Information processing method and device

Abstract

The present application provides an information processing method and device, which first obtains the dialogue information between the target user and the intelligent customer service, and determines whether the dialogue information has any abnormality; secondly, if the dialogue information has any abnormality, determines at least one associated question information corresponding to at least one user question information in the dialogue information; then, based on the dialogue information and multiple dialogue information samples, determines the first target question information from the associated question information; finally, based on the first target question information, determines the question answer information returned to the target user. The present application can effectively identify the user's intention by determining the target question information from multiple associated question information based on the dialogue information and dialogue information samples, and returns the question information recognition result that meets the user's intention, thereby reducing the situation of manual intervention and having high efficiency.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an information processing method and device. Background Art

[0002] At present, with the rapid development of the Internet and artificial intelligence, intelligent customer service has been widely used. Intelligent customer service can identify the user's intention based on the user's question and return the corresponding reply. Compared with traditional manual customer service, it has excellent feedback efficiency and convenience. It can efficiently identify some simple questions and quickly feedback, saving human resources. Sometimes, when a user talks with intelligent customer service, the user's question information may be vague and lack key information. At this time, the intelligent customer service will return one or more question recognition results based on the user's question information. The user then selects one of the question recognition results that meets his or her intention. The intelligent customer service will then return the answer information of the question based on the question corresponding to the question recognition result fed back by the user.

[0003] However, the problem recognition results returned by intelligent customer service may still not meet the user's expectations and are inefficient. Therefore, how to effectively identify the user's intentions and return the correct recognition results has become an urgent problem to be solved. Summary of the invention

[0004] In view of this, the purpose of the present application is to provide an information processing method and device that can effectively identify the user's intention and return a question information recognition result that meets the user's intention with high efficiency.

[0005] According to one aspect of the present application, an electronic device is provided, which may include one or more storage media, one or more processors communicating with the storage media, and a bus. The one or more storage media store machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform one or more of the following operations:

[0006] Obtaining the conversation information between the target user and the intelligent customer service, and determining whether there is any abnormality in the conversation information;

[0007] If the dialogue information is abnormal, determining at least one associated question information corresponding to at least one user question information in the dialogue information;

[0008] Determining first target question information from the associated question information according to the dialogue information and a plurality of dialogue information samples;

[0009] Determine, based on the first target question information, question answer information to be returned to the target user.

[0010] In some embodiments of the present application, the dialogue information also includes at least one question information recognition result fed back by the intelligent customer service according to the user question information; when the processor executes the machine-readable instructions, one or more of the following operations may be performed:

[0011] Determining the amount of the user question information contained in the dialogue information;

[0012] If the number of the user question information contained in the dialogue information is greater than a preset threshold, and in the dialogue information, the type of feedback information of the target user for the recognition result of the question information is negative information, it is determined that there is an abnormality in the dialogue information.

[0013] In some embodiments of the present application, when the processor executes the machine-readable instruction, one or more of the following operations may be performed:

[0014] According to a pre-established conversation information knowledge graph, associated question information associated with the user question information is determined.

[0015] In some embodiments of the present application, when the processor executes the machine-readable instruction, one or more of the following operations may be performed:

[0016] Determine, according to the dialogue information sample, historical user question information in the dialogue information sample, historical question information recognition results, and historical feedback information of the user corresponding to the dialogue information sample with respect to the historical question information recognition results;

[0017] According to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, first target question information is determined from the associated question information.

[0018] In some embodiments of the present application, when the processor executes the machine-readable instruction, one or more of the following operations may be performed:

[0019] Training the reinforcement learning model according to the historical user question information, the historical question information recognition result, and the historical feedback information;

[0020] Determine the relevance of each of the associated question information to the user question information by using the trained reinforcement learning model and the conversation information;

[0021] The related question information with the highest relevance is selected as the first target question information.

