Task execution method and device, electronic equipment and storage medium

By obtaining user's question text and business status information, and using prediction model and dialogue generation module to generate target reply text, the problem of low response accuracy of intelligent customer service is solved and more accurate user answers are achieved.

CN118917414APending Publication Date: 2024-11-08ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202410918053.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-09
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The accuracy of smart customer service generates reply text is low and cannot effectively answer users' questions.

Method used

By obtaining the user's problem text data and business status information, using a preset prediction model to predict the service target and its sequential relationship, the dialogue generation module is called to generate sub-reply text, and combined it into the target reply text to display it to the user.

Benefits of technology

It improves the accuracy of smart customer service to generate reply text, can answer user questions more accurately, and improves user experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118917414A_ABST
    Figure CN118917414A_ABST
Patent Text Reader

Abstract

The invention discloses a task execution method and device, electronic equipment and a storage medium, and can obtain supplementary data used for representing the current service state of a user while obtaining question text data sent by the user. Therefore, each service target required by the user can be determined through the prediction model according to the question text data sent by the user and the supplementary data corresponding to the question text data, and then a sub-reply text corresponding to each service target can be generated through the dialogue generation module; and the sub reply texts are combined into the target reply text replied to the user, so that the accuracy of the generated reply text is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present specification relates to the field of natural language processing technology, and in particular to a task execution method, device, electronic device and storage medium. Background Art

[0002] Intelligent customer service is a customer service system based on artificial intelligence technology. It can automatically, efficiently and accurately interact with users by simulating human dialogue to solve various questions raised by users, provide required information, perform related operations or guide users to complete specific tasks. For example, in financial services, risk warnings, permission restrictions and other strategies are usually adopted to protect users' personal privacy data. When users receive risk warnings or permission restrictions, they will inevitably have a series of questions about the current business status. At this time, users can seek answers to these questions through intelligent customer service.

[0003] Generally, when solving user questions through intelligent customer service, a large number of manual rules (such as polite language specifications, reply templates corresponding to keywords, session termination conditions, etc.) need to be set in advance so that the intelligent customer service can reply based on these manual rules. However, the accuracy of the reply text generated by the intelligent customer service in the above way is often low, resulting in the inability to accurately answer user questions.

[0004] Therefore, how to improve the accuracy of the reply texts generated by intelligent customer service is an urgent problem to be solved. Summary of the invention

[0005] This specification provides a task execution method, device, electronic device and storage medium to solve the problem of low response accuracy in the prior art.

[0006] This manual adopts the following technical solutions:

[0007] This specification provides a task execution method, including:

[0008] Get the question text data sent by the user;

[0009] According to the user identifier based on which the user sends the question text data, query the service status information of the user, so as to determine the supplementary data corresponding to the question text data according to the service status information, wherein the service status information is used to characterize the current service status of the user;

[0010] Inputting the question text data and the supplementary data into a preset prediction model, so as to predict the service objectives and the order relationship between the service objectives under the current business state of the user through the prediction model according to the question text data and the supplementary data;

[0011] By calling a preset dialogue generation module, a sub-reply text corresponding to each service target is generated in sequence according to the sequence relationship;

[0012] According to the sequential relationship, the sub-reply texts are combined into a target reply text, and the target reply text is displayed to the user for task execution.

[0013] Optionally, the question text data and the supplementary data are input into a preset prediction model, so that the prediction model predicts the service objectives and the order relationship between the service objectives under the current business state of the user according to the question text data and the supplementary data, specifically including:

[0014] Determine at least one intention tag matching the question text data from the preset intention tags as a first intention tag combination, and determine at least one intention tag matching the supplementary data from the preset intention tags as a second intention tag combination;

[0015] According to the first intention label combination and the second intention label combination, the intention vector corresponding to the question text data is determined, and the intention vector is input into a preset prediction model so that the prediction model can predict the service objectives for the user's current business state and the sequence relationship between the service objectives based on the intention vector.

[0016] Optionally, predicting each service target for the user's current business state based on the question text data and the supplementary data through the prediction model specifically includes:

[0017] The prediction model predicts the service objectives involved in answering the question text data and the supplementary data, as well as the service objectives involved in other question text data sent by the user after answering the question text data.

[0018] Optionally, the question text data and the supplementary data are input into a preset prediction model, so that the prediction model predicts the service objectives and the order relationship between the service objectives under the current business state of the user according to the question text data and the supplementary data, specifically including:

[0019] Acquire a sample conversation record, wherein the sample conversation record includes historical question text data sent by a user in the past and a standard reply text for the historical question text data;

[0020] The sample conversation record, the question text data and the supplementary data are input into a preset prediction model so that the prediction model can predict the service objectives for the user's current business status and the sequence relationship between the service objectives based on the sample conversation record, the question text data and the supplementary data.

