Intelligent customer service processing method and system

Abstract

The invention provides an intelligent customer service processing method and system, and is suitable for the technical field of data processing, and the method comprises the steps: carrying out the classification processing of a plurality of pieces of current user question information and a plurality of pieces of historical customer service dialogue information, and generating a plurality of pieces of user question category information; based on a plurality of preset user question analysis vectors, performing analysis processing on the plurality of pieces of user question category information to obtain a plurality of pieces of user question analysis information; and generating multiple pieces of current customer service processing information according to the multiple pieces of user question analysis information and the multiple pieces of user group attribute information. The present question intention and demand of the user are accurately grasped, more targeted and accurate services are provided for different user groups in combination with the past dialogue condition and group characteristics of the user, the semantic comprehension ability is effectively improved, the customer service better meets the actual demand of the user, and the user experience satisfaction degree is improved.

Description

Technical Field

[0001] This application belongs to the field of data processing technology, and in particular relates to intelligent customer service processing methods and systems. Background Technology

[0002] With the rapid development of artificial intelligence and big data technologies, intelligent customer service has gradually evolved from basic rule-matching services into a comprehensive service system that integrates natural language processing, knowledge graphs, voice interaction and other diverse technologies.

[0003] Existing technologies typically utilize natural language processing for intent recognition, analyze user input using deep learning models to attempt to determine user intent, and identify user emotional tendencies through sentiment analysis to assist in service decision-making.

[0004] However, in existing technologies, the accuracy of intent recognition is difficult to guarantee when faced with complex and ambiguous user questions, resulting in an inability to accurately grasp user needs and responses that often deviate from the core of the problem. While sentiment analysis can make basic emotion judgments, it struggles to accurately capture complex and subtle user emotions, making it impossible to flexibly and effectively adjust service strategies accordingly. Summary of the Invention

[0005] In view of this, the present application provides an intelligent customer service processing method and system, which aims to solve the problems of inaccurate understanding of intent and imprecise grasp of emotions in the prior art, resulting in poor customer service processing effects.

[0006] The first aspect of this application provides an intelligent customer service processing method, including:

[0007] Retrieve multiple current user question information, multiple historical customer service dialogue information, and multiple user group attribute information;

[0008] The multiple current user question information and multiple historical customer service dialogue information are classified and processed to generate multiple user question category information;

[0009] Based on multiple preset user question parsing vectors, the multiple user question category information is parsed and processed to obtain multiple user question parsing information;

[0010] Based on the parsing information of multiple user questions and the attribute information of multiple user groups, multiple current customer service processing information are generated.

[0011] A second aspect of this application provides an intelligent customer service processing system, including:

[0012] The information acquisition module is used to acquire multiple current user question information, multiple historical customer service dialogue information, and multiple user group attribute information;

[0013] The user question category information generation module is used to classify and process the multiple current user question information and multiple historical customer service dialogue information to generate multiple user question category information.

[0014] The user question parsing information generation module is used to parse and process the multiple user question category information based on multiple preset user question parsing vectors to obtain multiple user question parsing information;

[0015] The current customer service processing information generation module is used to generate multiple current customer service processing information based on the parsing information of multiple user questions and the attribute information of multiple user groups.

[0016] A third aspect of this application provides a terminal device, the terminal device including a memory and a processor, the memory storing a computer program executable on the processor, and the processor executing the computer program to implement the steps of the intelligent customer service processing method described in the first aspect above.

[0017] A fourth aspect of this application provides a computer-readable storage medium, comprising: storing a computer program, wherein when executed by a processor, the computer program implements the steps of the intelligent customer service processing method described in the first aspect above.

[0018] Compared with the prior art, the beneficial effects of this application's embodiments are: this application accurately grasps the user's current questioning intent and needs, combines the user's past dialogue situations and group characteristics, provides more targeted and accurate services for different user groups, effectively improves semantic understanding capabilities, simplifies the usage process for special groups, makes customer service more in line with the user's actual needs, and enhances the user experience. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the implementation process of the intelligent customer service processing method provided in Embodiment 1 of this application;

[0021] Figure 2 This is a schematic diagram illustrating the implementation process of the intelligent customer service processing method provided in Embodiment 2 of this application;

[0022] Figure 3This is a schematic diagram illustrating the implementation process of the intelligent customer service processing method provided in Embodiment 3 of this application;

[0023] Figure 4 This is a schematic diagram illustrating the implementation process of the intelligent customer service processing method provided in Embodiment 4 of this application;

[0024] Figure 5 This is a schematic diagram illustrating the implementation process of the intelligent customer service processing method provided in Embodiment 5 of this application;

[0025] Figure 6 This is a schematic diagram illustrating the implementation process of the intelligent customer service processing method provided in Embodiment Six of this application;

[0026] Figure 7 This is a schematic diagram of the implementation process of the intelligent customer service processing method provided in Embodiment 7 of this application;

[0027] Figure 8 This is a schematic diagram of the intelligent customer service processing system provided in the embodiments of this application;

[0028] Figure 9 This is a schematic diagram of the terminal device provided in the embodiments of this application. Detailed Implementation

[0029] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0030] To illustrate the technical solution described in this application, specific embodiments are provided below.

[0031] Figure 1 A flowchart illustrating the implementation of the intelligent customer service processing method provided in Embodiment 1 of this application is shown, and is described in detail below:

[0032] Step S101: Obtain multiple current user question information, multiple historical customer service dialogue information, and multiple user group attribute information.

[0033] In this embodiment, the current user question information can refer to multiple independent questions raised by the user in the form of text, voice, etc. during the current interaction with the customer service system. These questions may be supplementary explanations for the same issue, or they may be related questions from different dimensions, such as "What is the warranty period for this product?" "Can I get an invoice?" and other consecutive or related questions. These questions can be collected in real time through the interaction interface of the customer service system, including text information entered by the user on the web page or APP, voice question text converted by speech recognition technology, or real-time user question content synchronized by third-party platforms (such as social media, instant messaging tools). The system will temporarily store and organize this information according to time sequence or logical association. Historical customer service dialogue information refers to all dialogue records generated when a user interacts with the customer service system (or human customer service) in the past. This includes information such as the user's historical questions, customer service replies, the time and context of the dialogue, etc. For example, a dialogue from last month when the user inquired about product usage, or a communication record from six months ago when the user raised after-sales issues. This information can be retrieved from the system's dialogue log database. The customer service system automatically stores all users' historical interaction data and retrieves all past customer service-related dialogue content of the user by associating and querying through the user's unique identifier (such as account ID or mobile phone number), and then classifies and archives it according to timeline or business type. User group attribute information refers to group tag information categorized based on users' basic characteristics, behavioral habits, consumption preferences, etc., such as age group (youth group, elderly group), consumption level (high-net-worth customers, ordinary customers), geographical distribution (northern users, southern users), usage habits (users who prefer voice interaction, users who prefer text interaction), etc. This information is used to reflect the common characteristics of the user group. It can be collected through user registration information (such as age, address), behavioral data analysis (such as historical consumption records, interaction method preferences), third-party data cooperation (such as compliant user profile data sharing), etc. After data cleaning and tagging, a structured user group attribute database is formed. The system can match the attribute information of the user group by user identifier.

[0034] Step S102: Classify the multiple current user question information and multiple historical customer service dialogue information to generate multiple user question category information.

