CN115248851A

Movatterモバイル変換

Info

Publication number: CN115248851A
Application number: CN202210490232.4A
Authority: CN
Inventors: 杨学峰
Original assignee: Shanghai Wenzheng Management Consulting Co ltd
Current assignee: Shanghai Wenzheng Management Consulting Co ltd
Priority date: 2022-05-07
Filing date: 2022-05-07
Publication date: 2022-10-28

Abstract

Translated fromChinese

本发明涉及智能对话技术领域，具体公开了一种对话任务描述方法及系统，所述方法包括基于预设的基本话题类型拆分对话任务，得到话题树；接收用户输入的含有对话内容的对话请求，基于所述话题树执行对话过程；本发明提出话题树的设计方法：主题话题用于描述对话任务，包含若干个子话题；通过嵌套主题话题，可以形成话题树。通过主题话题拆分对话任务的粒度，提升话题复用率和对话配置的灵活性。本发明提出话题链接的设计方法：在配置话题时，采用话题引用的形式复用已定义话题，减轻重复配置工作量，减少配置文件冗余。基于该方法，可以把对话技能库和对话任务这两种配置工作独立开来，有利于对话技能沉淀，同时减轻对话任务配置难度。

The invention relates to the technical field of intelligent dialogue, and specifically discloses a dialogue task description method and system. The method includes splitting a dialogue task based on a preset basic topic type to obtain a topic tree; receiving a dialogue request containing dialogue content input by a user , the dialogue process is performed based on the topic tree; the present invention proposes a design method for the topic tree: topic topics are used to describe dialogue tasks and include several subtopics; topic trees can be formed by nesting topic topics. The granularity of dialogue tasks is split by topic topic, which improves the topic reuse rate and the flexibility of dialogue configuration. The invention proposes a design method for topic links: when configuring topics, the defined topics are reused in the form of topic references, so as to reduce the workload of repeated configuration and reduce the redundancy of configuration files. Based on this method, the two configuration tasks of the dialogue skill base and the dialogue task can be separated, which is conducive to the precipitation of dialogue skills, and at the same time reduces the difficulty of configuring dialogue tasks.

Description

Translated fromChinese

一种对话任务描述方法及系统A dialog task description method and system

技术领域technical field

本发明涉及智能对话技术领域，具体是一种对话任务描述方法及系统。The invention relates to the technical field of intelligent dialog, in particular to a dialog task description method and system.

背景技术Background technique

对话管理(Dialog Manager)是聊天机器人系统的重要功能模块，根据用户输入及上下文采取适当的策略输出响应。对话管理方法可以分为基于规则和基于统计两大类。其中，基于规则的对话管理方法又可以分为基于有限状态机、基于框架和基于目标。在工程实现上，现有的对话管理方法的任务配置比较冗长，任务复用粒度较粗，对复杂的对话流程缺乏灵活性。Dialog Manager is an important functional module of the chat robot system, which adopts appropriate strategies to output responses according to user input and context. Dialogue management methods can be divided into two categories: rule-based and statistics-based. Among them, the rule-based dialogue management methods can be divided into finite state machine-based, frame-based and goal-based. In terms of engineering implementation, the task configuration of the existing dialog management method is relatively lengthy, the granularity of task reuse is relatively coarse, and it lacks flexibility for complex dialog processes.

发明内容SUMMARY OF THE INVENTION

本发明的目的在于提供一种对话任务描述方法及系统，以解决上述背景技术中提出的问题。The purpose of the present invention is to provide a dialog task description method and system to solve the problems raised in the background art above.

为实现上述目的，本发明提供如下技术方案：To achieve the above object, the present invention provides the following technical solutions:

一种对话任务描述方法，所述方法包括：A dialog task description method, the method comprising:

基于预设的基本话题类型拆分对话任务，得到话题树；Split dialogue tasks based on preset basic topic types to obtain topic trees;

接收用户输入的含有对话内容的对话请求，基于所述话题树执行对话过程；Receive a dialogue request containing dialogue content input by the user, and execute a dialogue process based on the topic tree;

其中，所述基本话题类型包括：主题话题、表述话题、请求话题、执行话题和答复话题；Wherein, the basic topic types include: theme topic, expression topic, request topic, execution topic and reply topic;

所述主题话题用于描述一个对话任务；嵌套包含子话题；The theme topic is used to describe a dialogue task; nesting includes subtopics;

所述表述话题用于描述用户的陈述；无子话题；The statement topic is used to describe the user's statement; no subtopics;

所述请求话题用于描述对用户的询问；无子话题；包括若干槽位；The request topic is used to describe the query to the user; there is no subtopic; it includes several slots;

所述执行话题用于描述任务执行；无子话题；The execution topic is used to describe task execution; no subtopics;

所述答复话题用于描述对话任务对用户的响应；无子话题。The reply topic is used to describe the response of the dialog task to the user; there are no subtopics.

