WO2018000282A1 - 一种聊天对话系统的扩充学习方法及聊天对话系统 - Google Patents

一种聊天对话系统的扩充学习方法及聊天对话系统 Download PDF

Info

Publication number
WO2018000282A1
WO2018000282A1 PCT/CN2016/087774 CN2016087774W WO2018000282A1 WO 2018000282 A1 WO2018000282 A1 WO 2018000282A1 CN 2016087774 W CN2016087774 W CN 2016087774W WO 2018000282 A1 WO2018000282 A1 WO 2018000282A1
Authority
WO
WIPO (PCT)
Prior art keywords
question
answer
user
chat dialogue
learning method
Prior art date
Application number
PCT/CN2016/087774
Other languages
English (en)
French (fr)
Inventor
王昊奋
邱楠
杨新宇
Original Assignee
深圳狗尾草智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳狗尾草智能科技有限公司 filed Critical 深圳狗尾草智能科技有限公司
Priority to PCT/CN2016/087774 priority Critical patent/WO2018000282A1/zh
Priority to CN201680001735.3A priority patent/CN106663128A/zh
Publication of WO2018000282A1 publication Critical patent/WO2018000282A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/332—Query formulation
    • G06F16/3329—Natural language query formulation or dialogue systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Definitions

  • the invention relates to computer technology, in particular to an extended learning method and a chat dialogue system of a chat dialogue system.
  • Human-computer interaction is a technology that has been influenced by researchers and computing devices since the birth of the computer. Its goal is to enable machines to help people meet their mission needs efficiently, comfortably and safely.
  • the automatic chat dialogue system is a human-computer interaction system, which receives the input of the user's natural language form and gives corresponding feedback; however, such a dialogue system often has the problem that the user problem cannot be recognized and the error is recognized.
  • the technical problem to be solved by the present invention is to provide an extended learning method and chat pair of a chat dialogue system that can reduce the problem of user identification or identify errors.
  • the present invention provides an extended learning method for a chat dialogue system, including the steps of:
  • the correct answer to the user's question is found through the correction module, and the user question and the correct answer are added to the question answering system.
  • the user problem is automatically generated by the problem generation module.
  • the existence of the problem generation module will enable user problems to be generated in large quantities, and a large number of user questions are input into the question and answer system, which is beneficial to rapidly expanding the question and answer system.
  • the problem generation module is implemented based on any one of a state machine, a voice model, and an RNN neural network.
  • a state machine e.g., a voice model
  • an RNN neural network e.g., a neural network that uses a neural network to generate a plurality of problem generation modules corresponding to the question answering system and implemented based on different mechanisms to help the question answering system to perform extended learning.
  • the question and answer system is enhanced, and the question answering system is confirmed to enhance the confidence of the output of the question. If the answer is correct, the corresponding confidence level is increased, the resource occupancy of the confidence judgment module can be reduced, the next round of question and answer judgment can be entered as soon as possible, and the learning efficiency of the question and answer system can be improved.
  • the step of inputting the question into the question and answer system, the step of generating an answer according to the question system further includes the steps:
  • the preset condition is set by a rule system.
  • the preset conditions can be set and changed to adapt to different needs of different periods and different regions, and improve the applicability of the question and answer system.
  • the correction module comprises a manual interaction unit or an automatic correction unit.
  • the manual correction method for the user problem judged as the wrong answer, we can use the manual correction method or the automatic correction method. Even the combination of manual correction and automatic correction can be used to better match the user problem.
  • the correct answer is added to the Q&A system.
  • the user problem is input through a human-computer interaction unit.
  • the user problem may be to use the state machine, the voice model and the RNN neural network based problem generation module to perform rapid generation and expansion learning before the formal operation; or after the question and answer system is running, the user actually uses the process to gradually expand the learning expansion. .
  • the present invention also provides a chat dialogue system using the extended learning method according to any one of the present invention, comprising: a question answering system for receiving a user question and giving an answer;
  • Confidence judgment module for making confidence judgments on user questions and answers
  • a correction module for correcting the answers to user questions that are judged to be erroneously answered, and adding user questions and correct answers to the question and answer system.
  • the chat dialogue system further includes a question filtering module, configured to filter the user problem and remove the problem that meets the preset condition.
  • the setting of the problem filtering module can reduce the resource occupation problem of meaningless problems, improve the practicability of the question and answer system and the efficiency of learning expansion.
  • the invention has the beneficial effects that the invention has improved the correction function, so that the question answering system corrects and gives the correct answer when the correct answer to the input user question cannot be returned, so as to answer questions and answers.
