified_users = urllib.request.urlopen(
+    'https://raw.githubusercontent.com/cardiffnlp/timelms/main/data/verified_users.v091122.txt').readlines()
+verified_users = [x.decode().strip('\n').lower() for x in verified_users]
+
+# clean  datasets
+train['gold_label_list'] = train[sem_emotions].values.tolist()
+train['text'] = train['Tweet']
+train['text'] = train['text'].apply(clean_text)
+train[cols_to_keep].to_json('../data/tweet_emotion/train.jsonl', lines=True, orient='records')
+
+validation['gold_label_list'] = validation[sem_emotions].values.tolist()
+validation['text'] = validation['Tweet']
+validation['text'] = validation['text'].apply(clean_text)
+validation[cols_to_keep].to_json('../data/tweet_emotion/validation.jsonl', lines=True, orient='records')
+
+test['gold_label_list'] = test[sem_emotions].values.tolist()
+test['text'] = test['Tweet']
+test['text'] = test['text'].apply(clean_text)
+test[cols_to_keep].to_json('../data/tweet_emotion/test.jsonl', lines=True, orient='records')

super_tweeteval.py

@@ -2,7 +2,7 @@
 import json
 import datasets
 
-_VERSION = "0.1.42"
+_VERSION = "0.1.43"
 _SUPER_TWEETEVAL_CITATION = """TBA"""
 _SUPER_TWEETEVAL_DESCRIPTION = """TBA"""
 _TWEET_TOPIC_DESCRIPTION = """
@@ -151,6 +151,25 @@ _TWEET_SENTIMENT_CITATION = """\
     abstract = "This paper describes the fifth year of the Sentiment Analysis in Twitter task. SemEval-2017 Task 4 continues with a rerun of the subtasks of SemEval-2016 Task 4, which include identifying the overall sentiment of the tweet, sentiment towards a topic with classification on a two-point and on a five-point ordinal scale, and quantification of the distribution of sentiment towards a topic across a number of tweets: again on a two-point and on a five-point ordinal scale. Compared to 2016, we made two changes: (i) we introduced a new language, Arabic, for all subtasks, and (ii) we made available information from the profiles of the Twitter users who posted the target tweets. The task continues to be very popular, with a total of 48 teams participating this year.",
 }
 """
+_TWEET_EMOTION_DESCRIPTION = """TBA"""
+_TWEET_EMOTION_CITATION = """\
+@inproceedings{mohammad-etal-2018-semeval,
+    title = "{S}em{E}val-2018 Task 1: Affect in Tweets",
+    author = "Mohammad, Saif  and
+      Bravo-Marquez, Felipe  and
+      Salameh, Mohammad  and
+      Kiritchenko, Svetlana",
+    booktitle = "Proceedings of the 12th International Workshop on Semantic Evaluation",
+    month = jun,
+    year = "2018",
+    address = "New Orleans, Louisiana",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/S18-1001",
+    doi = "10.18653/v1/S18-1001",
+    pages = "1--17",
+    abstract = "We present the SemEval-2018 Task 1: Affect in Tweets, which includes an array of subtasks on inferring the affectual state of a person from their tweet. For each task, we created labeled data from English, Arabic, and Spanish tweets. The individual tasks are: 1. emotion intensity regression, 2. emotion intensity ordinal classification, 3. valence (sentiment) regression, 4. valence ordinal classification, and 5. emotion classification. Seventy-five teams (about 200 team members) participated in the shared task. We summarize the methods, resources, and tools used by the participating teams, with a focus on the techniques and resources that are particularly useful. We also analyze systems for consistent bias towards a particular race or gender. The data is made freely available to further improve our understanding of how people convey emotions through language.",
+}
+"""
 
 
 class SuperTweetEvalConfig(datasets.BuilderConfig):
@@ -170,7 +189,8 @@ class SuperTweetEvalConfig(datasets.BuilderConfig):
             'False' or 'True'.
           **kwargs: keyword arguments forwarded to super.
         """
-        super(SuperTweetEvalConfig, self).__init__(version=datasets.Version(_VERSION), **kwargs)
+        super(SuperTweetEvalConfig, self).__init__(
+            version=datasets.Version(_VERSION), **kwargs)
         self.features = features
         self.label_classes = label_classes
         self.data_url = data_url
@@ -243,7 +263,8 @@ class SuperTweetEval(datasets.GeneratorBasedBuilder):
             name="tweet_nerd",
             description=_TWEET_NERD_DESCRIPTION,
             citation=_TWEET_NERD_CITATION,
-            features=['gold_label_binary', 'target', 'context', 'definition', 'text_start', 'text_end'],
+            features=['gold_label_binary', 'target', 'context',
+                      'definition', 'text_start', 'text_end'],
             data_url="https://huggingface.co/datasets/cardiffnlp/super_tweet_eval/resolve/main/data/tweet_nerd",
         ),
         SuperTweetEvalConfig(
@@ -259,24 +280,35 @@ class SuperTweetEval(datasets.GeneratorBasedBuilder):
             citation=_TWEET_SENTIMENT_CITATION,
             features=['gold_label', 'target', 'text'],
             data_url="https://huggingface.co/datasets/cardiffnlp/super_tweet_eval/resolve/main/data/tweet_sentiment",
+        ),
+        SuperTweetEvalConfig(
+            name="tweet_emotion",
+            description=_TWEET_EMOTION_DESCRIPTION,
+            citation=_TWEET_EMOTION_CITATION,
+            features=["text", "gold_label_list"],
+            data_url="https://huggingface.co/datasets/cardiffnlp/super_tweet_eval/resolve/main/data/tweet_emotion",
         )
     ]
 