作为本发明进一步的方案：所述接收用户输入的含有对话内容的对话请求，基于所述话题树执行对话过程的步骤包括：As a further solution of the present invention: the step of receiving a dialog request input by a user containing dialog content, and executing the dialog process based on the topic tree includes:

接收用户输入的对话请求，基于请求话题请求并接收用户上传的图像信息；Receive the dialogue request input by the user, request and receive the image information uploaded by the user based on the request topic;

基于执行话题对所述图像信息进行处理，执行知识查询，得到反馈内容；Processing the image information based on the execution topic, performing knowledge query, and obtaining feedback content;

基于答复话题将所述反馈内容反馈至用户。The feedback content is fed back to the user based on the reply topic.

作为本发明进一步的方案：所述基于执行话题对所述图像信息进行处理，执行知识查询，得到反馈内容的步骤包括：As a further solution of the present invention: the steps of processing the image information based on the execution topic, performing knowledge query, and obtaining feedback content include:

基于根话题在话题树中查询图像文字提取子话题和知识图谱查询子话题；Query image text extraction subtopics and knowledge graph query subtopics in the topic tree based on the root topic;

基于所述图像文字提取子话题对所述图像信息进行文字识别，得到文本信息；performing text recognition on the image information based on the image text extraction subtopic to obtain text information;

基于所述知识图谱查询子话题对所述文本信息进行知识查询，得到知识查询结果。A knowledge query is performed on the text information based on the knowledge graph query subtopic to obtain a knowledge query result.

作为本发明进一步的方案：所述基于所述图像文字提取子话题对所述图像信息进行文字识别，得到文本信息的步骤包括：As a further solution of the present invention: the step of performing text recognition on the image information based on the image text extraction subtopic to obtain text information includes:

遍历所述图像信息的所有像素点，获取各像素点的色值；Traverse all the pixels of the image information to obtain the color value of each pixel;

将所述色值与预设的色值范围进行比对，根据比对结果得到若干个色值数组；所述色值数组与所述色值范围为映射关系，所述色值数组的值为像素点位置，所述色值数组的下标为像素点个数；Comparing the color value with a preset color value range, and obtaining several color value arrays according to the comparison result; the color value array and the color value range are in a mapping relationship, and the value of the color value array is Pixel position, the subscript of the color value array is the number of pixels;

根据所述若干个色值数组确定图像类型，根据所述图像类型确定区域识别模型，根据所述区域识别模型对所述图像信息进行区域识别，得到待识别文本框；Determine the image type according to the several color value arrays, determine a region recognition model according to the image type, perform region recognition on the image information according to the region recognition model, and obtain a text box to be recognized;

对所述待识别文本框中的内容进行内容识别，得到文本信息。Perform content identification on the content in the text box to be identified to obtain text information.

作为本发明进一步的方案：所述对所述待识别文本框中的内容进行内容识别，得到文本信息的步骤包括：As a further solution of the present invention: the step of performing content identification on the content in the text box to be identified to obtain text information includes:

采用神经网络模型识别所述待识别文本框中的文本信息，所述神经网络模型包括: 卷积层和池化层；Adopting neural network model to identify the text information in the text box to be identified, the neural network model includes: convolutional layer and pooling layer;

其中，所述卷积层包括交替连接的标准卷积核和扩张卷积核，所述扩张卷积核的感受野的宽度相对所述标准卷积核的感受野的宽度大；Wherein, the convolution layer includes alternately connected standard convolution kernels and expanded convolution kernels, and the width of the receptive field of the expanded convolution kernel is larger than the width of the receptive field of the standard convolution kernel;

所述池化层的分块窗口为矩形，且采用标准最大池化和平均池化加权混合池化，池化权重系数根据分块图片的全局最大值和平均值计算确定。The block window of the pooling layer is rectangular, and adopts standard maximum pooling and average pooling weighted mixed pooling, and the pooling weight coefficient is determined according to the global maximum value and average value of the block picture.

作为本发明进一步的方案：所述基于所述知识图谱查询子话题对所述文本信息进行知识查询，得到知识查询结果的步骤包括：As a further solution of the present invention: the step of performing knowledge query on the text information based on the query subtopic of the knowledge map, and obtaining the knowledge query result includes:

对所述文本信息进行词性识别，根据词性识别结果输出与文本信息对应的待检字表；Perform part-of-speech recognition on the text information, and output a list of characters to be checked corresponding to the text information according to the part-of-speech recognition result;

遍历所述待检字表，根据预设的内容转换模型将待检字表转换为参考字表；Traversing the word list to be checked, converting the word list to be checked into a reference word list according to a preset content conversion model;

基于所述参考字表在预设的查询数据库中查询目标内容及目标内容的查询参数；其中，所述查询参数至少包括查询次数；Querying the target content and query parameters of the target content in a preset query database based on the reference list; wherein, the query parameters include at least the number of queries;

根据目标内容及目标内容的查询参数生成知识图谱。Generate a knowledge map according to the target content and the query parameters of the target content.

