Talk skill recommendation method and related equipment

Abstract

The invention discloses a verbal skill recommendation method and related equipment, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining an input first statement in response to a verbal skill recommendation instruction, obtaining a client ID corresponding to a client inputting the first statement, obtaining a background tag corresponding to the client based on the client ID, analyzing the first statement, and obtaining a background tag corresponding to the client; determining a business scene and a business problem corresponding to the first statement, determining a target expert rule in a preset expert rule set based on the background label, the business scene, the business problem, a preset expert rule matching engine and the preset expert rule set, and determining a recommendation verbal skill corresponding to the first statement based on the target expert rule, and recommending the recommended verbal skill to the client. The expert rule is set according to the experience of the business personnel, the target expert rule is matched according to the customer type and the business scene when the customer service is carried out, and then the target verbal skill corresponding to the target expert rule is recommended to the customer, so that the verbal skill recommendation accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a dialogue recommendation method and related equipment. BACKGROUND

[0002] With the rapid development of the digital economy era, customer service is also facing major challenges. In related technologies, due to the customer diversity and business scenario diversity corresponding to customer service, personalized recommendation cannot be performed, resulting in low accuracy of dialogue recommendation and affecting user experience.

[0003] The above content is only used to assist in understanding the technical solutions of the present application and does not represent the acknowledgement of the above content as related technology. SUMMARY

[0004] The main purpose of the present application is to provide a dialogue recommendation method and related equipment, aiming to solve the technical problem of how to improve the accuracy of dialogue recommendation.

[0005] To achieve the above purpose, the present application provides a dialogue recommendation method, which comprises:

[0006] In response to a dialogue recommendation instruction, a first sentence is input and a customer ID corresponding to a customer inputting the first sentence is obtained;

[0007] Based on the customer ID, a background label corresponding to the customer is obtained, and the first sentence is parsed to determine a business scenario and a business problem corresponding to the first sentence;

[0008] Based on the background label, the business scenario, the business problem, a preset expert rule matching engine and a preset expert rule set, a target expert rule in the preset expert rule set is determined;

[0009] Based on the target expert rule, a recommended dialogue corresponding to the first sentence is determined, and the recommended dialogue is recommended to the customer.

[0010] In an embodiment, the step of parsing the first sentence to determine the business scenario and the business problem corresponding to the first sentence further comprises:

[0011] Parsing the first sentence to determine a target semantic corresponding to the first sentence;

[0012] Based on the target semantic, a business scenario corresponding to the first sentence is determined, and a standardized business problem is generated based on a preset sentence standardization strategy and the target semantic.

[0013] In an embodiment, the step of parsing the first sentence to determine the target semantic corresponding to the first sentence further comprises:

[0014] perform semantic recognition on the first sentence based on a preset semantic recognition engine to determine a target semantic corresponding to the first sentence.

[0015] In an embodiment, the step of determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, a preset expert rule matching engine and a preset expert rule set further comprises:

[0016] determining a priority of each expert rule in the preset expert rule set;

[0017] based on the background label, the business scenario, the business problem, a preset expert rule strategy and the priority, using the preset expert rule matching engine to sequentially match each expert rule in the preset expert rule set with the first sentence in descending order of priority until a successful match is obtained, to obtain a target expert rule matched with the first sentence.

[0018] In an embodiment, the step of determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, a preset expert rule matching engine and a preset expert rule set further comprises:

[0019] determining a first access frequency corresponding to the business problem based on the customer ID, the business scenario and a preset expert rule strategy, and determining an access frequency threshold corresponding to the business problem based on the business scenario, the customer type and a preset access frequency strategy;

[0020] determining whether the first access frequency is less than the access frequency threshold;

[0021] If it is less, determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, a preset expert rule matching engine and a preset expert rule set.

[0022] In an embodiment, the step of determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, a preset expert rule matching engine and a preset expert rule set further comprises:

[0023] determining a first recommended dialogue corresponding to the first sentence based on the target expert rule;

[0024] determining whether there is a reinforcement learning flag in the preset expert rule strategy;

[0025] If there is, determining a second recommended dialogue based on the reinforcement learning flag

[0026] based on the first return probability and the second return probability, setting one of the first recommended dialogue and the second recommended dialogue as a target recommended dialogue, and recommending the target recommended dialogue to the customer.

[0027] In an embodiment, after the step of determining the recommended dialogue corresponding to the first sentence based on the target expert rule and recommending the recommended dialogue to the customer, the method further comprises:

[0028] obtaining a customer feedback result corresponding to the recommended dialogue;

[0029] based on the customer feedback result, the first sentence, the target expert rule and the recommended dialogue, constructing a back-check database, wherein the back-check database is used to adjust the expert rules in the preset expert rule set, and the back-check database is also used to provide training data for reinforcement learning.

[0030] In addition, to achieve the above-mentioned purposes, the present application further provides a dialogue recommendation device, which comprises:

[0031] an obtaining module, configured to obtain an input first sentence and a customer ID corresponding to a customer inputting the first sentence in response to a dialogue recommendation instruction;

[0032] an analyzing module, configured to obtain a background label corresponding to the customer based on the customer ID, and analyze the first sentence to determine a business scenario and a business problem corresponding to the first sentence;

[0033] a determining module, configured to determine a target expert rule in a preset expert rule set based on the background label, the business scenario, the business problem, a preset expert rule matching engine and the preset expert rule set;

[0034] a recommending module, configured to determine a recommended dialogue corresponding to the first sentence based on the target expert rule, and recommend the recommended dialogue to the customer.

