Distributed conversation processing method, device, equipment and medium

Abstract

The present disclosure provides a shunted dialogue processing method, device, equipment and medium, relating to the fields of voice technology, deep learning, natural language processing, Internet of Vehicles, intelligent cockpit and the like in artificial intelligence. The method comprises: in response to a received dialogue inquiry, determining associated information of the dialogue inquiry; wherein the associated information comprises: scene information of the dialogue inquiry, historical shunting information of the associated inquiry and semantic information of the dialogue inquiry; determining a target model to be called this time among a plurality of dialogue models according to the associated information; and processing the dialogue inquiry according to the target model to obtain a processing result, so as to improve the accuracy and timeliness of the dialogue processing result and reduce the processing resources consumed in the dialogue processing process.

Description

Technical Field

[0001] This disclosure relates to fields such as voice technology, deep learning, natural language processing, vehicle networking, and smart cockpit in artificial intelligence, and in particular to a method, apparatus, device, and medium for offloading dialogue processing. Background Technology

[0002] Currently, with the continuous development of artificial intelligence technology in the field of speech processing, voice dialogue systems have been applied to various fields. Voice dialogue systems understand the intent of the dialogue data sent by the speaker in order to provide feedback to the user that matches their intent.

[0003] Therefore, how to accurately and promptly provide feedback on the results of dialogue processing is an urgent problem that needs to be solved. Summary of the Invention

[0004] This disclosure provides a dialogue processing method, apparatus, device, and medium for accurately and timely determining the results of dialogue processing.

[0005] According to a first aspect of this disclosure, a diversionary dialogue processing method is provided, comprising:

[0006] In response to a received dialogue query, the association information of the dialogue query is determined; the association information includes: the scenario information of the dialogue query, the historical routing information of the associated queries, and the semantic information of the dialogue query; the scenario information is used to characterize the scenario in which the dialogue query is initiated; the associated query is a request that has a semantic relationship with the dialogue query; the historical routing information indicates the processing record information of the associated query; the semantic information characterizes the semantic features of the dialogue query;

[0007] Based on the associated information, the target model to be invoked in this instance is determined from multiple dialogue models; wherein, the dialogue model is used for dialogue query processing;

[0008] Based on the target model, the dialogue query is processed to obtain the processing result.

[0009] According to a second aspect of this disclosure, a split-flow dialogue processing apparatus is provided, comprising:

[0010] The first determining unit is configured to determine the associated information of the dialogue inquiry in response to a received dialogue inquiry; the associated information includes: scenario information of the dialogue inquiry, historical routing information of the associated inquiry, and semantic information of the dialogue inquiry; the scenario information is used to characterize the scenario in which the dialogue inquiry is initiated; the associated inquiry is a request that has semantic relevance to the dialogue inquiry; the historical routing information indicates the processing record information of the associated inquiry; the semantic information characterizes the semantic features of the dialogue inquiry.

[0011] The second determining unit is used to determine the target model to be invoked in this instance from multiple dialogue models based on the associated information; wherein the dialogue model is used for dialogue query processing.

[0012] The processing unit is used to process the dialogue query according to the target model and obtain the processing result.

[0013] According to a third aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method of the first aspect.

[0014] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the method described in the first aspect.

[0015] According to a fifth aspect of this disclosure, a computer program product is provided, the computer program product comprising: a computer program stored in a readable storage medium, wherein at least one processor of an electronic device can read the computer program from the readable storage medium, and the at least one processor executes the computer program to cause the electronic device to perform the method described in the first aspect.

[0016] The technology disclosed herein is beneficial for improving the accuracy of dialogue processing results and enhancing the user's dialogue experience.

[0017] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0018] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0019] Figure 1This is a schematic diagram based on the first embodiment of the present disclosure;

[0020] Figure 2 This is a schematic diagram according to the second embodiment of the present disclosure;

[0021] Figure 3 This is a schematic diagram according to the third embodiment of the present disclosure;

[0022] Figure 4 This is a schematic diagram according to the fourth embodiment of the present disclosure;

[0023] Figure 5 This is a schematic diagram according to the fifth embodiment of the present disclosure;

[0024] Figure 6 This is a schematic diagram according to the sixth embodiment of the present disclosure;

[0025] Figure 7 This is a block diagram of an electronic device used to implement the traffic splitting dialogue processing method of the embodiments of this disclosure. Detailed Implementation

[0026] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0027] Currently, traditional dialogue models are typically rule-based. Developers pre-write dialogue processing rules, and when a user sends a dialogue query, the appropriate processing result is determined based on the query and the pre-defined rules. However, in this approach, when a user's query cannot match the pre-written rules, only feedback that doesn't understand the user's intent can be provided. In other words, relying solely on dialogue processing rules limits the types of queries that can be understood and processed, and when the user's query is complex, an accurate response cannot be provided.

[0028] Another possible implementation involves combining large-scale modeling techniques from related technologies. Based on massive parameter processing, these large-scale models can process and respond to user queries. By learning parameter processing from vast amounts of data, these models possess a higher level of understanding compared to rule-based dialogue processing, enabling them to handle more complex dialogue queries. However, a drawback of this technique is the long processing time per dialogue and the significant consumption of processing resources, which can lead to users not receiving timely feedback.

[0029] To avoid at least one of the aforementioned technical problems, the inventors of this disclosure, through creative effort, arrived at the inventive concept of this disclosure: After obtaining a dialogue query, firstly, relevant information indicating the comprehension complexity of the dialogue query is determined. Then, based on the relevant information, a target model matching the current comprehension complexity is selected from multiple dialogue models to improve the accuracy of the final dialogue processing result. Furthermore, the relevant information includes not only the voice information of the dialogue query itself, but also the scene information corresponding to the dialogue query and the historical triage information of the relevant queries, to improve the accuracy of the selected model.

[0030] This disclosure provides a diversion dialogue processing method, apparatus, device, and medium, which can be applied to fields such as voice technology, deep learning, natural language processing, vehicle networking, and smart cockpit in artificial intelligence, to ensure the diversity of product pages obtained by users, thereby ensuring the accuracy of user feedback before the page goes live.