[0035] In this embodiment, the collected information is first comprehensively organized and preprocessed to remove invalid content such as garbled characters and repetitive expressions. Simultaneously, the audio information is converted to text to standardize the format and ensure consistency in subsequent processing. Then, key content is extracted from this preprocessed information, such as product names, service types, and keywords related to the question. This key content reflects the core focus of the user's inquiry. Next, based on preset classification standards—which are established according to common customer service scenarios and question types, such as inquiries, consultations, complaints, and suggestions—the extracted key content is compared and matched with these preset classification standards. Specifically, for each current user question and historical customer service dialogue, its category is determined according to the classification standards corresponding to its key content. If the key content of a certain piece of information simultaneously matches the characteristics of multiple categories, further judgment is made based on contextual information to select the most suitable category. Through this comparison, matching, and judgment process, each current user question and historical customer service dialogue is assigned to a corresponding category, thereby generating multiple user question category information. Each category corresponds to a batch of user questions or historical dialogue content with similar characteristics.

[0036] Step S103: Based on multiple preset user question parsing vectors, the multiple user question category information is parsed and processed to obtain multiple user question parsing information.

[0037] In this embodiment, the multiple preset user question parsing vectors can be manually set. It can begin by calling multiple pre-set user question parsing vectors in the system. These parsing vectors are constructed for different user question categories and contain common semantic features, core intents, and related contextual elements common to the corresponding category. Then, each user question category is compared one by one with the multiple preset user question parsing vectors. During the comparison, the key expressions, potential intents, and contextual logic contained in the user question category information are matched with the semantic features and intents covered in the parsing vectors to determine which parsing vector is most closely associated with the category information. Then, based on the matching results, the preset user question parsing vector with the highest correlation is used to perform in-depth analysis of the user question category information. During the parsing process, the core elements in the parsing vector are combined to extract key information points from the user's question category information, clarifying the user's specific needs, the core of the problem, and the possible direction of subsequent services. At the same time, the connection between this category information and related content in historical dialogues is sorted out. Then, each user's question category information is transformed into structured information containing specific intent, core needs, and related information. This structured information is the parsing information for multiple user questions, providing an accurate basis for generating the current customer service processing information.

[0038] Step S104: Generate multiple current customer service processing information based on the multiple user question parsing information and multiple user group attribute information.

[0039] In this embodiment, multiple user question parsing information can be integrated and associated with multiple user group attribute information to match each user's question parsing information, including its core intent, specific needs, and related information, with the corresponding group attribute information, such as age group, usage habits, and geographical distribution, forming a set of comprehensive information containing the core of the user's question and the characteristics of the group. Then, each set of integrated information is analyzed, combined with preset customer service processing rules and service strategies. These rules and strategies are formulated based on the characteristics of different user groups and common handling methods for various types of questions. During the analysis, the impact of user group attributes on the handling method will be given special consideration. For example, for question parsing information of the elderly, more concise language and clearer step guidance will be preferred; for consultations of high-net-worth customers, more professional service content will be prioritized. Then, based on the analysis results, the most suitable content will be selected from the preset reply template library and service process library. The response templates are matched based on the user's core intent, while the style and level of detail are adjusted according to group attributes. The service process is determined based on user needs and group habits, considering factors such as whether voice guidance is needed and whether steps should be simplified. Finally, the selected responses and service processes are combined to form the current customer service processing information for each user. This processing information includes both direct responses to user questions and service methods and steps, ensuring accurate fulfillment of user needs while adapting to their group characteristics, thus improving the relevance and convenience of the service.

[0040] The intelligent customer service processing method provided in this application accurately grasps the user's current questioning intent and needs, and combines the user's past dialogue and group characteristics to provide more targeted and accurate services for different user groups. It effectively improves semantic understanding capabilities, simplifies the usage process for special groups, makes customer service more in line with the user's actual needs, and enhances the user experience.

[0041] Figure 2 The flowchart illustrating the implementation of the intelligent customer service processing method provided in Embodiment 2 of this application is shown. The difference between this method and Embodiment 1 is that step S102 specifically includes:

[0042] Step S201: Based on the preset customer service dialogue standard information database, standardize the multiple current user question information and multiple historical customer service dialogue information to obtain multiple current user question standardized information and multiple historical customer service dialogue standardized information.

[0043] In this embodiment, the preset standard information library for customer service conversations can be set manually and can include common normative expressions in the customer service field, a synonym mapping table, sentence pattern conversion rules, and other content. It is possible to adjust the non-standard expressions in multiple current user question information and multiple historical customer service conversation information according to the standard information library. For example, convert the colloquial "how to do" to the standard "how to operate", and correct the misspelling "bill one" to "bill", so that all information conforms to the unified expression standard, thereby obtaining multiple standardized current user question information and multiple standardized historical customer service conversation information.

[0044] Step S202: Perform word segmentation on the multiple standardized current user question information and the multiple standardized historical customer service conversation information to obtain multiple customer service conversation vocabulary information.

[0045] In this embodiment, it is possible to use a word segmentation tool in natural language processing to split the multiple standardized current user question information and the multiple standardized historical customer service conversation information into independent word units according to semantic logic. For example, split "query product warranty period" into "query", "product", and "warranty period". Each split word is a customer service conversation vocabulary information, and finally multiple customer service conversation vocabulary information is obtained.

[0046] Step S203: Calculate multiple customer service conversation vocabulary number information and multiple customer service conversation vocabulary frequency information based on the multiple customer service conversation vocabulary information; the customer service conversation vocabulary information corresponds to the customer service conversation vocabulary number information one by one.

[0047] In this embodiment, for each customer service conversation vocabulary information, the total number of times it appears in the multiple standardized current user question information and the multiple standardized historical customer service conversation information can be counted, and this number is the corresponding customer service conversation vocabulary number information; at the same time, calculate the ratio of the number of times each customer service conversation vocabulary information appears in a single standardized information to the total number of words in this information to obtain the corresponding customer service conversation vocabulary frequency information.

[0048] Step S204: Count the quantity information of the multiple current user question information and the quantity information of the multiple historical customer service conversation information to obtain the current user question quantity information and the historical customer service conversation quantity information.

[0049] In this embodiment, count the multiple current user question information collected, and the obtained quantity is the current user question quantity information; similarly, count the multiple historical customer service conversation information, and the obtained quantity is the historical customer service conversation quantity information.

[0050] Step S205: Calculate the sum of the customer service conversation information quantities based on the current user question quantity information and the historical customer service conversation quantity information.

[0051] In this embodiment, the current number of user questions is added to the historical number of customer service conversations, and the sum of the two is the total number of customer service conversations.

[0052] Step S206: Based on the number of multiple customer service dialogue terms and the sum of customer service dialogue information, perform a division operation to obtain the generality representation information of multiple customer service dialogue terms.

[0053] In this embodiment, for each customer service dialogue term, the number of its corresponding customer service dialogue terms is divided by the sum of the number of customer service dialogue terms. The result can reflect the prevalence of the term in all dialogue terms, that is, the prevalence representation information of multiple customer service dialogue terms.

[0054] Step S207: Perform a logarithmic operation on the prevalence representation information of the multiple customer service dialogue terms to obtain multiple prevalence measurement values ​​of customer service dialogue terms.

[0055] In this embodiment, logarithmic operations are used to process the prevalence representation information of multiple customer service dialogue words. By leveraging the mathematical properties of logarithms, the numerical range of word prevalence is compressed, making it easier to distinguish and compare the differences in prevalence of different words, thereby obtaining multiple prevalence measurement values ​​of customer service dialogue words.