  • the system learns the user problem and adds the corresponding correct answer to the question and answer system, expands the question and answer database of the question and answer system, avoids the problem that the user problem still cannot be correctly fed back when the user problem occurs again, and gradually learns and expands, which will gradually improve the question and answer system. , thereby reducing the user's problem, unable to identify or identify the error, the user's intention can not be understood.
  • 1 is an extended learning method of a chat dialogue system
  • Figure 2 is a chat dialogue system.
  • Embodiment 1 is a diagrammatic representation of Embodiment 1:
  • FIG. 1 is an extended learning method of a chat dialogue system.
  • the extended learning method includes the following steps:
  • the invention has the beneficial effects that: the invention increases the correction function, so that the question answering system corrects and gives the correct answer when the input user question cannot feed back the correct answer, so that the question answering system learns the user problem and correspondingly
  • the correct answer is added to the question and answer system, and the question and answer library of the question and answer system is expanded to avoid the problem that the user's problem still cannot be correctly fed back.
  • the gradual learning and expansion will gradually improve the question and answer system, thereby reducing the user's problem or not being recognized. Identify errors and the user's intentions cannot be understood.
  • the user problem is automatically generated by the problem generation module.
  • the existence of the problem generation module will enable user problems to be generated in large quantities, and a large number of user questions are input into the question and answer system. Conducive to rapid expansion of the question and answer system.
  • the problem generation module is implemented based on any one of a state machine, a voice model, and an RNN neural network.
  • a state machine e.g., a voice model
  • an RNN neural network e.g., a neural network that uses neural network to generate a plurality of problem generation modules.
  • the question answering system is enhanced, and the question answering system is confirmed to enhance the confidence level of the question output. If the answer is correct, the corresponding confidence level is increased, the resource occupancy of the confidence judgment module can be reduced, the next round of question and answer judgment can be entered as soon as possible, and the learning efficiency of the question and answer system can be improved.
  • the question is input into the question answering system, and the step of generating an answer according to the question system further includes the steps:
  • System-related keywords, or inconsistencies, or other errors and meaningless issues, can be considered as meeting the pre-set conditions.
  • the preset condition is set by the rule system.
  • the preset conditions can be set and changed to adapt to different needs of different periods and different regions, and improve the applicability of the question and answer system.
  • the correction module comprises a human interaction unit or a networked correction unit.
  • a human interaction unit for users who are judged to be erroneously answering questions, we can use manual correction or network correction. Even manual correction and network correction can be used to better match the user problem. The correct answer is added to the Q&A system.
  • the user problem is input through the human-computer interaction unit.
  • User problem can be Before the formal operation, the problem generation module based on the state machine, the speech model and the RNN neural network is used for rapid generation and expansion learning; and after the question and answer system is running, the user actually uses the process to gradually expand and learn.
  • Embodiment 2 is a diagrammatic representation of Embodiment 1:
  • the chat dialogue system 100 includes: a question answering system 1 for receiving a user question and giving an answer;
  • Confidence judgment module 2 configured to perform confidence judgment on user questions and answers
  • the correcting module 3 is configured to correct the answer to the user question judged as the wrong answer, and add the user question and the correct answer to the question answering system.
  • the chat dialogue system further includes a question filtering module, configured to filter the user problem and remove the problem that meets the preset condition.
  • the setting of the problem filtering module can reduce the resource occupation problem of meaningless problems, improve the practicability of the question and answer system and the efficiency of learning expansion.
  • the invention has the beneficial effects that: the invention has a correction function added, and the correction function is added, so that the question answering system corrects and gives the correct answer when the input user question cannot feed back the correct answer, so that the question answering system can learn the User questions and the corresponding correct answers are added to the question and answer system, and the question and answer library of the question and answer system is expanded to avoid the problem that the user questions cannot be correctly fed back when the problem occurs again. Gradually learning and expanding will gradually improve the question and answer system and thus reduce User issues are not recognized or identified, and the user's intent is not understood.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Databases & Information Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Human Computer Interaction (AREA)
  • Computational Linguistics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Electrically Operated Instructional Devices (AREA)

