用于中文的Rasa NLU，从RasaHQ/rasa_nlu分支而来。

请参考Rasa NLU官方文档获取最新指南

中文博客

你应该拥有的文件：

data/total_word_feature_extractor_zh.dat

由MITIE wordrep工具从中文语料库训练而成（训练需要2-3天）

如需训练，请构建MITIE Wordrep Tool。注意，中文语料在输入工具进行训练前应先进行分词。最适合用户案例的领域相关语料效果最佳。

从中文维基百科和百度百科训练的模型可以从中文博客下载。

data/examples/rasa/demo-rasa_zh.json

应尽可能添加更多示例。

使用方法：

克隆此项目，并运行

python setup.py install

修改配置。

目前对于中文，我们有两种管道：

使用MITIE+Jieba (sample_configs/config_jieba_mitie.yml):

language: "zh"

pipeline:
- name: "nlp_mitie"
  model: "data/total_word_feature_extractor_zh.dat"
- name: "tokenizer_jieba"
- name: "ner_mitie"
- name: "ner_synonyms"
- name: "intent_entity_featurizer_regex"
- name: "intent_classifier_mitie"

推荐：使用MITIE+Jieba+sklearn (sample_configs/config_jieba_mitie_sklearn.yml):

language: "zh"

pipeline:
- name: "nlp_mitie"
  model: "data/total_word_feature_extractor_zh.dat"
- name: "tokenizer_jieba"
- name: "ner_mitie"
- name: "ner_synonyms"
- name: "intent_entity_featurizer_regex"
- name: "intent_featurizer_mitie"
- name: "intent_classifier_sklearn"

（可选）使用结巴用户自定义词典或切换结巴默认词典：

你可以将文件路径或目录路径作为"user_dicts"的值。(sample_configs/config_jieba_mitie_sklearn_plus_dict_path.yml)

language: "zh"

pipeline:
- name: "nlp_mitie"
  model: "data/total_word_feature_extractor_zh.dat"
- name: "tokenizer_jieba"
  default_dict: "./default_dict.big"
  user_dicts: "./jieba_userdict"
#  user_dicts: "./jieba_userdict/jieba_userdict.txt"
- name: "ner_mitie"
- name: "ner_synonyms"
- name: "intent_entity_featurizer_regex"
- name: "intent_featurizer_mitie"
- name: "intent_classifier_sklearn"

通过运行以下命令训练模型：

如果你在配置文件中指定了项目名称，这将把你的模型保存在/models/your_project_name。

否则，你的模型将被保存在/models/default

python -m rasa_nlu.train -c sample_configs/config_jieba_mitie_sklearn.yml --data data/examples/rasa/demo-rasa_zh.json --path models

运行rasa_nlu服务器：

python -m rasa_nlu.server -c sample_configs/config_jieba_mitie_sklearn.yml --path models

打开一个新终端，现在你可以从服务器获取结果，例如：

$ curl -XPOST localhost:5000/parse -d '{"q":"我发烧了该吃什么药？", "project": "rasa_nlu_test", "model": "model_20170921-170911"}' | python -mjson.tool
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100   652    0   552  100   100    157     28  0:00:03  0:00:03 --:--:--   157
{
    "entities": [
        {
            "end": 3,
            "entity": "disease",
            "extractor": "ner_mitie",
            "start": 1,
            "value": "发烧"
        }
    ],
    "intent": {
        "confidence": 0.5397186422631861,
        "name": "medical"
    },
    "intent_ranking": [
        {
            "confidence": 0.5397186422631861,
            "name": "medical"
        },
        {
            "confidence": 0.16206323981749196,
            "name": "restaurant_search"
        },
        {
            "confidence": 0.1212448457737397,
            "name": "affirm"
        },
        {
            "confidence": 0.10333600028547868,
            "name": "goodbye"
        },
        {
            "confidence": 0.07363727186010374,
            "name": "greet"
        }
    ],
    "text": "我发烧了该吃什么药？"
}