[0021] Optionally, predicting each service target for the user's current business state based on the question text data and the supplementary data through the prediction model specifically includes:

[0022] Predicting each candidate service target group corresponding to the question text data according to the question text data and the supplementary data by the prediction model, wherein for each candidate service target group, the candidate service target group includes each service target corresponding to the predicted question text data;

[0023] According to the priority of each service target included in each candidate service target group, a service target group to be output is screened out from each candidate service target group;

[0024] The service targets included in the service target group to be output are used as predicted service targets for the current service state of the user.

[0025] Optionally, by calling a preset dialogue generation module, a sub-reply text corresponding to each service target is generated in sequence according to the sequence relationship, specifically including:

[0026] By calling the preset dialogue generation module, the sub-reply text corresponding to each service target is generated in sequence according to the sequence relationship and the preset reply templates. The reply template is pre-constructed according to the historical conversation records, and the historical conversation records contain the historical question text data sent by the user in history, and the historical reply text for the historical question text data.

[0027] Optionally, according to the sequence relationship, the sub-reply texts are combined into a target reply text, and the target reply text is displayed to the user to perform the task, specifically including:

[0028] By means of the dialogue generation module, for at least part of the sub-reply text, a transition text corresponding to the at least part of the sub-reply text is generated;

[0029] According to the sequential relationship, each sub-reply text and the transition text corresponding to at least part of the sub-reply text are combined into a target reply text, and the target reply text is displayed to the user for task execution.

[0030] This specification provides a task execution device, including:

[0031] The acquisition module is used to obtain the question text data sent by the user;

[0032] A query module, used to query the service status information of the user based on the user identifier based on which the user sends the question text data, so as to determine the supplementary data corresponding to the question text data according to the service status information, wherein the service status information is used to characterize the current service status of the user;

[0033] A prediction module, used for inputting the question text data and the supplementary data into a preset prediction model, so as to predict the service objectives and the order relationship between the service objectives under the current business state of the user through the prediction model according to the question text data and the supplementary data;

[0034] A generation module, configured to generate sub-reply texts corresponding to each service target in sequence by calling a preset dialogue generation module according to the sequence relationship;

[0035] An execution module is used to combine the sub-reply texts into a target reply text according to the sequence relationship, and display the target reply text to the user to perform the task.

[0036] Optionally, the prediction module is specifically used to determine at least one intention label matching the question text data from the preset intention labels as a first intention label combination, and to determine at least one intention label matching the supplementary data from the preset intention labels as a second intention label combination;

[0037] According to the first intention label combination and the second intention label combination, the intention vector corresponding to the question text data is determined, and the intention vector is input into a preset prediction model so that the prediction model can predict the service objectives for the user's current business state and the sequence relationship between the service objectives based on the intention vector.

[0038] Optionally, the prediction module is specifically used to predict, through the prediction model based on the question text data and the supplementary data, the service objectives involved in replying to the question text data, as well as the service objectives involved in other question text data sent by the user after replying to the question text data.

[0039] Optionally, the prediction module is specifically used to obtain a sample conversation record, wherein the sample conversation record includes historical question text data sent by the user in the past and a standard reply text for the historical question text data;

[0040] The sample conversation record, the question text data and the supplementary data are input into a preset prediction model so that the prediction model can predict the service objectives for the user's current business status and the sequence relationship between the service objectives based on the sample conversation record, the question text data and the supplementary data.

[0041] Optionally, the prediction module is specifically used to predict, through the prediction model, each candidate service target group corresponding to the question text data according to the question text data and the supplementary data, and for each candidate service target group, the candidate service target group includes each service target corresponding to the predicted question text data;

[0042] According to the priority of each service target included in each candidate service target group, a service target group to be output is screened out from each candidate service target group;

[0043] The service targets included in the service target group to be output are used as predicted service targets for the current service state of the user.

[0044] Optionally, the generation module is specifically used to generate sub-reply texts corresponding to each service target in sequence according to the chronological relationship and preset reply templates by calling a preset dialogue generation module, wherein the reply templates are pre-constructed based on historical conversation records, and the historical conversation records contain historical question text data sent by users in history, as well as historical reply texts for the historical question text data.

[0045] This specification provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned task execution method is implemented.

[0046] This specification provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned task execution method when executing the program.

