Rasa_nlu returns intent as null for training samples using tensorflow classifier(Chinese)

The language of my system is Chinese. Most of the training samples are classified correctly, but a few samples return an intent as null. The configuration is as follows:

language: “zh” pipeline:

  • name: “nlp_mitie” model: “data/total_word_feature_extractor_zh.dat”
  • name: “tokenizer_jieba” dictionary_path: “data/userdict.txt”
  • name: “intent_featurizer_count_vectors”
  • name: “intent_classifier_tensorflow_embedding”
  • name: “ner_mitie”
  • name: “ner_synonyms”

The model failed to classify the sample “在吗” which means “are you there” and “是” which means “yes”. The model “data/total_word_feature_extractor_zh.dat” is not trained based on my corpus and I just copy it from another system, is it the problem?

Hi @zhuzhu, thanks for posting your question on the forum. @Ghostvv from Rasa will try to help you on this one.

could it be that the ones that are classified with intent null never present in your training data? then in this case it is expected behavior, because classifier never saw this word