Datasets:

Modalities:
Text
Formats:
json
ArXiv:
Libraries:
Datasets
pandas
File size: 2,794 Bytes
1b1262c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
826e083
e07d280
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
> 上述数据集为ABSA(Aspect-Based Sentiment Analysis)领域数据集,基本形式为从句子中抽取:方面术语、方面类别(术语类别)、术语在上下文中情感极性以及针对该术语的观点词,不同数据集抽取不同的信息,这点在jsonl文件的“instruction”键中有分别提到,在此我将其改造为了生成任务,需要模型按照一定格式生成抽取结果。

#### 以acos数据集中抽取的jsonl文件一条数据举例:

```
{
    "task_type": "generation",
    "dataset": "acos", 
    "input": ["the computer has difficulty switching between tablet and computer ."], 
    "output": "[['computer', 'laptop usability', 'negative', 'difficulty']]",
    "situation": "none", 
    "label": "", 
    "extra": "", 
    "instruction": "    
        Task: Extracting aspect terms and their corresponding aspect categories, sentiment polarities, and opinion words. 
        Input: A sentence
        Output: A list of 4-tuples, where each tuple contains the extracted aspect term, its aspect category, sentiment polarity, and opinion words (if any). Supplement: \"Null\" means that there is no occurrence in the sentence.
        Example:  
            Sentence: \"Also it's not a true SSD drive in there but eMMC, which makes a difference.\"  
            Output: [['SSD drive', 'hard_disc operation_performance', 'negative', 'NULL']]' 
    "
}
```
> 此处未设置label和extra,在instruction中以如上所示的字符串模板,并给出一个例子进行one-shot,ABSA领域数据集(absa-quad,acos,arts,aste-data-v2,mams,semeval-2014,semeval-2015,semeval-2016,towe)每个数据集对应instruction模板相同,内容有细微不同,且部分数据集存在同一数据集不同数据instruction内容不同的情况。


#### 原始数据集
- 数据[链接](https://github.com/IsakZhang/ABSA-QUAD)
- Paper: [Aspect-Category-Opinion-Sentiment Quadruple Extraction with Implicit Aspects and Opinions](https://aclanthology.org/2021.acl-long.29.pdf)
- 说明:原始数据集由Restaurant-ACOS和Laptop-ACOS两个文件夹的数据组成,本次改造我将两个数据集的数据合并并区分为train、validation与test


#### 当前SOTA
*数据来自[论文](https://arxiv.org/abs/2305.09193)*

- 评价指标:F1 score
- SOTA模型:E2H-large (Restaurant-ACOS上F1 Score:**63.50** , Laptop-ACOS上F1 Score:**44.51**)
- Paper:[Easy-to-Hard Learning for Information Extraction](https://arxiv.org/pdf/2305.09193.pdf)
- 说明:该论文来自[Google Scholar](https://scholar.google.com/scholar?as_ylo=2023&hl=zh-CN&as_sdt=2005&sciodt=0,5&cites=5295149944344120368&scipsc=)检索到的引用ACOS原论文的论文之一,我比较了2023年的一些论文工作后筛选了一个最优指标以及模型。