[0047] At least one of the above technical solutions adopted in this specification can achieve the following beneficial effects:

[0048] In the task execution method provided in the present specification, the question text data sent by the user is first obtained, and the business status information of the user is queried based on the user identifier based on which the user sends the question text data, so as to determine the supplementary data corresponding to the question text data based on the business status information, wherein the business status information is used to characterize the current business status of the user, and then the question text data and the supplementary data are input into a preset prediction model, so as to predict the service goals and the sequence relationship between the service goals under the current business status of the user based on the question text data and the supplementary data through the prediction model, and by calling the preset dialogue generation module, the sub-reply text corresponding to each service goal is generated in turn according to the sequence relationship, and the sub-reply texts are combined into the target reply text according to the sequence relationship, and the target reply text is displayed to the user for task execution.

[0049] It can be seen from the above method that while obtaining the question text data sent by the user, supplementary data used to characterize the user's current business status can be obtained, so that the prediction model can be used to determine the service goals required by the user based on the question text data sent by the user and the supplementary data corresponding to the question text data. Then, the dialogue generation module can be used to generate sub-reply texts corresponding to each service goal, and the sub-reply texts can be combined into a target reply text to the user, so as to improve the accuracy of the generated reply text. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of this specification and constitute a part of this specification. The exemplary embodiments and descriptions of this specification are used to explain this specification and do not constitute an improper limitation on this specification.

[0051] In the figure:

[0052] Figure 1 A schematic diagram of the risk control process for the user's business behavior provided in this manual;

[0053] Figure 2 A flowchart of a task execution method provided in this specification;

[0054] Figure 3 A schematic diagram of a process for determining an intent vector provided in this specification;

[0055] Figure 4 A schematic diagram of the process of determining the service objectives provided in this specification;

[0056] Figure 5A schematic diagram of the generation process of the sub-response text provided in this specification;

[0057] Figure 6 A schematic diagram of a task execution device provided in this specification;

[0058] Figure 7 A method corresponding to the Figure 1 Schematic diagram of an electronic device. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solutions and advantages of this specification more clear, the technical solutions of this specification will be clearly and completely described below in combination with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.

[0060] At present, in order to ensure the security of users' personal privacy data, business platforms can usually prevent users' personal privacy data from being leaked through methods such as risk warnings, security education, and permission restrictions when monitoring users' risky business behaviors. For example, Figure 1 shown.

[0061] Figure 1 This is a schematic diagram of the risk control process for the user's business behavior provided in this manual.

[0062] Combination Figure 1 It can be seen that the business platform can restrict the permissions of the user's account when it detects that the user's account currently has business risks. In the process of the business platform protecting the user's personal privacy data, the user will inevitably have questions about risk warnings, security education, permission restrictions, and permission restriction removal. At this time, the business platform can generate a reply text based on the question text data sent by the user to answer the user's questions. The accuracy of the reply text generated by the business platform is of great significance to whether it can provide effective help to the user. Therefore, it is particularly important to generate accurate reply text for questions raised by users.

[0063] The technical solutions provided by the embodiments of this specification are described in detail below in conjunction with the accompanying drawings.

[0064] Figure 2 A flowchart of a task execution method provided in this specification includes the following steps:

[0065] S200: Acquire question text data sent by the user.

[0066] S202: querying the service status information of the user based on the user identifier based on which the user sends the question text data, and determining the supplementary data corresponding to the question text data based on the service status information, wherein the service status information is used to characterize the current service status of the user.

[0067] In this specification, after receiving the question text data sent by the user through the client of the business platform, the business platform can query the user's business status information based on the user identifier based on which the user sends the question text data, so as to determine the supplementary data corresponding to the question text data based on the user's business status information.

[0068] The above-mentioned question text data may refer to the text written by the user in the client to describe the question that needs to be answered by the business platform.

[0069] The above-mentioned business status information is used to characterize the current business status of the user. For example, in a business risk control scenario, the above-mentioned business status information may refer to the types of restrictions on the permissions of the account used by the user obtained by the business platform (such as restrictions on receiving payments, restrictions on transactions, restrictions on sending information, etc.), and the types of risks that the account used by the user may currently involve (such as being defrauded, participating in risky transactions, etc.).

[0070] For another example, in an e-commerce scenario, the business status information may be order information about a user's account that may be in dispute and obtained by the business platform, historical transaction information about the user's account and obtained by the business platform, etc.

[0071] It should be noted that the above-mentioned supplementary data is data obtained by performing operations such as format conversion and filtering on the obtained business status information.

