基于Rasa框架搭建中文机器人对话系统
Rasa是一个能用于构建机器人对话系统的框架，基于Rasa框架搭建机器人对话系统，可以使用于工业各类语音智能服务场景，如：远程医疗问诊、智能客户服务、保险产品销售、金融催收服务、手机智能助手等领域。支持基于规则、填槽和机器学习来构建对话系统。主要模块包括：NLU（意图识别和实体提取）和Core（基于模型及规则进行回复）。提供了搭建对话系统的脚手架。
常用命令有：
rasa init 创建项目
rasa train 训练NLU和Core模型
rasa run actions 启动动作服务（主要是自己写的动作，系统默认的动作，不需要单独启动）
rasa shell 启动对话机器人 （会自动启动默认动作服务）
涉及配置文件有：
config.yml 即对NLU和Core模型的配置
# Configuration for Rasa NLU.
# https://rasa.com/docs/rasa/nlu/components/
language: zh
pipeline:
# # No configuration for the NLU pipeline was provided. The following default pipeline was used to train your model.
# # If you'd like to customize it, uncomment and adjust the pipeline.
# # See https://rasa.com/docs/rasa/tuning-your-model for more information.
# - name: WhitespaceTokenizer
- name: JiebaTokenizer #支持中文
- name: RegexFeaturizer
- name: LexicalSyntacticFeaturizer
- name: CountVectorsFeaturizer
- name: CountVectorsFeaturizer
analyzer: char_wb
min_ngram: 1
max_ngram: 4
- name: DIETClassifier
epochs: 100
constrain_similarities: true
- name: EntitySynonymMapper
- name: ResponseSelector
epochs: 100
constrain_similarities: true
- name: FallbackClassifier
threshold: 0.3
ambiguity_threshold: 0.1
# Configuration for Rasa Core.
# https://rasa.com/docs/rasa/core/policies/
policies:
# # No configuration for policies was provided. The following default policies were used to train your model.
# # If you'd like to customize them, uncomment and adjust the policies.
# # See https://rasa.com/docs/rasa/policies for more information.
- name: MemoizationPolicy
- name: TEDPolicy
max_history: 5
epochs: 100
constrain_similarities: true
- name: RulePolicy
domain.yml 领域配置
version: '2.0'
config:
store_entities_as_slots: true
session_config:
session_expiration_time: 60
carry_over_slots_to_new_session: true
intents:
- greet:
use_entities: true
- deny:
use_entities: true
- request_names:
use_entities: true
- goodbye:
use_entities: true
- affirm:
use_entities: true
- mood_great:
use_entities: true
- mood_unhappy:
use_entities: true
- bot_challenge:
use_entities: true
entities: []
slots:
first_name:
type: rasa.shared.core.slots.TextSlot
initial_value: null
auto_fill: true
influence_conversation: true
last_name:
type: rasa.shared.core.slots.TextSlot
initial_value: null
auto_fill: true
influence_conversation: true
name_spelled_correctly:
type: rasa.shared.core.slots.BooleanSlot
initial_value: null
auto_fill: true
influence_conversation: true
requested_slot:
type: rasa.shared.core.slots.UnfeaturizedSlot
initial_value: null
auto_fill: true
influence_conversation: false
responses:
utter_greet:
- text: 嗨，你好吗?
utter_cheer_up:
- image: https://i.imgur.com/nGF1K8f.jpg
text: '这是某些使你嗨起来的东西:'
utter_did_that_help:
- text: 这对你有帮助吗?
utter_happy:
- text: 好极了，继续!
utter_goodbye:
- text: 再见
utter_iamabot:
- text: 我是个机器人，由Rasa框架提供支持.
utter_ask_first_name:
- text: 你贵姓?
utter_ask_last_name:
- text: 你的名字?
utter_ask_name_spelled_correctly:
- buttons:
- payload: /affirm
title: 是
- payload: /deny
title: 否
text: 姓 {first_name} 拼写对了吗?
utter_submit:
- text: 好的，谢谢！
utter_slots_values:
- text: 我记住你了， {first_name} {last_name}!