Abstract

一种聊天对话系统的扩充学习方法及聊天对话系统，该扩充学习方法包括步骤：将用户问题输入问答系统，问答系统生成答案（S1）；对用户问题和答案的关联度进行置信度判断，判断回答是否正确（S2）；若判断问答系统回答错误，则通过纠正模块找到用户问题的正确答案，并将该用户问题和正确答案添加到问答系统中（S3）。通过该方法可以减少用户问题无法识别或识别错误的问题。

Description

一种聊天对话系统的扩充学习方法及聊天对话系统 技术领域
本发明涉及计算机技术，特别是涉及一种聊天对话系统的扩充学习方法及聊天对话系统。
背景技术
人机交互是计算机诞生以来产生的研究人和计算设备之间相互影响的技术.其目标是使机器帮助人高效、舒适、安全地完成任务需求。
而其中，自动聊天对话系统是一种人机交互系统，通过接收用户自然语言形式的输入，给出相应反馈；但是这类对话系统时常出现用户问题无法识别、识别错误的问题。
应该注意，上面对技术背景的介绍只是为了方便对本申请的技术方案进行清楚、完整的说明，并方便本领域技术人员的理解而阐述的。不能仅仅因为这些方案在本申请的背景技术部分进行了阐述而认为上述技术方案为本领域技术人员所公知。
发明内容
有鉴于现有技术的上述缺陷，本发明所要解决的技术问题是提供了一种减少用户问题无法识别或识别错误问题的聊天对话系统的扩充学习方法及聊天对 话系统。
为实现上述目的，本发明提供了一种聊天对话系统的扩充学习方法，包括步骤：
将用户问题输入问答系统，问答系统生成答案；
对用户问题和答案的关联度进行置信度判断，判断回答是否正确；
若判断问答系统回答错误，则通过纠正模块找到用户问题的正确答案，并将该用户问题和正确答案添加到问答系统中。
进一步的，所述用户问题是由问题生成模块自动生成的。问题生成模块的存在，将使得用户问题可以大量的产生，大量用户问题输入到问答系统中，有利于快速的扩充问答系统。
进一步的，所述问题生成模块是基于状态机、语音模型和RNN神经网络中的任一一种实现的。当然，如果条件允许，设置多个对应于问答系统的，且基于不同机制实现的问题生成模块来帮助问答系统进行扩充学习也是可以的。进一步的，若判断回答系统回答正确，则对问答系统进行增强，让问答系统确认，提升该问题输出的置信度。若答案正确，则提高相对应的置信度，可以减少置信度判断模块的资源占用，尽快的进入下一轮的问答判断，提高问答系统的学习效率。
进一步的，所述将问题输入问答系统，根据问题系统生成答案的步骤之前还包括步骤：
对生成的问题进行过滤，去除符合预设条件的问题。对于一些无意义的问题，进行过滤，不仅可以提高问答系统的学习扩充效率，而且在问答系统后期的人机交互时，减少用户进行无意义问答的情况；例如，该生成的用户问题没 有出现问答系统相关的关键词，或者不通顺时，或者是其他错误和无意义问题的情况，可以认定为是符合预设条件的问题。
进一步的，所述预设条件是通过规则系统设置的。预设条件可以进行设置和更改，适应不同时期和不同地区不同的需求，提高该问答系统的可适用度。
进一步的，所述纠正模块包括人工交互单元或自动纠正单元。对于被判断为回答错误的用户问题，我们可以采用人工纠正的方式，也可以采用自动纠正的方式，甚至，可以采用人工纠正和自动纠正结合的方式，以更好的给出匹配于用户问题的正确答案，并添加到问答系统中。