[0056] Step S208: Based on the frequency information of the multiple customer service dialogue words and the prevalence measurement values ​​of the multiple customer service dialogue words, generate multiple customer service dialogue word importance representation information; the customer service dialogue word information and the customer service dialogue word importance representation information correspond one-to-one.

[0057] In this embodiment, the frequency information of each customer service dialogue word is multiplied by the prevalence metric of the customer service dialogue word. The result can comprehensively reflect the frequency of the word in a specific context and its prevalence in the overall dialogue, thus serving as the representation of the importance of the word, i.e., the importance representation information of multiple customer service dialogue words.

[0058] Step S209: Encode the multiple customer service dialogue vocabulary information and the multiple customer service dialogue vocabulary importance representation information to generate multiple customer service dialogue vocabulary vector information.

[0059] In this embodiment, an encoding algorithm can be used to convert multiple customer service dialogue vocabulary information into a computable numerical form, while incorporating the corresponding customer service dialogue vocabulary importance representation information, so that each vocabulary corresponds to a vector containing its semantic and importance information, i.e., multiple customer service dialogue vocabulary vector information. The encoding algorithm can be lossless compression encoding or lossy compression encoding.

[0060] Step S210: Classify the multiple customer service dialogue word vector information to generate multiple user question category information.

[0061] In this embodiment, vectors with similar semantics and importance features can be grouped into one category based on the similarity between multiple customer service dialogue word vectors. The customer service dialogue word information corresponding to each category of vectors collectively reflects a certain type of user question content, thereby generating multiple user question category information.

[0062] The intelligent customer service processing method provided in this application improves the accuracy of user question classification, deeply mines the key information in user questions, and, combined with user group attribute information, can generate current customer service processing information that is more in line with user needs. This effectively enhances the semantic understanding ability and service targeting of intelligent customer service, and improves the user experience of different user groups.

[0063] Figure 3 The flowchart illustrating the implementation of the intelligent customer service processing method provided in Embodiment 3 of this application is shown. The difference between this method and Embodiment 2 is that step S210 specifically includes:

[0064] Step S301: Based on the preset number of customer service dialogue word categories, randomly select from the multiple customer service dialogue word vector information to obtain multiple customer service dialogue word central vector information.

[0065] In this embodiment, the preset number of customer service dialogue vocabulary categories can be manually set, or it can be a pre-set number based on common user question types, such as the number of categories corresponding to queries, inquiries, and complaints. A corresponding number of vectors can be randomly selected from multiple customer service dialogue vocabulary vector information according to this number; these selected vectors are the central vector information of multiple customer service dialogue vocabulary categories.

[0066] Step S302: Based on the central vector information of the multiple customer service dialogue words and the vector information of the multiple customer service dialogue words, calculate the spacing information of the multiple customer service dialogue words.

[0067] In this embodiment, for each customer service dialogue word vector, the distance between it and the central vector of each customer service dialogue word is calculated. This distance is used to measure the similarity between the two, and all the calculated distances constitute the spacing information between multiple customer service dialogue words. The distance can be calculated using Euclidean distance.

[0068] Step S303: Based on the spacing information between the multiple customer service dialogue words and the central vector information of the multiple customer service dialogue words, classify the vector information of the multiple customer service dialogue words to generate multiple customer service dialogue word vector category information.

[0069] In this embodiment, for each customer service dialogue word vector information, the distance information between it and the customer service dialogue word central vector information of each customer service dialogue word is compared, and the vector information is assigned to the category to which the nearest customer service dialogue word central vector information belongs, thereby dividing all customer service dialogue word vector information into different categories and forming multiple customer service dialogue word vector category information.

[0070] Step S304: Calculate the mean of multiple customer service dialogue word vector category information to obtain the central vector information of multiple customer service dialogue word vector categories.

[0071] In this embodiment, for each customer service dialogue word vector category information, the average vector of all customer service dialogue word vector information in that category is calculated, which is the central vector information of multiple customer service dialogue word vector categories.

[0072] Step S305: Determine whether the central vector information of the multiple customer service dialogue word vector categories is the same as the central vector information of the multiple customer service dialogue words; if yes, proceed to step S306; if no, proceed to step S307.

[0073] In this embodiment, the category central vector information of multiple customer service dialogue vocabulary vectors is compared with the category central vector information of multiple customer service dialogue vocabulary vectors to see if the two are consistent, thereby determining whether the category central is stable.

[0074] Step S306: Generate multiple user question category information based on the multiple customer service dialogue word vector category information.

[0075] In this embodiment, when the central vector information of the customer service dialogue word vector category is the same as the central vector information of the customer service dialogue word, it indicates that the category division has been stable. At this time, each customer service dialogue word vector category information corresponds to a batch of semantically related customer service dialogue word information. The user question types reflected by these word information are summarized to form multiple user question category information.

[0076] Step S307: The central vector information of the multiple customer service dialogue word vector categories is used as the central vector information of multiple customer service dialogue words, and the process returns to step S302.

[0077] In this embodiment, when the customer service dialogue word vector category center vector information is different from the customer service dialogue word center vector information, it indicates that the category center still needs to be adjusted. At this time, the newly obtained customer service dialogue word vector category center vector information is used as the new customer service dialogue word center vector information until the category center is stable.

[0078] The intelligent customer service processing method provided in this application improves the accuracy and stability of user question category information, making subsequent parsing and processing based on category information more closely aligned with the user's actual question intent, effectively enhancing the service accuracy and reliability of intelligent customer service, and improving user experience.

[0079] Figure 4 The flowchart illustrating the implementation of the intelligent customer service processing method provided in Embodiment 4 of this application is shown. The difference between this method and Embodiment 3 above is that step S303 specifically includes:

[0080] Step S401: Sort the spacing information of multiple customer service dialogue words corresponding to the multiple customer service dialogue word vector information in ascending order to obtain multiple customer service dialogue word spacing sequence information.

[0081] In this embodiment, the multiple customer service dialogue word spacing information corresponding to each customer service dialogue word vector information are arranged in ascending order of value to form an ordered sequence. Each such sequence is a customer service dialogue word spacing sequence information, and finally multiple customer service dialogue word spacing sequence information are obtained.

[0082] Step S402: Extract the first element of the multiple customer service dialogue word spacing sequence information to obtain multiple customer service dialogue word category confirmation spacing information.

[0083] In this embodiment, since the spacing information of each customer service dialogue word is arranged in ascending order, its first element is the minimum distance between the word vector information and the central vector information of each customer service dialogue word. This minimum distance is used to confirm the category to which the word vector information belongs. Therefore, the first element of all the sequences is extracted to obtain multiple customer service dialogue word category confirmation spacing information.

[0084] Step S403: Use the customer service dialogue word central vector information corresponding to the multiple customer service dialogue word category confirmation spacing information as the customer service dialogue word category confirmation central vector information.

[0085] In this embodiment, the confirmation gap information for each customer service dialogue word category is calculated by a certain customer service dialogue word vector information and a specific customer service dialogue word central vector information. The specific customer service dialogue word central vector information is found and used as the confirmation central vector of the category to which the word vector information belongs. All such central vector information constitutes the customer service dialogue word category confirmation central vector information.

[0086] Step S404: Use the customer service dialogue word vector information corresponding to the multiple customer service dialogue word category confirmation spacing information as the customer service dialogue word category confirmation vector information.