actions:
- utter_greet
- utter_slots_values
- utter_submit
- validate_name_form
forms:
name_form:
first_name:
- type: from_text
last_name:
- type: from_text
e2e_actions: []credentials.yml 证书配置，用于调用语音通道的接口
# This file contains the credentials for the voice & chat platforms
# which your bot is using.
# https://rasa.com/docs/rasa/messaging-and-voice-channels
rest:
# # you don't need to provide anything here - this channel doesn't
# # require any credentials
#facebook:
# verify: ""
# secret: ""
# page-access-token: ""
#slack:
# slack_token: ""
# slack_channel: ""
# slack_signing_secret: ""
#socketio:
# user_message_evt:
# bot_message_evt:
# session_persistence:
#mattermost:
# url: "https:///api/v4"
# token: ""
# webhook_url: ""
# This entry is needed if you are using Rasa X. The entry represents credentials
# for the Rasa X "channel", i.e. Talk to your bot and Share with guest testers.
rasa:
url: "http://localhost:5002/api"
endpoints.yml 端点配置，如:机器人要使用的模型、动作、存储服务等。
# This file contains the different endpoints your bot can use.
# Server where the models are pulled from.
# https://rasa.com/docs/rasa/model-storage#fetching-models-from-a-server
#models:
# url: http://my-server.com/models/default_core@latest
# wait_time_between_pulls: 10 # [optional](default: 100)
# Server which runs your custom actions.
# https://rasa.com/docs/rasa/custom-actions
#action_endpoint:
# url: "http://localhost:5055/webhook"
action_endpoint:
url: "http://localhost:5055/webhook"
# Tracker store which is used to store the conversations.
# By default the conversations are stored in memory.
# https://rasa.com/docs/rasa/tracker-stores
#tracker_store:
# type: redis
# url:
# port:
# db:
# password:
# use_ssl:
#tracker_store:
# type: mongod
# url:
# db:
# username:
# password:
tracker_store:
type: SQL
dialect: sqlite
db: trackers.db
# Event broker which all conversation events should be streamed to.
# https://rasa.com/docs/rasa/event-brokers
#event_broker:
# url: localhost
# username: username
# password: password
# queue: queue
event_broker:
type: SQL
dialect: sqlite
db: events.db
names.txt 本案例使用的姓名列表。
孙悟空
猪八戒
唐三藏
沙悟净
诸葛青云数据，主要包括：
nlu.yml 用于训练nlu模型的训练数据
version: "2.0"
nlu:
- intent: greet
examples: |
- 你好
- 上午好
- 中午好
- 嗨
- intent: goodbye
examples: |
- 再见
- 回头见
- 晚安
- intent: affirm
examples: |
- 是的
- 有
- 当然
- 听上去不错
- intent: deny
examples: |
- 不
- 不要
- 没有
- 没有
- 我不喜欢
- intent: mood_great
examples: |
- 太好了
- 感觉不错
- 非常好
- intent: mood_unhappy
examples: |
- 不开心
- 感到沮丧
- 不高兴
- 不太好
- intent: bot_challenge
examples: |
- 你是个机器人吗?
- 你是人类吗?
- 我是在和机器人讲话吗?
- 我是在和人类讲话吗?