[0022] In some embodiments of the present application, when the processor executes the machine-readable instruction, one or more of the following operations may be performed:

[0023] If it is detected that the type of the feedback information of the target user in response to the question answer information is negative information, a greedy selection strategy and the relevance of each of the associated question information and the user question information are used to determine the second target question information from the associated question information.

[0024] In some embodiments of the present application, when the processor executes the machine-readable instruction, one or more of the following operations may be performed:

[0025] If it is detected that the type of the feedback information of the target user in response to the question answer information is affirmative information, the dialogue information knowledge graph is updated using the feedback information of the target user in response to the question answer information and the dialogue information.

[0026] According to one aspect of the present application, there is provided an information processing method, comprising:

[0027] Obtaining the conversation information between the target user and the intelligent customer service, and determining whether there is any abnormality in the conversation information;

[0028] If the dialogue information is abnormal, determining at least one associated question information corresponding to at least one user question information in the dialogue information;

[0029] Determining first target question information from the associated question information according to the dialogue information and a plurality of dialogue information samples;

[0030] Determine, based on the first target question information, question answer information to be returned to the target user.

[0031] In some embodiments of the present application, the dialogue information also includes at least one question information recognition result fed back by the intelligent customer service according to the user question information;

[0032] The determining whether the dialogue information is abnormal includes:

[0033] Determining the amount of the user question information contained in the dialogue information;

[0034] If the number of the user question information contained in the dialogue information is greater than a preset threshold, and in the dialogue information, the type of feedback information of the target user for the recognition result of the question information is negative information, it is determined that there is an abnormality in the dialogue information.

[0035] In some embodiments of the present application, determining at least one associated question information corresponding to at least one user question information in the dialog information includes:

[0036] According to a pre-established conversation information knowledge graph, associated question information associated with the user question information is determined.

[0037] In some embodiments of the present application, determining the first target question information from the associated question information according to the dialogue information and the plurality of dialogue information samples includes:

[0038] Determine, according to the dialogue information sample, historical user question information in the dialogue information sample, historical question information recognition results, and historical feedback information of the user corresponding to the dialogue information sample with respect to the historical question information recognition results;

[0039] According to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, first target question information is determined from the associated question information.

[0040] In some embodiments of the present application, determining the first target question information from the associated question information according to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model includes:

[0041] Training the reinforcement learning model according to the historical user question information, the historical question information recognition result, and the historical feedback information;

[0042] Determine the relevance of each of the associated question information to the user question information by using the trained reinforcement learning model and the conversation information;

[0043] The related question information with the highest relevance is selected as the first target question information.

[0044] In some embodiments of the present application, the method further includes:

[0045] If it is detected that the type of the feedback information of the target user in response to the question answer information is negative information, a greedy selection strategy and the relevance of each of the associated question information and the user question information are used to determine the second target question information from the associated question information.

[0046] In some embodiments of the present application, the method further includes:

[0047] If it is detected that the type of the feedback information of the target user in response to the question answer information is affirmative information, the dialogue information knowledge graph is updated using the feedback information of the target user in response to the question answer information and the dialogue information.

[0048] According to another aspect of the present application, there is provided an information processing device, including:

[0049] The first determination module is used to obtain the conversation information between the target user and the intelligent customer service, and determine whether there is any abnormality in the conversation information;

[0050] A second determination module, configured to determine at least one associated question information corresponding to at least one user question information in the dialogue information if there is an abnormality in the dialogue information;

[0051] a third determination module, configured to determine first target question information from the associated question information according to the dialogue information and a plurality of dialogue information samples;

[0052] The fourth determination module is used to determine the question answer information returned to the target user according to the first target question information.

[0053] In some embodiments of the present application, the dialogue information also includes at least one question information recognition result fed back by the intelligent customer service according to the user question information;

[0054] When determining whether the dialogue information is abnormal, the first determination module is specifically used to:

[0055] Determining the amount of the user question information contained in the dialogue information;

[0056] If the number of the user question information contained in the dialogue information is greater than a preset threshold, and in the dialogue information, the type of feedback information of the target user for the recognition result of the question information is negative information, it is determined that there is an abnormality in the dialogue information.