作为本发明进一步的方案：在配置话题时，采用话题引用的形式复用已定义话题；对话技能库和对话任务两种配置工作独立进行。As a further solution of the present invention: when configuring the topic, the defined topic is reused in the form of topic reference; the two configuration tasks of the dialog skill library and the dialog task are performed independently.

本发明技术方案还提供了一种对话任务描述系统，所述系统包括：The technical solution of the present invention also provides a dialogue task description system, the system comprising:

话题树配置模块，用于基于预设的基本话题类型拆分对话任务，得到话题树；Topic tree configuration module, used to split dialogue tasks based on preset basic topic types to obtain topic trees;

对话执行模块，用于接收用户输入的含有对话内容的对话请求，基于所述话题树执行对话过程；A dialog execution module, configured to receive a dialog request input by a user containing dialog content, and execute a dialog process based on the topic tree;

作为本发明进一步的方案：所述对话执行模块包括：As a further solution of the present invention: the dialogue execution module includes:

图像接收单元，用于接收用户输入的对话请求，基于请求话题请求并接收用户上传的图像信息；An image receiving unit, configured to receive a dialogue request input by a user, request and receive image information uploaded by the user based on the request topic;

内容生成单元，用于基于执行话题对所述图像信息进行处理，执行知识查询，得到反馈内容；A content generation unit, configured to process the image information based on the execution topic, perform knowledge query, and obtain feedback content;

反馈执行单元，用于基于答复话题将所述反馈内容反馈至用户。A feedback executing unit, configured to feed back the feedback content to the user based on the reply topic.

作为本发明进一步的方案：所述内容生成单元包括：As a further solution of the present invention: the content generation unit includes:

话题查询子单元，用于基于根话题在话题树中查询图像文字提取子话题和知识图谱查询子话题；The topic query sub-unit is used to query the image text extraction sub-topic and the knowledge map query sub-topic in the topic tree based on the root topic;

识别子单元，用于基于所述图像文字提取子话题对所述图像信息进行文字识别，得到文本信息；A recognition subunit, configured to perform text recognition on the image information based on the image text extraction subtopic to obtain text information;

图谱查询子单元，用于基于所述知识图谱查询子话题对所述文本信息进行知识查询，得到知识查询结果。The graph query sub-unit is configured to perform knowledge query on the text information based on the knowledge graph query subtopic to obtain a knowledge query result.

与现有技术相比，本发明的有益效果是：本发明采用了话题链接方法能够有效降低对话任务配置的复杂度，提高话题复用度，降低对话任务配置难度，提高对话任务配置灵活性。本发明采用话题树的形式描述对话任务，通过组合复用已有配置，减少对话配置冗余，提升对话配置质量和效率。Compared with the prior art, the beneficial effects of the present invention are: the present invention adopts the topic linking method, which can effectively reduce the complexity of dialog task configuration, improve topic reuse, reduce the difficulty of dialog task configuration, and improve the flexibility of dialog task configuration. The present invention describes the dialog task in the form of a topic tree, reduces the redundancy of the dialog configuration and improves the quality and efficiency of the dialog configuration by combining and reusing existing configurations.

附图说明Description of drawings

为了更清楚地说明本发明实施例中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例。In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings that need to be used in the descriptions of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only of the present invention. some examples.

图1为对话任务描述方法的流程框图。Fig. 1 is a flow chart of the dialog task description method.

图2为对话任务描述方法的第一子流程框图。Fig. 2 is a block diagram of the first sub-flow of the dialog task description method.

图3为对话任务描述方法的第二子流程框图。Fig. 3 is a block diagram of the second sub-flow of the dialog task description method.

图4为对话任务描述系统的组成结构框图。Fig. 4 is a structural block diagram of the dialogue task description system.

图5为对话任务描述系统中对话执行模块的组成结构框图。Fig. 5 is a structural block diagram of the dialog execution module in the dialog task description system.

图6为对话执行模块中内容生成单元的组成结构框图。Fig. 6 is a structural block diagram of the content generation unit in the dialog execution module.

图7为对话任务描述方法的图片问答技能示意图。Figure 7 is a schematic diagram of the picture question answering skill of the dialogue task description method.

图8为对话任务描述方法的图片问答技能简化示意图。Fig. 8 is a simplified schematic diagram of the picture question answering skill of the dialogue task description method.

具体实施方式Detailed ways

为了使本发明所要解决的技术问题、技术方案及有益效果更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

实施例1Example 1

图1为对话任务描述方法的流程框图，本发明实施例中，一种对话任务描述方法，所述方法包括步骤S100至步骤S200：FIG. 1 is a flowchart of a method for describing a dialog task. In an embodiment of the present invention, a method for describing a dialog task includes steps S100 to S200:

步骤S100：基于预设的基本话题类型拆分对话任务，得到话题树；Step S100: splitting the dialogue task based on the preset basic topic type to obtain a topic tree;

步骤S200：接收用户输入的含有对话内容的对话请求，基于所述话题树执行对话过程；Step S200: receiving a dialog request input by a user containing dialog content, and executing a dialog process based on the topic tree;

本发明把对话管理中的对话任务抽象为话题(Topic)，以便统一设计概念。The invention abstracts the dialogue task in the dialogue management into a topic (Topic), so as to unify the design concept.