[0087] In this embodiment, each customer service dialogue vocabulary category confirmation gap information corresponds to a customer service dialogue vocabulary vector information. These corresponding vector information are collected to form customer service dialogue vocabulary category confirmation vector information, and these vector information will be classified into their corresponding categories.

[0088] Step S405: Generate customer service dialogue vocabulary vector category information based on the central vector information for confirming customer service dialogue vocabulary categories and the confirmation vector information for customer service dialogue vocabulary categories.

[0089] In this embodiment, the customer service dialogue vocabulary category confirmation vector information is grouped according to its corresponding customer service dialogue vocabulary category confirmation central vector information. All vector information corresponding to each central vector information constitutes a category. These categories are the customer service dialogue vocabulary vector category information. The vector information in each category is semantically similar and is most similar to the central vector information of that category.

[0090] The intelligent customer service processing method provided in this application improves the accuracy and stability of classification, making the user question category information generated based on the category information more reliable. This enhances the intelligent customer service's ability to understand and process user questions, providing users with a more accurate and efficient service experience.

[0091] Figure 5 The flowchart illustrating the implementation of the intelligent customer service processing method provided in Embodiment 5 of this application is shown. Its difference from Embodiment 1 described above lies in:

[0092] Multiple preset user question parsing vectors include preset user question intent association parsing vectors, preset user question content matching parsing vectors, and preset user question core semantic parsing vectors;

[0093] Step S103 specifically includes:

[0094] Step S501: Based on the preset user question intent association parsing vector, the multiple user question category information is parsed and processed to obtain multiple user question intent association parsing information.

[0095] In this embodiment, the preset user question intent association parsing vector can be manually set and constructed by integrating the "question-intent" mapping relationship in historical customer service dialogues. First, a large amount of historical user question information and its manually labeled intent tags (such as "query," "complaint," and "consultation") are collected. For example, 10,000 historical dialogues are selected, where "How much phone credit do I have left?" is labeled as "query" and "Why was I charged?" is labeled as "complaint." Next, these question information are segmented and key information is extracted. High-frequency feature words in each intent category are statistically analyzed (e.g., the frequency of "check," "how much," and "balance" in "query"), and each feature word is assigned a weight (the higher the frequency, the greater the weight). Finally, the feature words and their weights of the same intent category are combined into a vector. For example, the "query" intent association parsing vector contains elements such as "check: 0.8," "balance: 0.7," and "how much: 0.6," forming a vector that can represent the association between the intent and the question features. Each user's question category information can be compared with a preset user question intent association parsing vector to analyze the potential user intent contained in the category information, identify the user's purpose and expected result, such as whether the user wants to query product information, seek technical support, or make a complaint or suggestion, thereby generating multiple user question intent association parsing information, which clarifies the intent behind each user's question category information.

[0096] In this embodiment, the preset user question intent association parsing vector can focus on the intent association features behind user questions, including common intent directions, contextual logical connections, and potential relationships between intents in different question categories. During processing, each user question category information is compared with this parsing vector to mine the implicit intent tendencies in the category information, analyze the possible purpose and expected results of the user's question, for example, parsing the user's intent to focus on after-sales service from the category information of "checking the product warranty period," thereby forming multiple user question intent association parsing information, clearly presenting the intent direction corresponding to each category information.

[0097] Step S502: Based on the preset user question content matching parsing vector, the multiple user question category information is parsed to obtain multiple user question content matching parsing information.

[0098] In this embodiment, the preset user question content matching parsing vector can be manually set and constructed based on standardized "question-answer" pairing data. First, common standard questions and corresponding answers from the customer service knowledge base are organized. For example, "How to check the bill" corresponds to "Log in to the APP → My → Bill," and "How long is the product warranty period?" corresponds to "The warranty period for electronic products is 1 year," forming a standardized dataset containing 5000 sets of "question-answer." Then, the standard questions are segmented and semantically encoded, converting each question into a vector containing keywords and their location information (e.g., "How to check the bill" is converted into a vector containing "How: 0.5," "Query: 0.9," and "Bill: 1.0"). Simultaneously, core information from the answer (e.g., "APP path," "1 year") is recorded. Finally, the question vector is bound to the core information of the answer to form a content matching parsing vector. For example, the parsing vector for "How to check the bill" includes both the question keyword weights and the answer features associated with "APP → My → Bill," ensuring accurate matching between user question content and standard answers. Each user's question category information can be compared with a preset user question content matching parsing vector to find the standard questions and answers that best match the category information. The specific content of the user's question can be analyzed to determine the product or service details that the user is concerned about, such as product functions, usage methods, after-sales service, etc., and then multiple user question content matching parsing information can be generated. This information provides reference content that is highly relevant to the user's question content.

[0099] In this embodiment, the preset user question content matching parsing vector focuses on the specific content features of user questions, covering common core vocabulary, expression methods, and matching relationships with business knowledge in various question categories. The system compares each user question category information with this parsing vector, extracts specific content elements from the category information, and matches them with related business information. For example, for the category information of "consulting on invoice issuance," it parses out the invoice types and issuance processes that the user is interested in, thereby generating multiple user question content matching parsing information that accurately reflects the specific content details contained in each category information.

[0100] Step S503: Based on the preset user question core semantic parsing vector, the multiple user question category information is parsed and processed to obtain multiple user question core semantic parsing information.

[0101] In this embodiment, the preset core semantic parsing vector for user questions can be manually set and constructed by capturing common core semantics under different expressions. First, diverse question instances with the same core semantics are collected. For example, for the core semantic of "check phone bill," 500 different expressions such as "how much phone bill is left," "my phone bill balance," and "check my phone bill" are collected. Next, these instances are processed by synonym conversion, mapping non-standard expressions to standard core semantics (e.g., "how much phone bill is left" is mapped to "check phone bill balance"), and the variant forms of core words in each expression are counted (e.g., "phone bill," "phone bill balance," and "phone bill" are considered synonymous core words). Finally, based on the standard core semantics, the feature words of all synonymous expressions and their mapping relationships are integrated into a vector. For example, the core semantic parsing vector of "check phone bill" includes core word weights such as "phone bill: 1.0," "balance: 0.9," and "check: 0.8," as well as synonym mapping rules such as "how much is left → check balance" and "phone bill → phone bill," achieving accurate capture of core semantics under different expressions. Each user's question category information can be compared with a preset user question core semantic parsing vector to extract the core semantic elements in that category information. Ignoring superficial differences in expression, the core needs of the user's question can be explored. For example, the core problem can be extracted from the user's complex expression to clarify the problem that the user really wants to solve. This generates multiple user question core semantic parsing information that accurately reflects the core content of the user's question.

[0102] In this embodiment, the preset core semantic parsing vector for user questions focuses on capturing the core semantic features of user questions, including the essential meaning under different expressions, key semantic elements, and the inherent relationships between semantics. During processing, each user question category information is compared with this parsing vector, stripping away surface expression differences to extract the core semantics of the category information. For example, from similar questions such as "Can this be issued with a formal invoice?" and "How do I get an invoice?", the core semantics of "consulting on invoice issuance" are parsed out, thereby obtaining multiple user question core semantic parsing information and accurately grasping the essential meaning of each category information.

[0103] Step S504: Generate joint information of multiple user question intents and contents based on the multiple user question intent association parsing information and the multiple user question content matching parsing information.