[0057] In some embodiments of the present application, the second determination module is specifically used to:

[0058] According to a pre-established conversation information knowledge graph, associated question information associated with the user question information is determined.

[0059] In some embodiments of the present application, the third determination module is specifically used to:

[0060] Determine, according to the dialogue information sample, historical user question information in the dialogue information sample, historical question information recognition results, and historical feedback information of the user corresponding to the dialogue information sample with respect to the historical question information recognition results;

[0061] According to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, first target question information is determined from the associated question information.

[0062] In some embodiments of the present application, when the third determination module determines the first target question information from the associated question information according to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, it is specifically used to:

[0063] Training the reinforcement learning model according to the historical user question information, the historical question information recognition result, and the historical feedback information;

[0064] Determine the relevance of each of the associated question information to the user question information by using the trained reinforcement learning model and the conversation information;

[0065] The related question information with the highest relevance is selected as the first target question information.

[0066] In some embodiments of the present application, the device further comprises:

[0067] The fifth determination module is used to determine the second target question information from the associated question information by using a greedy selection strategy and the relevance of each associated question information to the user question information if it is detected that the type of the feedback information of the target user to the question answer information is negative information.

[0068] In some embodiments of the present application, the device further comprises:

[0069] An updating module is used to update the conversation information knowledge graph using the feedback information of the target user regarding the question answer information and the conversation information if it is detected that the type of the feedback information of the target user regarding the question answer information is affirmative information.

[0070] According to another aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned information processing method are executed.

[0071] The information processing method and device provided in the embodiment of the present application first obtain the dialogue information between the target user and the intelligent customer service, and determine whether there is an abnormality in the dialogue information; secondly, if there is an abnormality in the dialogue information, determine at least one related question information corresponding to at least one user question information in the dialogue information; then, based on the dialogue information and multiple dialogue information samples, determine the first target question information from the related question information; finally, based on the first target question information, determine the question answer information returned to the target user. The present application can effectively identify the user's intention by determining the target question information from multiple related question information based on the dialogue information and dialogue information samples, return the question information recognition result that meets the user's intention, reduce the situation of manual intervention, and has high efficiency.

[0072] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0074] Figure 1 A flowchart of an information processing method provided by an embodiment of the present application is shown;

[0075] Figure 2 A schematic diagram of the structure of an information processing device provided in an embodiment of the present application is shown;

[0076] Figure 3 A schematic diagram showing the structure of another information processing device provided in an embodiment of the present application is shown;

[0077] Figure 4 A schematic structural diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0078] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application usually described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, each other embodiment obtained by those skilled in the art without making creative work belongs to the scope of protection of the present application.

[0079] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0080] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0081] At present, with the rapid development of the Internet and artificial intelligence, intelligent customer service has been widely used. When a user talks to an intelligent customer service, the user's question information may be vague and lack key information. At this time, the intelligent customer service will return one or more question recognition results based on the user's question information. The user then selects one of the question recognition results that meets his or her intention. The intelligent customer service will then return the answer information based on the question corresponding to the question recognition result fed back by the user.

[0082] However, the problem recognition results returned by intelligent customer service may still not meet the user's expectations and are inefficient. Therefore, how to effectively identify the user's intentions and return the correct recognition results has become an urgent problem to be solved.

[0083] Based on this, an embodiment of the present application provides an information processing method to improve the accuracy of identifying user intentions and improve efficiency.

[0084] See also Figure 1 , Figure 1 This is a flow chart of an information processing method provided by an embodiment of the present application. Figure 1As shown in , the information processing method provided by the embodiment of the present application includes:

[0085] S101. Acquire conversation information between a target user and an intelligent customer service representative, and determine whether there is any abnormality in the conversation information.