     def _info(self):
-        features = {feature: datasets.Value("string") for feature in self.config.features}
+        features = {feature: datasets.Value(
+            "string") for feature in self.config.features}
         if self.config.name == "tweet_topic":
             names = [
                 "arts_&_culture", "business_&_entrepreneurs", "celebrity_&_pop_culture", "diaries_&_daily_life",
                 "family", "fashion_&_style", "film_tv_&_video", "fitness_&_health", "food_&_dining", "gaming",
                 "learning_&_educational", "music", "news_&_social_concern", "other_hobbies", "relationships",
                 "science_&_technology", "sports", "travel_&_adventure", "youth_&_student_life"]
-            features["gold_label_list"] = datasets.Sequence(datasets.features.ClassLabel(names=names))
+            features["gold_label_list"] = datasets.Sequence(
+                datasets.features.ClassLabel(names=names))
         if self.config.name == "tweet_ner7":
             names = [
                 'B-corporation', 'B-creative_work', 'B-event', 'B-group', 'B-location', 'B-person', 'B-product',
                 'I-corporation', 'I-creative_work', 'I-event', 'I-group', 'I-location', 'I-person', 'I-product', 'O']
-            features["gold_label_sequence"] = datasets.Sequence(datasets.features.ClassLabel(names=names))
-            features["text_tokenized"] = datasets.Sequence(datasets.Value("string"))
+            features["gold_label_sequence"] = datasets.Sequence(
+                datasets.features.ClassLabel(names=names))
+            features["text_tokenized"] = datasets.Sequence(
+                datasets.Value("string"))
         if self.config.name in ["tweet_intimacy", "tweet_similarity"]:
             features["gold_score"] = datasets.Value("float32")
         if self.config.name == "tempo_wic":
@@ -285,15 +317,18 @@ class SuperTweetEval(datasets.GeneratorBasedBuilder):
             features["text_start_2"] = datasets.Value("int32")
             features["text_end_1"] = datasets.Value("int32")
             features["text_end_2"] = datasets.Value("int32")
-            features["text_1_tokenized"] = datasets.Sequence(datasets.Value("string"))
-            features["text_2_tokenized"] = datasets.Sequence(datasets.Value("string"))
+            features["text_1_tokenized"] = datasets.Sequence(
+                datasets.Value("string"))
+            features["text_2_tokenized"] = datasets.Sequence(
+                datasets.Value("string"))
             features['date_1'] = datasets.Value("string")
             features['date_2'] = datasets.Value("string")
         if self.config.name == "tweet_hate":
             label_classes = [
-                    'hate_gender', 'hate_race', 'hate_sexuality', 'hate_religion','hate_origin', 'hate_disability',
-                    'hate_age', 'not_hate']
-            features['gold_label'] = datasets.features.ClassLabel(names=label_classes)
+                'hate_gender', 'hate_race', 'hate_sexuality', 'hate_religion', 'hate_origin', 'hate_disability',
+                'hate_age', 'not_hate']
+            features['gold_label'] = datasets.features.ClassLabel(
+                names=label_classes)
             features["text"] = datasets.Value("string")
         if self.config.name == "tweet_nerd":
             features['target'] = datasets.Value("string")
@@ -309,13 +344,17 @@ class SuperTweetEval(datasets.GeneratorBasedBuilder):
             with open(dl_manager.download(url_map)) as f:
                 label_classes = f.readlines()
             label_classes = [x.strip('\n') for x in label_classes]
-            features['gold_label'] = datasets.features.ClassLabel(names=label_classes)
+            features['gold_label'] = datasets.features.ClassLabel(
+                names=label_classes)
             features["text"] = datasets.Value("string")
         if self.config.name == "tweet_sentiment":
-            label_classes = ["strongly negative", "negative", "negative or neutral", "positive", "strongly positive"]
-            features['gold_label'] = datasets.features.ClassLabel(names=label_classes)
+            label_classes = ["strongly negative", "negative",
+                             "negative or neutral", "positive", "strongly positive"]
             features["text"] = datasets.Value("string")
-            features["target"] = datasets.Value("string")
+            names = ['anger', 'anticipation', 'disgust', 'fear', 'joy',
+                      'love', 'optimism', 'pessimism', 'sadness', 'surprise', 'trust']
+            features["gold_label_list"] = datasets.Sequence(
+                datasets.features.ClassLabel(names=names))
 
         return datasets.DatasetInfo(
             description=_SUPER_TWEETEVAL_DESCRIPTION + "\n" + self.config.description,
@@ -324,8 +363,9 @@ class SuperTweetEval(datasets.GeneratorBasedBuilder):
         )
 
     def _split_generators(self, dl_manager):
-        splits = ['train', 'test', 'validation']        
-        downloaded_file = dl_manager.download_and_extract({s: f"{self.config.data_url}/{s}.jsonl" for s in splits})
+        splits = ['train', 'test', 'validation']
+        downloaded_file = dl_manager.download_and_extract(
+            {s: f"{self.config.data_url}/{s}.jsonl" for s in splits})
         return [datasets.SplitGenerator(name=s, gen_kwargs={"filepath": downloaded_file[s]}) for s in splits]
 
     def _generate_examples(self, filepath):