- intent: request_names
examples: |
- 我想告诉你姓名
- 你知道我的姓名吗？
- lookup: names
examples: |
- 猪八戒
- 孙悟空
- 沙悟净
- 唐三藏
rules.yml 规则，根据用户意图，进行具体的动作（包括查询、填槽、回复等）
version: "2.0"
rules:
- rule: Say goodbye anytime the user says goodbye
steps:
- intent: goodbye
- action: utter_goodbye
- rule: Say 'I am a bot' anytime the user challenges
steps:
- intent: bot_challenge
- action: utter_iamabot
- rule: Activate form
steps:
- intent: request_names
- action: name_form
- active_loop: name_form
- rule: Submit form
condition:
- active_loop: name_form
steps:
- action: name_form
- active_loop: null
- slot_was_set:
- requested_slot: null
- action: utter_submit
- action: utter_slots_valuesstories.yml 故事情节，描述对话的流程
version: "2.0"
stories:
- story: happy path
steps:
- intent: greet
- action: utter_greet
- intent: mood_great
- action: utter_happy
- story: sad path 1
steps:
- intent: greet
- action: utter_greet
- intent: mood_unhappy
- action: utter_cheer_up
- action: utter_did_that_help
- intent: affirm
- action: utter_happy
- story: sad path 2
steps:
- intent: greet
- action: utter_greet
- intent: mood_unhappy
- action: utter_cheer_up
- action: utter_did_that_help
- intent: deny
- action: utter_goodbye
- story: interactive_story_1
steps:
- intent: greet
- action: utter_greet
- intent: request_names
- action: name_form
- active_loop: name_form
- slot_was_set:
- requested_slot: first_name
- slot_was_set:
- name_spelled_correctly: None
- slot_was_set:
- first_name: None
- slot_was_set:
- requested_slot: last_name
- slot_was_set:
- name_spelled_correctly: None
- slot_was_set:
- last_name: None
- slot_was_set:
- requested_slot: null
- active_loop: null
- action: utter_submit
- action: utter_slots_values下面举例创建mybot的过程：
1、使用rasa init 创建mybot项目，目录结果如下：
涉及的配置文件在上文已经逐一列出，下面看看自定义的action是如何写的，主要完成了输入验证的过程，默认的action是以utter_打头，系统默认支持。
actions.py
import yaml
import pathlib
from typing import Text, List, Any, Dict, Optional
from rasa_sdk import Tracker, FormValidationAction
from rasa_sdk.executor import CollectingDispatcher
from rasa_sdk.types import DomainDict
names = pathlib.Path("names.txt").read_text().split("\n")
class ValidateNameForm(FormValidationAction):
def name(self) -> Text:
return "validate_name_form"
async def required_slots(
self,
slots_mapped_in_domain: List[Text],
dispatcher: "CollectingDispatcher",
tracker: "Tracker",
domain: "DomainDict",
) -> Optional[List[Text]]:
first_name = tracker.slots.get("first_name")
if first_name is not None:
if first_name not in names:
return ["name_spelled_correctly"] + slots_mapped_in_domain
return slots_mapped_in_domain
async def extract_name_spelled_correctly(
self, dispatcher: CollectingDispatcher, tracker: Tracker, domain: Dict
) -> Dict[Text, Any]:
intent = tracker.get_intent_of_latest_message()
return {"name_spelled_correctly": intent == "affirm"}
def validate_name_spelled_correctly(
self,
slot_value: Any,
dispatcher: CollectingDispatcher,
tracker: Tracker,
domain: DomainDict,
) -> Dict[Text, Any]:
"""Validate `first_name` value."""
if tracker.get_slot("name_spelled_correctly"):
return {"first_name": tracker.get_slot("first_name"), "name_spelled_correctly": True}
return {"first_name": None, "name_spelled_correctly": None}
def validate_first_name(
self,
slot_value: Any,
dispatcher: CollectingDispatcher,
tracker: Tracker,
domain: DomainDict,
) -> Dict[Text, Any]:
"""Validate `first_name` value."""
# If the name is super short, it might be wrong.
print(f"姓 = {slot_value} 长度 = {len(slot_value)}")
if len(slot_value) <=1:
dispatcher.utter_message(text=f"姓太短了，你确定拼写对了？")
return {"first_name": None}
else:
return {"first_name": slot_value}
def validate_last_name(
self,
slot_value: Any,
dispatcher: CollectingDispatcher,
tracker: Tracker,
domain: DomainDict,
) -> Dict[Text, Any]:
"""Validate `last_name` value."""
# If the name is super short, it might be wrong.
print(f"名字 = {slot_value} 长度 = {len(slot_value)}")
if len(slot_value) <= 1:
dispatcher.utter_message(text=f"名字太短了，你确定拼写对了？")
return {"last_name": None}
else:
return {"last_name": slot_value}2、启动自动定义的actions服务：
3、训练NLU和Core模型：
4、启动机器人进行对话
分期业务效果：
实际尝试：可以自动完成正常的分期业务流程，中间可以穿插其它的对话内容，如：分期时间长、手续费太贵、担心被套路等。