对话系统 Dialogue system
在逻辑要求都特别复杂的情况下，我们将任务的逻辑进行拆解。
In the case where the logical requirements are particularly complex, we will break down the logic of the task.
1．将人类在输入框输入的语言，转化成结构化的信息。比如，转化成json的格式化，或者是转化成某个表达式。
1. Convert the language input by humans into input boxes into structured information. For example, it can be converted into a format of json, or converted into an expression.
2．然后，我们拿这个表达式，去查我们的数据库。查出数据库再扔给大模型
2. Then, we take this expression and look up our database. Find out the database and throw it to the big model
3．我们告诉大模型，我们查出来的数据是这个，让他推荐套餐。然后，大模型根据你的查询结果，再去做自然语言的封装返回给用户。
3. We tell the big model that the data we found out is this, and ask him to recommend the package. Then, based on your query results, the large model will wrap the natural language and return it to the user.
这样得到的好处就是，这样的结果会比较准确。因为大模型它这样就只做了一个动作，就是理解用户的需求。相当于我们只是用大模型去解决部分问题。
The advantage of this is that the results will be more accurate. Because of the large model, it only does one action, which is to understand the needs of users. It's like we're just using a big model to solve part of the problem.
In the case where the logical requirements are particularly complex, we will break down the logic of the task.
1．将人类在输入框输入的语言，转化成结构化的信息。比如，转化成json的格式化，或者是转化成某个表达式。
1. Convert the language input by humans into input boxes into structured information. For example, it can be converted into a format of json, or converted into an expression.
2．然后，我们拿这个表达式，去查我们的数据库。查出数据库再扔给大模型
2. Then, we take this expression and look up our database. Find out the database and throw it to the big model
3．我们告诉大模型，我们查出来的数据是这个，让他推荐套餐。然后，大模型根据你的查询结果，再去做自然语言的封装返回给用户。
3. We tell the big model that the data we found out is this, and ask him to recommend the package. Then, based on your query results, the large model will wrap the natural language and return it to the user.
这样得到的好处就是，这样的结果会比较准确。因为大模型它这样就只做了一个动作，就是理解用户的需求。相当于我们只是用大模型去解决部分问题。
The advantage of this is that the results will be more accurate. Because of the large model, it only does one action, which is to understand the needs of users. It's like we're just using a big model to solve part of the problem.