本发明提出5个基本的话题类型：主题话题(Subject Topic)、表述话题(InformTopic)、请求话题(Request Topic)、执行话题(Execute)和答复话题(Answer)。 SubjectTopic：主题话题，用于描述一个对话任务；可以嵌套包含子话题。The present invention proposes five basic topic types: Subject Topic, Inform Topic, Request Topic, Execute and Answer. SubjectTopic: Subject topic, which is used to describe a dialogue task; subtopics can be nested.

Inform Topic：表述话题，用于描述用户或聊天机器人的陈述；无子话题。Inform Topic: Descriptive topic, used to describe the statement of the user or chatbot; no subtopics.

Request Topic：请求话题，用于描述对用户的询问；无子话题；包括若干槽位。Request Topic: request topic, used to describe the query to the user; no sub-topics; including several slots.

Execute Topic：执行话题，用于描述任务执行；无子话题。Execute Topic: Execution topic, used to describe task execution; no sub-topics.

Answer Topic：答复话题，用于描述对话任务对用户的响应；无子话题。Answer Topic: answer topic, used to describe the response of the dialog task to the user; no subtopics.

本发明提出话题树的设计方法：主题话题用于描述对话任务，包含若干个子话题；通过嵌套主题话题，可以形成话题树。通过主题话题拆分对话任务的粒度，提升话题复用率和对话配置的灵活性。The invention proposes a method for designing a topic tree: a topic topic is used to describe a dialogue task and includes several subtopics; a topic tree can be formed by nesting topic topics. Split the granularity of dialogue tasks through topics and topics, improve the topic reuse rate and the flexibility of dialogue configuration.

在上述内容的基础上，本发明技术方案还提出了话题链接的设计方法：在配置话题时，采用话题引用的形式复用已定义话题，减轻重复配置工作量，减少配置文件冗余。基于该方法，可以把对话技能库和对话任务这两种配置工作独立开来，有利于对话技能沉淀，同时减轻对话任务配置难度。On the basis of the above content, the technical solution of the present invention also proposes a topic link design method: when configuring topics, reuse defined topics in the form of topic references, reduce the workload of repeated configuration, and reduce the redundancy of configuration files. Based on this method, the two configuration tasks of dialogue skill database and dialogue task can be separated, which is conducive to the accumulation of dialogue skills and reduces the difficulty of dialogue task configuration.

在本发明技术方案的一个实例中，用户在使用业务系统的时候遇到了操作问题，把问题截图发送给问答机器人来获得操作帮助，无需文字描述即可快速获得反馈。问答机器人的图片问答技能有效改善了业务系统的用户体验。In an example of the technical solution of the present invention, the user encounters an operation problem when using the business system, and sends a screenshot of the problem to the Q&A robot to obtain operation help, and can quickly obtain feedback without a text description. The picture question answering skills of the question answering robot effectively improve the user experience of the business system.

如图7所示，左侧是相对独立的子技能；便于组合复用；右侧是业务技能，通过自定义子技能和链接已有技能的方式实现业务技能。As shown in Figure 7, the left side is a relatively independent sub-skill; it is easy to combine and reuse; the right side is a business skill, which realizes business skills by customizing sub-skills and linking existing skills.

在图7中，图片问答是主题话题Subject Topic，用于描述一个相对完整的对话任务；该对话任务由2个子主题话题组成：图像文字提取主话题，知识图谱查询主话题。In Figure 7, the picture question and answer is a subject topic, which is used to describe a relatively complete dialogue task; the dialogue task consists of two sub-topics: the main topic of image text extraction, and the main topic of knowledge graph query.

图像文字提取(子主话题)用于描述图像文字提取任务，分别由请求上传图片(请求话题)、执行文字识别(执行话题)和返回识别结果(答复话题)组成。请求上传图片是一个相对通用的话题，通过话题关键词REQUEST_IMAGE进行链接引用，无需重复定义。返回识别结果也是一个相对通用的话题，通过话题关键词SIMPLE_ANSWER进行链接引用，无需重复定义。Image text extraction (sub-main topic) is used to describe the image text extraction task, which consists of requesting to upload pictures (request topic), performing text recognition (execution topic) and returning recognition results (reply topic). Requesting to upload an image is a relatively common topic, and the topic keyword REQUEST_IMAGE is used for link reference without repeated definitions. Returning the recognition result is also a relatively common topic, which can be linked and referenced through the topic keyword SIMPLE_ANSWER without repeated definitions.

