File size: 5,060 Bytes
3f3ed66
f6e0c90
3f3ed66
f8f8214
 
 
3f3ed66
 
 
25a9e24
3f3ed66
25a9e24
 
3f3ed66
1941b34
 
3f3ed66
 
 
 
 
 
25a9e24
 
3f3ed66
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ba3d44
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a720574
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
---
language: en
widget:
- text: Covid cases are increasing fast!
datasets:
- tweet_eval
---


# Twitter-roBERTa-base for Sentiment Analysis - UPDATED (2022)

This is a RoBERTa-base model trained on ~124M tweets from January 2018 to December 2021, and finetuned for sentiment analysis with the TweetEval benchmark. 
The original Twitter-based RoBERTa model can be found [here](https://huggingface.co./cardiffnlp/twitter-roberta-base-2021-124m) and the original reference paper is [TweetEval](https://github.com/cardiffnlp/tweeteval). This model is suitable for English. 

- Reference Paper: [TimeLMs paper](https://arxiv.org/abs/2202.03829). 
- Git Repo: [TimeLMs official repository](https://github.com/cardiffnlp/timelms).

<b>Labels</b>: 
0 -> Negative;
1 -> Neutral;
2 -> Positive

This sentiment analysis model has been integrated into [TweetNLP](https://github.com/cardiffnlp/tweetnlp). You can access the demo [here](https://tweetnlp.org).

## Example Pipeline
```python
from transformers import pipeline
sentiment_task = pipeline("sentiment-analysis", model=model_path, tokenizer=model_path)
sentiment_task("Covid cases are increasing fast!")
```
```
[{'label': 'Negative', 'score': 0.7236}]
```

## Full classification example

```python
from transformers import AutoModelForSequenceClassification
from transformers import TFAutoModelForSequenceClassification
from transformers import AutoTokenizer, AutoConfig
import numpy as np
from scipy.special import softmax
# Preprocess text (username and link placeholders)
def preprocess(text):
    new_text = []
    for t in text.split(" "):
        t = '@user' if t.startswith('@') and len(t) > 1 else t
        t = 'http' if t.startswith('http') else t
        new_text.append(t)
    return " ".join(new_text)
MODEL = f"cardiffnlp/twitter-roberta-base-sentiment-latest"
tokenizer = AutoTokenizer.from_pretrained(MODEL)
config = AutoConfig.from_pretrained(MODEL)
# PT
model = AutoModelForSequenceClassification.from_pretrained(MODEL)
#model.save_pretrained(MODEL)
text = "Covid cases are increasing fast!"
text = preprocess(text)
encoded_input = tokenizer(text, return_tensors='pt')
output = model(**encoded_input)
scores = output[0][0].detach().numpy()
scores = softmax(scores)
# # TF
# model = TFAutoModelForSequenceClassification.from_pretrained(MODEL)
# model.save_pretrained(MODEL)
# text = "Covid cases are increasing fast!"
# encoded_input = tokenizer(text, return_tensors='tf')
# output = model(encoded_input)
# scores = output[0][0].numpy()
# scores = softmax(scores)
# Print labels and scores
ranking = np.argsort(scores)
ranking = ranking[::-1]
for i in range(scores.shape[0]):
    l = config.id2label[ranking[i]]
    s = scores[ranking[i]]
    print(f"{i+1}) {l} {np.round(float(s), 4)}")
```

Output: 

```
1) Negative 0.7236
2) Neutral 0.2287
3) Positive 0.0477
```


### References 
```
@inproceedings{camacho-collados-etal-2022-tweetnlp,
    title = "{T}weet{NLP}: Cutting-Edge Natural Language Processing for Social Media",
    author = "Camacho-collados, Jose  and
      Rezaee, Kiamehr  and
      Riahi, Talayeh  and
      Ushio, Asahi  and
      Loureiro, Daniel  and
      Antypas, Dimosthenis  and
      Boisson, Joanne  and
      Espinosa Anke, Luis  and
      Liu, Fangyu  and
      Mart{\'\i}nez C{\'a}mara, Eugenio" and others,
    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, UAE",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.emnlp-demos.5",
    pages = "38--49"
}

```

```
@inproceedings{loureiro-etal-2022-timelms,
    title = "{T}ime{LM}s: Diachronic Language Models from {T}witter",
    author = "Loureiro, Daniel  and
      Barbieri, Francesco  and
      Neves, Leonardo  and
      Espinosa Anke, Luis  and
      Camacho-collados, Jose",
    booktitle = "Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics: System Demonstrations",
    month = may,
    year = "2022",
    address = "Dublin, Ireland",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.acl-demo.25",
    doi = "10.18653/v1/2022.acl-demo.25",
    pages = "251--260"
}

```


## Bias, Risks, and Limitations

The model was analyzed with [Giskard](github.com/giskard-AI/giskard) and the following limitations were identified:

- Typos and misspellings can have a negative impact on the model performances, potentially affecting non-native speakers or people with limited educational background.
- The model is sensitive to letter case and could classify the same text differently if transformed to lowercase or uppercase. Please make sure that your usage takes this into account.

A full report is available [here](https://htmlpreview.github.io/?https://github.com/Giskard-AI/ml-scan-reports/blob/main/scan_reports/huggingface/cardiffnlp/twitter-roberta-base-sentiment-latest/ea9d76e8-a9f2-429a-b053-ed6da8ef6df/report.html).