NLU训练完，意图识别不对

[HCIS2020]

Rasa NLU version (e.g. 0.7.3): '0.10.5'

Used backend / pipeline (mitie, spacy_sklearn, ...): { "name": "rasa_nlu_test", "pipeline": ["nlp_mitie", "tokenizer_jieba", "ner_mitie", "ner_synonyms", "intent_entity_featurizer_regex", "intent_featurizer_mitie", "intent_classifier_sklearn"], "language": "zh", "mitie_file": "../data/total_word_feature_extractor_zh.dat", "path" : "../models", "data" : "../data/nlu.md" }

Operating system (windows, osx, ...): WINDOWS 10 x64 + Python 3.6 Issue:

Building prefix dict from the default dictionary ... Loading model from cache C:\Users\Benny\AppData\Local\Temp\jieba.cache Loading model cost 1.053 seconds. Prefix dict has been built succesfully. Fitting 2 folds for each of 6 candidates, totalling 12 fits C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) C:\Users\Benny\AppData\Local\Programs\Python\Python36\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples. 'precision', 'predicted', average, warn_for) [Parallel(n_jobs=1)]: Done 12 out of 12 | elapsed: 0.0s finished

Content of configuration file (if used & relevant):

intent:greet

• hey
• hello
• hi
• 在吗？
• 早上好
• 晚上好

intent:goodbye

• bye
• 再见
• 一天好心情
• 再会
• 拜拜

intent:mood_affirm

• yes
• 是的
• 当然
• 听起来不错
• 正确

intent:mood_deny

• no
• 不好
• 我不觉得
• 不喜欢那样
• 没办法

intent:mood_great

• 棒极了
• 我感觉很好
• 非常好
• 太好了

intent:mood_unhappy

• 很糟糕
• 很伤心
• 心情不好
• 我很失望

---

意图识别情况:

interpreter = Interpreter.load(model_dir, config) intent_entities = interpreter.parse('没办法')

Building prefix dict from the default dictionary ... Loading model from cache C:\Users\Benny\AppData\Local\Temp\jieba.cache Loading model cost 0.873 seconds. Prefix dict has been built succesfully. {'intent': {'name': 'mood_affirm', 'confidence': 0.22176080230671813}, 'entities': [], 'intent_ranking': [{'name': 'mood_affirm', 'confidence': 0.22176080230671813}, {'name': 'greet', 'confidence': 0.21021923238681889}, {'name': 'goodbye', 'confidence': 0.1875789668700675}, {'name': 'mood_unhappy', 'confidence': 0.13687120608411668}, {'name': 'mood_deny', 'confidence': 0.13181582565934799}, {'name': 'mood_great', 'confidence': 0.11175396669293076}], 'text': '没办法'}

应该识别出的意图为：mood_deny

但是识别出来的为： mood_affirm

问题可能出在哪里？

[crownpku]

每种类别的训练数据太少了。

[HCIS2020]

谢谢，确实是训练数据太少。每个类别增加到12个样本，就正常了。

还有关于Storied.md的设计，是否有可视化的对话流程工具来生成。 使用RASA官网的mood的例子，storied.md比较简单，很容易就进入Action_listen的状态，会话就无法进行下去了。

[floating-cloud]

@HCIS2020 请问怎么把configuration file （.md）文件生成demo-rasa_zh.json的呢？谢谢~~

[HCIS2020]

@floating-cloud \rasa_nlu\utils\md_to_json.py 这个工具是来进行MD和JSON转换的， 实际调用是在Agent的Training_data里面，

[floating-cloud]

@HCIS2020 太感谢了~~