[0035] In an embodiment, the analyzing module further comprises:

[0036] an analyzing unit, configured to analyze the first sentence to determine a target semantic corresponding to the first sentence;

[0037] The first determining unit is configured to determine a business scenario corresponding to the first sentence based on the target semantics, and generate a standardized business question based on a preset sentence standardization strategy and the target semantics.

[0038] In an embodiment, the parsing module further includes:

[0039] The semantic recognition unit is configured to perform semantic recognition on the first sentence based on a preset semantic recognition engine to determine target semantics corresponding to the first sentence.

[0040] In an embodiment, the determining module further includes:

[0041] The second determining unit is configured to determine a priority of each expert rule in the preset expert rule set.

[0042] The first matching unit is configured to sequentially match each expert rule in the preset expert rule set with the first sentence in descending order of priority based on the background label, the business scenario, the business question, a preset expert rule matching engine and the priority, until a match is successful, to obtain a target expert rule matched with the first sentence.

[0043] In an embodiment, the determining module further includes:

[0044] The third determining unit is configured to determine a first access frequency corresponding to the business question based on the customer ID, the business scenario and a preset expert rule strategy, and determine an access frequency threshold corresponding to the business question based on the business scenario, the customer type and a preset access frequency strategy.

[0045] The judging unit is configured to judge whether the first access frequency is less than the access frequency threshold.

[0046] The second matching unit is configured to, if the first access frequency is less than the access frequency threshold, determine a target expert rule in the preset expert rule set based on the background label, the business scenario, the business question, a preset expert rule matching engine and the preset expert rule set.

[0047] In an embodiment, the recommendation module further includes:

[0048] The fourth determining unit is configured to determine a first recommended dialogue corresponding to the first sentence based on the target expert rule.

[0049] The fifth determining unit is configured to determine whether there is a reinforcement learning flag in the preset expert rule strategy.

[0050] The third matching unit is configured to, if there is the reinforcement learning flag, determine a second recommended dialogue based on the reinforcement learning flag.

[0051] The sixth determining unit is configured to determine a first return probability corresponding to the first recommended dialogue and a second return probability corresponding to the second recommended dialogue based on a preset shunt strategy, set one of the first recommended dialogue and the second recommended dialogue as a target recommended dialogue based on the first return probability and the second return probability, and recommend the target recommended dialogue to the customer.

[0052] In an embodiment, the dialogue recommendation device further comprises a construction module, and the construction module comprises:

[0053] The acquisition unit is configured to acquire a customer feedback result corresponding to the recommended dialogue.

[0054] The construction unit is configured to construct a back-check database based on the customer feedback result, the first sentence, the target expert rule, and the recommended dialogue, wherein the back-check database is used to adjust an expert rule in the preset expert rule set, and the back-check database is further used to provide training data for reinforcement learning.

[0055] In addition, to achieve the above-mentioned purposes, the present application further provides a dialogue recommendation device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the dialogue recommendation method as described above.

[0056] In addition, to achieve the above-mentioned purposes, the present application further provides a storage medium, which is a computer readable storage medium, and a computer program is stored on the storage medium, and the computer program is executed by a processor to implement the steps of the dialogue recommendation method as described above.

[0057] In addition, to achieve the above-mentioned purposes, the present application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the steps of the dialogue recommendation method as described above.

[0058] The one or more technical solutions provided by the present application have at least the following technical effects:

[0059] The application provides a dialogue recommendation method and related equipment, and relates to the field of artificial intelligence. In the related art, due to the diversity of customers and the diversity of business scenarios corresponding to customer service, personalized recommendation cannot be performed, which leads to low accuracy of dialogue recommendation and affects user experience. In the application, first, in response to a dialogue recommendation instruction, a first sentence is input and a customer ID corresponding to the customer inputting the first sentence is obtained. Then, based on the customer ID, a background label corresponding to the customer is obtained, and the first sentence is parsed to determine a business scenario and a business problem corresponding to the first sentence. Further, based on the background label, the business scenario, the business problem, a preset expert rule matching engine and a preset expert rule set, a target expert rule in the preset expert rule set is determined. Finally, based on the target expert rule, a recommended dialogue corresponding to the first sentence is determined, and the recommended dialogue is recommended to the customer.

[0060] It can be understood that, according to the experience of business personnel, the application sets expert rules, and when providing customer service, a target expert rule is matched according to the type of the customer and the business scenario, and then a target dialogue corresponding to the target expert rule is recommended to the customer, thereby improving the accuracy of dialogue recommendation. BRIEF DESCRIPTION OF DRAWINGS

[0061] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the application and, together with the specification, serve to explain the principles of the application.

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the application or the related art, the accompanying drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0063] Figure 1 A flowchart is provided for the dialogue recommendation method embodiment one of the application;

[0064] Figure 2 A flowchart is provided for the dialogue recommendation method embodiment two of the application;

[0065] Figure 3 A flowchart is provided for the dialogue recommendation method embodiment three of the application;

[0066] Figure 4 A module structure diagram of the dialogue recommendation device of the application embodiment is provided;

[0067] Figure 5 A device structure diagram of the hardware running environment related to the dialogue recommendation method in the application embodiment is provided.