知识图谱查询(子主话题)用于描述基于知识图谱的问答任务，分别由接收文本表述(请求话题)、执行知识查询(执行话题)和封装返回答案(答复话题)组成。知识图谱查询是一个相对通用的话题，通过话题关键词SIMPLE_KG_QUERY进行链接引用，无需重复定义。返回相关答案也是一个相对通用的话题，通过话题关键词 SIMPLE_ANSWER进行链接引用，无需重复定义。Knowledge graph query (sub-main topic) is used to describe the question answering task based on knowledge graph, which consists of receiving text representation (request topic), executing knowledge query (execution topic) and encapsulating the returned answer (reply topic). Knowledge graph query is a relatively common topic, which can be linked and referenced through the topic keyword SIMPLE_KG_QUERY without repeated definitions. Returning related answers is also a relatively common topic, which can be linked and referenced through the topic keyword SIMPLE_ANSWER without repeated definitions.

在对话系统运维过程中，把一些复杂的业务技能进一步拆分和提炼，形成若干相对独立的规模较小的子技能，沉淀到基础技能库，便于复用，提高业务技能配置灵活性，降低新技能开发成本。In the process of dialogue system operation and maintenance, some complex business skills are further split and refined to form a number of relatively independent and small-scale sub-skills, which are deposited into the basic skill library to facilitate reuse, improve the flexibility of business skill configuration, and reduce New skills development costs.

在本示例中，图像文字提取话题和知识图谱查询话题也可以作为通用话题，分别定义为SIMPLE_OCR和SIMPLE_QA，通过话题链接方式进行引用，简化图片问答技能配置。In this example, the image text extraction topic and the knowledge graph query topic can also be used as general topics, which are defined as SIMPLE_OCR and SIMPLE_QA respectively, and are referenced through topic links to simplify the configuration of image question-and-answer skills.

图片问答技能配置可以进一步简化为图8所示，在图8中，左侧是相对独立的子技能，便于组合复用；右侧是业务技能，通过链接引用已有的技能，即可实现业务技能，既简化了配置工作，又节省了子技能开发工作。The picture question and answer skill configuration can be further simplified as shown in Figure 8. In Figure 8, the left side is a relatively independent sub-skill, which is convenient for combination and reuse; the right side is a business skill, and the existing skills can be referenced through links to realize business skills, which not only simplifies the configuration work, but also saves the development work of sub-skills.

作为本发明技术方案的一个优先实施例，在配置话题时，采用话题引用的形式复用已定义话题；对话技能库和对话任务两种配置工作独立进行。As a preferred embodiment of the technical solution of the present invention, when configuring the topic, the defined topic is reused in the form of topic reference; the two configuration tasks of the dialog skill library and the dialog task are performed independently.

图2为对话任务描述方法的第一子流程框图，所述接收用户输入的含有对话内容的对话请求，基于所述话题树执行对话过程的步骤包括步骤S201至步骤S203：Fig. 2 is the first sub-flow diagram of the dialog task description method, the step of receiving the dialog request input by the user containing the dialog content, and executing the dialog process based on the topic tree includes steps S201 to S203:

步骤S201：接收用户输入的对话请求，基于请求话题请求并接收用户上传的图像信息；Step S201: receiving a dialog request input by the user, requesting and receiving image information uploaded by the user based on the request topic;

步骤S202：基于执行话题对所述图像信息进行处理，执行知识查询，得到反馈内容；Step S202: Process the image information based on the execution topic, perform knowledge query, and obtain feedback content;

步骤S203：基于答复话题将所述反馈内容反馈至用户。Step S203: Feedback the feedback content to the user based on the reply topic.

在本发明技术方案的一个实例中，用户的对话请求以图像的形式发送，系统在接收到用户上传的图像信息时，进行识别，然后确定相关内容，最后将相关内容反馈给用户。In an example of the technical solution of the present invention, the user's dialogue request is sent in the form of an image. When the system receives the image information uploaded by the user, it recognizes it, then determines the relevant content, and finally feeds back the relevant content to the user.

图3为对话任务描述方法的第二子流程框图，所述基于执行话题对所述图像信息进行处理，执行知识查询，得到反馈内容的步骤包括步骤S2021至步骤S2023：Fig. 3 is a second sub-flow diagram of the dialogue task description method, the steps of processing the image information based on the execution topic, performing knowledge query, and obtaining feedback content include steps S2021 to S2023:

步骤S2021：基于根话题在话题树中查询图像文字提取子话题和知识图谱查询子话题；Step S2021: Query the image text extraction subtopic and the knowledge graph query subtopic in the topic tree based on the root topic;

步骤S2022：基于所述图像文字提取子话题对所述图像信息进行文字识别，得到文本信息；Step S2022: Perform text recognition on the image information based on the image text extraction subtopic to obtain text information;

步骤S2023：基于所述知识图谱查询子话题对所述文本信息进行知识查询，得到知识查询结果。Step S2023: Perform a knowledge query on the text information based on the knowledge graph query subtopic, and obtain a knowledge query result.

