使用PYABSA批量情感分析自定义文件时遇RuntimeError求助

我按照指定教程完成情感推理，单句子列表的推理代码可正常运行，但修改批量推理代码使用自定义文件test.dat.apc（共2行，每行含2个句子）时，触发以下错误：

修改后的代码 #

# inference_sets = ABSADatasetList.Phone # original code
inference_sets = 'test.dat.apc' # this is my own file that I want to infer sentiment for each sentence
results = sent_classifier.batch_infer(target_file=inference_sets,
                     print_result=True,
                     save_result=True,
                     ignore_error=False,
                     )

报错信息 #

RuntimeError                              Traceback (most recent call last)
Input In [56], in <cell line: 2>()
      1 test = 'test.dat.apc'
----> 2 results = sent_classifier.batch_infer(target_file=test,
      3                      print_result=False,
      4                      save_result=True,
      5                      ignore_error=False,
      6                      )

File ~\Anaconda3\envs\spacy\lib\site-packages\pyabsa-1.16.15-py3.9.egg\pyabsa\core\apc\prediction\sentiment_classifier.py:197, in SentimentClassifier.batch_infer(self, target_file, print_result, save_result, ignore_error, clear_input_samples)
    193     self.clear_input_samples()
    195 save_path = os.path.join(os.getcwd(), 'apc_inference.result.json')
--> 197 target_file = detect_infer_dataset(target_file, task='apc')
    198 if not target_file:
    199     raise FileNotFoundError('Can not find inference datasets!')

File ~\Anaconda3\envs\spacy\lib\site-packages\pyabsa-1.16.15-py3.9.egg\pyabsa\functional\dataset\dataset_manager.py:302, in detect_infer_dataset(dataset_path, task)
    300     if os.path.isdir(dataset_path.dataset_name):
    301         print('No inference set found from: {}, unrecognized files: {}'.format(dataset_path, ', '.join(os.listdir(dataset_path.dataset_name))))
--> 302     raise RuntimeError(
    303         'Fail to locate dataset: {}. If you are using your own dataset, you may need rename your dataset according to {}'.format(
    304             dataset_path,
    305             'https://github.com/yangheng95/ABSADatasets#important-rename-your-dataset-filename-before-use-it-in-pyabsa')
    306     )
    307 if len(dataset_path) > 1:
    308     print(colored('Please DO NOT mix datasets with different sentiment labels for training & inference !', 'yellow'))

RuntimeError: Fail to locate dataset: ['test.dat.apc']. If you are using your own dataset, you may need rename your dataset according to https://github.com/yangheng95/ABSADatasets#important-rename-your-dataset-filename-before-use-it-in-pyabsa

问题原因与解决办法 #

1. 文件名不符合PyABSA识别规则 #

PyABSA的detect_infer_dataset函数会严格校验数据集文件名，test.dat.apc不符合它的命名规范，导致无法识别为有效的APC任务数据集。

• 解决：将文件名改为test.apc，去掉多余的.dat后缀，让系统能识别这是测试集文件。

2. 文件内容格式不符合APC任务要求 #

APC（基于方面的情感分类）要求每行输入为「句子 + 目标方面词」的组合，用[SEP]分隔，示例格式：

The camera quality is excellent [SEP] camera quality
This phone gets hot quickly [SEP] heat issue

你当前文件每行包含2个句子，完全不符合输入格式，即使文件名正确，后续也会报错。

• 解决：调整文件内容，每行对应一个待推理的「句子-方面」对，用[SEP]分隔开。

3. 确保文件路径正确 #

如果文件不在当前工作目录，需要传入完整的绝对路径，比如r"C:\your_path\test.apc"，避免系统找不到文件。

内容的提问来源于stack exchange，提问作者Nemo