[0104] In this embodiment, the parsing information associated with each user's question intent is integrated with the parsing information matching the corresponding user's question content. This combines the user's intent with the specific question content, forming a set of comprehensive information containing both user intent and detailed content. This integration allows for a more comprehensive understanding of user needs. For example, combining a user's intent to inquire about a product's warranty period with specific product models, purchase dates, and other information generates combined information on multiple user question intents. This information provides richer evidence for generating more accurate customer service processing information subsequently.

[0105] In this embodiment, the parsing information associated with each user's question intent can be integrated with the parsing information matching the corresponding user's question content. This combines the intent direction with specific content elements to form comprehensive information that possesses both intent and content characteristics. For example, combining the intent information of "paying attention to after-sales service" with the content information of "product warranty period and warranty scope" generates joint user question intent content information that reflects both user intent and specific content, providing richer evidence for a comprehensive understanding of user needs.

[0106] Step S505: Based on the core semantic parsing information of the multiple user questions and the joint information of the intent content of the multiple user questions, obtain the parsing information of the multiple user questions.

[0107] In this embodiment, the core semantic parsing information of each user's question can be deeply integrated with the corresponding user's question intent content information. By comprehensively considering the core semantics, intent, and specific content of the user's question, the specific needs of the user, the core of the problem, and the possible directions of subsequent services can be further clarified. Through this integration, ambiguity between information can be eliminated, and the user's true needs can be grasped more accurately. For example, in the case of vague expression by the user, the specific question of the user can be determined by combining the core semantics, intent, and content information, thereby obtaining multiple user question parsing information. This information provides a solid foundation for generating accurate customer service processing information.

[0108] In this embodiment, the core semantic parsing information of each user's question can be deeply integrated with the corresponding user's question intent content. By combining the core semantics, intent direction, and specific content, the user's true needs, the core points of the problem, and the possible service directions can be further identified. For example, by combining the core semantics of "consulting on invoice issuance," the intent of "needing reimbursement," and the content information of "invoice type and issuance process," user question parsing information containing the user's complete needs can be formed, providing an accurate and comprehensive basis for generating the current customer service processing information.

[0109] The intelligent customer service processing method provided in this application analyzes user question category information from three dimensions: intent association, content matching, and core semantics. The analysis results are gradually integrated to achieve a more comprehensive and in-depth understanding of user questions, effectively improving the accuracy and richness of user question analysis information. This makes the subsequent generated current customer service processing information more in line with the user's actual needs, thereby enhancing the service targeting and reliability of intelligent customer service and improving user experience.

[0110] Figure 6 The flowchart illustrating the implementation of the intelligent customer service processing method provided in Embodiment Six of this application is shown. Its difference from Embodiment One described above lies in:

[0111] Multiple preset user question parsing vectors include preset user question logic parsing vectors, preset user question reasoning parsing vectors, and preset user question dialogue interaction parsing vectors;

[0112] Step S103 specifically includes:

[0113] Step S601: Based on the multiple user question category information and the preset user question logic parsing vector, obtain multiple user question logic parsing information.

[0114] In this embodiment, the preset user question logic parsing vector can be manually set or constructed by analyzing the logical structure of user questions in historical dialogues. First, 5000 historical user questions containing clear logical relationships are collected, and their logical types (such as causal, parallel, and progressive relationships) and corresponding logical markers (such as "because…therefore…", "both…and…", "not only…but also…") are manually labeled. For example, "Because the product cannot be turned on, I want to apply for after-sales service" is labeled as a "causal relationship," with the logical marker "because…therefore…"; "I want to know the price and also want to know about promotional activities" is labeled as a "parallel relationship," with the logical marker "both…and…". Next, the frequently occurring logical markers and related words in each logical type are statistically analyzed, and weights are assigned to the logical markers (the higher the frequency, the greater the weight), such as "because: 0.9", "therefore: 0.8", and "leads to: 0.7" in the "causal relationship". Finally, the marker words, weights, and corresponding logical structure patterns of the same logical type are combined into a vector to form a preset user question logic parsing vector that can represent the characteristics of that logical type.

[0115] In this embodiment, the preset user question logic parsing vector can focus on the logical structure features of user questions, including common logical association patterns such as causal relationships, progressive relationships, and parallel relationships in different types of questions, as well as the connecting words and sentence structures that embody these logics. During processing, each user question category information is compared with the parsing vector to analyze the logical relationships between the content elements in the category information. For example, from the category information "The product is broken and the warranty period has not expired, what should I do?", the parallel relationship between "product damage" and "within the warranty period" is parsed, as well as the causal logic that both point to "after-sales service", thereby obtaining multiple user question logic parsing information and clearly presenting the internal logical architecture of each category information.

[0116] Step S602: Based on the multiple user question category information and the preset user question reasoning parsing vector, obtain multiple user question reasoning parsing information.

[0117] In this embodiment, the preset user question inference parsing vector can be manually set, or it can be constructed based on the mapping relationship of "question-implicit need" in historical dialogues. First, 8,000 historical user questions and manually labeled implicit needs are selected. For example, "Does this product drain the battery quickly?" corresponds to the implicit need "concerned about the product's battery life and possibly concerned about daily use convenience"; "Are there any offline stores nearby?" corresponds to the implicit need "want to experience the product in person and may need offline after-sales service". Then, the question content is segmented, core words are extracted (such as "drains the battery quickly" and "offline stores"), and the association frequency between core words and implicit needs is counted. High weights are assigned to closely related words (such as the association weight of "drains the battery quickly" and "battery life" is 0.85). At the same time, common inference rules are sorted out (such as "asking about store location → may need on-site service"). Finally, the core vocabulary, associated weights, and inference rules are integrated into vectors. For example, the inference parsing vector for "asking about power consumption-related issues" includes inference rules such as "power consumption: 0.9", "fast: 0.8", "battery life: 0.85", and "focus on ease of use", forming a preset user question inference parsing vector.

[0118] In this embodiment, the preset user question reasoning parsing vector focuses on features that allow for reasonable inferences based on the content of user questions. This includes the underlying preconditions, potential needs, and possible follow-up questions implied in different question categories. The system compares each user question category with this parsing vector, combining common sense and business knowledge to extend the reasoning. For example, from the category information "Does this product support seven-day no-reason returns?", it infers that the user may have just purchased the product, is dissatisfied with it, or wants to understand the return process, thus generating multiple user question reasoning parsing messages and uncovering implicit content related to the question that the user did not directly state.

[0119] Step S603: Based on the multiple user question category information and the preset user question dialogue interaction parsing vector, obtain multiple user question dialogue interaction parsing information.

[0120] In this embodiment, the preset user question dialogue interaction parsing vector can be manually set or constructed by analyzing the interaction flow characteristics of historical dialogues. First, 10,000 complete customer service dialogue records are collected, and the interaction patterns of user questions and customer service responses are extracted. The interaction requirements corresponding to user questions are labeled (such as "step-by-step answer needed," "confirmation needed," and "guided operation needed"). For example, "explain how to get a refund first, then explain the arrival time" corresponds to the "step-by-step answer" requirement; "Is the address you mentioned XX Road?" corresponds to the "confirmation" requirement. Next, the expression characteristics of user questions in different interaction requirements are statistically analyzed (such as "first...then..." corresponding to step-by-step answer, "is it...?" corresponding to confirmation), and the frequency of these expression characteristics is calculated, assigning weights to high-frequency characteristics (such as "first...then..." having a weight of 0.9). At the same time, the corresponding customer service response strategy is associated (such as requiring step-by-step answers to list the steps in sequence). Finally, the expression features, weights, and response strategies are combined into vectors to form a preset user question dialogue interaction parsing vector. For example, the "step-by-step answer" interaction parsing vector includes the strategies of "first: 0.85", "second: 0.85", and "respond in sequence according to steps".