[0072] Furthermore, after obtaining the question text data sent by the user and the supplementary data corresponding to the question text data, the business platform can perform standardization processing on the obtained question text data and the supplementary data corresponding to the question text data to obtain the standardized question text data and the standardized supplementary data, and then determine the intent vector corresponding to the question text data based on the standardized question text data and the standardized supplementary data, as shown in the following example: Figure 3 shown.

[0073] Figure 3 A schematic diagram of the process of determining the intent vector provided in this specification.

[0074] Combination Figure 3It can be seen that the business platform can determine at least one intention label that matches the standardized problem text data from each preset intention label as the first intention label combination, and determine at least one intention label that matches the standardized supplementary data from each preset intention label as the second intention label combination. Furthermore, the first intention label combination and the second intention label combination determined based on the obtained problem text data and the supplementary data corresponding to the problem text data can be converted into feature vectors to obtain the intention vector corresponding to the problem text data input by the user.

[0075] Specifically, the method for the business platform to convert the first intention label combination and the second intention label combination into feature vectors can be to convert each label included in the first intention label combination and the second intention label combination into the corresponding vector encoding according to the corresponding relationship between each preset intention label and each vector encoding, so as to obtain the intention vector corresponding to the problem text data input by the user.

[0076] Among them, the above-mentioned standardization process is used to eliminate the noise in the problem text data and the supplementary data corresponding to the problem text data, and to unify the data formats of the problem text data and the supplementary data of the problem text data, and standardize the expression forms of the problem text data and the supplementary data of the problem text data. By performing standardization processing on the problem text data and the supplementary data corresponding to the problem text data, it is convenient to analyze the problem text data and the supplementary data corresponding to the problem text data subsequently.

[0077] There are various methods for the above-mentioned standardization process. For example: removing irrelevant characters (that is, removing non-text elements such as special symbols, whitespace characters (such as extra spaces, line breaks), Hyper Text Markup Language (HTML) tags, and emojis) in the text, converting case (that is, uniformly converting to lowercase or uppercase to eliminate the influence caused by case differences), removing non-ASCII characters (filtering out characters in non-English character sets or performing appropriate transcoding according to requirements), and other text cleaning methods.

[0078] Another example: removing stop words (that is, removing common meaningless or rarely informative words in the question (such as "de", "le", etc.)), removing punctuation and only retaining text information, and other text filtering methods.

[0079] Another example: for non-English texts, appropriate word segmentation processing is performed to split sentences into words or word units, and words are restored to their basic form through stem extraction algorithms (such as Porter Stemmer, Snowball Stemmer, etc.) to reduce the size and redundancy of the vocabulary. For languages ​​with complex morphological changes (such as English, German, etc.), word form restoration is performed to convert words into their dictionary form (such as converting "running" to "run", etc.) to retain more semantic information and other text filtering methods.

[0080] In the above content, the preset intention tags can be pre-set tags corresponding to key information related to task execution that may be contained in the question text data, or tags corresponding to key information related to task execution that may be contained in the supplementary data corresponding to the question text data.

[0081] For example, the intent labels corresponding to the key information related to task execution that may be included in the question text data may be the current complaint progress, the current complaint options, query current restriction information, etc. Another example: the intent labels corresponding to the key information related to task execution that may be included in the supplementary data corresponding to the question text data may be the restriction type of the account used by the user, the risk type that may exist in the account used by the user, etc.

[0082] From the above content, it can be seen that the business platform can map the question text data sent by the user and the supplementary data corresponding to the question text data into intent vectors corresponding to each intent label, so that the subsequent prediction model can better understand the user's intention based on the intent labels matching the question text data and the intent vectors corresponding to each intent label matching the supplementary data corresponding to the question text data, so as to generate a more accurate reply text for the user to solve the problem currently faced by the user.

[0083] In this specification, the execution subject used to implement the task execution method can refer to a designated device such as a server set up on a business platform, or it can refer to a terminal device such as a desktop computer or a laptop computer. For the sake of ease of description, the task execution method provided in this specification is explained below using the server as the execution subject as an example.

[0084] S204: Input the question text data and the supplementary data into a preset prediction model, so that the prediction model can predict the service objectives for the user's current business status and the sequence relationship between the service objectives based on the question text data and the supplementary data.

[0085] Furthermore, the server can input the intent vector corresponding to the question text data input by the user into a preset prediction model, so as to predict the service goals and the order relationship between the service goals under the current business state of the user according to the intent vector corresponding to the question text data through the prediction model, such as Figure 4 shown.

[0086] Figure 4 A schematic diagram of the process of determining the service objectives provided in this specification.