[0068] The purposes, functional features and advantages of the present application will be further illustrated in conjunction with the embodiments, with reference to the accompanying drawings. DETAILED DESCRIPTION

[0069] It should be understood that the specific embodiments described herein are merely intended to explain the technical solutions of the present application, and are not intended to limit the present application.

[0070] In order to better understand the technical solutions of the present application, the specific embodiments will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0071] The main solution of the embodiment of the present application is:

[0072] In the present embodiment, for the sake of description, the following is described with the execution subject of the dialogue recommendation device.

[0073] Due to the related art: due to the diversity of customers corresponding to customer service, the diversity of business scenarios, the personalized recommendation cannot be performed, resulting in low accuracy of dialogue recommendation, affecting user experience.

[0074] The present application provides a solution, which is: first, in response to a dialogue recommendation instruction, input a first sentence and obtain a customer ID corresponding to a customer inputting the first sentence, then, based on the customer ID, obtain the background label corresponding to the customer, and parse the first sentence to determine the business scenario and business problem corresponding to the first sentence, further, based on the background label, the business scenario, the business problem, the preset expert rule matching engine and the preset expert rule set, determine the target expert rule in the preset expert rule set, finally, based on the target expert rule, determine the recommended dialogue corresponding to the first sentence, and recommend the recommended dialogue to the customer.

[0075] It can be understood that, according to the experience of business personnel, the present application sets expert rules, and when providing customer service, the target expert rule is matched according to the type of the customer and the business scenario, and then the target dialogue corresponding to the target expert rule is recommended to the customer, thereby improving the accuracy of dialogue recommendation.

[0076] It should be noted that the execution subject of the present embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device, a dialogue recommendation device, etc. capable of realizing the above functions. The present embodiment and the following embodiments will be described below with the dialogue recommendation device as an example.

[0077] Based on this, the present embodiment of the present application provides a dialogue recommendation method, which is described in detail with reference to Figure 1 , Figure 1A flowchart of a first embodiment of a dialogue recommendation method of the present application.

[0078] In this embodiment, the dialogue recommendation method comprises steps S10-S40:

[0079] In step S10, in response to a dialogue recommendation instruction, a first sentence input is obtained, and a customer ID corresponding to a customer inputting the first sentence is obtained.

[0080] It should be noted that the execution subject of the present embodiment can be a computing service device having data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device, a dialogue recommendation device, etc. capable of realizing the above functions. In the following, the dialogue recommendation device is taken as an example to describe the present embodiment and the following embodiments.

[0081] It should be noted that the dialogue recommendation instruction is an instruction issued by a user or a system, requiring the system to recommend appropriate answers or dialogues. For example, when a customer service personnel is in conversation with a customer, the system may be required to help generate appropriate answers. In the present application, the dialogue recommendation device receives an instruction triggering the dialogue recommendation function, such as a user clicking a "recommend reply" button, or the dialogue recommendation device detecting a scenario requiring dialogue recommendation.

[0082] It should be noted that the first sentence refers to the text to be processed in the current conversation, which can be a keyword, a descriptive sentence, etc.

[0083] It should be noted that in the present application, the system obtains data through a specific interface (communication interface). The communication interface is a channel for data exchange between systems or between internal modules of a system, which can be an API interface, a database interface or other forms of data transmission channels.

[0084] The customer ID is a unique identifier of a user in the system, used to distinguish different users. The system needs to obtain the user's historical data, preference settings, etc. through the customer ID, in order to provide more accurate recommendations.

[0085] In the present embodiment, when the user triggers the "dialogue recommendation" function, the dialogue recommendation device immediately captures the customer message that needs to be replied in the current conversation, i.e. the "first sentence", as the input basis for subsequent reply recommendation.

[0086] For example, taking the promotion scenario of an e-commerce platform as an example, when the customer first enters the e-commerce platform, the customer sees the relevant activities about promotion on the homepage of the platform. Further, the customer opens the customer service chat and inputs "Is there a coupon for this product" after entering the customer service chat interface and clicking send.

[0087] The dialogue recommendation device receives the customer's "Is there a coupon for this product?" At this time, the dialogue recommendation device detects the scene that needs to recommend the dialogue, and the dialogue recommendation device processes "Is there a coupon for this product?" as the "first sentence" to generate a recommended reply.

[0088] In step S20, based on the customer ID, the background label corresponding to the customer is obtained, and the first sentence is parsed to determine the business scenario and business problem corresponding to the first sentence.

[0089] Specifically, the step of parsing the first sentence to determine the business scenario and business problem corresponding to the first sentence further comprises steps S21-S22:

[0090] In step S21, the first sentence is parsed to determine the target semantics corresponding to the first sentence.

[0091] Specifically, the step of parsing the first sentence to determine the target semantics corresponding to the first sentence further comprises:

[0092] Based on the preset semantic recognition engine, the semantic recognition of the first sentence is performed to determine the target semantics corresponding to the first sentence.

[0093] It should be noted that the semantic recognition engine is a tool based on natural language processing (NLP) technology, which can understand the meaning of text. It usually includes lexical analysis, syntactic analysis, semantic understanding and other modules.

[0094] The preset semantic recognition engine is usually a trained machine learning model, such as a deep learning-based Transformer model (such as BERT, GPT, etc.).

[0095] Through the analysis of the semantic recognition engine, the specific intention or meaning expressed by the first sentence input by the user is determined.

