实体识别和意图识别之后怎么生成回答？

[HwarLee]

实体识别和意图识别之后，我们能得到问题的意图和实体信息（如下例子），然后怎么生成回答呢？有没有开发的工具子类的？ { "entities": [ { "extractor": "ner_mitie", "confidence": null, "end": 2, "value": "明天", "entity": "weatherDate", "start": 0 } ], "intent": { "confidence": 0.99102248555852168, "name": "getWeather" }, "text": "明天会下雨吗？", "intent_ranking": [ { "confidence": 0.99102248555852168, "name": "getWeather" }, { "confidence": 0.0048164909262445208, "name": "baike" }, { "confidence": 0.0024581628351175943, "name": "assistant" }, { "confidence": 0.0017028606801161609, "name": "playMusic" } ] }

[floating-cloud]

构建自己的知识库, 或者简单点调用网上已有的问答api

[feichaohao]

rasa core 最近我在看用这个做回答