[0086] In this step, the conversation information between the user and the intelligent customer service will be saved in the database. The server can query the conversation information between the target user and the intelligent customer service from the database. The conversation information may include the user's question information, the question information recognition result or question answer information fed back by the intelligent customer service for the user's question information, and the user's feedback information on the question information recognition result. The server can determine whether there is any abnormality in the conversation information based on the content of the conversation information.

[0087] For example, in some application scenarios, the user's question information may be "service points". The intelligent customer service cannot directly determine the user's intention through the "service points". It can identify the user's question information and feedback the identification result to the user. The question information identification result can be "how to calculate service points", "how to improve service points", etc. The intelligent customer service can feedback one or more question information identification results to the user. The user selects a question information identification result that meets his or her intention. The server then returns corresponding feedback information based on the question information identification result selected by the user. If the question information identification results do not meet the user's intention, the user can ask the question again or choose to give up asking the question.

[0088] Furthermore, in a group of dialogue information, if the user asks questions multiple times, but the question information recognition results returned by the intelligent customer service do not match the user's intention, it can be determined that there is an abnormality in the dialogue information.

[0089] Specifically, the server may determine that the dialogue information is abnormal when it is detected that the dialogue information contains more than a preset number of user question information and each user's feedback information on the question information recognition result in the dialogue information is negative feedback information.

[0090] Among them, the negative type can be feedback such as the user did not make a selection, or selected an option without a correct answer, etc.

[0091] S102: If there is an abnormality in the dialogue information, determine at least one associated question information corresponding to at least one user question information in the dialogue information.

[0092] In this step, the user can determine the related question information corresponding to the user's question information based on the conversation information knowledge graph, and each user's question information can correspond to multiple related question information.

[0093] Among them, the conversation information knowledge graph can be generated based on historical user question information and the business type of the current application scenario.

[0094] Specifically, after a user completes a conversation with an intelligent customer service representative, the conversation information can be tagged with the type of question according to the business type. A set of conversation information can have multiple question type tags. For example, in the online car-hailing application scenario, the tagged content can be "special car-listening order-listening order failure". The server can obtain these historical conversation information with tags, and then classify multiple questions according to the content of the tags to obtain multiple topics, and then build a knowledge graph based on the obtained topics and conversation information.

[0095] Among them, the conversation information knowledge graph contains multiple types of nodes, which may include domain nodes (such as order categories), topic nodes (such as order cancellation), question nodes (such as "how to query service points"), and the edges between nodes identify the relationships between nodes, such as inclusion relationships, synonym / synonymous relationships, question type relationships (why, how, where), etc.

[0096] After establishing the conversation information knowledge graph, you can also use dynamic data to update it in real time to make the conversation information knowledge graph more timely.

[0097] In this way, the node to which the current dialogue information belongs in the dialogue information knowledge graph can be determined, and then the related question information can be determined based on the content of the adjacent nodes.

[0098] S103: Determine first target question information from the associated question information according to the dialogue information and a plurality of dialogue information samples.

[0099] In this step, there are usually multiple related question information determined in the conversation information knowledge graph, but the user's intention is usually single. The strategy retrieval method in reinforcement learning can be used to select one from multiple related question information.

[0100] Specifically, multiple dialogue information samples and the current dialogue information can be used to determine the first target question information through reinforcement learning. The reinforcement learning strategy can include three dimensions, namely, utilization, exploration, and feedback dimensions. The utilization dimension is to use dialogue information samples to determine the question information that users ask more frequently; the exploration dimension is to explore the question information that has not been fed back to the user in the question information recognition results, which can avoid blind spots; the feedback dimension is to determine the relevance of related types of related question information and user intentions based on the user's feedback information on the question information recognition results. Through the above three dimensions, machine learning is performed to determine the user's intentions and determine the related question information with the highest relevance to the user's question information.

[0101] The dialogue information samples may be multiple historical dialogue information, which can be used as training samples for the reinforcement learning model.

[0102] S104: Determine question answer information to be returned to the target user based on the first target question information.

[0103] In this step, after determining the first target question information, the server may return question answer information corresponding to the first target question information from the business database.

[0104] Specifically, question answer information can be determined in the conversation information knowledge graph.

[0105] In a possible implementation, the conversation information further includes at least one question information recognition result fed back by the intelligent customer service according to the user question information;

[0106] The determining whether the dialogue information is abnormal includes:

[0107] Determining the amount of the user question information contained in the dialogue information;

[0108] If the number of the user question information contained in the dialogue information is greater than a preset threshold, and in the dialogue information, the type of feedback information of the target user for the recognition result of the question information is negative information, it is determined that there is an abnormality in the dialogue information.

[0109] In a possible implementation manner, determining at least one associated question information corresponding to at least one user question information in the dialog information includes:

[0110] According to a pre-established conversation information knowledge graph, associated question information associated with the user question information is determined.

[0111] In a possible implementation manner, determining the first target question information from the associated question information according to the dialogue information and the plurality of dialogue information samples includes:

[0112] Determine, according to the dialogue information sample, historical user question information in the dialogue information sample, historical question information recognition results, and historical feedback information of the user corresponding to the dialogue information sample with respect to the historical question information recognition results;

[0113] According to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, first target question information is determined from the associated question information.

[0114] In this step, the historical user question information in the conversation information sample, the historical question information recognition result and the historical feedback information of the user corresponding to the conversation information sample regarding the historical question information recognition result can be determined first, wherein the conversation information sample is the conversation information between the historical user and the intelligent customer service stored in the database, the historical user question information is the user question information in the conversation information sample, the historical question information recognition result is the question information recognition result fed back by the intelligent customer service based on the historical user question information, and the historical feedback information is the feedback information of the historical user regarding the historical question information recognition result.

[0115] Furthermore, a reinforcement learning model can be trained based on historical user question information, historical question information recognition results, and historical feedback information. The utilization dimension in the reinforcement learning model is determined based on historical user question information; the exploration dimension is determined based on historical question information recognition results; and the feedback dimension is determined based on historical feedback information.

[0116] After the reinforcement learning model is trained, the user question information and related question information of the target user can be input into the reinforcement learning model. The reinforcement learning model determines the correlation between each related question information and the user question information based on the trained parameters, and outputs the related question information with the highest correlation as the first target question information.

[0117] In a possible implementation, determining the first target question information from the associated question information according to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model includes:

[0118] Training the reinforcement learning model according to the historical user question information, the historical question information recognition result, and the historical feedback information;

[0119] Determine the relevance of each of the associated question information to the user question information by using the trained reinforcement learning model and the conversation information;

[0120] The related question information with the highest relevance is selected as the first target question information.

[0121] In a possible implementation, the method further includes:

[0122] If it is detected that the type of the feedback information of the target user in response to the question answer information is negative information, a greedy selection strategy and the relevance of each of the associated question information and the user question information are used to determine the second target question information from the associated question information.

[0123] In this step, the server can determine whether the question answer information meets the user's intention based on the feedback information of the target user. If it still does not meet the intention, the server uses a greedy selection strategy to select the question information with the second highest relevance information in the associated question information as the second target question information, and feeds back the question answer information corresponding to the second target question information to the target user.

[0124] In a possible implementation, the method further includes:

[0125] If it is detected that the type of the feedback information of the target user in response to the question answer information is affirmative information, the dialogue information knowledge graph is updated using the feedback information of the target user in response to the question answer information and the dialogue information.

[0126] In this step, when the type of feedback information of the target user in response to the question is affirmative information, it can be determined that the reply is in line with the user's intention, and the dialogue information can be used to update the dialogue information knowledge graph, enhance the correlation strength of relevant information, and optimize the dialogue information knowledge graph.

[0127] The information processing method provided in the embodiment of the present application first obtains the dialogue information between the target user and the intelligent customer service, and determines whether the dialogue information has any abnormality; secondly, if the dialogue information has any abnormality, determines at least one associated question information corresponding to at least one user question information in the dialogue information; then, based on the dialogue information and multiple dialogue information samples, determines the first target question information from the associated question information; finally, based on the first target question information, determines the question answer information returned to the target user. The present application can effectively identify the user's intention by determining the target question information from multiple associated question information based on the dialogue information and dialogue information samples, and returns the question information recognition result that meets the user's intention, thereby reducing the situation of manual intervention and having high efficiency.

[0128] See also Figure 2 , Figure 3 , Figure 2 A schematic diagram of the structure of an information processing device provided in an embodiment of the present application is shown in FIG. Figure 3 This is a schematic diagram of the structure of another information processing device provided in an embodiment of the present application. Figure 2 As shown in , the information processing device 200 includes:

[0129] The first determination module 210 is used to obtain the conversation information between the target user and the intelligent customer service, and determine whether there is any abnormality in the conversation information;

[0130] A second determination module 220, configured to determine at least one associated question information corresponding to at least one user question information in the dialogue information if there is an abnormality in the dialogue information;

[0131] A third determination module 230, configured to determine first target question information from the associated question information according to the dialogue information and a plurality of dialogue information samples;

[0132] The fourth determination module 240 is used to determine the question answer information returned to the target user according to the first target question information.

[0133] In some embodiments of the present application, the dialogue information also includes at least one question information recognition result fed back by the intelligent customer service according to the user question information;

[0134] When determining whether the dialog information is abnormal, the first determination module 210 is specifically used to:

[0135] Determining the amount of the user question information contained in the dialogue information;

[0136] If the number of the user question information contained in the dialogue information is greater than a preset threshold, and in the dialogue information, the type of feedback information of the target user for the recognition result of the question information is negative information, it is determined that there is an abnormality in the dialogue information.

[0137] In some embodiments of the present application, the second determining module 220 is specifically used to:

[0138] According to a pre-established conversation information knowledge graph, associated question information associated with the user question information is determined.

[0139] In some embodiments of the present application, the third determination module 230 is specifically used to:

[0140] Determine, according to the dialogue information sample, historical user question information in the dialogue information sample, historical question information recognition results, and historical feedback information of the user corresponding to the dialogue information sample with respect to the historical question information recognition results;

[0141] According to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, first target question information is determined from the associated question information.

[0142] In some embodiments of the present application, when the third determination module 230 determines the first target question information from the associated question information according to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, it is specifically used to:

[0143] Training the reinforcement learning model according to the historical user question information, the historical question information recognition result, and the historical feedback information;

[0144] Determine the relevance of each of the associated question information to the user question information by using the trained reinforcement learning model and the conversation information;

[0145] The related question information with the highest relevance is selected as the first target question information.

[0146] like Figure 3 As shown, in some embodiments of the present application, the information processing device 300 includes a first determination module 310, a second determination module 320, a third determination module 330, a fourth determination module 340 and:

[0147] The fifth determination module 350 is used to determine the second target question information from the associated question information by using a greedy selection strategy and the relevance of each associated question information to the user question information if it is detected that the type of the feedback information of the target user to the question answer information is negative information.

[0148] In some embodiments of the present application, the information processing device 300 further includes:

[0149] The updating module 360 ​​is used to update the conversation information knowledge graph using the feedback information of the target user regarding the question answer information and the conversation information if it is detected that the type of the feedback information of the target user regarding the question answer information is affirmative information.

[0150] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 4 As shown in , the electronic device 400 includes a processor 410 , a memory 420 and a bus 430 .

[0151] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 communicates with the memory 420 via the bus 430. When the machine-readable instructions are executed by the processor 410, one or more of the following operations may be performed:

[0152] Obtaining the conversation information between the target user and the intelligent customer service, and determining whether there is any abnormality in the conversation information;

[0153] If the dialogue information is abnormal, determining at least one associated question information corresponding to at least one user question information in the dialogue information;

[0154] Determining first target question information from the associated question information according to the dialogue information and a plurality of dialogue information samples;

[0155] Determine, based on the first target question information, question answer information to be returned to the target user.