[0096] The target semantics refers to the core intention or demand expressed by the user's input sentence. The semantic recognition engine maps the user's input to a predefined semantic category by analyzing the structure, vocabulary and context of the sentence.

[0097] In step S22, based on the target semantics, the business scenario corresponding to the first sentence is determined, and based on the preset sentence standardization strategy and the target semantics, a standardized business problem is generated.

[0098] The business scenario refers to the business field to which the user's demand belongs, such as finance, insurance, loan, etc. By classifying the user's demand into a specific business scenario, the system can more accurately handle the user's problem.

[0099] The sentence standardization strategy is a set of rules or algorithms for converting user's natural language input into standardized business questions. These strategies usually include keyword extraction, template matching, semantic mapping and other techniques.

[0100] The standardized business question is a formatted question that the system can understand and process, facilitating subsequent business logic processing.

[0101] In this application, the natural language input of the user is quickly converted into a standardized question that the system can understand and process, thereby providing accurate services or responses.

[0102] Step S30, based on the background label, the business scenario, the business question, the preset expert rule matching engine and the preset expert rule set, determining the target expert rule in the preset expert rule set;

[0103] It should be noted that the preset expert rule matching engine is a pre-defined matching engine for guiding how to select appropriate expert rules according to business scenarios and customer types. It should also be noted that in this application, the expert rule is set by the business personnel based on relevant business experience.

[0104] It should be noted that the preset expert rule set is a library containing multiple expert rules, each rule defining the action or decision to be taken under certain conditions.

[0105] It should be noted that the target expert rule refers to the rule in the preset expert rule set that best matches the current sentence.

[0106] Specifically, the step of determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business question, the preset expert rule matching engine and the preset expert rule set further includes steps S31-S32:

[0107] Step S31, determining the priority of each expert rule in the preset expert rule set;

[0108] It should be noted that each expert rule has a priority value, and the priority can be a number, such as 1, 2, 3, where 1 is the highest; the priority can also be a label, such as "high", "medium", "low".

[0109] In this embodiment, the priority can be defined based on various factors, such as business importance, customer type, rule urgency, etc. Among them, the business importance refers to the rules that may be more important in certain business scenarios; the customer type refers to the customers that may need to be handled first; the rule urgency refers to the rules that may need to be matched first.

[0110] Step S32, based on the background label, the business scenario, the business problem, the preset expert rule strategy and the priority, using a preset expert rule matching engine, each expert rule in the preset expert rule set is matched with the first statement in order of priority from high to low until a successful match is obtained, and a target expert rule matched with the first statement is obtained.

[0111] In this embodiment, the expert rule set is sorted according to the priority from high to low. Based on the sorting result, starting from the rule with the highest priority, each rule is matched with the first statement using a preset expert rule matching engine. If a matching rule is found, the matching process is stopped, and the rule is the target expert rule.

[0112] In this embodiment, by defining the priority and matching the rules in order of priority, the system can efficiently find the most suitable rule according to the business needs and customer characteristics. This method not only improves the matching efficiency, but also ensures that important rules are processed first, thereby improving user experience and service quality.

[0113] Step S40, based on the target expert rule, determining the recommended dialogue corresponding to the first statement, and recommending the recommended dialogue to the customer.

[0114] It should be noted that the recommended dialogue refers to a specific answer or suggestion generated according to the target expert rule, which is used to respond to the customer's question.

[0115] In this embodiment, first, according to the definition in the target expert rule, a suitable answer is generated or selected. Then, the generated recommended dialogue is sent to the current customer as the system's answer or suggestion.

[0116] It can be understood that in this embodiment, the dialogue recommendation device generates a suitable answer through the matched target expert rule, and performs personalized processing according to the customer portrait and context information, and finally sends the recommended dialogue to the customer. This process not only improves the accuracy and relevance of the answer, but also improves the customer's satisfaction and experience.

[0117] Specifically, after the step of determining the recommended dialogue corresponding to the first statement based on the target expert rule and recommending the recommended dialogue to the customer, the steps S50-S60 are further included:

[0118] Step S50, obtaining a customer feedback result corresponding to the recommended dialogue;

[0119] Step S60, based on the customer feedback result, the first sentence, the target expert rule and the recommended dialogue, a back-check database is constructed, wherein the back-check database is used to adjust the expert rules in the preset expert rule set, and the back-check database is also used to provide training data for reinforcement learning.

[0120] It should be noted that the back-check database is a database for storing feedback data and related information, which is used for subsequent rule adjustment.

[0121] In this embodiment, according to the data in the back-check database, the preset expert rule set is optimized to improve the accuracy and effect of the recommended dialogue.

[0122] It should be noted that in this embodiment, the feedback collection includes explicit feedback and implicit feedback. Explicit feedback refers to the evaluation given directly by the customer, such as "satisfied", "dissatisfied", "helpful", "not helpful", etc. Implicit feedback refers to feedback inferred from the customer's behavior, such as whether the customer continues to ask questions or completes the transaction. Further, the customer's feedback result is stored in the database, and the feedback result includes customer ID, recommended dialogue, feedback result, and timestamp.

[0123] After storing the feedback result, the customer feedback result, the first sentence, the target expert rule and the recommended dialogue are integrated into a database, which is the back-check database.

[0124] It can be understood that the dialogue recommendation device can find out the rules with poor effect by analyzing the data in the back-check database and optimize them.