[0087] Combination Figure 4 It can be seen that the server can input the intention vector corresponding to the question text data input by the user into the preset prediction model, so that the prediction model can predict the candidate service targets that the user may need from the preset candidate service targets based on the intention vector corresponding to the question text data input by the user, as the service targets corresponding to the question text data.

[0088] Specifically, the server may use a prediction model to predict the candidate service targets that the user may need from the preset candidate service targets according to the question text data and the supplementary data, as a candidate service target group corresponding to the question text data.

[0089] Furthermore, the server can repeat the above method multiple times to predict the candidate service target groups corresponding to the question text data according to the question text data and the supplementary data through the prediction model. Then, for each candidate service target group, according to the priority of each service target included in each candidate service target group, the service target group to be output can be screened from each candidate service target group, and then each service target included in the service target group to be output can be used as the predicted service targets for the user's current business state.

[0090] The above service goals are used to represent the business requirements that users may need to be implemented by the server under different business conditions. For example, the above service goal may be to display information representing the user's current restriction status to the user. Another example is to answer the user's questions about the solution to the restriction.

[0091] The order of precedence between the above service objectives is used to reflect the logical order of each service objective in the conversation process under the current business state of the user. For example, when answering the current problem of the user, the hierarchical order of the various aspects of information that the user needs to understand to solve the problem. Another example is: when answering the current problem of the user, the execution order of the various steps that the user needs to perform to solve the problem.

[0092] The priorities of the above service objectives may be determined in advance based on multiple factors of the business scenario in which the user is located.

[0093] For example: the priority is determined in advance based on the degree of dependency between different service objectives for answering different questions in different business scenarios, wherein the higher the degree of dependency between answering a certain question in a certain business scenario and a service objective, the higher the priority of this service objective for answering this question. The degree of dependency here is used to characterize the clarity of the server's answer to this question based on the reply content generated by the service objective.

[0094] It should be noted that the service objectives output by the above-mentioned prediction model not only include the service objectives involved in the answering question text data, but also include the service objectives involved in other question text data that the user may send after answering the question text data.

[0095] For example: after showing the user the information used to represent the user's current restriction status, the user may also have questions about how to remove the restriction status. At this time, "showing the user the information used to represent the user's current restriction status" is the service goal involved in answering the question text data, and "answering the user's questions about how to remove the restriction status" is the service goal involved in other question text data that may appear after answering the question text data.

[0096] In actual application scenarios, in order to improve the accuracy of the service targets corresponding to the question text data determined by the server, the server can also obtain sample conversation records, and input the obtained sample conversation records, question text data, and supplementary data into a preset prediction model, so as to predict the service targets corresponding to the question text data based on the sample conversation records, question text data, and supplementary data through the prediction model, wherein the sample conversation records include historical question text data sent by users in history and standard reply texts for the historical question text data.

[0097] From the above content, it can be seen that in order to improve the accuracy of the service objectives corresponding to the question text data determined by the server, the server can also input the sample conversation records as prompt information into the prediction model, so that the prediction model can predict the service objectives corresponding to the question text data based on the sample conversation records, thereby improving the accuracy of the predicted service objectives.

[0098] It should be noted that in addition to obtaining sample conversation records and inputting the sample conversation records into the prediction model, the server can also obtain dialogue path rules extracted in advance based on the sample conversation records, and input the obtained dialogue path rules, question text data, and supplementary data into the prediction model, so as to predict the service objectives corresponding to the question text data based on the dialogue path rules, question text data, and supplementary data through the prediction model.

[0099] In the above content, the dialogue path rule is used to characterize the flow dialogue state that may appear in the process of the server answering the question text data sent by the user. The flow dialogue state here is used to characterize the change of the intention state caused by the change of the user's known information in the process of answering the question text data sent by the user.

[0100] S206: By calling a preset dialogue generation module, a sub-reply text corresponding to each service target is generated in sequence according to the sequence relationship.

[0101] In this specification, the server can generate sub-reply texts corresponding to each service target in sequence according to the sequence relationship between the service targets through a preset dialogue generation module, as shown in the following example: Figure 5 shown.

[0102] Figure 5 A schematic diagram of the process of generating the sub-reply text provided in this manual.

[0103] Combination Figure 5 It can be seen that the server can generate sub-reply texts corresponding to each service target in sequence through a preset dialogue generation module according to the sequence relationship between the service targets and the preset reply templates.

[0104] Among them, the above-mentioned reply template can be pre-constructed according to each historical conversation record. The historical conversation record here contains each historical question text data sent by each historical user, and the historical reply text for the historical question text data. Different reply templates can include a set of sample reply texts required to answer different service goals. The sample reply texts here can be: restriction type description text, restriction lifting explanation text, etc.