[0121] In this embodiment, the preset user question dialogue interaction parsing vector can focus on the interaction characteristics of user questions in the dialogue scenario, including common response directions corresponding to different types of questions, the interaction methods that users may expect, and the association rules of the dialogue context. During processing, the parsing vector is compared with the parsing information of each user question category to analyze the positioning of the user question in the dialogue flow and the required interaction response. For example, for the category information "Tell me the warranty period first, and then explain how to apply for warranty", the user's interaction requirement of obtaining information in sequence and the feature that the dialogue needs to be promoted step by step are parsed to obtain multiple user question dialogue interaction parsing information, and to clarify the dialogue interaction logic and response method corresponding to each category of information.

[0122] Step S604: Generate multiple user question logic reasoning parsing information based on the multiple user question logic parsing information and the multiple user question reasoning parsing information.

[0123] In this embodiment, the logical parsing information for each user question can be integrated with its corresponding inference parsing information for the user question. This combines the inherent logic of the user question with logically-based extended reasoning to form comprehensive information that integrates logical structure and implicit needs. For example, combining the logical relationship of "the product is damaged and within the warranty period" with the inference content of "the user may need to repair or replace the product" generates logical inference parsing information for the user question that reflects both the question logic and the inference result, providing a deeper basis for comprehensively understanding user needs.

[0124] Step S605: Based on the multiple user question dialogue interaction parsing information and the multiple user question logical reasoning parsing information, obtain multiple user question parsing information.

[0125] In this embodiment, the interactive dialogue parsing information for each user's question can be deeply integrated with the corresponding logical reasoning parsing information for the user's question. By combining the three aspects of dialogue interaction needs, question logic structure, and implicit reasoning content, the complete user needs, the logical chain of the problem, and the appropriate dialogue response strategy can be further identified. For example, by combining the interaction feature of "users asking for warranty information in sequence," the causal logic of "product under warranty and after-sales processing," and the reasoning result that "users may need specific operation guidance," user question parsing information containing user needs, logical relationships, and interaction methods can be formed, providing an accurate and comprehensive basis for generating current customer service processing information.

[0126] The intelligent customer service processing method provided in this application analyzes user question category information from three dimensions: logical structure, extended reasoning, and dialogue interaction. The analysis results are gradually integrated to achieve a more comprehensive and in-depth understanding of user questions, effectively improving the completeness and accuracy of user question analysis information. This makes the subsequently generated current customer service processing information more in line with the user's logical habits and interaction needs, thereby enhancing the service logic and interactivity of intelligent customer service and significantly improving the user experience.

[0127] Figure 7 The flowchart illustrating the implementation of the intelligent customer service processing method provided in Embodiment Seven of this application is shown. The difference between this method and Embodiment One is that step S104 specifically includes:

[0128] Step S701: The multiple user question parsing information and the multiple user group attribute information are fused together to obtain multiple user group question parsing information.

[0129] In this embodiment, each user's question analysis information can be linked one-to-one with its corresponding user group attribute information. This combines the user's specific needs and core intentions with the age, usage habits, and consumption level of the user's group. For example, the analysis information for a user's question about "consulting about product warranty periods" can be integrated with the group attribute information of "elderly group" and "preference for voice interaction," clarifying the elderly user's concern about after-sales service and their need for voice services. Similarly, the analysis information for "inquiring about high-end product prices" can be integrated with the group attribute information of "high-net-worth clients," highlighting this group's potential demand for product value and exclusive services. Through this integration, multiple user group question analysis information are formed, containing both individual user question details and reflecting common group characteristics, providing a comprehensive basis for subsequent targeted processing.

[0130] Step S702: Generate weighted information for multiple user group questions based on the parsing information of the multiple user group questions and the preset user customer service processing weight information.

[0131] In this embodiment, the preset user customer service processing weight information can be manually set or constructed by analyzing the differences in attention given to customer service elements by different user groups. First, 10,000 user feedback data on customer service are collected, and user group attributes (such as age group, consumption level, interaction preference) and corresponding satisfaction / dissatisfaction items are labeled. For example, the satisfaction rate of the elderly group for "clarity of operation steps" is 85%, while "confused steps" accounts for 60% of the dissatisfaction items; the satisfaction rate of high-net-worth customers for "service professionalism" is 90%, while "unprofessional response" accounts for 55% of the dissatisfaction items; the satisfaction rate of ordinary users for "response speed" is 80%, while "long waiting time" accounts for 70% of the dissatisfaction items. Next, the frequency of attention paid to service elements (such as clarity, professionalism, efficiency, and personalization) by each group is statistically analyzed, and weighted according to frequency (the total weight sums to 1). For example, for the elderly: clarity of operation steps 0.4, conciseness of language 0.3, voice guidance 0.2, other 0.1; for high-net-worth clients: professional depth 0.4, exclusive service 0.3, response speed 0.2, other 0.1; for ordinary users: response efficiency 0.4, information completeness 0.3, ease of interaction 0.2, other 0.1. Finally, the group attributes are associated with the corresponding weight values ​​to form preset user customer service processing weight information, ensuring that the priority of service elements matches the needs of the group.

[0132] In this embodiment, the preset user customer service processing weight information is based on pre-defined importance parameters according to different user group attributes and question types. For example, a higher weight is assigned to "clarity of operation steps" for the elderly (e.g., 0.9), a higher weight is assigned to "service professionalism" for high-net-worth clients (e.g., 0.85), and a higher weight is assigned to "response efficiency" for ordinary users (e.g., 0.8). The system will correlate and calculate the key elements in the question analysis information of each user group with the corresponding preset weights. For example, for after-sales consultations for the elderly, the weight of "step guidance" will be emphasized; for product consultations for high-net-worth clients, the weight of "professional parameter interpretation" will be emphasized. Through this weighted processing, multiple user group question weight information is generated, highlighting the key processing elements in each piece of information according to the importance of the group's needs.

[0133] Step S703: Generate multiple user group question conversion information based on the weighted information of the multiple user group questions and the preset user customer service processing bias information.

[0134] In this embodiment, the preset user customer service processing bias information can be manually set or constructed based on the behavioral habits and preference characteristics of user groups. First, 50,000 user-customer service interaction records are collected, and their usage habits are statistically analyzed according to group attributes. For example, among "users who prefer voice interaction," 90% of their questions are expressed in colloquial language, and 70% hope that the reply includes voice options; among "young people," 85% prefer short sentence replies, and 60% habitually use emoticons to aid understanding; among "northern users," 75% of their questions contain regional dialect words (such as "zha zheng" and "she shi hui"). Next, adjustment rules are set for the habit characteristics of different groups, such as setting a "colloquialization conversion bias" for voice-preferred groups (converting written language to "do you think this operation is okay?"), setting a "short sentence splitting bias" for young people (breaking long sentences into short sentences of no more than 3 sentences), and setting a "dialect mapping bias" for northern users (converting "zha zheng" to "how to operate" while retaining the dialect's friendliness). Simultaneously, trigger thresholds are set for each bias rule (e.g., if a certain habit accounts for more than 60% of the group, it is activated). For example, when it is detected that a user "prefers voice interaction" and 90% of the group uses colloquial expressions, a colloquial conversion bias is automatically triggered. Finally, the group attributes, habit characteristics, bias rules, and trigger thresholds are integrated to form preset user customer service handling bias information, enabling adaptive adjustments to service content.

