对话系统-“任务型”多轮对话（二）：对话状态追踪（DST）【基于规则；基于模型】【输入：当前意图和槽值对+历史槽值对；输出：State（槽值对集合）或State Vector 】【为DP做数据准备】

快来打我* 2023-09-30 08:08 29阅读 0赞

在任务型的对话系统中，对话状态跟踪（DST）的目标是从对话历史中监控对话的状态。

DST的输入：Intent+Slot+History；输出：State或State Vector

DST中的State用一组Slot-Value键值对表示；

![在这里插入图片描述][watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16]

![在这里插入图片描述][watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16 1]

![在这里插入图片描述][watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16 2]

## 一、基于规则的DST ##

## 二、基于模型的DST ##

At a high level, given a dialog context and a candidate slot- value pair, our model outputs a score indicating the relevance of the candidate.

In other words, the approach is similar to a sentence pair classiﬁcation task.

*  The ﬁrst input corresponds to the dialog context, and it consists of the system utterance from the previous turn and the user utterance from the cur- rent turn. The two utterances are separated by a \[SEP\] token.
 *  The second input is the candidate slot-value pair. We simply represent the candidate pair as a sequence of tokens (words or pieces of words).

At each turn, the proposed BERT-based model is used to estimate the probability score of every candidate slot-value pair.After that, only pairs with predicted probability equal to at least 0.5 are chosen as the ﬁnal prediction for the turn.  
![在这里插入图片描述][watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16 3]  
![在这里插入图片描述][watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16 4]

[watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16]: https://img-blog.csdnimg.cn/6a386d6362bf4381b823c0b7f76a3f10.png?x-oss-process=image/watermark,type_d3F5LXplbmhlaQ,shadow_50,text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO,size_20,color_FFFFFF,t_70,g_se,x_16
[watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16 1]: https://img-blog.csdnimg.cn/34a470d97b924a07ada961cc52d5decc.png?x-oss-process=image/watermark,type_d3F5LXplbmhlaQ,shadow_50,text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO,size_20,color_FFFFFF,t_70,g_se,x_16
[watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16 2]: https://img-blog.csdnimg.cn/3f8d0642bee642b1ad593f3166b40202.png?x-oss-process=image/watermark,type_d3F5LXplbmhlaQ,shadow_50,text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO,size_20,color_FFFFFF,t_70,g_se,x_16
[watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16 3]: https://img-blog.csdnimg.cn/d12c41faaa3e4342a40ea95318ebb005.png?x-oss-process=image/watermark,type_d3F5LXplbmhlaQ,shadow_50,text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO,size_20,color_FFFFFF,t_70,g_se,x_16
[watermark_type_d3F5LXplbmhlaQ_shadow_50_text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO_size_20_color_FFFFFF_t_70_g_se_x_16 4]: https://img-blog.csdnimg.cn/a23b1e085d544b2188a82d685e6e2145.png?x-oss-process=image/watermark,type_d3F5LXplbmhlaQ,shadow_50,text_Q1NETiBA5b-N6ICF44Gu5Lmx5aSq6YOO,size_20,color_FFFFFF,t_70,g_se,x_16