Complaint service intelligent identification processing method and system based on large model

Abstract

The invention relates to the technical field of computers, in particular to a complaint service intelligent identification processing method and system based on a large model, which can remarkably improve the complaint service processing efficiency. The complaint service intelligent identification processing method based on the large model comprises the steps of performing vectorization processing on customer information according to a complaint service to form a structured text of the customer information and determining semantic coordinates of vector similarity; converting the complaint description data into a high-order vector form according to a customer database and a vectorization representation technology based on a large model, and extracting complaint service key information; performing query screening by utilizing a structured query language according to the key information of the complaint service so as to reduce a matching search range; according to the query screening result, vector similarity calculation is carried out relative to the customer information structured text, and a product instance candidate list matched or close to complaint description data semantics is obtained according to the size; and selecting a product instance from the product instance candidate list as a reference for processing and replying the complaint service.

Description

Technical Field

[0001] One or more embodiments of the present invention relate to the field of computer technology, and in particular to a method and system for intelligent identification and processing of complaint business based on a large model. Background Technology

[0002] With the rapid development of knowledge bases and large model technologies, information retrieval methods based on vector representation have become the core technology of the new generation of search systems. However, their application in the communications industry has not been entirely satisfactory, especially in scenarios that contain a lot of communications terminology (customer and customer abbreviation, fault type, service type, address and abbreviated address, etc.), where accurate search matching is often impossible.

[0003] When dealing with complaints from corporate clients, installation and maintenance systems or personnel often lack the means to quickly locate specific fault reports. They often need to seek support through multiple channels or maintain and record customer information offline, resulting in poor overall efficiency and support effectiveness. They are often unable to respond to customer complaints in a timely manner, which leads to poor complaint handling efficiency and fails to meet the goal of improving the quality and efficiency of government and enterprise business support. Summary of the Invention

[0004] One or more embodiments of the present invention describe a method and system for intelligent identification and processing of complaint business based on a large model, which can significantly improve the efficiency of complaint business processing.

[0005] According to an embodiment of a first aspect of the present invention, a method for intelligent identification and processing of complaint procedures based on a large model is provided, comprising: Based on the received customer complaints, the customer information related to the complaints is vectorized to form structured text of the customer information, and the semantic coordinates of vector similarity are determined based on the result of the vectorization process. Based on the customer database and vectorization representation technology based on large models, the complaint description data of the complaint business is transformed into a high-bit vector form, and then the key information of the complaint business is extracted. The key information of the complaint business includes at least one of the following: customer name, business type, product instance, and fault address. Based on the key information of the complaint business extracted, the structured query language (SQL) is used to filter the query and narrow down the matching search scope; Based on the results obtained after query filtering, vector similarity is calculated relative to the structured text of the customer information, and a candidate list of product instances that are semantically matched or close to the complaint description data is obtained based on the magnitude of the vector similarity. Select a suitable product instance from the candidate product instance list as a reference for processing or responding to the complaint.

[0006] Preferably, in any embodiment, The customer information related to the complaint service includes at least one of the following: customer name, customer abbreviation, A / Z address, and product instance.

[0007] Preferably, in any embodiment, The key information of the complaint service includes at least one of the following: service type, city, district / county.

[0008] Preferably, in any embodiment, The vector similarity calculation includes performing vector similarity calculations for at least one of customer name, business type, product instance, and running status.

[0009] Preferably, in any embodiment, The product instance candidate list includes multiple product instances arranged in descending order of semantic vector similarity to the complaint description data.

[0010] Preferably, in any embodiment, The customer database includes: a communication industry-specific vector knowledge base, which includes at least one of the following: communication industry terminology, commonly used customer expressions, history, and fault cases.

[0011] Preferably, in any embodiment, The step of calculating vector similarity relative to the structured text of the customer information includes: using the communication industry-specific vector knowledge base, employing vector retrieval technology to perform semantic similarity calculation on the complaint description data to obtain a candidate list of matching or close product instances.