进一步的，所述用户问题是通过人机交互单元输入的。用户问题可以是在正式运行之前使用基于状态机、语音模型和RNN神经网络的问题生成模块进行快速生成和扩充学习；也可以在问答系统运行后，用户实际使用的过程中，逐渐的进行学习扩充。
本发明还提供了一种使用如本发明任一所述的扩充学习方法的聊天对话系统，包括：问答系统，用于接收用户问题并给出答案；
置信度判断模块，用于对用户问题和答案进行置信度判断；
纠正模块，用于对被判断为回答错误的用户问题进行答案纠正，并将用户问题和正确答案添加到问答系统中。
进一步的，所述聊天对话系统还包括问题过滤模块，用于对用户问题进行过滤，去除符合预设条件的问题。问题过滤模块的设置，可以减少无意义问题的资源占用问题，提高问答系统的实用性和学习扩充效率。
本发明的有益效果是：本发明由于增加了纠正功能，使得问答系统在对输入的用户问题无法反馈出正确答案的时候，会纠正并给出正确答案，以便问答 系统学习该用户问题并将对应的正确答案添加到问答系统中，扩充问答系统的问答库，避免该用户问题再次出现时仍然无法正确反馈的问题，逐步的学习和扩充，将使得问答系统逐渐完善，进而减少用户问题无法识别或识别错误，用户的意图无法得到理解的情况。
参照后文的说明和附图，详细公开了本申请的特定实施方式，指明了本申请的原理可以被采用的方式。应该理解，本申请的实施方式在范围上并不因而受到限制。在所附权利要求的精神和条款的范围内，本申请的实施方式包括许多改变、修改和等同。
针对一种实施方式描述和/或示出的特征可以以相同或类似的方式在一个或更多个其它实施方式中使用，与其它实施方式中的特征相组合，或替代其它实施方式中的特征。
应该强调，术语“包括/包含”在本文使用时指特征、整件、步骤或组件的存在，但并不排除一个或更多个其它特征、整件、步骤或组件的存在或附加。
附图说明
所包括的附图用来提供对本申请实施例的进一步的理解，其构成了说明书的一部分，用于例示本申请的实施方式，并与文字描述一起来阐释本申请的原理。显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。在附图中：
图1是一种聊天对话系统的扩充学习方法；
图2是一种聊天对话系统。
具体实施方式
为了使本技术领域的人员更好地理解本申请中的技术方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例，都应当属于本申请保护的范围。
实施例一：
图1是一种聊天对话系统的扩充学习方法，参见图1，该扩充学习方法包括步骤：
S1:将用户问题输入问答系统，问答系统生成答案；
S2:对用户问题和答案的关联度进行置信度判断，判断回答是否正确；
S3:若判断问答系统回答错误，则通过纠正模块找到用户问题的正确答案，并将该用户问题和正确答案添加到问答系统中。
本发明的有益效果是：本发明由于增加了纠正功能，使得问答系统在对输入的用户问题无法反馈出正确答案的时候，会纠正并给出正确答案，以便问答系统学习该用户问题并将对应的正确答案添加到问答系统中，扩充问答系统的问答库，避免该用户问题再次出现时仍然无法正确反馈的问题，逐步的学习和扩充，将使得问答系统逐渐完善，进而减少用户问题无法识别或识别错误，用户的意图无法得到理解的情况。