[0125] It can be understood that the dialogue recommendation device realizes the continuous optimization of the preset expert rule set by collecting customer feedback and constructing the back-check database. This method not only improves the accuracy and effect of the recommended dialogue, but also dynamically adjusts the rules according to the actual feedback, thereby improving the overall performance of the system and the user experience.

[0126] The application provides a dialogue recommendation method and related equipment, and relates to the field of artificial intelligence. In the related art, due to the diversity of customers and the diversity of business scenarios corresponding to customer service, personalized recommendation cannot be performed, which leads to low accuracy of dialogue recommendation and affects user experience. In the application, first, in response to a dialogue recommendation instruction, a first sentence is input and a customer ID corresponding to the customer inputting the first sentence is obtained. Then, based on the customer ID, a background label corresponding to the customer is obtained, and the first sentence is parsed to determine a business scenario and a business problem corresponding to the first sentence. Further, based on the background label, the business scenario, the business problem, a preset expert rule matching engine and a preset expert rule set, a target expert rule in the preset expert rule set is determined. Finally, based on the target expert rule, a recommended dialogue corresponding to the first sentence is determined, and the recommended dialogue is recommended to the customer.

[0127] It can be understood that, according to the experience of business personnel, the application sets expert rules, and when performing customer service, a target expert rule is matched according to the type of the customer and the business scenario, and then a target dialogue corresponding to the target expert rule is recommended to the customer, thereby improving the accuracy of dialogue recommendation.

[0128] Based on the first embodiment of the application, in the second embodiment of the application, the same or similar contents as the above embodiment one can refer to the above introduction, and the subsequent will not be described in detail. On this basis, please refer to Figure 2 , the step of determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, the preset expert rule matching engine and the preset expert rule set further includes steps A1-A3:

[0129] Step A1, based on the customer ID, the business scenario and a preset expert rule strategy, a first access frequency corresponding to the business problem is determined, and based on the business scenario, the customer type and a preset access frequency strategy, an access frequency threshold corresponding to the business problem is determined.

[0130] It should be noted that the first access frequency refers to the number of times that the business problem is accessed or queried in historical data. In this embodiment, the dialogue recommendation equipment queries a database or data warehouse to count the number of times that the business problem appears in past dialogue records.

[0131] Step A2, determining whether the first access frequency is less than the access frequency threshold;

[0132] Step A3, if less, determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, the preset expert rule matching engine and the preset expert rule set.

[0133] For example, assume that this is an e-commerce customer service robot system, and the customer is a customer with user ID 12345. The customer inputs a sentence: “Does your store offer discount coupons?” After standardization, the first sentence is “Does your store offer discount coupons?” Then, the e-commerce customer service robot system queries the database and finds that “Does your store offer discount coupons?” has been accessed 150 times in the past month. According to the preset policy, the access frequency threshold for old customers in the e-commerce scenario is 50. Because 120 > 50, the condition of “access frequency less than threshold” is not met. Therefore, the e-commerce customer service robot system will not reply to this question or return “access frequency exceeds threshold”.

[0134] Assume that another customer inputs a sentence: “Do you have any discount coupons?” After standardization, the first sentence is “Does your store offer discount coupons?” Query the database and find that this sentence has been accessed 30 times in the past month. Because 30 < 50, the e-commerce customer service robot system will reply to this question.

[0135] It can be understood that in this embodiment, the frequency threshold is used to determine the commonality of a certain sentence. If the access frequency is lower than the threshold, the system will select appropriate expert rules according to the business scenario and customer type, thereby providing more accurate recommended dialogues. This method not only improves the flexibility of the system, but also dynamically adjusts the recommendation strategy according to the actual situation, improving user experience and service quality.

[0136] Based on the first and second embodiments of the present application, in the third embodiment of the present application, the same or similar contents as the above embodiments one and two can refer to the above introduction, and will not be repeated hereinafter. On this basis, please refer to Figure 3 , the step of determining the recommended dialogue corresponding to the first sentence based on the target expert rule and recommending the recommended dialogue to the customer further includes steps B1-B4:

[0137] Step B1, determining the first recommended dialogue corresponding to the first sentence based on the target expert rule;

[0138] Step B2, determining whether there is a reinforcement learning flag in the preset expert rule strategy;

[0139] Step B3, if there is, determining the second recommended dialogue based on the reinforcement learning flag;

[0140] Step B4, based on the preset shunt strategy, respectively determining a first return probability corresponding to the first recommended dialogue and a second return probability corresponding to the second recommended dialogue, based on the first return probability and the second return probability, setting one of the first recommended dialogue and the second recommended dialogue as a target recommended dialogue, and recommending the target recommended dialogue to the customer.

[0141] It should be noted that the reinforcement learning flag is a Boolean value (true or false) indicating whether the reinforcement learning mechanism is enabled.

[0142] It should be further noted that in the present embodiment, the reinforcement learning training result is stored in HBASE. HBase refers to an open source, distributed, versioned non-relational database (NoSQL) commonly used for storing large-scale data. The specific steps include:

[0143] Firstly, through the data migration service, the address, file path and pulling method of the corresponding data source in the exchange system SFTP are configured, and the target data is pulled from the SFTP to the HDFS.

[0144] Then, based on the pre-created HIVE original table corresponding to the standard file on the SFTP, the data migration service is configured to load the file on the HDFS into the corresponding HIVE original table.

[0145] Further, the HIVE label loading task and the corresponding prelude are configured, and through the timing task, the corresponding data is cleaned every day to filter out abnormal values, and the result of the original field mapping conversion is written into the HIVE result table.

[0146] Finally, through the data migration service, the LOADER task is configured to load the result data from the HIVE table to the specified HBASE, wherein the HBASE adopts two sets of clusters with mutual trust, the data writing mode is double writing, and the ROWKEY of the HBASE table is set as the reverse of the customer ID.

[0147] It should be noted that by storing the training result in HBase, efficient management and utilization of large-scale data can be achieved, providing persistent support for intelligent systems.

[0148] It should be noted that in the present application, reinforcement learning is used to optimize the mapping relationship between the business problem and the recommended dialogue, and after optimization, a new mapping relationship set is established to optimize the final recommended dialogue. When the reinforcement learning flag is detected, the dialogue recommendation device will obtain the recommended dialogue corresponding to the business problem according to the previously established new mapping relationship set and the business problem, i.e. the second recommended dialogue.

[0149] It should be noted that the shunting strategy is a preset rule, and in this application, the shunting strategy is used to determine the probability of returning the result.

[0150] For example, the predicted shunting strategy is "second recommended script: first recommended script = 1:4", and the probability of returning the second recommended script is 20%.

[0151] In addition, it should be noted that the shunting strategy can be dynamically adjusted. For example, when the script recommendation is just started, the preset shunting strategy is "second recommended script: first recommended script = 0:1", that is, all the returned results are first recommended scripts, then the preset shunting strategy is gradually adjusted, for example, it is adjusted to "second recommended script: first recommended script = 1:9", then the satisfaction degree of the customer to the script recommendation result is observed, and based on the satisfaction degree, the preset shunting strategy is gradually optimized until the returned result of the script recommendation result is optimal.

[0152] In this embodiment, if reinforcement learning is enabled, the returned recommended script is optimized by the reinforcement learning algorithm, which can better meet the customer's needs.

[0153] It can be understood that in this embodiment, by checking the reinforcement learning flag and the preset shunting strategy, the returned recommended script is shunted, which not only improves the flexibility and adaptability of the system, but also optimizes the rules through reinforcement learning to improve the overall performance and user experience of the system.

[0154] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the script recommendation method of the present application. More forms of simple changes based on this technical concept are within the protection scope of the present application.

[0155] It should be noted that the user-related data involved in the present application is obtained after obtaining the user's permission or consent; that is, when the present application is applied to a specific product or technology, the user's permission is required to realize the acquisition and processing of related data, and the processing of related data needs to comply with relevant laws, regulations and regulatory standards of relevant countries and regions.

[0156] For example, when the current geographic location of the user needs to be obtained, a location acquisition prompt can be displayed in the terminal of the user, and after receiving the confirmation operation of the user for the location acquisition prompt, the terminal can obtain the current geographic location of the user.

[0157] The present application also provides a script recommendation device, please refer to Figure 4 , the script recommendation device comprises:

[0158] An acquisition module 10 is configured to acquire an input first sentence and acquire a customer ID corresponding to a customer inputting the first sentence in response to a script recommendation instruction.

[0159] An analysis module 20 is configured to acquire a background label corresponding to the customer based on the customer ID and analyze the first sentence to determine a business scenario and a business problem corresponding to the first sentence.

[0160] A determination module 30 is configured to determine a target expert rule in a preset expert rule set based on the background label, the business scenario, the business problem, a preset expert rule matching engine, and a preset expert rule set.

[0161] A recommendation module 40 is configured to determine a recommended script corresponding to the first sentence based on the target expert rule and recommend the recommended script to the customer.

[0162] In an embodiment, the analysis module further includes:

[0163] An analysis unit is configured to analyze the first sentence to determine a target semantic corresponding to the first sentence.

[0164] A first determination unit is configured to determine a business scenario corresponding to the first sentence based on the target semantic and generate a standardized business problem based on a preset sentence standardization strategy and the target semantic.

[0165] In an embodiment, the analysis module further includes:

[0166] A semantic recognition unit is configured to perform semantic recognition on the first sentence based on a preset semantic recognition engine to determine a target semantic corresponding to the first sentence.

[0167] In an embodiment, the determination module further includes:

[0168] A second determination unit is configured to determine a priority of each expert rule in the preset expert rule set.

[0169] A first matching unit is configured to use a preset expert rule matching engine to sequentially match each expert rule in the preset expert rule set with the first sentence in a descending order of priority until a match is successful, to obtain a target expert rule matched with the first sentence, based on the background label, the business scenario, the business problem, a preset expert rule strategy, and the priority.

[0170] In an embodiment, the determination module further includes:

[0171] The third determining unit is configured to determine a first access frequency corresponding to the business problem based on the customer ID, the business scenario, and a preset expert rule strategy, and determine an access frequency threshold corresponding to the business problem based on the business scenario, the customer type, and a preset access frequency strategy.

[0172] The judging unit is configured to judge whether the first access frequency is less than the access frequency threshold.

[0173] The second matching unit is configured to, if the first access frequency is less than the access frequency threshold, determine a target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, a preset expert rule matching engine, and the preset expert rule set.

[0174] In an embodiment, the recommendation module further comprises:

[0175] The fourth determining unit is configured to determine a first recommended dialogue corresponding to the first statement based on the target expert rule.