步骤S2021至步骤S2023对反馈内容的生成过程进行了具体的限定，反馈内容的生成过程主要分为两个部分，一是识别过程，二是知识查询过程，这两个过程分别由两个独立的子话题完成，识别过程的输出作为知识查询过程的输入。Steps S2021 to S2023 specifically define the process of generating feedback content. The process of generating feedback content is mainly divided into two parts, one is the recognition process, and the other is the knowledge query process. These two processes are composed of two independent The subtopic is completed, and the output of the recognition process is used as the input of the knowledge query process.

具体的，在本发明技术方案的一个实例中，所述基于所述图像文字提取子话题对所述图像信息进行文字识别，得到文本信息的步骤包括：Specifically, in an example of the technical solution of the present invention, the step of performing text recognition on the image information based on the image text extraction subtopic to obtain text information includes:

进一步的，所述对所述待识别文本框中的内容进行内容识别，得到文本信息的步骤包括：Further, the step of performing content identification on the content in the text box to be identified to obtain text information includes:

上述内容提供了一种具体的生成文本信息的技术方案，文本信息的生成过程包括两个步骤，一是文本框选取步骤，二是对文本框中的内容进行识别的步骤，文本框的选取过程包括对图像的色值分析过程，对图像进行色值分析的目的是，确定不同精度的区域识别模型，提高计算资源的利用率。其中，图像的色值种类数越少(白纸黑字)，文本框的选取过程越容易。The above content provides a specific technical solution for generating text information. The generation process of text information includes two steps, one is the step of selecting the text box, and the other is the step of identifying the content in the text box. The selection process of the text box Including the process of analyzing the color value of the image, the purpose of analyzing the color value of the image is to determine the area recognition model with different precision and improve the utilization rate of computing resources. Among them, the fewer types of color values of the image (black and white characters), the easier the selection process of the text box.

文本信息的识别过程是神经网络模型的一个应用，这在现有技术背景下，是很容易完成的。The recognition process of text information is an application of the neural network model, which can be easily accomplished under the background of the existing technology.

作为本发明技术方案的一个优先实施例，所述基于所述知识图谱查询子话题对所述文本信息进行知识查询，得到知识查询结果的步骤包括：As a preferred embodiment of the technical solution of the present invention, the step of performing knowledge query on the text information based on the query subtopic of the knowledge graph, and obtaining the knowledge query result includes:

上述内容对知识图谱的生成过程进行了具体的描述，首先，需要将文本信息转换为一张字表，然后对这些字表进行归一化处理，所述归一化处理就是将字表中的元素用标准的语言代替。最后，根据参考字表在预设的查询数据库中查询相应的内容并显示即可，查询的内容被查询的次数越多，在生成的知识图谱中的显示占比就越大。The above content specifically describes the generation process of the knowledge map. First, it is necessary to convert the text information into a word list, and then perform normalization processing on these word lists. The normalization process is to convert the word list Elements are replaced with standard language. Finally, according to the reference list, the corresponding content can be queried in the preset query database and displayed. The more times the queried content is queried, the larger the proportion of display in the generated knowledge graph will be.

所述查询数据库可以是本地数据库，也可以是云端数据库，可以是本运营商提供的数据库，也可以是其他运营商提供的数据库，具体不做限定。The query database may be a local database, or a cloud database, or a database provided by the operator, or a database provided by other operators, which is not specifically limited.

实施例2Example 2

图4为对话任务描述系统的组成结构框图，本发明实施例中，一种对话任务描述系统，所述系统10包括：Fig. 4 is a structural block diagram of a dialogue task description system. In an embodiment of the present invention, a dialogue task description system, thesystem 10 includes:

话题树配置模块11，用于基于预设的基本话题类型拆分对话任务，得到话题树；Topictree configuration module 11, for splitting dialogue tasks based on preset basic topic types to obtain topic trees;

对话执行模块12，用于接收用户输入的含有对话内容的对话请求，基于所述话题树执行对话过程；Dialogue execution module 12, configured to receive a dialog request input by a user containing dialog content, and execute a dialog process based on the topic tree;

图5为对话任务描述系统中对话执行模块12的组成结构框图，所述对话执行模块12包括：Fig. 5 is a structural block diagram of thedialog execution module 12 in the dialog task description system, and thedialog execution module 12 includes:

图像接收单元121，用于接收用户输入的对话请求，基于请求话题请求并接收用户上传的图像信息；Theimage receiving unit 121 is configured to receive a dialog request input by the user, request and receive image information uploaded by the user based on the request topic;

内容生成单元122，用于基于执行话题对所述图像信息进行处理，执行知识查询，得到反馈内容；Acontent generating unit 122, configured to process the image information based on the execution topic, perform knowledge query, and obtain feedback content;

反馈执行单元123，用于基于答复话题将所述反馈内容反馈至用户。Thefeedback execution unit 123 is configured to feed back the feedback content to the user based on the reply topic.