[0156] In some embodiments of the present application, the dialogue information also includes at least one question information recognition result fed back by the intelligent customer service according to the user question information; when the processor 410 executes the machine-readable instruction, one or more of the following operations may be performed:

[0157] Determining the amount of the user question information contained in the dialogue information;

[0158] If the number of the user question information contained in the dialogue information is greater than a preset threshold, and in the dialogue information, the type of feedback information of the target user for the recognition result of the question information is negative information, it is determined that there is an abnormality in the dialogue information.

[0159] In some embodiments of the present application, when the processor 410 executes the machine-readable instructions, one or more of the following operations may be performed:

[0160] According to a pre-established conversation information knowledge graph, associated question information associated with the user question information is determined.

[0161] In some embodiments of the present application, when the processor 410 executes the machine-readable instructions, one or more of the following operations may be performed:

[0162] Determine, according to the dialogue information sample, historical user question information in the dialogue information sample, historical question information recognition results, and historical feedback information of the user corresponding to the dialogue information sample with respect to the historical question information recognition results;

[0163] According to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, first target question information is determined from the associated question information.

[0164] In some embodiments of the present application, when the processor 410 executes the machine-readable instructions, one or more of the following operations may be performed:

[0165] Training the reinforcement learning model according to the historical user question information, the historical question information recognition result, and the historical feedback information;

[0166] Determine the relevance of each of the associated question information to the user question information by using the trained reinforcement learning model and the conversation information;

[0167] The related question information with the highest relevance is selected as the first target question information.

[0168] In some embodiments of the present application, when the processor 410 executes the machine-readable instructions, one or more of the following operations may be performed:

[0169] If it is detected that the type of the feedback information of the target user in response to the question answer information is negative information, a greedy selection strategy and the relevance of each of the associated question information and the user question information are used to determine the second target question information from the associated question information.

[0170] In some embodiments of the present application, when the processor 410 executes the machine-readable instructions, one or more of the following operations may be performed:

[0171] If it is detected that the type of the feedback information of the target user in response to the question answer information is affirmative information, the dialogue information knowledge graph is updated using the feedback information of the target user in response to the question answer information and the dialogue information.

[0172] Correspondingly, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by the processor, the steps of the above-mentioned information processing method are executed.

[0173] Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, etc. When the computer program on the storage medium is run, the above-mentioned high-capacity lane determination method can be executed to solve the technical problem of the lack of authenticity of the high-capacity lane determination results in the prior art.

[0174] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0175] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0176] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0177] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0178] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application can essentially be embodied in the form of a software product, or in other words, the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0179] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. An information processing method, characterized in that: The method comprises: Acquire conversation information between the target user and the intelligent customer service, and determine whether the conversation information has an abnormality; the conversation information also includes at least one question information recognition result fed back by the intelligent customer service according to the user question information; the determining whether the conversation information has an abnormality includes: determining the number of the user question information contained in the conversation information; if the number of the user question information contained in the conversation information is greater than a preset threshold, and in the conversation information, the type of feedback information of the target user for the question information recognition result is negative information, then determine that the conversation information has an abnormality; If the dialogue information is abnormal, at least one associated question information corresponding to at least one user question information in the dialogue information is determined; the determining of at least one associated question information corresponding to at least one user question information in the dialogue information includes: determining the associated question information associated with the user question information according to a pre-established dialogue information knowledge graph; wherein the dialogue information knowledge graph includes multiple types of nodes, the nodes include domain nodes, subject nodes and question nodes, and the relationship between the edges between the nodes includes a relationship, a synonym relationship, a near-synonymous relationship and a question type relationship; According to the dialogue information and multiple dialogue information samples, determining the first target question information from the related question information; including: using multiple dialogue information samples and current dialogue information, through a strategy retrieval method in reinforcement learning, selecting one from multiple related question information as the first target question information; wherein the dialogue information samples are multiple historical dialogue information; the reinforcement learning strategy includes a utilization dimension, an exploration dimension and a feedback dimension; the utilization dimension is determined according to historical user question information; the exploration dimension is determined according to historical question information recognition results; and the feedback dimension is determined according to historical feedback information; Determine, based on the first target question information, question answer information to be returned to the target user.