[0176] The fifth determining unit is configured to determine whether there is a reinforcement learning flag in the preset expert rule strategy.

[0177] The third matching unit is configured to, if there is the reinforcement learning flag, determine a second recommended dialogue based on the reinforcement learning flag.

[0178] The sixth determining unit is configured to respectively determine a first return probability corresponding to the first recommended dialogue and a second return probability corresponding to the second recommended dialogue based on a preset shunt strategy, set one of the first recommended dialogue and the second recommended dialogue as a target recommended dialogue based on the first return probability and the second return probability, and recommend the target recommended dialogue to the customer.

[0179] In an embodiment, the dialogue recommendation device further comprises a construction module, and the construction module comprises:

[0180] The acquisition unit is configured to acquire a customer feedback result corresponding to the recommended dialogue.

[0181] The construction unit is configured to construct a back-check database based on the customer feedback result, the first statement, the target expert rule, and the recommended dialogue, wherein the back-check database is used to adjust expert rules in the preset expert rule set, and the back-check database is also used to provide training data for reinforcement learning.

[0182] The technical problems of the dialogue recommendation method are solved by the dialogue recommendation device provided in the present application. The dialogue recommendation device provided in the present application has the same beneficial effects as the dialogue recommendation method provided in the above embodiments, and other technical features of the dialogue recommendation device are the same as the features disclosed in the above embodiments, which will not be repeated here.

[0183] The present application provides a dialogue recommendation device, which comprises at least one processor and a memory connected with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the dialogue recommendation method in the above embodiment one.

[0184] Reference will be made to the following description of the embodiments of the present application with reference to the drawings. Figure 5 The dialogue recommendation device in the embodiments of the present application can include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (for example, vehicle-mounted navigation terminals), and the like, and fixed terminals such as digital TVs, desktop computers, and the like. Figure 5 The dialogue recommendation device shown is only an example, and should not bring any limitation to the functions and use range of the embodiments of the present application.

[0185] As Figure 5As shown, the dialogue recommendation device can include a processing apparatus 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to programs stored in a read only memory (ROM) 1002 or loaded from a storage apparatus 1003 into a random access memory (RAM) 1004. Various programs and data required for the dialogue recommendation device to operate are also stored in the RAM 1004. The processing apparatus 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: input apparatuses 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output apparatuses 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; the storage apparatus 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication apparatus 1009. The communication apparatus 1009 can allow the dialogue recommendation device to communicate with other devices wirelessly or by wire to exchange data. Although the dialogue recommendation device with various systems is shown in the figure, it should be understood that all the shown systems are not required to be implemented or possessed. More or fewer systems can be alternatively implemented or possessed.

[0186] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program codes for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through the communication apparatus, or installed from the storage apparatus 1003, or installed from the ROM 1002. When the computer program is executed by the processing apparatus 1001, the above-mentioned functions defined in the methods of the embodiments disclosed in the present application are performed.

[0187] The dialogue recommendation device provided in the present application adopts the dialogue recommendation method in the above-mentioned embodiments, and can solve the technical problems. Compared with the related art, the dialogue recommendation device provided in the present application has the same beneficial effects as the dialogue recommendation method provided in the above-mentioned embodiments, and other technical features in the dialogue recommendation device are the same as the features disclosed in the above-mentioned embodiments, which will not be described here.

[0188] It is to be understood that the various parts of the disclosure can be implemented (e.g., as portions of an apparatus) in hardware, software, firmware or any combination thereof. In the description of the embodiments described above, specific features, structures, materials or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0189] The above description is merely illustrative of the application and is not intended to limit the scope of the application. Any changes and modifications that can be made to the application in light of the teachings described herein are considered to be within the scope of the application. Therefore, the scope of the application should be determined not with reference to the above description, but should be determined with reference to the appended claims along with their full scope of equivalents.

[0190] The application provides a computer readable storage medium having stored thereon computer readable program instructions (i.e., a computer program) for performing the dialogue recommendation method in the above-described embodiments.

[0191] The computer readable storage medium provided by the application may, for example, be a U disk, but is not limited to an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system or device, or any combination thereof. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more conductive wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present embodiment, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system or device. The program code contained on the computer readable storage medium can be transmitted using any suitable medium, including but not limited to a wire, an optical fiber, an RF (Radio Frequency) cable, etc., or any suitable combination thereof.

[0192] The above-described computer readable storage medium can be included in the dialogue recommendation device; or can exist separately and not be assembled into the dialogue recommendation device.

[0193] The above-described computer readable storage medium carries one or more programs, which, when executed by the dialogue recommendation device, cause the dialogue recommendation device to:

[0194] In response to the dialogue recommendation instruction, a first sentence input by the user is obtained, and a customer ID corresponding to a customer inputting the first sentence is obtained;

[0195] Based on the customer ID, a background label corresponding to the customer is obtained, and the first sentence is parsed to determine a business scenario and a business problem corresponding to the first sentence;

[0196] Based on the background label, the business scenario, the business problem, a preset expert rule matching engine, and a preset expert rule set, a target expert rule in the preset expert rule set is determined;

[0197] Based on the target expert rule, a recommended dialogue corresponding to the first sentence is determined, and the recommended dialogue is recommended to the customer.