本实施例优选的，用户问题是由问题生成模块自动生成的。问题生成模块的存在，将使得用户问题可以大量的产生，大量用户问题输入到问答系统中， 有利于快速的扩充问答系统。
本实施例优选的，问题生成模块是基于状态机、语音模型和RNN神经网络中的任一一种实现的。当然，如果条件允许，设置多个对应于问答系统的，且基于不同机制实现的问题生成模块来帮助问答系统进行扩充学习也是可以的。
本实施例优选的，若判断回答系统回答正确，则对问答系统进行增强，让问答系统确认，提升该问题输出的置信度。若答案正确，则提高相对应的置信度，可以减少置信度判断模块的资源占用，尽快的进入下一轮的问答判断，提高问答系统的学习效率。
本实施例优选的，将问题输入问答系统，根据问题系统生成答案的步骤之前还包括步骤：
对生成的问题进行过滤，去除符合预设条件的问题。对于一些无意义的问题，进行过滤，不仅可以提高问答系统的学习扩充效率，而且在问答系统后期的人机交互时，减少用户进行无意义问答的情况；例如，该生成的用户问题没有出现问答系统相关的关键词，或者不通顺时，或者是其他错误和无意义问题的情况，可以认定为是符合预设条件的问题。
本实施例优选的，预设条件是通过规则系统设置的。预设条件可以进行设置和更改，适应不同时期和不同地区不同的需求，提高该问答系统的可适用度。
本实施例优选的，纠正模块包括人工交互单元或联网纠正单元。对于被判断为回答错误的用户问题，我们可以采用人工纠正的方式，也可以采用联网纠正的方式，甚至，可以采用人工纠正和联网纠正结合的方式，以更好的给出匹配于用户问题的正确答案，并添加到问答系统中。
本实施例优选的，用户问题是通过人机交互单元输入的。用户问题可以是 在正式运行之前使用基于状态机、语音模型和RNN神经网络的问题生成模块进行快速生成和扩充学习；也可以在问答系统运行后，用户实际使用的过程中，逐渐的进行学习扩充。
实施例二：
图2是本发明一种使用如本发明任一所述的扩充学习方法的聊天对话系统，该聊天对话系统100包括：问答系统1，用于接收用户问题并给出答案；
置信度判断模块2，用于对用户问题和答案进行置信度判断；
纠正模块3，用于对被判断为回答错误的用户问题进行答案纠正，并将用户问题和正确答案添加到问答系统中。
本实施例优选的，聊天对话系统还包括问题过滤模块，用于对用户问题进行过滤，去除符合预设条件的问题。问题过滤模块的设置，可以减少无意义问题的资源占用问题，提高问答系统的实用性和学习扩充效率。
本发明的有益效果是：本发明由于设置了纠正模块，增加了纠正功能，使得问答系统在对输入的用户问题无法反馈出正确答案的时候，会纠正并给出正确答案，以便问答系统学习该用户问题并将对应的正确答案添加到问答系统中，扩充问答系统的问答库，避免该用户问题再次出现时仍然无法正确反馈的问题，逐步的学习和扩充，将使得问答系统逐渐完善，进而减少用户问题无法识别或识别错误，用户的意图无法得到理解的情况。
以上详细描述了本发明的较佳具体实施例。应当理解，本领域的普通技术人员无需创造性劳动就可以根据本发明的构思作出诸多修改和变化。因此，凡本技术领域中技术人员依本发明的构思在现有技术的基础上通过逻辑分析、推 理或者有限的实验可以得到的技术方案，皆应在由权利要求书所确定的保护范围内。