图6为对话执行模块中内容生成单元122的组成结构框图，所述内容生成单元122包括：Fig. 6 is a block diagram of the composition structure of thecontent generation unit 122 in the dialog execution module, and thecontent generation unit 122 includes:

话题查询子单元1221，用于基于根话题在话题树中查询图像文字提取子话题和知识图谱查询子话题；Thetopic query subunit 1221 is used to query the image text extraction subtopic and the knowledge map query subtopic in the topic tree based on the root topic;

识别子单元1222，用于基于所述图像文字提取子话题对所述图像信息进行文字识别，得到文本信息；Therecognition subunit 1222 is configured to perform text recognition on the image information based on the image text extraction subtopic to obtain text information;

图谱查询子单元1223，用于基于所述知识图谱查询子话题对所述文本信息进行知识查询，得到知识查询结果。Thegraph query subunit 1223 is configured to perform knowledge query on the text information based on the knowledge graph query subtopic, and obtain a knowledge query result.

所述对话任务描述方法所能实现的功能均由计算机设备完成，所述计算机设备包括一个或多个处理器和一个或多个存储器，所述一个或多个存储器中存储有至少一条程序代码，所述程序代码由所述一个或多个处理器加载并执行以实现所述对话任务描述方法的功能。The functions that can be realized by the dialog task description method are all completed by computer equipment, and the computer equipment includes one or more processors and one or more memories, and at least one program code is stored in the one or more memories, The program code is loaded and executed by the one or more processors to realize the functions of the dialog task description method.

处理器从存储器中逐条取出指令、分析指令，然后根据指令要求完成相应操作，产生一系列控制命令，使计算机各部分自动、连续并协调动作，成为一个有机的整体，实现程序的输入、数据的输入以及运算并输出结果，这一过程中产生的算术运算或逻辑运算均由运算器完成；所述存储器包括只读存储器(Read-Only Memory，ROM)，所述只读存储器用于存储计算机程序，所述存储器外部设有保护装置。The processor takes out the instructions one by one from the memory, analyzes the instructions, and then completes the corresponding operations according to the instruction requirements, and generates a series of control commands, so that the various parts of the computer can automatically, continuously and coordinate actions to form an organic whole, and realize the input of programs and the exchange of data. Input and calculation and output results, the arithmetic operation or logic operation generated in this process is completed by the arithmetic unit; the memory includes a read-only memory (Read-Only Memory, ROM), and the read-only memory is used to store computer programs , a protection device is provided outside the memory.

示例性的，计算机程序可以被分割成一个或多个模块，一个或者多个模块被存储在存储器中，并由处理器执行，以完成本发明。一个或多个模块可以是能够完成特定功能的一系列计算机程序指令段，该指令段用于描述计算机程序在终端设备中的执行过程。Exemplarily, a computer program can be divided into one or more modules, and one or more modules are stored in a memory and executed by a processor to implement the present invention. One or more modules may be a series of computer program instruction segments capable of accomplishing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

本领域技术人员可以理解，上述服务设备的描述仅仅是示例，并不构成对终端设备的限定，可以包括比上述描述更多或更少的部件，或者组合某些部件，或者不同的部件，例如可以包括输入输出设备、网络接入设备、总线等。Those skilled in the art can understand that the above description of the service device is only an example, and does not constitute a limitation on the terminal device, and may include more or less components than the above description, or combine certain components, or different components, such as It can include input and output devices, network access devices, buses, etc.

所称处理器可以是中央处理单元(Central Processing Unit，CPU)，还可以是其他通用处理器、数字信号处理器(Digital Signal Processor，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现成可编程门阵列 (Field-Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等，上述处理器是上述终端设备的控制中心，利用各种接口和线路连接整个用户终端的各个部分。The so-called processor can be a central processing unit (Central Processing Unit, CPU), and can also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf Programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The above-mentioned processor is the control center of the above-mentioned terminal equipment, and uses various interfaces and lines to connect various parts of the entire user terminal.

上述存储器可用于存储计算机程序和/或模块，上述处理器通过运行或执行存储在存储器内的计算机程序和/或模块，以及调用存储在存储器内的数据，实现上述终端设备的各种功能。存储器可主要包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需的应用程序(比如信息采集模板展示功能、产品信息发布功能等)等；存储数据区可存储根据泊位状态显示系统的使用所创建的数据(比如不同产品种类对应的产品信息采集模板、不同产品提供方需要发布的产品信息等) 等。此外，存储器可以包括高速随机存取存储器，还可以包括非易失性存储器，例如硬盘、内存、插接式硬盘，智能存储卡(Smart Media Card，SMC)，安全数字(Secure Digital，SD)卡，闪存卡(Flash Card)、至少一个磁盘存储器件、闪存器件、或其他易失性固态存储器件。The above-mentioned memory can be used to store computer programs and/or modules, and the above-mentioned processor realizes various functions of the above-mentioned terminal device by running or executing the computer programs and/or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a storage program area and a storage data area, wherein the storage program area can store an operating system, at least one application program required by a function (such as information collection template display function, product information release function, etc.); Store the data created according to the use of the berth status display system (such as product information collection templates corresponding to different product categories, product information to be released by different product providers, etc.), etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card , a flash memory card (Flash Card), at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