[0198] Computer program code for carrying out operations of the present application can be written in one or more programming languages or combinations of languages including object oriented languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0199] The flow diagrams and the block diagrams in the drawings are illustrations of architectures, functionalities, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of present application. In this regard, each block in the flow diagrams or block diagrams can represent a module, a procedure, or a part of code, which comprises one or more executable instructions for implementing the specified functions. It should also be noted that, in some alternative implementations, the functions noted in the blocks can occur in a different order than that noted in the figures. For example, two blocks noted in succession can in fact be executed substantially concurrently or in the opposite order, depending on the functionality involved. It should also be noted that each block in the block diagrams and / or flow diagrams, and combinations of blocks in the block diagrams and / or flow diagrams, can be implemented by dedicated hardware-based systems that perform the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0200] The modules described in the embodiments of the present application can be implemented in the form of software or in the form of hardware. In some cases, the name of the module does not constitute a limitation on the module itself.

[0201] The computer readable storage medium provided by the present application is a computer readable storage medium, which stores computer readable program instructions (i.e., a computer program) for executing the above-mentioned dialogue recommendation method, and can solve the technical problems of the dialogue recommendation method. Compared with the related art, the computer readable storage medium provided by the present application has the same beneficial effects as the dialogue recommendation method provided by the above-mentioned embodiments, which will not be repeated here.

[0202] The present application also provides a computer program product, comprising a computer program, which is executed by a processor to implement the steps of the dialogue recommendation method as described above.

[0203] The computer program product provided by the present application can solve the technical problems of the dialogue recommendation method. Compared with the related art, the computer program product provided by the present application has the same beneficial effects as the dialogue recommendation method provided by the above-mentioned embodiments, which will not be repeated here.

[0204] The above only describes some embodiments of the present application, and does not limit the protection scope of the present application. Any equivalent structural transformation, direct / indirect application in other related technical fields based on the technical concept of the present application, and the contents of the present application and the accompanying drawings are included in the protection scope of the present application.

Claims

1. A method for recommending sales scripts, characterized in that, The recommended scripts include: In response to the script recommendation instruction, obtain the first input statement and the customer ID corresponding to the customer who input the first statement; Based on the customer ID, obtain the background tag corresponding to the customer, parse the first statement, and determine the business scenario and business problem corresponding to the first statement; Based on the background label, the business scenario, the business problem, the preset expert rule matching engine, and the preset expert rule set, the target expert rule in the preset expert rule set is determined; Based on the target expert rules, the recommended script corresponding to the first statement is determined, and the recommended script is recommended to the customer.

2. The script recommendation method as described in claim 1, characterized in that, The step of parsing the first statement and determining the business scenario and business problem corresponding to the first statement further includes: Parse the first statement to determine the target semantics corresponding to the first statement; Based on the target semantics, the business scenario corresponding to the first statement is determined, and a standardized business question is generated based on the preset statement standardization strategy and the target semantics.

3. The script recommendation method as described in claim 2, characterized in that, The step of parsing the first statement and determining the target semantics corresponding to the first statement further includes: Based on a preset semantic recognition engine, the first statement is semantically recognized to determine the target semantics corresponding to the first statement.

4. The script recommendation method as described in claim 1, characterized in that, The step of determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, the preset expert rule matching engine, and the preset expert rule set further includes: Determine the priority of each expert rule in the preset expert rule set; Based on the background label, the business scenario, the business problem, the preset expert rule strategy, and the priority, the preset expert rule matching engine is used to match each expert rule in the preset expert rule set with the first statement in descending order of priority until a match is successful, thereby obtaining the target expert rule that matches the first statement.

5. The script recommendation method as described in claim 1, characterized in that, The step of determining the target expert rule in the preset expert rule set based on the background label, the business scenario, the business problem, the preset expert rule matching engine, and the preset expert rule set further includes: Based on the customer ID, the business scenario, and the preset expert rule strategy, determine the first access frequency corresponding to the business problem, and based on the business scenario, the customer type, and the preset access frequency strategy, determine the access frequency threshold corresponding to the business problem. Determine whether the first access frequency is less than the access frequency threshold; If the value is less than the specified value, the target expert rule in the preset expert rule set is determined based on the background label, the business scenario, the business problem, the preset expert rule matching engine, and the preset expert rule set.

6. The script recommendation method as described in claim 1, characterized in that, The step of determining the recommended script corresponding to the first statement based on the target expert rule and recommending the recommended script to the customer further includes: Based on the target expert rules, the first recommended script corresponding to the first statement is determined; Determine whether a reinforcement learning flag exists in the preset expert rule strategy; If it exists, a second recommended script is determined based on the reinforcement learning flag; Based on a preset traffic splitting strategy, a first return probability corresponding to the first recommended message and a second return probability corresponding to the second recommended message are determined respectively. Based on the first return probability and the second return probability, one of the recommended messages, the first recommended message and the second recommended message, is set as the target recommended message, and the target recommended message is recommended to the customer.

7. The script recommendation method as described in claim 1, characterized in that, After the step of determining the recommended script corresponding to the first statement based on the target expert rule and recommending the recommended script to the customer, the method further includes: Obtain customer feedback results corresponding to the recommended script; Based on the customer feedback results, the first statement, the target expert rule, and the recommended script, a feedback database is constructed. The feedback database is used to adjust the expert rules in the preset expert rule set, and it is also used to provide training data for reinforcement learning.

8. A script recommendation device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the script recommendation method as described in any one of claims 1 to 7.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the script recommendation method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the script recommendation method as described in any one of claims 1 to 7.

Images