[0031] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0032] Figure 1 This is a schematic diagram based on the first embodiment of the present disclosure, as shown below. Figure 1 As shown, the traffic splitting dialogue processing method of this disclosure embodiment includes:

[0033] S101. In response to the received dialogue query, determine the associated information of the dialogue query; the associated information includes: the scenario information of the dialogue query, the historical routing information of the associated query, and the semantic information of the dialogue query; the scenario information is used to characterize the scenario in which the dialogue query is initiated; the associated query is a request that has a semantic relationship with the dialogue query; the historical routing information indicates the processing result of the associated query; the semantic information characterizes the semantic features of the dialogue query.

[0034] For example, the execution subject of this embodiment can be a traffic-sharing dialogue processing device. The dialogue processing device can be a server (such as a local server or a cloud server), a computer, a processor, a chip, etc. This embodiment does not limit the scope of the device.

[0035] In one example, the dialogue processing method received in this embodiment can be applied to dialogue processing in the travel field. For instance, the above-mentioned dialogue processing method can be applied to destination information query, destination navigation, travel mode recommendation, and other travel-related information, such as weather and traffic information, which are not specifically limited in this embodiment.

[0036] For example, the above method can be applied to scenarios where dialogue queries received by in-vehicle navigation devices are processed, as well as scenarios where dialogue queries received by various smart terminal devices are processed. No specific limitations are imposed in this embodiment.

[0037] In one example, the above method can also be applied to different scenarios such as product search and voice-controlled terminal devices. This embodiment does not impose specific limitations on the application areas of the above method.

[0038] The received dialogue query in this embodiment can be used to indicate dialogue needs. When applied to the travel field, it can specifically refer to travel needs. For example, in practical applications, the dialogue query could be a user's voice message asking "What's the current weather like?" or a voice message asking "Recommend the most popular nearby attractions," etc.

[0039] In one example, the associated information corresponding to the dialogue query is information that can indicate the level of understanding of the current dialogue query. For example, the associated information may specifically include a level or rating of understanding complexity.

[0040] In this embodiment, the related information of the dialogue is characterized from three dimensions: scene information, semantic information of the dialogue, and historical diversion information of related questions.

[0041] The semantic information of the dialogue query is used to characterize the semantic features indicated by the dialogue query itself. In this embodiment, there are no specific restrictions on the method of obtaining the semantic information. For example, multiple intent recognition models provided in related technologies can be used to process the dialogue query, and the semantic information of the current dialogue query can be determined based on the intents identified by each intent recognition model.

[0042] Furthermore, the context information of the dialogue inquiry can be used to indicate the context in which the dialogue inquiry was initiated. For example, the user can be asked about their location and the weather information corresponding to that location. In this embodiment, there are no specific limitations on the context information.

[0043] Furthermore, this embodiment not only considers the semantic information of the dialogue query itself and the scenario information corresponding to the dialogue query, but also analyzes the historical routing information corresponding to the associated queries. The associated queries can be understood as dialogue queries received during historical periods that are semantically related to the dialogue query.

[0044] Furthermore, the historical routing information corresponding to related queries refers to the processing records during the processing of related queries in historical time periods. For example, the processing records may include the time consumed in processing the related query, the processing model used in processing the related query, the output of the processing model used in processing the related query, and the processing resource consumption used in processing the related query. In practical applications, the processing records of the related queries can generally be obtained from the system's stored log information. This embodiment does not impose specific restrictions on the method of obtaining the historical routing information of related queries.

[0045] For example, when the above dialogue processing method is applied to a travel scenario, the historical dialogue queries issued by the current user and related users can be considered as related queries. Related users can be understood as users with the same travel characteristics as the current user. These travel characteristics can be characterized from multiple perspectives, such as historical destinations, the number of times historical destinations were visited, and the type of historical destinations (e.g., leisure, education, office, etc.). The historical traffic diversion results can represent the processing records of the related queries during historical periods, such as the dialogue model used and the actual processing time for obtaining the feedback results.

[0046] S102. Based on the association information, determine the target model to be called this time from multiple dialogue models; wherein, the dialogue model is used to perform dialogue query processing.

[0047] For example, in this embodiment, after obtaining the aforementioned association information, a model matching the comprehension complexity indicated by the association information can be selected from multiple dialogue models as the target model to be scheduled. Here, a dialogue model can be understood as a model used to process dialogue queries to obtain feedback results.

[0048] In one example, a first complexity value can be set for each dialogue model. After acquiring the associated information, each piece of information in the acquired associated information is scored according to a pre-set scoring rule, and a second complexity value is obtained by combining the scores. Then, the final selected dialogue model is determined by comparing the second complexity value and the first complexity value. For example, the dialogue model corresponding to the first complexity value, which is greater than the second complexity value, can be used as the target model. In another example, in addition to considering the first complexity value for each dialogue model, the execution time of each dialogue model can also be considered, and the final target model is selected by combining the execution time and the complexity value.

[0049] S103. Based on the target model, process the dialogue query to obtain the processing result.

[0050] For example, after determining the target model, the received dialogue query is processed based on the determined target model to determine the processing result corresponding to the current dialogue query.

[0051] It should be noted that in practical applications, since different dialogue models have different processing rules for their corresponding input data, after determining the target model, it is also necessary to process the currently received dialogue query based on the input processing rules corresponding to the target model, and use the processed data as the input data of the target model to ensure the accuracy of the processing results.

[0052] Understandably, in this embodiment, by acquiring the association information of the received dialogue queries and filtering models among multiple dialogue models based on this association information, a suitable target model for the current dialogue query is determined as the final model to be processed, thereby improving the accuracy and timeliness of dialogue processing. This allows dialogue queries of varying complexity to be routed to different dialogue models, thus avoiding the problem in related technologies where using a single dialogue model for dialogue processing can easily lead to inaccurate processing results due to the model's inability to handle highly complex dialogues. Furthermore, it avoids the high time consumption and resource consumption associated with using only a single large model to process dialogue queries of varying complexity.

[0053] To help readers gain a deeper understanding of the implementation principles of this disclosure, the following will be discussed in conjunction with... Figures 2-3 right Figure 1 The illustrated embodiments are further refined.

[0054] Figure 2 This is a schematic diagram based on the second embodiment of the present disclosure, specifically a flowchart illustrating another traffic-sharing dialogue processing method provided by the embodiments of the present disclosure, as shown below. Figure 2 The method shown includes the following steps:

[0055] S201. In response to the received dialogue query, determine the associated information of the dialogue query; the associated information includes: the scenario information of the dialogue query, the historical routing information of the associated query, and the semantic information of the dialogue query; the scenario information is used to characterize the scenario in which the dialogue query is initiated; the associated query is a request that has a semantic relationship with the dialogue query; the historical routing information indicates the processing record information of the associated query; the semantic information characterizes the semantic features of the dialogue query.

[0056] For example, the execution subject of this embodiment can be a traffic-sharing dialogue processing device. The dialogue processing device can be a server (such as a local server or a cloud server), a computer, a processor, a chip, etc. This embodiment does not limit the scope of the device.

[0057] The technical principle of step S201 can be found in step S101, and will not be repeated here.

[0058] S202. Obtain preset rule information, wherein the preset rule information includes feature information of the dialogue model; the feature information is used to indicate the understanding complexity of the dialogue that the dialogue model can handle.

[0059] For example, in this embodiment, the preset rule information includes feature information corresponding to each dialogue model. It should be noted that the feature information corresponding to a dialogue model can indicate the complexity of understanding the dialogue that the dialogue model can accurately process. For instance, the rule information in this embodiment can be specifically configured to the scenarios and semantics that the dialogue model is suitable for processing.

[0060] S203. If it is determined that there is feature information in the preset rule information that matches the associated information, then the dialogue model corresponding to the feature information that matches the associated information is determined as the target model. The dialogue model is used for dialogue query processing.

[0061] For example, by comparing the associated information and the preset rule information, it can be determined whether the feature information that matches the associated information can be obtained based on the preset rule information, and then the target model can be determined according to the dialogue model corresponding to the matched rule information.

[0062] In one example, in practical applications, the "matching" mentioned in this embodiment can be understood as the feature information being completely contained within the feature information. For example, the scenarios of the dialogue model represented by the feature information are scenario A, scenario B, and scenario C, and the semantics that the dialogue model is applicable to process include semantics A and semantics B. If the association information represents the current dialogue query corresponding to scenario B, semantics B, and semantics C, by comparing the above association information and rule information, since the rule information does not include semantics C, it can be determined that the two do not match. If the association information represents the current dialogue query corresponding to scenario B and semantics B, by comparing the above association information and rule information, it can be determined that the two match. Furthermore, among the multiple matched dialogue models, the historical routing information of the association queries can be combined to further filter the final dialogue model.

[0063] It is understood that in this embodiment, by prioritizing the matching of current association information with pre-set preset rule information and using the dialogue model corresponding to the successfully matched feature information as the target model, the above matching method is more efficient and consumes less processing resources compared to directly processing association information based on a trained model to determine the target model.

[0064] In one example, when performing step S203, "determine that there is feature information in the preset rule information that matches the associated information", it can be achieved through the following steps: "Perform keyword matching processing on the feature information and associated information in the preset rule information to determine the matching similarity; wherein, the matching similarity is used to indicate the degree of similarity between the feature information and the associated information; if the matching similarity is greater than the preset threshold, it is determined that there is feature information in the preset rule information that matches the associated information."

[0065] For example, in this embodiment, when comparing preset rule information and related information, key information extraction processing can first be performed on the related information to filter out keyword fields contained in the related information, thereby improving the matching efficiency of the related information and rule information. Furthermore, after obtaining the matching similarity between the related information and feature information in this embodiment, the similarity between the two is determined by comparing the matching similarity with a preset threshold. When it is determined that the matching similarity is greater than the preset threshold, it can be determined that the preset rule information includes feature information that matches the current related information.

[0066] For example, the scenarios represented by the feature information in the dialogue model are scenario A, scenario B, and scenario C, and the semantics that the dialogue model is applicable to process include semantic A and semantic B. If the association information represents the current dialogue query corresponding to scenario B, semantic B, and semantic C, then semantic C can be compared with semantic A and semantic B by keyword similarity. If the matching similarity between semantic C and any of the semantics in semantic A and semantic B is greater than a preset threshold, then it can be determined that the association information and the feature information match.

[0067] It is understandable that in this embodiment, comparing keywords can improve the efficiency of comparing feature information and related information. Furthermore, this embodiment can use dialogue models with a similarity greater than a preset threshold as target models, thereby avoiding the large number of matching failures that can easily occur when comparing based on a complete inclusion method, which could lead to users not receiving timely feedback.

[0068] S204. If it is determined that there is no feature information matching the associated information in the preset rule information, then the diversion model is invoked; wherein, the diversion model is trained based on the historical processing results corresponding to the historical queries under the historical time period; the diversion model is used to determine the model to be invoked.

[0069] For example, in this embodiment, when it is determined that the feature information in the preset rule information cannot match the current association information, this embodiment will further use a pre-trained diversion model to determine the target model that needs to be called in the end.

[0070] The traffic splitting model in this embodiment is trained based on historical queries received during historical time periods and the corresponding historical processing results. In practical applications, the historical processing results may include the dialogue model invoked when processing the aforementioned historical queries, and the user's satisfaction with the feedback information obtained from the aforementioned dialogue model. Satisfaction can be indicated by whether the user accepts or rejects the information corresponding to the aforementioned dialogue model. Alternatively, satisfaction can also be determined directly by the type of feedback information; if the feedback information type is one that cannot understand the user's intent, it can be directly determined as dissatisfaction.

[0071] S205. Based on the traffic splitting model, process the associated information to obtain the target model that needs to be called this time.

[0072] For example, in this embodiment, after obtaining the splitting model, the association information can be directly used as the input data corresponding to the splitting model, and the model indicated by the model output result obtained by the splitting model based on the above association information can be determined as the target model.

[0073] Understandably, in this embodiment, to avoid the problem of not being able to provide more accurate processing feedback to users when matching target models based solely on rule information is limited by the rule information configuration and a matching failure occurs, a pre-trained traffic splitting model can be used to determine the target model. It is understood that the traffic splitting model, through training, learns the model selection results indicated by the rule information and the user's satisfaction with the feedback information after the final model selection, continuously improving its own traffic splitting processing. This allows the traffic splitting model to accurately determine the target model for various scenarios (i.e., not limited to the configured rule information), facilitating accurate and timely feedback of processing results to the user.

[0074] S206. Based on the target model, process the dialogue query to obtain the processing result.

[0075] For example, the technical principle of step S206 can be found in step S103, and will not be repeated here.

[0076] In this embodiment, by prioritizing the matching of current association information with pre-set rule information and using the dialogue model corresponding to the successfully matched feature information as the target model, this matching method is more efficient and consumes fewer processing resources compared to directly processing association information based on a trained model to determine the target model. Furthermore, to avoid scenarios where matching the target model based on rule information fails, the target model is further determined by combining it with a trained triage model, which facilitates accurate and timely feedback of the processing results to the user.

[0077] Figure 3 This is a schematic diagram based on the third embodiment of the present disclosure, specifically a flowchart illustrating another traffic-splitting dialogue processing method provided by the embodiments of the present disclosure, as shown below. Figure 3 The method shown includes the following steps:

[0078] S301. In response to the received dialogue query, determine the time when the terminal device sends the dialogue query.

[0079] For example, the execution subject of this embodiment can be a traffic-sharing dialogue processing device. The dialogue processing device can be a server (such as a local server or a cloud server), a computer, a processor, a chip, etc. This embodiment does not limit the scope of the device.

[0080] This embodiment provides a method for determining scene information in the associated information of a dialogue inquiry. The historical routing information of the associated inquiry and the semantic information of the dialogue inquiry mentioned in the associated information can be referred to the description in step S101. The scene information is used to characterize the scene in which the dialogue inquiry is initiated; the associated inquiry is a request that has a semantic connection with the dialogue inquiry; the historical routing information indicates the processing record information of the associated inquiry; and the semantic information characterizes the semantic features of the dialogue inquiry.

[0081] Furthermore, in this embodiment, the dialogue query is sent by the terminal device. When determining the scenario information corresponding to the dialogue query, the sending time of the dialogue query by the terminal device can be determined first. It should be noted that when determining the sending time, the sending time can be included in the dialogue query, or it can be based on the time when the dialogue query is received. Alternatively, the sending time can be determined as the difference between the time of receiving the dialogue information and the estimated time of sending the dialogue query.

[0082] In one example, the related query includes: previous conversation information, and / or, historical conversation information;

[0083] The above dialogue information refers to dialogues initiated by terminal devices within adjacent time periods; the terminal device is the device that initiates the dialogue inquiry; the adjacent time period is the time period adjacent to the time when the terminal device sends the dialogue inquiry; the historical dialogue information refers to dialogue information received in historical time periods whose semantic similarity to the dialogue inquiry is greater than a preset value.

[0084] For example, in this embodiment, when determining the historical routing information of associated queries in the associated information, it is first necessary to be able to determine the associated queries related to the currently received dialogue query. Specifically, when determining the associated queries, since there is a contextual relationship between adjacent dialogue statements in a round-robin dialogue when a user initiates multiple rounds of dialogue in a short period of time, in this embodiment, the preceding dialogue information corresponding to the adjacent time period adjacent to the sending time of the currently received dialogue query (i.e., dialogue queries sent by the same terminal device in adjacent time periods) can be used as the associated queries here.

[0085] In addition, based on the dialogue query received this time, historical dialogue information with a semantic similarity greater than a preset value can be searched among the dialogue queries received in the historical time period before the current time, and used as the related query here.

[0086] It should be noted that in practical applications, related queries can include one or more of the following: historical dialogue information and previous dialogue information, in order to improve the accuracy of the target model determined subsequently.

[0087] In one example, historical triage information includes: the dialogue model that processed the related queries, and / or, satisfaction information; wherein the satisfaction information is used to indicate satisfaction with the query result; the query result is the result obtained after processing the related queries based on the dialogue model.

[0088] For example, in this embodiment, the historical triage information of the related queries included in the related information may specifically be the dialogue model for processing the aforementioned related queries, and / or the satisfaction level of the query results corresponding to the related queries in historical cases.

[0089] For example, when the related query is the preceding dialogue information, the dialogue model corresponding to the processing of the preceding dialogue information can be used as a factor in determining the target model. For instance, the same dialogue model can be prioritized for processing multiple rounds of dialogue from the same terminal device to ensure the consistency of model processing. Alternatively, when the related query is the preceding dialogue information, the dialogue model corresponding to the processing of the preceding dialogue information and the satisfaction level of the query result can be considered as factors in determining the target model. For example, if the terminal device sends the same dialogue query multiple times, and the text content of the two dialogue queries has a high degree of similarity, it can be determined to be the same dialogue query, indicating that the user is dissatisfied with the feedback corresponding to the previous model. Therefore, the aforementioned satisfaction information and the previously used dialogue model can be considered as factors in determining the target model. This is to further improve the accuracy of the subsequent target model and thus improve the accuracy of the subsequent processing results.

[0090] Alternatively, when the associated query is historical dialogue information, and there are multiple historical dialogue information, if it is determined that the dialogue models corresponding to each historical dialogue information are different in the historical period, but the satisfaction level corresponding to each historical dialogue information represents satisfaction, then the dialogue model with the lower processing time among the dialogue models corresponding to the above historical dialogue information can be used as the candidate model for the current target model selection, so that the target model determined later can consume less processing resources and processing time.

