路易斯模式不能识别所有实体话语 [英] luis patterns does not identify all entity utterances

问题描述

我们有一个使用C#SDK V3设计的机器人. 该功能使用户输入后，查询将传递到qnamaker并将答案提供给用户.如果在qnamaker中找不到答案，则搜索为 定向到共享点.

We have a bot designed in C# SDK V3. The functionality is such that upon user input, the query is passed to qnamaker and the answers are given to the user. If there is no answer found in the qnamaker, the search is directed to sharepoint.

有一个新的要求，其中要求机器人在请求任何搜索关键字时，必须向用户提供一个已处理的文件-为此目的实现了一个网络api.在尝试将此Web api集成到bot中时，问题就开始了.

There was a new requirement wherein upon asking for any search keyword, the bot has to give the user a processed file - a web api was implemented for this purpose. The issue started while trying to integrate this web api in the bot.

添加了一个新的意图来调用此Web api.但是，添加此意图会干扰现有功能.因此，我决定继续使用模式-添加简单实体并拥有一个模式.

A new intent was added to invoke this web api. However, adding this intent has disturbed the existing functionality. Hence, I have decided to proceed with the patterns- add simple entity and have a patterns.

我不确定添加模式是否是此问题的正确解决方案.我现在面临的问题是，我在添加所有实体话语之后在qnamaker中进行测试时，即使经过多次训练，基于模式的输入也无法显示某些话语的实体或模式预测.

I am not sure if adding patterns is the correct solution for this issue. The problem I face now is when I test within the qnamaker after adding all the entity utterances, the pattern based input does not show the entity or pattern prediction for some of the utterances even after training it several times.

推荐答案

通过遵循Zeryth关于训练模型并再次发布Luis模型的建议，我能够解决此问题.

I was able to resolve the issue by following Zeryth's suggestion of training the model and publishing the Luis model again.

这篇关于路易斯模式不能识别所有实体话语的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！