2. The method according to claim 1, characterized in that The negative information means that the user did not provide feedback, or the user selected an option that has no correct answer.

3. The method according to claim 1, characterized in that The determining, based on the pre-established conversation information knowledge graph, the associated question information associated with the user question information includes: Determine the node to which the conversation information belongs in the conversation information knowledge graph; The associated question information is determined according to the content of the adjacent nodes.

4. The method according to claim 1, characterized in that: The determining, based on the dialogue information and the plurality of dialogue information samples, first target question information from the associated question information includes: Determine, according to the dialogue information sample, historical user question information in the dialogue information sample, historical question information recognition results, and historical feedback information of the user corresponding to the dialogue information sample with respect to the historical question information recognition results; According to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model, first target question information is determined from the associated question information.

5. The method according to claim 4, characterized in that The determining the first target question information from the associated question information according to the historical user question information, the historical question information recognition result, the historical feedback information, the dialogue information, and the reinforcement learning model includes: Training the reinforcement learning model according to the historical user question information, the historical question information recognition result, and the historical feedback information; Determine the relevance of each of the associated question information to the user question information by using the trained reinforcement learning model and the conversation information; The related question information with the highest relevance is selected as the first target question information.

6. The method according to claim 5, characterized in that The method further comprises: If it is detected that the type of the feedback information of the target user in response to the question answer information is negative information, a greedy selection strategy and the relevance of each of the associated question information and the user question information are used to determine the second target question information from the associated question information.

7. The method according to claim 6, characterized in that The method further comprises: If it is detected that the type of the feedback information of the target user in response to the question answer information is affirmative information, the dialogue information knowledge graph is updated using the feedback information of the target user in response to the question answer information and the dialogue information.

8. An information processing device, characterized in that: The device comprises: A first determination module is used to obtain the dialogue information between the target user and the intelligent customer service, and determine whether there is an abnormality in the dialogue information; the dialogue information also includes at least one question information recognition result fed back by the intelligent customer service according to the user question information, and the first determination module is specifically used to: determine the number of the user question information contained in the dialogue information; if the number of the user question information contained in the dialogue information is greater than a preset threshold, and in the dialogue information, the type of feedback information of the target user for the question information recognition result is negative information, then determine that there is an abnormality in the dialogue information; A second determination module is used to determine at least one associated question information corresponding to at least one user question information in the dialogue information if there is an abnormality in the dialogue information; the second determination module is specifically used to: determine the associated question information associated with the user question information according to a pre-established dialogue information knowledge graph; wherein the dialogue information knowledge graph contains multiple types of nodes, the nodes include domain nodes, subject nodes and question nodes, and the relationship between the edges between the nodes includes a relationship, a synonym relationship, a near-synonymous relationship and a question type relationship; A third determination module is used to determine the first target question information from the related question information according to the dialogue information and multiple dialogue information samples; specifically, using multiple dialogue information samples and current dialogue information, through a strategy retrieval method in reinforcement learning, select one of the multiple related question information as the first target question information; wherein the dialogue information samples are multiple historical dialogue information; the reinforcement learning strategy includes a utilization dimension, an exploration dimension and a feedback dimension; the utilization dimension is determined according to historical user question information; the exploration dimension is determined according to historical question information recognition results; and the feedback dimension is determined according to historical feedback information; The fourth determining module is used to determine the question answer information returned to the target user according to the first target question information.

9. An electronic device, characterized in that: include: A processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the storage medium communicate via the bus, and the processor executes the machine-readable instructions to perform the steps of the information processing method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the information processing method according to any one of claims 1 to 7 are executed.

Images