[0091] It is understood that in this embodiment, combining the dialogue model indicated in the historical triage information and the satisfaction information as reference factors for determining the target model is beneficial to improving the accuracy and timeliness of subsequent processing results.

[0092] In one example, the semantic information of a dialogue query includes at least one of the following: length information, keyword information, intent information, and slot information; the length information is the text length corresponding to the dialogue query; the keyword information is used to indicate the key content in the dialogue query; the intent information represents the purpose of the dialogue query; and the slot information is used to indicate the key information needed to achieve the purpose of the dialogue query.

[0093] For example, in this embodiment, semantic information corresponding to the dialogue query can be obtained from multiple different dimensions, such as length information representing the text length of the text corresponding to the dialogue query, and keyword information (e.g., time, location, etc.) representing the keywords contained in the dialogue query.

[0094] In addition, it may also include the intent information of the current dialogue query, that is, the purpose of the dialogue query (e.g., finding a destination that meets the user's requirements, planning a route to the destination, etc.). It should be noted that the intent information in this embodiment can be determined solely based on the currently received dialogue query, or it can be determined in combination with the preceding query in the dialogue query; no specific limitation is made in this embodiment.

[0095] In addition, the semantic information may also include slot information corresponding to the dialogue query. Slot information can be understood as the key information needed to realize the intent corresponding to the dialogue query. For example, when the dialogue query is "What will the weather be like tomorrow?", the corresponding slot information may include the date of tomorrow, the weather, and the location of the user who made the dialogue query, so as to complete the corresponding weather information query later.

[0096] It should be noted that in practical applications, semantic information may include one or more of the following: length information, keyword information, intent information, and slot information. This embodiment does not impose specific limitations. Specifically, the content corresponding to the semantic information can be increased or decreased according to the actual processing resource consumption.

[0097] Understandably, analyzing the semantic information of the dialogue itself and using it as a factor in determining the target model helps ensure that the determined model can accurately understand the user needs indicated by the dialogue, thereby improving the accuracy of the final processing results.

[0098] S302. Determine the display data of the terminal device's display interface at the transmission time; wherein the display data includes: the interface identifier of the display interface, and / or, the display information in the display interface; the display information is used to indicate the display content in the display interface.

[0099] For example, after determining the sending time, the display data displayed on the terminal device's interface at that sending time can be further obtained. It should be noted that in this embodiment, the display data corresponding to the terminal device can be obtained from the terminal device after receiving the dialogue query. In another possible case, it can also be data actively reported by the terminal device when sending the dialogue query.

[0100] Furthermore, the display data obtained in this embodiment may include an interface identifier corresponding to the display interface, wherein the interface identifier is used to uniquely identify the displayed interface. Alternatively, it may also include display information displayed in the display interface, that is, the content displayed in the display interface.

[0101] For example, let's take a terminal device with a travel application as an example. The travel application can have multiple display interfaces, such as a main page, a destination search page, a navigation route search page, and a navigation page, each with its own interface identifier. By providing feedback on these page identifiers, we can characterize the current user scenario when using the travel application.

[0102] Furthermore, in addition to using interface identifiers as display data, the actual content displayed on the interface can also be used as display data.

[0103] For example, when a user is on a destination search page but has not yet entered a destination or destination filter, only the interface icon corresponding to the destination search page can be used as the display data. When the destination search page displays an entered destination or destination filter, both the interface icon of the destination search page and the entered and displayed content can be used as the display data; alternatively, only the content currently displayed on the interface can be used as the display data.

[0104] S303. Determine that the displayed data is the information included in the scenario information of the dialogue inquiry.

[0105] For example, after determining the display data, the display data can be identified as the scene information corresponding to the aforementioned dialogue query.

[0106] It is understood that in this embodiment, the display data in the interface corresponding to the terminal device at the time of sending is used as the information included in the scene information, so that the obtained display data can be combined to accurately match the target model suitable for handling the current dialogue inquiry, thereby improving the accuracy of the subsequent push processing results.

[0107] S304. If it is determined that there is a preceding dialogue information initiated by the terminal device, then obtain the preceding feedback result of the preceding dialogue information; the preceding dialogue information is a dialogue initiated within an adjacent time period; the adjacent time period is the time period adjacent to the sending time; the preceding feedback result is the result fed back to the terminal device after the preceding dialogue information has been processed.

[0108] For example, in this embodiment, after obtaining the aforementioned sending time, it is further determined whether the terminal device has sent other dialogues (i.e., the aforementioned dialogue information) in the adjacent time period before sending the dialogue inquiry, based on the obtained sending time. If the aforementioned dialogue information exists, it indicates that the current situation is a multi-turn dialogue scenario, and the aforementioned feedback result corresponding to the aforementioned dialogue information is further obtained.

[0109] S305. Determine the information included in the dialogue and feedback results above as contextual information for the dialogue inquiry.

[0110] For example, the obtained dialogue information and feedback results can also be included in the scenario information of the dialogue inquiry currently received in this embodiment.

[0111] It should be noted that in practical applications, steps S304-S305 and steps S302-S303 can be selectively executed. That is, the scene information may only include the display data determined in steps S302-S303, or only the contextual dialogue information and contextual feedback information determined in steps S304-S305. Alternatively, all steps S302-S305 may be executed. Figure 3 The execution order shown is for illustrative purposes only. In practical applications, execution can also be performed in parallel. No specific restrictions are imposed in this embodiment.

[0112] It is understood that in this embodiment, the scene information may include not only the interface display data at the time of sending, but also, if it is determined that the above dialogue information was sent in the adjacent time period before the dialogue inquiry was sent, since the dialogue inquiries sent by the same terminal device in adjacent time periods usually have contextual relationships, this embodiment can also combine the above dialogue information and the feedback results of the above dialogue information to further determine the target model applicable to the current dialogue inquiry, so as to improve the accuracy of the target model determination.

[0113] S306. Based on the associated information, determine the target model to be invoked in this instance from among multiple dialogue models; wherein, the dialogue model is used for dialogue query processing.

[0114] S307. Based on the target model, process the dialogue query to obtain the processing result.

[0115] For example, steps S306-S307 in this embodiment can be referred to steps S102-S103, and will not be repeated here.

[0116] In this embodiment, the display data of the terminal device at the time of transmission is combined with the information contained in the scene information so that the obtained display data can be used as a reference to identify the user's intent and accurately match the target model suitable for handling the current dialogue inquiry.

[0117] Figure 4 This is a schematic diagram of the fourth embodiment of the present disclosure, as shown below. Figure 4 As shown, this embodiment provides a traffic splitting dialog processing device 400, which includes:

[0118] The first determining unit 401 is used to determine the associated information of the dialogue inquiry in response to the received dialogue inquiry. The associated information includes: the scenario information of the dialogue inquiry, the historical routing information of the associated inquiry, and the semantic information of the dialogue inquiry. The scenario information is used to characterize the scenario in which the dialogue inquiry is initiated. The associated inquiry is a request that has a semantic relationship with the dialogue inquiry. The historical routing information indicates the processing record information of the associated inquiry. The semantic information characterizes the semantic features of the dialogue inquiry.

[0119] The second determining unit 402 is used to determine the target model to be invoked in this instance from multiple dialogue models based on the associated information; wherein, the dialogue model is used for dialogue query processing.

[0120] The processing unit 403 is used to process the dialogue query according to the target model and obtain the processing result.

[0121] The apparatus provided in this embodiment is used to implement the technical solution provided by the above method. Its implementation principle and technical effect are similar, and will not be described again.

[0122] Figure 5 This is a schematic diagram of the fifth embodiment of the present disclosure, as shown below. Figure 5 As shown, this embodiment provides a traffic-splitting dialogue processing device 500, including:

[0123] The first determining unit 501 is used to determine the associated information of the dialogue inquiry in response to the received dialogue inquiry; the associated information includes: the scenario information of the dialogue inquiry, the historical routing information of the associated inquiry, and the semantic information of the dialogue inquiry; the scenario information is used to characterize the scenario in which the dialogue inquiry is initiated; the associated inquiry is a request that has a semantic relationship with the dialogue inquiry; the historical routing information indicates the processing record information of the associated inquiry; the semantic information characterizes the semantic features of the dialogue inquiry.

[0124] The second determining unit 502 is used to determine the target model to be invoked in this instance from multiple dialogue models based on the associated information; wherein, the dialogue model is used for dialogue query processing.

[0125] The processing unit 503 is used to process the dialogue query according to the target model and obtain the processing result.

[0126] In one example, the second determining unit 502 includes:

[0127] The first acquisition module 5021 is used to acquire preset rule information, wherein the preset rule information includes feature information of the dialogue model; the feature information is used to indicate the understanding complexity of the dialogue that the dialogue model can handle.

[0128] The first determining module 5022 is used to determine the dialogue model corresponding to the feature information that matches the associated information as the target model if the preset rule information contains feature information that matches the associated information.

[0129] In one example, the first determining module 5022 includes:

[0130] The first determining submodule 50221 is used to perform keyword matching processing on the feature information and related information in the preset rule information to determine the matching similarity; wherein, the matching similarity is used to indicate the degree of similarity between the feature information and the related information.

[0131] The second determining submodule 50222 is used to determine, if the matching similarity is greater than a preset threshold, that there is feature information in the preset rule information that matches the associated information.

[0132] The third determination submodule 50223 is used to determine the dialogue model corresponding to the feature information that matches the associated information as the target model.

[0133] In one example, the device also includes:

[0134] The second determining module 5023 is used to call the diversion model if the preset rule information does not contain feature information that matches the associated information; wherein, the diversion model is trained based on the historical processing results corresponding to the historical queries under the historical time period; the diversion model is used to determine the model to be called.

[0135] The processing module 5024 is used to process the associated information according to the diversion model to obtain the target model that needs to be called this time.

[0136] In one example, the first determining unit 501 includes:

[0137] The third determining module 5011 is used to determine the time when the terminal device sends the dialogue query in response to the received dialogue query.

[0138] The fourth determination module 5012 is used to determine the historical diversion information of the associated queries.

[0139] The fifth determining module 5013 is used to determine the semantic information of the dialogue query.

[0140] The sixth determining module 5014 is used to determine the display data of the terminal device's display interface at the transmission time; wherein, the display data includes: the interface identifier of the display interface, and / or, the display information in the display interface; the display information is used to indicate the display content in the display interface.

[0141] The seventh determination module 5015 is used to determine the information included in the scenario information of the dialogue inquiry in the displayed data.

[0142] In one example, the device also includes:

[0143] The second acquisition module 5016 is used to acquire the feedback result of the preceding dialogue information if it is determined that there is preceding dialogue information initiated by the terminal device; the preceding dialogue information is a dialogue initiated within an adjacent time period; the adjacent time period is the time period adjacent to the sending time; the preceding feedback result is the result of the preceding dialogue information being processed and fed back to the terminal device.

[0144] The eighth determining module 5017 is used to determine the information included in the preceding dialogue information and the preceding feedback result as the scenario information of the dialogue inquiry.

[0145] In one example, the related query includes: previous conversation information, and / or, historical conversation information;

[0146] The above dialogue information refers to dialogues initiated by terminal devices within adjacent time periods; the terminal device is the device that initiates the dialogue inquiry; the adjacent time period is the time period adjacent to the time when the terminal device sends the dialogue inquiry.

[0147] Historical dialogue information refers to dialogue information received during a historical period whose semantic similarity to the dialogue query is greater than a preset value.

[0148] In one example, historical triage information includes: the dialogue model that processed the related queries, and / or, satisfaction information; wherein the satisfaction information is used to indicate satisfaction with the query result; the query result is the result obtained after processing the related queries based on the dialogue model.

[0149] In one example, the semantic information of a dialogue query includes at least one of the following: length information, keyword information, intent information, and slot information; the length information is the text length corresponding to the dialogue query; the keyword information is used to indicate the key content in the dialogue query; the intent information represents the purpose of the dialogue query; and the slot information is used to indicate the key information needed to achieve the purpose of the dialogue query.

[0150] The apparatus provided in this embodiment is used to implement the technical solution provided by the above method. Its implementation principle and technical effect are similar, and will not be described again.

[0151] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0152] This disclosure provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method provided in any of the above embodiments.

[0153] Figure 6 This is a schematic diagram based on the sixth embodiment of the present disclosure, as shown below. Figure 6 As shown, the electronic device 600 in this disclosure may include a processor 601 and a memory 602.

[0154] Memory 602 is used to store programs. Memory 602 may include volatile memory, such as random-access memory (RAM), such as static random-access memory (SRAM), double data rate synchronous dynamic random-access memory (DDR SDRAM), etc.; memory may also include non-volatile memory, such as flash memory. Memory 602 is used to store computer programs (such as application programs, functional modules, etc. that implement the above methods), computer instructions, etc. The computer programs, computer instructions, etc., can be partitioned and stored in one or more memories 602. Furthermore, the computer programs, computer instructions, data, etc., can be accessed by processor 601.

[0155] The aforementioned computer programs and instructions can be stored in one or more partitions of memory 602. Furthermore, the aforementioned computer programs and instructions can be invoked by processor 601.

[0156] The processor 601 is configured to execute the computer program stored in the memory 602 to implement the various steps in the methods described in the above embodiments.

[0157] For details, please refer to the relevant descriptions in the preceding method embodiments.

[0158] The processor 601 and the memory 602 can be independent structures or integrated structures. When the processor 601 and the memory 602 are independent structures, the memory 602 and the processor 601 can be coupled together via bus 603.

[0159] The electronic device in this embodiment can execute the technical solution in the above method. Its specific implementation process and technical principle are the same, and will not be repeated here.

[0160] This disclosure provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to perform the methods provided in any of the above embodiments.

[0161] According to embodiments of this disclosure, this disclosure also provides a computer program product comprising: a computer program stored in a readable storage medium, at least one processor of an electronic device being able to read the computer program from the readable storage medium, and the at least one processor executing the computer program causing the electronic device to perform the scheme provided in any of the above embodiments.

[0162] Figure 7 A schematic block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0163] like Figure 7 As shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 702 or a computer program loaded from storage unit 708 into random access memory (RAM) 703. RAM 703 may also store various programs and data required for the operation of device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0164] Multiple components in device 700 are connected to I / O interface 705, including: input unit 706, such as keyboard, mouse, etc.; output unit 707, such as various types of monitors, speakers, etc.; storage unit 708, such as disk, optical disk, etc.; and communication unit 709, such as network card, modem, wireless transceiver, etc. Communication unit 709 allows device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0165] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as the offloading dialogue processing method. For example, in some embodiments, the offloading dialogue processing method can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the offloading dialogue processing method described above can be performed. Alternatively, in other embodiments, the computing unit 701 can be configured to perform the offloading dialogue processing method by any other suitable means (e.g., by means of firmware).

[0166] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0167] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0168] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0169] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0170] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0171] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service ecosystem, addressing the shortcomings of traditional physical hosts and VPS (Virtual Private Server, or simply "VPS") services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0172] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0173] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for splitting dialogue processing, comprising: In response to a received dialogue query, determine the associated information of the dialogue query; The associated information includes: scenario information of the dialogue inquiry, historical routing information of associated inquiries, and semantic information of the dialogue inquiry; the scenario information is used to characterize the scenario in which the dialogue inquiry is initiated; the associated inquiry is a request that has semantic relevance to the dialogue inquiry; the historical routing information indicates the processing record information of the associated inquiry; the semantic information characterizes the semantic features of the dialogue inquiry; and the associated information is used to characterize the comprehension complexity of the dialogue inquiry. Based on the associated information, the target model to be invoked in this instance is determined from multiple dialogue models; wherein, the dialogue model is used for dialogue query processing; Based on the target model, the dialogue query is processed to obtain the processing result; Based on the aforementioned association information, the target model to be invoked in this instance is determined from multiple dialogue models, including: Obtain preset rule information, wherein the preset rule information includes feature information corresponding to each of the plurality of dialogue models; the feature information is used to indicate the understanding complexity of the dialogue that the dialogue model can process; If it is determined that there is feature information in the preset rule information that matches the association information, then the dialogue model corresponding to the feature information that matches the association information is determined as the target model; If it is determined that there is no feature information in the preset rule information that matches the associated information, then the diversion model is invoked; wherein, the diversion model is trained based on the historical processing results corresponding to historical queries under historical time periods; the diversion model is used to determine the model to be invoked; Based on the aforementioned traffic splitting model, the associated information is processed to obtain the target model that needs to be called this time.

2. The method according to claim 1, wherein, Determining that the preset rule information contains feature information that matches the associated information includes: Keyword matching is performed on the feature information and the associated information in the preset rule information to determine the matching similarity; wherein, the matching similarity is used to indicate the degree of similarity between the feature information and the associated information; If the matching similarity is greater than a preset threshold, then it is determined that there is feature information in the preset rule information that matches the associated information.

3. The method according to any one of claims 1-2, wherein, Determining the associated information in the dialogue query includes: Determine the time at which the terminal device sends the dialogue query; Determine the display data of the terminal device's display interface at the transmission time; wherein the display data includes: the interface identifier of the display interface, and / or, the display information in the display interface; the display information is used to indicate the display content in the display interface; The displayed data is determined to be the information included in the scenario information of the dialogue inquiry.

4. The method according to claim 3, further comprising: If it is determined that there is a preceding dialogue message initiated by the terminal device, then the preceding feedback result of the preceding dialogue message is obtained; The above-mentioned dialogue information refers to dialogues initiated within adjacent time periods; The adjacent time period is the time period adjacent to the transmission time; The above feedback result is the result of processing the above dialogue information and sending it back to the terminal device. The aforementioned dialogue information and the aforementioned feedback result are determined to be information included in the scenario information of the dialogue inquiry.

5. The method according to any one of claims 1-2 and 4, wherein, The associated queries include: previous dialogue information, and / or, historical dialogue information; The above-mentioned dialogue information refers to dialogues initiated by terminal devices within adjacent time periods; the terminal device is the device that initiates the dialogue inquiry; the adjacent time period is the time period adjacent to the time when the terminal device sends the dialogue inquiry; The historical dialogue information refers to dialogue information received during a historical period whose semantic similarity to the dialogue query is greater than a preset value.

6. The method according to any one of claims 1-2, 4, wherein, The historical triage information includes: a dialogue model for processing related queries, and / or, satisfaction information; wherein the satisfaction information is used to indicate satisfaction with the query result; the query result is the result obtained after processing the related query based on the dialogue model.

7. The method according to any one of claims 1-2 and 4, wherein, The semantic information of the dialogue query includes at least one of the following: length information, keyword information, intent information, and slot information; the length information is the text length corresponding to the dialogue query; the keyword information is used to indicate the key content in the dialogue query; the intent information represents the purpose of the dialogue query; and the slot information is used to indicate the key information needed to achieve the purpose of the dialogue query.

8. A traffic-sharing dialogue processing device, comprising: The first determining unit is configured to determine the associated information of the dialogue inquiry in response to the received dialogue inquiry; The associated information includes: scenario information of the dialogue inquiry, historical routing information of associated inquiries, and semantic information of the dialogue inquiry; the scenario information is used to characterize the scenario in which the dialogue inquiry is initiated; the associated inquiry is a request that has semantic relevance to the dialogue inquiry; the historical routing information indicates the processing record information of the associated inquiry; the semantic information characterizes the semantic features of the dialogue inquiry; and the associated information is used to characterize the understanding complexity of the dialogue inquiry. The second determining unit is used to determine the target model to be invoked in this instance from multiple dialogue models based on the associated information; wherein the dialogue model is used for dialogue query processing. The processing unit is used to process the dialogue query according to the target model and obtain the processing result; The second determining unit includes: The first acquisition module is used to acquire preset rule information, wherein the preset rule information includes feature information corresponding to each of the plurality of dialogue models; the feature information is used to indicate the understanding complexity of the dialogue that the dialogue model can process. The first determining module is used to determine the dialogue model corresponding to the feature information that matches the association information as the target model if it is determined that there is feature information in the preset rule information that matches the association information. The second determining module is used to invoke a traffic splitting model if it is determined that there is no feature information matching the associated information in the preset rule information; wherein the traffic splitting model is trained based on the historical processing results corresponding to historical queries in historical time periods; the traffic splitting model is used to determine the model to be invoked. The processing module is used to process the associated information according to the traffic splitting model to obtain the target model that needs to be called this time.

9. The apparatus according to claim 8, wherein, The first determining module includes: The first determining submodule is used to perform keyword matching processing on the feature information and the associated information in the preset rule information to determine the matching similarity; wherein, the matching similarity is used to indicate the degree of similarity between the feature information and the associated information; The second determining submodule is used to determine, if the matching similarity is greater than a preset threshold, that there is feature information in the preset rule information that matches the association information; The third determining submodule is used to determine the dialogue model corresponding to the feature information that matches the associated information as the target model.

10. The apparatus according to any one of claims 8-9, wherein, The first defined unit includes: The third determining module is used to determine the time when the terminal device sends the dialogue query in response to the received dialogue query; The fourth determination module is used to determine the historical routing information of related queries; The fifth determining module is used to determine the semantic information of the dialogue query; The sixth determining module is used to determine the display data of the terminal device's display interface at the transmission time; wherein, the display data includes: the interface identifier of the display interface, and / or, the display information in the display interface; the display information is used to indicate the display content in the display interface; The seventh determining module is used to determine that the displayed data is the information included in the scenario information of the dialogue inquiry.

11. The apparatus of claim 10, further comprising: The second acquisition module is used to acquire the previous feedback result of the previous dialogue information if it is determined that there is previous dialogue information initiated by the terminal device. The above-mentioned dialogue information refers to dialogues initiated within adjacent time periods; The adjacent time period is the time period adjacent to the transmission time; The above feedback result is the result of processing the above dialogue information and sending it back to the terminal device. The eighth determining module is used to determine that the preceding dialogue information and the preceding feedback result are information included in the scenario information of the dialogue inquiry.

12. The apparatus according to any one of claims 8-9 and 11, wherein, The associated queries include: previous dialogue information, and / or, historical dialogue information; The above-mentioned dialogue information refers to dialogues initiated by terminal devices within adjacent time periods; the terminal device is the device that initiates the dialogue inquiry; the adjacent time period is the time period adjacent to the time when the terminal device sends the dialogue inquiry; The historical dialogue information refers to dialogue information received during a historical period whose semantic similarity to the dialogue query is greater than a preset value.

13. The apparatus according to any one of claims 8-9 and 11, wherein, The historical triage information includes: a dialogue model for processing related queries, and / or, satisfaction information; wherein the satisfaction information is used to indicate satisfaction with the query result; the query result is the result obtained after processing the related query based on the dialogue model.

14. The apparatus according to any one of claims 8-9, 11, wherein, The semantic information of the dialogue query includes at least one of the following: length information, keyword information, intent information, and slot information; the length information is the text length corresponding to the dialogue query; the keyword information is used to indicate the key content in the dialogue query; the intent information represents the purpose of the dialogue query; and the slot information is used to indicate the key information needed to achieve the purpose of the dialogue query.

15. An electronic device comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

16. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-7.

17. A computer program product comprising a computer program that, when executed by a processor, implements the steps of the method according to any one of claims 1-7.

Images