[0105] There may be multiple dialogue generation modules as mentioned above, wherein each dialogue generation module may respond to each service target, and different dialogue generation modules may have different response templates.

[0106] The above-mentioned reply template can not only be text data, but also pre-edited video data, voice data, picture data, etc. for answering different service goals. Similarly, the sub-reply text generated by the dialogue generation model can not only be text data, but also composite data consisting of text data and the video data, voice data, and picture data provided in the reply template.

[0107] It should be noted that the above-mentioned dialogue generation module can be a separate dialogue generation model, or can be a module in the above-mentioned prediction model.

[0108] In addition, the server can also generate transition text corresponding to at least part of the sub-reply text through the dialogue generation module, combine the sub-reply texts and the transition text into a target reply text, and display the target reply text to the user to execute the task.

[0109] S208: combining the sub-reply texts into a target reply text according to the sequence relationship, and displaying the target reply text to the user for task execution.

[0110] Furthermore, the server may combine the sub-reply texts into a target reply text according to a sequential relationship, and display the target reply text to the user for task execution.

[0111] Specifically, the execution of the above tasks can be determined according to the actual business scenario. For example, in a business risk control scenario, the above tasks can be tasks of answering questions raised by users about risk control solutions, or tasks of providing safety education to risky users, etc.

[0112] In addition, the server may also perform compliance detection on the target reply text before displaying the generated target reply text to the user.

[0113] Among them, the method for the server to perform compliance detection can be keyword screening (i.e., detecting whether there are words or phrases in the preset banned word library in the text. These banned words may involve non-compliant content such as copyright infringement and advertising fraud), semantic analysis, and semantic understanding using natural language processing technology (NLP).

[0114] The above-mentioned server compliance detection method can also be format and structure checking (that is, for specific types of text, such as legal documents, contracts, news releases, etc., compliance detection may include verification of specific elements such as text structure, reference integrity, copyright statement, disclaimer, contact information, etc.)

[0115] The above-mentioned server compliance detection method can also protect personal privacy (that is, detect whether the text contains personal sensitive information that has not been desensitized, such as ID number, mobile phone number, bank account number, address, etc., to ensure the security of personal information).

[0116] In addition, after displaying the target reply text to the user, the server can also receive conversation text sent by the user in response to the target reply text. At this time, the server can use the user's question text, the target reply text generated by the server, and the conversation text sent by the user in response to the target reply text as conversation text data, and input the conversation text data and supplementary data into the prediction model, so as to regenerate the target reply text based on the conversation text data and supplementary data through the prediction model and the dialogue generation module and display it to the user.

[0117] From the above content, it can be seen that the server can obtain supplementary data used to characterize the user's current business status while obtaining the question text data sent by the user, so that the prediction model can be used to determine the service goals required by the user based on the question text data sent by the user and the supplementary data corresponding to the question text data. Then, the dialogue generation module can be used to generate sub-reply texts corresponding to each service goal, and the sub-reply texts can be combined into a target reply text to the user, so as to improve the accuracy of the generated reply text.

[0118] The above is a task execution method provided in one or more embodiments of this specification. Based on the same idea, this specification also provides a corresponding task execution device, such as Figure 6 shown.

[0119] Figure 6 A schematic diagram of a task execution device provided for this specification includes:

[0120] The acquisition module 601 is used to acquire the question text data sent by the user;

[0121] A query module 602 is used to query the service status information of the user based on the user identifier based on which the user sends the question text data, so as to determine the supplementary data corresponding to the question text data according to the service status information, wherein the service status information is used to characterize the current service status of the user;

[0122] Prediction module 603, used for inputting the question text data and the supplementary data into a preset prediction model, so as to predict the service objectives and the order relationship between the service objectives under the current business state of the user according to the question text data and the supplementary data through the prediction model;

[0123] A generation module 604 is used to generate sub-reply texts corresponding to each service target in sequence according to the sequence relationship by calling a preset dialogue generation module;

[0124] The execution module 605 is used to combine the sub-reply texts into a target reply text according to the sequence relationship, and display the target reply text to the user to perform the task.

[0125] Optionally, the prediction module 603 is specifically used to determine at least one intention label that matches the question text data from the preset intention labels as a first intention label combination, and to determine at least one intention label that matches the supplementary data from the preset intention labels as a second intention label combination; determine the intention vector corresponding to the question text data based on the first intention label combination and the second intention label combination, and input the intention vector into a preset prediction model, so as to predict the service objectives for the user's current business state and the sequential relationship between the service objectives through the prediction model based on the intention vector.

[0126] Optionally, the prediction module 603 is specifically used to predict, through the prediction model based on the question text data and the supplementary data, the service objectives involved in replying to the question text data, as well as the service objectives involved in other question text data sent by the user after replying to the question text data.

[0127] Optionally, the prediction module 603 is specifically used to obtain sample conversation records, wherein the sample conversation records contain historical question text data sent by the user in the past and standard reply texts for the historical question text data; the sample conversation records, the question text data and the supplementary data are input into a preset prediction model, so as to predict the service objectives for the user's current business state and the sequence relationship between the service objectives through the prediction model based on the sample conversation records, the question text data and the supplementary data.

[0128] Optionally, the prediction module 603 is specifically used to predict, through the prediction model, each candidate service target group corresponding to the question text data according to the question text data and the supplementary data, and for each candidate service target group, the candidate service target group includes each predicted service target corresponding to the question text data; according to the priority of each service target included in each candidate service target group, filter out the service target group to be output from the candidate service target groups; and use each service target included in the service target group to be output as the predicted service target for the user's current business state.

[0129] Optionally, the generation module 604 is specifically used to generate sub-reply texts corresponding to each service target in sequence according to the chronological relationship and preset reply templates by calling a preset dialogue generation module, wherein the reply templates are pre-constructed based on historical conversation records, and the historical conversation records contain historical question text data sent by users in history, as well as historical reply texts for the historical question text data.

[0130] Optionally, the execution module 605 is specifically used to generate, through the dialogue generation module, a transition text corresponding to at least part of the sub-reply text; combine each sub-reply text and the transition text corresponding to at least part of the sub-reply text into a target reply text according to the sequence relationship, and display the target reply text to the user for task execution.

[0131] This specification also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 Provides a task execution method.

[0132] This manual also provides Figure 7 The one shown corresponds to Figure 1 A schematic diagram of the electronic device. Figure 7 At the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 Of course, in addition to the software implementation, this specification does not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0133] In the 1990s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0134] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.

[0135] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0136] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0137] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0138] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0139] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0141] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0142] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0143] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0144] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0145] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0146] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0147] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0148] The above are only embodiments of this specification and are not intended to limit this specification. For those skilled in the art, this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this specification should be included in the scope of the claims of this specification.

Claims

1. A task execution method, comprising: Get the question text data sent by the user; According to the user identifier based on which the user sends the question text data, query the service status information of the user, so as to determine the supplementary data corresponding to the question text data according to the service status information, wherein the service status information is used to characterize the current service status of the user; Inputting the question text data and the supplementary data into a preset prediction model, so as to predict the service objectives and the order relationship between the service objectives under the current business state of the user through the prediction model according to the question text data and the supplementary data; By calling a preset dialogue generation module, a sub-reply text corresponding to each service target is generated in sequence according to the sequence relationship; According to the sequential relationship, the sub-reply texts are combined into a target reply text, and the target reply text is displayed to the user for task execution.

2. The method according to claim 1, wherein the question text data and the supplementary data are input into a preset prediction model, so as to predict the service objectives and the order relationship between the service objectives under the current business state of the user according to the question text data and the supplementary data through the prediction model, specifically comprising: Determine at least one intention tag matching the question text data from the preset intention tags as a first intention tag combination, and determine at least one intention tag matching the supplementary data from the preset intention tags as a second intention tag combination; According to the first intention label combination and the second intention label combination, the intention vector corresponding to the question text data is determined, and the intention vector is input into a preset prediction model so that the prediction model can predict the service objectives for the user's current business state and the sequence relationship between the service objectives based on the intention vector.

3. The method according to claim 1, wherein the prediction model predicts the service objectives for the user's current business status based on the question text data and the supplementary data, specifically comprising: The prediction model predicts the service objectives involved in answering the question text data and the supplementary data, as well as the service objectives involved in other question text data sent by the user after answering the question text data.

4. The method according to claim 1, wherein the question text data and the supplementary data are input into a preset prediction model, so as to predict the service objectives and the order relationship between the service objectives under the current business state of the user according to the question text data and the supplementary data through the prediction model, specifically comprising: Acquire a sample conversation record, wherein the sample conversation record includes historical question text data sent by a user in the past and a standard reply text for the historical question text data; The sample conversation record, the question text data and the supplementary data are input into a preset prediction model so that the prediction model can predict the service objectives for the user's current business status and the sequence relationship between the service objectives based on the sample conversation record, the question text data and the supplementary data.

5. The method according to claim 1, wherein the prediction model predicts the service objectives for the user's current business status based on the question text data and the supplementary data, specifically comprising: Predicting each candidate service target group corresponding to the question text data according to the question text data and the supplementary data by the prediction model, wherein for each candidate service target group, the candidate service target group includes each service target corresponding to the predicted question text data; According to the priority of each service target included in each candidate service target group, a service target group to be output is screened out from each candidate service target group; The service targets included in the service target group to be output are used as predicted service targets for the current service state of the user.

6. The method according to claim 1, by calling a preset dialogue generation module, sequentially generates a sub-reply text corresponding to each service target according to the sequence relationship, specifically comprising: By calling the preset dialogue generation module, the sub-reply text corresponding to each service target is generated in sequence according to the sequence relationship and the preset reply templates. The reply template is pre-constructed according to the historical conversation records, and the historical conversation records contain the historical question text data sent by the user in history, and the historical reply text for the historical question text data.

7. The method according to claim 1, combining the sub-reply texts into a target reply text according to the sequence relationship, and presenting the target reply text to the user for task execution, specifically comprising: By means of the dialogue generation module, for at least part of the sub-reply text, a transition text corresponding to the at least part of the sub-reply text is generated; According to the sequential relationship, each sub-reply text and the transition text corresponding to at least part of the sub-reply text are combined into a target reply text, and the target reply text is displayed to the user for task execution.

8. A task execution device, comprising: The acquisition module is used to obtain the question text data sent by the user; A query module, used to query the service status information of the user based on the user identifier based on which the user sends the question text data, so as to determine the supplementary data corresponding to the question text data according to the service status information, wherein the service status information is used to characterize the current service status of the user; A prediction module, used for inputting the question text data and the supplementary data into a preset prediction model, so as to predict the service objectives and the order relationship between the service objectives under the current business state of the user through the prediction model according to the question text data and the supplementary data; A generation module, configured to generate sub-reply texts corresponding to each service target in sequence by calling a preset dialogue generation module according to the sequence relationship; An execution module is used to combine the sub-reply texts into a target reply text according to the sequence relationship, and display the target reply text to the user to perform the task.

9. The device of claim 8, wherein the prediction module is specifically configured to determine at least one intention label that matches the question text data from the preset intention labels as a first intention label combination, and to determine at least one intention label that matches the supplementary data from the preset intention labels as a second intention label combination; According to the first intention label combination and the second intention label combination, the intention vector corresponding to the question text data is determined, and the intention vector is input into a preset prediction model so that the prediction model can predict the service objectives for the user's current business state and the sequence relationship between the service objectives based on the intention vector.

10. In the device as described in claim 8, the prediction module is specifically used to predict, through the prediction model and based on the question text data and the supplementary data, the service objectives involved in answering the question text data, as well as the service objectives involved in other question text data sent by the user after answering the question text data.

11. The device according to claim 8, wherein the prediction module is specifically used to obtain a sample conversation record, wherein the sample conversation record includes historical question text data sent by a user in the past and a standard reply text for the historical question text data; The sample conversation record, the question text data and the supplementary data are input into a preset prediction model so that the prediction model can predict the service objectives for the user's current business status and the sequence relationship between the service objectives based on the sample conversation record, the question text data and the supplementary data.

12. The device according to claim 8, wherein the prediction module is specifically used to predict, through the prediction model, each candidate service target group corresponding to the question text data according to the question text data and the supplementary data, and for each candidate service target group, the candidate service target group includes each service target corresponding to the predicted question text data; According to the priority of each service target included in each candidate service target group, a service target group to be output is screened out from each candidate service target group; The service targets included in the service target group to be output are used as predicted service targets for the current service state of the user.

13. In the device as described in claim 8, the generation module is specifically used to generate sub-reply texts corresponding to each service target in sequence according to the chronological relationship and the preset reply templates by calling a preset dialogue generation module, and the reply templates are pre-constructed according to various historical conversation records, and the historical conversation records contain various historical question text data sent by the user in history, as well as historical reply texts for the historical question text data.

14. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the method according to any one of claims 1 to 7.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 7 when executing the program.

Citation Information

Patent Citations

  • Information recommendation method of intelligent customer service, server and computer readable medium

    CN110019693A

  • Inquiry session processing method and device based on artificial intelligence, and computer equipment

    CN112307168A

  • Dialogue processing method and device

    CN112735407A

  • Recommendation-oriented emotional dialogue method

    CN114168707A

  • Customer service session processing method, device and system, electronic equipment and storage medium

    CN115129846A