Claims (10)

  1. 一种聊天对话系统的扩充学习方法，其特征是：包括步骤：
    将用户问题输入问答系统，问答系统生成答案；
    对用户问题和答案的关联度进行置信度判断，判断回答是否正确；
    若判断问答系统回答错误，则通过纠正模块找到用户问题的正确答案，并将该用户问题和正确答案添加到问答系统中。
  2. 如权利要求1所述的聊天对话系统的扩充学习方法，其特征是：所述用户问题是由问题生成模块自动生成的。
  3. 如权利要求3所述的聊天对话系统的扩充学习方法，其特征是：所述问题生成模块是基于状态机、语音模型和RNN神经网络中的任一一种实现的。
  4. 如权利要求1所述的聊天对话系统的扩充学习方法，其特征是：若判断回答系统回答正确，则对问答系统进行增强，让问答系统确认，提升该问题输出的置信度。
  5. 如权利要求1所述的聊天对话系统的扩充学习方法，其特征是：所述将问题输入问答系统，根据问题系统生成答案的步骤之前还包括步骤：
    对生成的问题进行过滤，去除符合预设条件的问题。
  6. 如权利要求5所述的聊天对话系统的扩充学习方法，其特征是：所述预设条件是通过规则系统设置的。
  7. 如权利要求1所述的聊天对话系统的扩充学习方法，其特征是：所述纠正模块包括人工交互单元或自动纠正单元。
  8. 如权利要求1所述的聊天对话系统的扩充学习方法，其特征是：所述用 户问题是通过人机交互单元输入的。
  9. 一种使用如权利要求1-8任一所述的扩充学习方法的聊天对话系统，其特征是：包括：问答系统，用于接收用户问题并给出答案；
    置信度判断模块，用于对用户问题和答案进行置信度判断；
    纠正模块，用于对被判断为回答错误的用户问题进行答案纠正，并将用户问题和正确答案添加到问答系统中。
  10. 如权利要求8所述的聊天对话系统，其特征是：所述聊天对话系统还包括问题过滤模块，用于对用户问题进行过滤，去除符合预设条件的问题。
PCT/CN2016/087774 2016-06-29 2016-06-29 一种聊天对话系统的扩充学习方法及聊天对话系统 WO2018000282A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2016/087774 WO2018000282A1 (zh) 2016-06-29 2016-06-29 一种聊天对话系统的扩充学习方法及聊天对话系统
CN201680001735.3A CN106663128A (zh) 2016-06-29 2016-06-29 一种聊天对话系统的扩充学习方法及聊天对话系统

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2016/087774 WO2018000282A1 (zh) 2016-06-29 2016-06-29 一种聊天对话系统的扩充学习方法及聊天对话系统

Publications (1)

Publication Number Publication Date
WO2018000282A1 true WO2018000282A1 (zh) 2018-01-04

Family

ID=58838104

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/087774 WO2018000282A1 (zh) 2016-06-29 2016-06-29 一种聊天对话系统的扩充学习方法及聊天对话系统

Country Status (2)

Country Link
CN (1) CN106663128A (zh)
WO (1) WO2018000282A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109271495A (zh) * 2018-08-14 2019-01-25 阿里巴巴集团控股有限公司 问答识别效果检测方法、装置、设备及可读存储介质
CN109741824A (zh) * 2018-12-21 2019-05-10 质直(上海)教育科技有限公司 一种基于机器学习的医疗问诊方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107463301A (zh) * 2017-06-28 2017-12-12 北京百度网讯科技有限公司 基于人工智能的对话系统构建方法、装置、设备及计算机可读存储介质
CN107562863A (zh) * 2017-08-30 2018-01-09 深圳狗尾草智能科技有限公司 聊天机器人回复自动生成方法及系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101257512A (zh) * 2008-02-02 2008-09-03 黄伟才 用于问答系统的问答匹配方法及问答方法和系统
CN102903008A (zh) * 2011-07-29 2013-01-30 国际商业机器公司 用于计算机问答的方法及系统
CN103279528A (zh) * 2013-05-31 2013-09-04 俞志晨 一种基于人机结合的问答系统及方法
CN104050256A (zh) * 2014-06-13 2014-09-17 西安蒜泥电子科技有限责任公司 基于主动学习的问答方法及采用该方法的问答系统
CN105378731A (zh) * 2013-07-16 2016-03-02 国际商业机器公司 从被回答问题关联语料库/语料值