[0012] Preferably, in any embodiment, Before or after the step of calculating vector similarity relative to the structured text of the customer information, the method includes: performing semantic noise reduction processing on the complaint description data, identifying and filtering errors, ambiguities, or non-standard information in the complaint description data from the customer, and improving the complaint description data through semantic compensation processing.

[0013] Preferably, in any embodiment, Before the step of calculating vector similarity relative to the structured text of the customer information, the method includes: performing verification, adjustment and semantic correction processing on the results of the semantic noise reduction and semantic compensation processing according to the business logic rules of the communications industry or training a dedicated classification model.

[0014] According to a second aspect of the present invention, a large-model-based intelligent identification and analysis system for complaint handling is provided, for implementing the large-model-based intelligent identification and processing method for complaint handling as described above, including: The complaint business vectorization processing module is used to vectorize customer information related to the complaint business received from the customer to form structured text of the customer information, and determine the semantic coordinates of vector similarity based on the result of the vectorization processing. The complaint business key information extraction module is used to convert the complaint description data of the complaint business into a high-bit vector form based on the customer database and vectorization representation technology based on a large model, and then extract the key information of the complaint business. The key information of the complaint business includes at least one of the following: customer name, business type, product instance, and fault address. The complaint business key information filtering module is used to perform query filtering using structured query language SQL based on the extracted key information of the complaint business in order to narrow the matching search range. The complaint business similarity calculation and matching module is used to perform vector similarity calculation relative to the structured text of the customer information based on the results obtained after query filtering, and to obtain a candidate list of product instances that are semantically matched or close to the complaint description data based on the magnitude of the vector similarity. The complaint service instance selection module is used to select a suitable product instance from the product instance candidate list as a reference for processing or responding to the complaint service.

[0015] The intelligent identification and processing method and system for complaint handling based on a large model provided by one or more embodiments of the present invention can significantly improve the efficiency of complaint handling. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a large-model-based intelligent identification and processing method for complaint handling according to an embodiment of the present invention.

[0018] Figure 2 This is a schematic diagram of the structure of a large-model-based intelligent identification and processing system for complaint handling, according to an embodiment of the present invention. Detailed Implementation

[0019] One or more embodiments of the present invention describe a method and system for intelligent identification and processing of complaint business based on a large model, which can significantly improve the efficiency of complaint business processing.

[0020] The technical solutions of various embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments described in the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] According to an embodiment of a first aspect of the present invention, a method for intelligent identification and processing of complaint procedures based on a large model is provided, comprising: Based on the received customer complaints, the customer information related to the complaints is vectorized to form structured text of the customer information, and the semantic coordinates of vector similarity are determined based on the result of the vectorization process. Based on the customer database and vectorization representation technology based on large models, the complaint description data of the complaint business is transformed into a high-bit vector form, and then the key information of the complaint business is extracted. The key information of the complaint business includes at least one of the following: customer name, business type, product instance, and fault address. Based on the key information of the complaint business extracted, the structured query language (SQL) is used to filter the query and narrow down the matching search scope; Based on the results obtained after query filtering, vector similarity is calculated relative to the structured text of the customer information, and a candidate list of product instances that are semantically matched or close to the complaint description data is obtained based on the magnitude of the vector similarity. Select a suitable product instance from the candidate product instance list as a reference for processing or responding to the complaint.