[0135] In this embodiment, the preset user customer service processing bias information consists of adjustment parameters set for the habits and needs of different user groups. These parameters optimize the presentation of processed information. For example, a "concise text template" bias is set for groups that prefer text interaction, a "colloquial expression" bias is set for groups that prefer voice interaction, and a "regionalized vocabulary" bias (such as using commonly used local terms) is set for users in northern regions. The system adjusts the processing direction based on the group attributes in the parsed information of each user group's question, using the corresponding preset bias information. For example, the weighted information for elderly users is converted into a "voice broadcast + large font steps" format, and the weighted information for users in southern regions is converted into an expression style adapted to their language habits. This conversion generates multiple user group question conversion information, making the processed content more aligned with the user groups' usage habits and preferences.

[0136] Step S704: Integrate and process the question conversion information of the multiple user groups to generate multiple current customer service processing information.

[0137] In this embodiment, the various processing elements (such as response content, service methods, and operation instructions) in the conversion information of each user group's inquiry can be sorted and combined to form a complete customer service response plan. For example, for the conversion information of the elderly, the content of "voice reply on warranty duration + step-by-step voice guidance on the application process" can be integrated; for the conversion information of high-net-worth customers, the content of "dedicated consultant contact information + detailed product parameter manual + on-site demonstration appointment channel" can be integrated; and for the conversion information of ordinary young users, the content of "concise text reply on price + APP ordering link" can be integrated. Through this integration, it is ensured that each current customer service processing information not only contains the core content of responding to the user's inquiry, but also covers the service form and additional content adapted to the attributes of the group, forming a targeted, complete and efficient processing plan.

[0138] The intelligent customer service processing method provided in this application embodiment enables the generated current customer service processing information to not only accurately respond to the individual needs of users, but also to deeply adapt to the common characteristics and habits of their respective groups, thereby improving the personalization and accuracy of intelligent customer service, effectively meeting the differentiated needs of different groups, enhancing user satisfaction and acceptance of services, and significantly improving the overall service quality.

[0139] Corresponding to the method in the above embodiments, Figure 8 The diagram shows a structural block diagram of the intelligent customer service processing system provided in the embodiments of this application. For ease of explanation, only the parts related to the embodiments of this application are shown. Figure 8 The example intelligent customer service processing system can be the executing entity of the intelligent customer service processing method provided in the aforementioned embodiment one.

[0140] Reference Figure 8 The intelligent customer service system includes:

[0141] The information acquisition module 810 is used to acquire multiple current user question information, multiple historical customer service dialogue information, and multiple user group attribute information.

[0142] The user question category information generation module 820 is used to classify and process the multiple current user question information and multiple historical customer service dialogue information to generate multiple user question category information.

[0143] The user question parsing information generation module 830 is used to parse and process the multiple user question category information based on multiple preset user question parsing vectors to obtain multiple user question parsing information.

[0144] The current customer service processing information generation module 840 is used to generate multiple current customer service processing information based on the multiple user question parsing information and multiple user group attribute information.

[0145] For details on how each module in the intelligent customer service processing system provided in this application implements its respective function, please refer to the foregoing. Figure 1 The description of Embodiment 1 shown will not be repeated here.

[0146] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0147] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0148] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0149] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0150] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance. It should also be understood that although the terms "first," "second," etc., are used in the text to describe various elements in some embodiments of this application, these elements should not be limited by these terms. These terms are merely used to distinguish one element from another. For example, a first table may be named a second table, and similarly, a second table may be named a first table, without departing from the scope of the various described embodiments. Both the first table and the second table are tables, but they are not the same table.

[0151] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0152] The intelligent customer service processing method provided in this application embodiment can be applied to terminal devices such as mobile phones, tablets, wearable devices, in-vehicle devices, augmented reality (AR) / virtual reality (VR) devices, laptops, ultra-mobile personal computers (UMPCs), netbooks, and personal digital assistants (PDAs). This application embodiment does not impose any restrictions on the specific type of terminal device.

[0153] For example, the terminal device may be a station (STAION, ST) in a WLAN, a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA) device, a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, an in-vehicle device, a vehicle networking terminal, a computer, a laptop computer, a handheld communication device, a handheld computing device, a satellite wireless device, a wireless modem card, a set-top box (STB), customer premises equipment (CPE), and / or other devices used for communication over a wireless system, as well as next-generation communication systems, such as mobile terminals in 5G networks or mobile terminals in future evolved Public Land Mobile Network (PLMN) networks.

[0154] As an example and not a limitation, when the terminal device is a wearable device, the term "wearable device" can also refer to any device that utilizes wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices worn directly on the body or integrated into a user's clothing or accessories. Wearable devices are not merely hardware devices; they achieve powerful functions through software support, data interaction, and cloud interaction. Broadly defined, wearable smart devices include those with comprehensive functions, large sizes, and the ability to perform complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses, as well as those focused on a specific application function that require interaction with other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0155] Figure 9 This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application. For example... Figure 9 As shown, the terminal device 9 of this embodiment includes: at least one processor 90 ( Figure 9 Only one is shown in the image), and a memory 91 stores a computer program 92 that can run on the processor 90. When the processor 90 executes the computer program 92, it implements the steps in the various intelligent customer service processing method embodiments described above, for example... Figure 1 Steps S101 to S104 are shown. Alternatively, when the processor 90 executes the computer program 92, it implements the functions of each module / unit in the above system embodiments, for example... Figure 8The functions of modules 810 to 840 are shown.

[0156] The terminal device 9 can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor 90 and a memory 91. Those skilled in the art will understand that... Figure 9 This is merely an example of terminal device 9 and does not constitute a limitation on terminal device 9. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal device may also include input transmission devices, network access devices, buses, etc.

[0157] The processor 90 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0158] In some embodiments, the memory 91 may be an internal storage unit of the terminal device 9, such as a hard disk or memory of the terminal device 9. The memory 91 may also be an external storage device of the terminal device 9, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal device 9. Furthermore, the memory 91 may include both internal and external storage units of the terminal device 9. The memory 91 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory 91 can also be used to temporarily store data that has been sent or will be sent.

[0159] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0160] This application also provides a terminal device, which includes at least one memory, at least one processor, and a computer program stored in the at least one memory and executable on the at least one processor. When the processor executes the computer program, it causes the terminal device to implement the steps in any of the above method embodiments.

[0161] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0162] This application provides a computer program product that, when run on a terminal device, enables the terminal device to implement the steps described in the various method embodiments above.

[0163] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0164] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0165] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0166] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0167] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. An intelligent customer service processing method, characterized by, The method comprises the following steps: Obtain a plurality of current user question information, a plurality of historical customer service dialogue information and a plurality of user group attribute information; Classify the plurality of current user question information and the plurality of historical customer service dialogue information to generate a plurality of user question category information; Based on a plurality of preset user question analysis vectors, the plurality of user question category information is analyzed to obtain a plurality of user question analysis information; According to the plurality of user question analysis information and the plurality of user group attribute information, a plurality of current customer service processing information is generated.