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104714954A (zh) * 2013-12-13 2015-06-17 中国电信股份有限公司 基于上下文理解的信息搜索方法和系统
CN103744843B (zh) * 2013-12-25 2017-01-04 北京百度网讯科技有限公司 一种在线语音翻译方法及装置
US10180988B2 (en) * 2014-12-02 2019-01-15 International Business Machines Corporation Persona-based conversation
CN105591882B (zh) * 2015-12-10 2018-03-06 北京中科汇联科技股份有限公司 一种智能机器人与人混合客服的方法及系统
CN105677783A (zh) * 2015-12-31 2016-06-15 上海智臻智能网络科技股份有限公司 智能问答系统的信息处理方法及装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101257512A (zh) * 2008-02-02 2008-09-03 黄伟才 用于问答系统的问答匹配方法及问答方法和系统
CN102903008A (zh) * 2011-07-29 2013-01-30 国际商业机器公司 用于计算机问答的方法及系统
CN103279528A (zh) * 2013-05-31 2013-09-04 俞志晨 一种基于人机结合的问答系统及方法
CN105378731A (zh) * 2013-07-16 2016-03-02 国际商业机器公司 从被回答问题关联语料库/语料值
CN104050256A (zh) * 2014-06-13 2014-09-17 西安蒜泥电子科技有限责任公司 基于主动学习的问答方法及采用该方法的问答系统

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109271495A (zh) * 2018-08-14 2019-01-25 阿里巴巴集团控股有限公司 问答识别效果检测方法、装置、设备及可读存储介质
CN109271495B (zh) * 2018-08-14 2023-02-17 创新先进技术有限公司 问答识别效果检测方法、装置、设备及可读存储介质
CN109741824A (zh) * 2018-12-21 2019-05-10 质直(上海)教育科技有限公司 一种基于机器学习的医疗问诊方法
CN109741824B (zh) * 2018-12-21 2023-08-04 质直(上海)教育科技有限公司 一种基于机器学习的医疗问诊方法

Also Published As

Publication number Publication date
CN106663128A (zh) 2017-05-10

Similar Documents

Publication Publication Date Title
WO2018000282A1 (zh) 一种聊天对话系统的扩充学习方法及聊天对话系统
US10755046B1 (en) Applied artificial intelligence technology for conversational inferencing
CN105931638B (zh) 面向智能机器人的对话系统数据处理方法及装置
US8818926B2 (en) Method for personalizing chat bots
WO2022121257A1 (zh) 模型训练方法、语音识别方法、装置、设备及存储介质
JP2019528512A (ja) 人工知能に基づくヒューマンマシンインタラクション方法及び装置
CN110175229B (zh) 一种基于自然语言进行在线培训的方法和系统
CN106528530A (zh) 一种确定句子类型的方法及装置
WO2020113918A1 (zh) 基于语义解析的语句合理性判断方法、装置、计算机设备
KR101575276B1 (ko) 가상 상담 시스템
CN105955963A (zh) 一种机器人问答交互开放平台及交互方法
CN114897163A (zh) 预训练模型数据处理方法、电子设备及计算机存储介质
CN116561538A (zh) 问答评分方法、问答评分装置、电子设备及存储介质
CN114168619A (zh) 语言转换模型的训练方法及装置
CN114564595A (zh) 知识图谱更新方法、装置及电子设备
CN106708950B (zh) 用于智能机器人自学习系统的数据处理方法及装置
CN106776557B (zh) 情感机器人的情感状态记忆识别方法及装置
Dethlefs et al. Optimising incremental generation for spoken dialogue systems: Reducing the need for fillers
CN116402057A (zh) 一种基于自然语言的人机交互系统
CN115114404A (zh) 智能客服的问答方法、装置、电子设备及计算机存储介质
US8051026B2 (en) Rules collector system and method with user interaction
CN108648749B (zh) 基于声控系统与vr的医疗语音识别构建方法及系统
CN106886430A (zh) 机器人应用程序快速启动的方法及装置
Conway Induction in Aristotle and St. Thomas
CN111797244B (zh) 基于知识图谱和对话机器人的智能情境教学方法及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16906683

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16906683

Country of ref document: EP

Kind code of ref document: A1