[0022] Thus, for received customer complaints, the relevant customer information (e.g., customer name, customer address, service type, complaint / fault description, etc.) is extracted and vectorized to form structured text of the customer information. Based on the vectorization result, semantic coordinates of vector similarity are determined for subsequent vector similarity calculations. For complaint description data from customer complaints, information recorded in the customer database is used, and vectorization representation technology based on a large model is applied to transform the complaint description data into high-order vector form. Key information from this vector form (e.g., customer name, service type, product instance, fault address, etc.) can then be extracted. This key information can be used for subsequent... Continuing vector similarity calculation, product instances with high similarity are matched for the complaint business. For efficiency considerations, based on the extracted key information of the complaint business, Structured Query Language (SQL) can be used for query filtering to narrow the matching search range, thereby effectively reducing the workload in subsequent similarity matching processes and improving efficiency. Based on the results obtained after query filtering, vector similarity is calculated relative to the structured text of the customer information. It should be understood that the higher the vector similarity, the higher the matching degree, and the more suitable it is as a reference for responding to or resolving the business complaint. Therefore, based on the magnitude of the vector similarity, a semantic match or... A candidate list of similar product instances is provided. It should be understood that this candidate list may contain multiple product instances with different high matching or similarity scores. However, in some cases, only one suitable product instance with a high matching or similarity score may exist. Selecting a suitable product instance with a high matching or similarity score for inclusion in the candidate list may depend on a pre-set matching or similarity score threshold range (e.g., only product instances with a matching or similarity score of 90% or higher are eligible for inclusion in the candidate list). Thus, one suitable product instance (e.g., the one with the highest matching or similarity score) or multiple product instances (e.g., matching or similarity scores ranked from highest to lowest) can be selected from the candidate list. The two product examples ranked first and second in the lower category can be used as a reference for processing or responding to the complaint. Of course, when necessary (e.g., after further communication with the customer or based on supplementary explanations provided by the customer), other more suitable product examples can be selected from the candidate list of product examples as a reference. Furthermore, the matching degree or similarity threshold range can be adjusted (e.g., product examples with a matching degree or similarity of more than 80% can be re-evaluated to qualify for inclusion in the candidate list) to obtain the product example that best matches the actual needs of the customer's business complaint as a reference for processing or responding to the complaint. This can significantly improve the efficiency of complaint processing and provide customers with strong technical support.

[0023] The intelligent complaint identification and processing method based on a large model, as provided in this invention, enables rapid and intelligent location and matching of customer complaints. The average manual response time in existing technologies is 10-15 minutes, while the intelligent complaint identification and processing method based on a large model provided in this invention significantly optimizes this process, enabling complaint location within one minute. This greatly improves complaint location efficiency, significantly reduces customer waiting time, and can even achieve a zero-wait experience of "responding to every request." Furthermore, it can reduce repetitive manpower input (such as basic information queries and cross-system coordination) by more than 60%, effectively reducing operating costs while improving efficiency.

[0024] Therefore, the intelligent identification and processing method for complaint handling based on a large model provided by one or more embodiments of the present invention can significantly improve the efficiency of complaint handling.

[0025] Preferably, in any embodiment, The customer information related to the complaint service includes at least one of the following: customer name, customer abbreviation, A / Z address, and product instance.

[0026] In this way, text fields containing various types of customer information related to complaint handling, especially unstructured text information, can be transformed into standardized structured text information through the vectorization process. This allows the customer information to be converted into a form that artificial intelligence (AI) systems can understand, facilitating subsequent semantic similarity calculations by AI systems.

[0027] Preferably, in any embodiment, The key information of the complaint service includes at least one of the following: service type, city, district / county.

[0028] In this way, by further locating the extracted key information, complaints about business within a specific area can be processed. For example, by using Structured Query Language (SQL) for query filtering, the matching search scope can be significantly narrowed, and the matching search efficiency can be improved.

[0029] Preferably, in any embodiment, The vector similarity calculation includes performing vector similarity calculations for at least one of customer name, business type, product instance, and running status.

[0030] In this way, it is possible to search for results that are semantically closest to or match the complaint description data of the complaint business as accurately as possible. For example, one or more product instances that the customer has encountered in the past regarding the same or similar business type. This allows complaint handling personnel to quickly and accurately select product instances that are suitable for the current complaint business as a reference, thereby enabling them to efficiently and accurately process or respond to customer complaints and provide professional solutions to the complaint issues.

[0031] Preferably, in any embodiment, The product instance candidate list includes multiple product instances arranged in descending order of semantic vector similarity to the complaint description data.

[0032] In this way, in practical applications, multiple candidate product instances with high similarity or matching degree can be provided for complaint handling personnel to choose from. The complaint handling personnel can first choose the product instance with the highest similarity or matching degree as a reference to provide a solution to the customer. However, if it is found in the subsequent processing (for example, when the customer provides supplementary information, i.e., the complaint description data is updated based on the customer's supplementary information) that the product instance is not completely suitable or does not fully meet the customer's needs in at least some aspects, then the product instance with the second highest similarity or matching degree or other candidate product instances can be selected as a reference to provide a solution to the customer as efficiently and accurately as possible.