2. The intelligent customer service processing method of claim 1, wherein, The step of classifying the plurality of current user question information and the plurality of historical customer service dialogue information to generate a plurality of user question category information specifically comprises: According to the preset customer service dialogue standard information library, the plurality of current user question information and the plurality of historical customer service dialogue information are standardized to obtain a plurality of current user question standardized information and a plurality of historical customer service dialogue standardized information; The plurality of current user question standardized information and the plurality of historical customer service dialogue standardized information are segmented to obtain a plurality of customer service dialogue vocabulary information; According to the plurality of customer service dialogue vocabulary information, the number of customer service dialogue vocabulary information and the frequency of customer service dialogue vocabulary information are calculated; the customer service dialogue vocabulary information and the customer service dialogue vocabulary information correspond one by one; The number of the plurality of current user question information and the number of the plurality of historical customer service dialogue information are counted to obtain the current user question number information and the historical customer service dialogue number information; According to the current user question number information and the historical customer service dialogue number information, the sum of customer service dialogue information is calculated; According to the plurality of customer service dialogue vocabulary number information and the sum of customer service dialogue information, the division operation is carried out to obtain a plurality of customer service dialogue vocabulary universality representation information; The plurality of customer service dialogue vocabulary universality representation information is logarithmically operated to obtain a plurality of customer service dialogue vocabulary universality measurement values; According to the plurality of customer service dialogue vocabulary frequency information and the plurality of customer service dialogue vocabulary universality measurement values, a plurality of customer service dialogue vocabulary importance representation information is generated; the customer service dialogue vocabulary information and the customer service dialogue vocabulary importance representation information correspond one by one; The plurality of customer service dialogue vocabulary information and the plurality of customer service dialogue vocabulary importance representation information are encoded to generate a plurality of customer service dialogue vocabulary vector information; The plurality of customer service dialogue vocabulary vector information is classified to generate a plurality of user question category information.

3. The intelligent customer service processing method of claim 2, wherein, The step of classifying the plurality of customer service dialogue vocabulary vector information to generate a plurality of user question category information specifically comprises: According to the preset customer service dialogue vocabulary category number information, the plurality of customer service dialogue vocabulary vector information is randomly selected to obtain a plurality of customer service dialogue vocabulary hub vector information; According to the plurality of customer service dialogue vocabulary hub vector information and the plurality of customer service dialogue vocabulary vector information, a plurality of customer service dialogue vocabulary distance information is calculated; According to the plurality of customer service dialogue word distance information and the plurality of customer service dialogue word hub vector information, the plurality of customer service dialogue word vector information is classified and processed to generate plurality of customer service dialogue word vector category information; The mean of the plurality of customer service dialogue word vector category information is calculated to obtain the plurality of customer service dialogue word vector category hub vector information; It is judged whether the plurality of customer service dialogue word vector category hub vector information is the same as the plurality of customer service dialogue word hub vector information; If yes, the plurality of user question category information is generated according to the plurality of customer service dialogue word vector category information; If no, the plurality of customer service dialogue word vector category hub vector information is taken as the plurality of customer service dialogue word hub vector information, and the step of calculating the plurality of customer service dialogue word distance information according to the plurality of customer service dialogue word hub vector information and the plurality of customer service dialogue word vector information is returned.

4. The intelligent customer handling method of claim 3, wherein, The step of classifying and processing the plurality of customer service dialogue word vector information according to the plurality of customer service dialogue word distance information and the plurality of customer service dialogue word hub vector information to generate the plurality of customer service dialogue word vector category information specifically includes: The plurality of customer service dialogue word distance information corresponding to the plurality of customer service dialogue word vector information is arranged in ascending order to obtain the plurality of customer service dialogue word distance sequence information; The first element of the plurality of customer service dialogue word distance sequence information is extracted to obtain the plurality of customer service dialogue word category confirmation distance information; The customer service dialogue word hub vector information corresponding to the plurality of customer service dialogue word category confirmation distance information is taken as the customer service dialogue word category confirmation hub vector information; The customer service dialogue word vector information corresponding to the plurality of customer service dialogue word category confirmation distance information is taken as the customer service dialogue word category confirmation vector information; The customer service dialogue word vector category information is generated according to the customer service dialogue word category confirmation hub vector information and the customer service dialogue word category confirmation vector information.

5. The intelligent customer service processing method of claim 1, wherein The plurality of preset user question analysis vectors includes a preset user question intention association analysis vector, a preset user question content matching analysis vector, and a preset user question core semantic analysis vector; The step of analyzing and processing the plurality of user question category information based on the plurality of preset user question analysis vectors to obtain the plurality of user question analysis information specifically includes: The plurality of user question intention association analysis information is obtained by analyzing and processing the plurality of user question category information according to the preset user question intention association analysis vector; The plurality of user question content matching analysis information is obtained by analyzing and processing the plurality of user question category information according to the preset user question content matching analysis vector; The plurality of user question core semantic analysis information is obtained by analyzing and processing the plurality of user question category information according to the preset user question core semantic analysis vector; According to the multiple user question intention association analysis information and the multiple user question content matching analysis information, multiple user question intention content joint information is generated; According to the multiple user question core semantic analysis information and the multiple user question intention content joint information, multiple user question analysis information is obtained.

6. The intelligent customer service processing method of claim 1, wherein, The multiple preset user question analysis vectors include a preset user question logic analysis vector, a preset user question reasoning analysis vector, and a preset user question dialogue interaction analysis vector; The step of analyzing and processing the multiple user question category information based on the multiple preset user question analysis vectors to obtain multiple user question analysis information specifically includes: According to the multiple user question category information and the preset user question logic analysis vector, multiple user question logic analysis information is obtained; According to the multiple user question category information and the preset user question reasoning analysis vector, multiple user question reasoning analysis information is obtained; According to the multiple user question category information and the preset user question dialogue interaction analysis vector, multiple user question dialogue interaction analysis information is obtained; According to the multiple user question logic analysis information and the multiple user question reasoning analysis information, multiple user question logic reasoning analysis information is generated; According to the multiple user question dialogue interaction analysis information and the multiple user question logic reasoning analysis information, multiple user question analysis information is obtained.

7. The intelligent customer handling method of claim 1, wherein, The step of generating multiple current customer service processing information according to the multiple user question analysis information and the multiple user group attribute information specifically includes: The multiple user question analysis information and the multiple user group attribute information are fused to obtain multiple user group question analysis information; According to the multiple user group question analysis information and preset user customer service processing weight information, multiple user group question weighted information is generated; According to the multiple user group question weighted information and preset user customer service processing bias information, multiple user group question conversion information is generated; The multiple user group question conversion information is integrated to generate multiple current customer service processing information.

8. An intelligent customer service processing system, characterized by, It includes: An information acquisition module is configured to acquire multiple current user question information, multiple historical customer service dialogue information, and multiple user group attribute information; A user question category information generation module is configured to classify the multiple current user question information and the multiple historical customer service dialogue information to generate multiple user question category information; A user question analysis information generation module is configured to analyze and process the multiple user question category information based on multiple preset user question analysis vectors to obtain multiple user question analysis information; A current customer service processing information generation module is configured to generate multiple current customer service processing information according to the multiple user question analysis information and the multiple user group attribute information.

9. A terminal device, comprising: The terminal device comprises a memory, a processor, the memory has a computer program stored thereon, the computer program is executable on the processor, and the processor implements the steps of the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1-9. The computer program is executable on the processor, and the processor implements the steps of the method according to any one of claims 1 to 7 when executing the computer program.

Images