终端设备集成的模块/单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本发明实现上述实施例系统中的全部或部分模块/单元，也可以通过计算机程序来指令相关的硬件来完成，上述的计算机程序可存储于计算机可读存储介质中，该计算机程序在被处理器执行时，可实现上述各个系统实施例的功能。其中，计算机程序包括计算机程序代码，计算机程序代码可以为源代码形式、对象代码形式、可执行文件或某些中间形式等。计算机可读介质可以包括：能够携带计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random AccessMemory)、电载波信号、电信信号以及软件分发介质等。If the integrated module/unit of the terminal device is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention realizes all or part of the modules/units in the system of the above-mentioned embodiments, and it can also be completed by instructing related hardware through a computer program. The above-mentioned computer program can be stored in a computer-readable storage medium. The computer When the program is executed by the processor, the functions of the above-mentioned various system embodiments can be realized. Wherein, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, removable hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access Memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。It should be noted that, in this document, the term "comprising", "comprising" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article or apparatus comprising a set of elements includes not only those elements, It also includes other elements not expressly listed, or elements inherent in the process, method, article, or device. Without further limitations, an element defined by the phrase "comprising a ..." does not preclude the presence of additional identical elements in the process, method, article, or apparatus comprising that element.

以上仅为本发明的优选实施例，并非因此限制本发明的专利范围，凡是利用本发明说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本发明的专利保护范围内。The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process conversion made by using the description of the present invention and the contents of the accompanying drawings, or directly or indirectly used in other related technical fields , are all included in the scope of patent protection of the present invention in the same way.

Claims

Translated fromChinese

1.一种对话任务描述方法，其特征在于，所述方法包括：1. A dialogue task description method, characterized in that the method comprises:

2.根据权利要求1所述的对话任务描述方法，其特征在于，所述接收用户输入的含有对话内容的对话请求，基于所述话题树执行对话过程的步骤包括：2. The dialogue task description method according to claim 1, wherein the dialogue request containing dialogue content input by the user is received, and the step of performing dialogue process based on the topic tree comprises:

3.根据权利要求2所述的对话任务描述方法，其特征在于，所述基于执行话题对所述图像信息进行处理，执行知识查询，得到反馈内容的步骤包括：3. The dialog task description method according to claim 2, wherein the step of processing the image information based on the execution topic, performing knowledge query, and obtaining feedback content comprises:

4.根据权利要求3所述的对话任务描述方法，其特征在于，所述基于所述图像文字提取子话题对所述图像信息进行文字识别，得到文本信息的步骤包括：4. The dialogue task description method according to claim 3, wherein the said image information extraction subtopic based on said image text is used for character recognition to obtain text information, and the step of obtaining text information comprises:

5.根据权利要求4所述的对话任务描述方法，其特征在于，所述对所述待识别文本框中的内容进行内容识别，得到文本信息的步骤包括：5. The dialog task description method according to claim 4, wherein the step of performing content identification to the content in the text box to be identified to obtain text information comprises:

采用神经网络模型识别所述待识别文本框中的文本信息，所述神经网络模型包括:卷积层和池化层；Adopt neural network model to identify the text information in the text box to be identified, described neural network model includes: convolutional layer and pooling layer;

6.根据权利要求3所述的对话任务描述方法，其特征在于，所述基于所述知识图谱查询子话题对所述文本信息进行知识查询，得到知识查询结果的步骤包括：6. The dialog task description method according to claim 3, wherein the step of performing knowledge query on the text information based on the query subtopic of the knowledge map, and obtaining the knowledge query result comprises:

7.根据权利要求1至6任一项所述的对话任务描述方法，其特征在于，在配置话题时，采用话题引用的形式复用已定义话题；对话技能库和对话任务两种配置工作独立进行。7. The dialog task description method according to any one of claims 1 to 6, characterized in that when configuring the topic, the defined topic is reused in the form of topic reference; the two configurations of the dialog skill library and the dialog task work independently conduct.

8.一种对话任务描述系统，其特征在于，所述系统包括：8. A dialogue task description system, characterized in that the system comprises:

9.根据权利要求8所述的对话任务描述系统，其特征在于，所述对话执行模块包括：9. The dialogue task description system according to claim 8, wherein the dialogue execution module comprises:

10.根据权利要求9所述的对话任务描述系统，其特征在于，所述内容生成单元包括：10. The dialogue task description system according to claim 9, wherein the content generating unit comprises: