对话系统Deep Learning lab-4 Dialogue Bot for goal-oriented task 面向目标任务的对话机器人 -详细

原文链接： http://docs.deeppavlov.ai/en/master/features/skills/go_bot.html#intro

Docs 文件 » Dialogue Bot for goal-oriented task
翻译自 Dialogue Bot for goal-oriented task

我在pycharm中尝试运行代码，运行结果为对每句输入的回复。

from deeppavlov import build_model, configs

bot1 = build_model(configs.go_bot.gobot_dstc2, download=True)

print(bot1(['hi, i want restaurant in the cheap pricerange']))
print(bot1(['bye']))

Introduction

The dialogue bot is based on 一篇论文 which introduces Hybrid Code Networks (HCNs) that combine an RNN with domain-specific knowledge and system action templates.

该对话机器人基于Jason d. Williams，Kavosh Asadi，Geoffrey Zweig 提出的混合代码网络(Hybrid Code Networks，HCNs)架构，混合代码网络: 实用且高效的端到端对话控制，带有监督和强化学习-2017 这篇论文，引入了混合编码网络(hcn) ，该混合编码网络将 RNN 与特定领域的知识和系统动作模板相结合。

虚线对应不需要的(可选的)模块，黑色方块对应训练模型，梯形是依赖于数据集的模块，必须由软件开发人员提供。
下面是一个与受过训练的对话机器人进行交互的简单示例:
（这些是框架中举的例子）

>> Hello, welcome to the Cambridge restaurant system. You can ask for restaurants by area, price range or food type. How may I help you?
x::hi, i want cheap food
2018-05-16 12:28:48.459 INFO in 'deeppavlov.models.go_bot.bot'['bot'] at line 346: Made api_call with {'pricerange': 'cheap'}, got 22 results.
>> The lucky star serves chinese food in the cheap price range.
x::is there anything with french food?
2018-05-16 12:29:02.795 INFO in 'deeppavlov.models.go_bot.bot'['bot'] at line 346: Made api_call with {'pricerange': 'cheap', 'food': 'french'}, got 0 results.
>> Sorry there is no french restaurant in the cheap price range.
x::ok, maybe there is something in expensive pricerange?
2018-05-16 12:29:23.337 INFO in 'deeppavlov.models.go_bot.bot'['bot'] at line 346: Made api_call with {'pricerange': 'expensive', 'food': 'french'}, got 2 results.
>> Cote serves french food in the expensive price range.
x::in what part of town is it?
2018-05-16 12:29:31.193 INFO in 'deeppavlov.models.go_bot.bot'['bot'] at line 346: Made api_call with {'pricerange': 'expensive', 'food': 'french'}, got 2 results.
>> Restaurant two two is a nice place in the north of town serving tasty french food.
x::ok, give me their phone number
>> The phone number of restaurant two two is 01223 351880.
x::address and post code too
>> Restaurant two two is on 22 chesterton road chesterton, c.b 4, 3 a.x.
x::ok, thank you!
>> Restaurant two two serves french food in the expensive price range.
x::and if i wanted an italian restuarant?
2018-05-16 12:30:15.98 INFO in 'deeppavlov.models.go_bot.bot'['bot'] at line 346: Made api_call with {'pricerange': 'expensive', 'food': 'italian'}, got 4 results.
>> Frankie and bennys serves italian food in the expensive price range.
x::more cheaply priced?
2018-05-16 12:30:36.560 INFO in 'deeppavlov.models.go_bot.bot'['bot'] at line 346: Made api_call with {'pricerange': 'cheap', 'food': 'italian'}, got 5 results.
>> Da vinci pizzeria serves italian food in the cheap price range.
x::ok, give me their phone number and address, please
>> The phone number of da vinci pizzeria is 01223 351707 and it is on 20 milton road chesterton.
x::thanks
>> Da vinci pizzeria serves italian food in the cheap price range.
x::bye
>> You are welcome!

下面是我的实际运行

Usages 用法

Requirements 要求

Configs 配置

A minimal model without slot_filler, intent_classifier and embedder is configured in configs/go_bot/gobot_dstc2_minimal.json.
没有槽填充、意图分类器和嵌入器的最小模型是在 configs / go_bot / gobot_dstc2 minimal.json中配置的。

除了chainer，用于训练管道的配置应该有三个额外的元素: dataset_reader数据集读取器、dataset_iterator 数据集迭代器和train训练器:

{
  "dataset_reader": {
    "class_name": ...,
    ...
  },
  "dataset_iterator": {
    "class_name": ...,
    ...
  },
  "chainer": {
    ...
  },
  "train": {
    ...
  }
}

如下图分析：

Usage example 使用示例

下面是我的使用样例 ，调整为当前的预训练模型gobot_dstc2.json(我已经复制到了当前文件夹下)

这个是通过python交互的形式，(,download=True)然后系统会自行下载需要的配置文件。

或者配置文件已经下载好之后，可以直接读取CONFIG_PATH

首先寻找cheap +south+chinese food ，并获取其phone number+ address + post code

然后寻找French food ，系统自动认为在原来的条件上，结合 cheap + french + north/south/east/west ，都got 0 results。

然后我意识到可能是因为价格的限制，于是调整为 expensive+french + centre/north ，都有相应的results。

Config parameters 配置参数

Datasets 数据集

DSTC2

Your data 你的数据

Dialogs

Templates 模板

Database（optional）数据库（可选）

Comparison 对比

对我们的机器人模型进行不同的修改，并与现有基准进行比较，结果如下

There were a few modifications to the original dataset. It differs from ours in train/valid/test split and intent/action labeling.

对原始数据集做了一些修改。 在训练 / 有效 / 测试分裂 和 意图 / 行动 标签上和我们的有所不同。