[0033] Preferably, in any embodiment, The customer database includes: a communication industry-specific vector knowledge base, which includes at least one of the following: communication industry terminology, commonly used customer expressions, history, and fault cases.

[0034] In this way, it is possible to provide a variety of relevant information based on the customer's communication business and common communication methods, such as past failure cases, so as to more accurately identify and analyze the content and needs of customer complaints.

[0035] Preferably, in any embodiment, The step of calculating vector similarity relative to the structured text of the customer information includes: using the communication industry-specific vector knowledge base, employing vector retrieval technology to perform semantic similarity calculation on the complaint description data to obtain a candidate list of matching or close product instances.

[0036] In this way, it is possible to provide multifaceted relevant information based on the customer's communication business situation and common communication methods, such as the customer's common communication or expression methods, so as to more accurately identify and analyze the content and needs of customer complaints. This allows for targeted semantic similarity calculations to provide more efficient and accurate semantically matching or similar product examples for customer complaint handling personnel to refer to.

[0037] Preferably, in any embodiment, Before or after the step of calculating vector similarity relative to the structured text of the customer information, the method includes: performing semantic noise reduction processing on the complaint description data, identifying and filtering errors, ambiguities, or non-standard information in the complaint description data from the customer, and improving the complaint description data through semantic compensation processing.

[0038] In practice, complaint description data from user complaint services may contain unclear or non-standard expressions, which may lead to semantic ambiguity or vagueness. Such semantic noise is identified and corrected through the semantic noise reduction process, and can be improved through appropriate semantic compensation processing. This makes the processed complaint description data clearer and more explicit, so that the system can identify, search, filter and match product instances that are conducive to solving the problem, thereby making the entire processing flow more efficient and accurate.

[0039] Preferably, in any embodiment, Before the step of calculating vector similarity relative to the structured text of the customer information, the method includes: performing verification, adjustment and semantic correction processing on the results of the semantic noise reduction and semantic compensation processing according to the business logic rules of the communications industry or training a dedicated classification model.

[0040] In practice, complaint descriptions from users may contain unclear or non-standard expressions, leading to semantic ambiguity or vagueness. Such semantic noise is identified and corrected through semantic noise reduction processing, and can be further improved through appropriate semantic compensation processing. However, in some cases, even after semantic noise reduction and semantic compensation processing, the results may still deviate significantly from the user's actual needs (e.g., due to unclear user statements or communication errors). Therefore, the results after semantic noise reduction and semantic compensation processing can be further verified and adjusted according to the business logic rules of the telecommunications industry or by training a dedicated classification model. Semantic correction processing can then be performed to correct erroneous matches caused by semantic bias or retrieval errors, ensuring the accuracy and reliability of the final output of business complaints or fault information. This facilitates the system's identification, searching, filtering, and matching of product instances that are conducive to resolving the problem, thereby making the entire processing flow more efficient and accurate.

[0041] Optionally, in any embodiment, it further includes: By systematically collecting and processing data such as professional terminology in the communications industry, commonly used customer expressions, historical work orders, and fault cases, a key information vector knowledge base optimized for the characteristics of the communications industry is constructed. This key information vector knowledge base includes or is associated with the customer database to support subsequent semantic understanding and information extraction.

[0042] Optionally, in any embodiment, the intelligent identification and processing method for complaint handling based on a large model includes: By utilizing the key information vector knowledge base, vector retrieval technology is used to perform semantic similarity calculation on user complaint or fault description texts, quickly locating the most relevant candidate fault types or business information, and achieving preliminary efficient matching.

[0043] In a preferred embodiment of the present invention, a method for intelligent identification and processing of complaint procedures based on a large model is provided, comprising: 1) Based on the complaints received from customers, the existing customer information (especially text fields such as customer name, customer abbreviation, A / Z address, product instance, etc.) is vectorized to transform the unstructured customer information into a structured text form that artificial intelligence (AI) can understand, and the semantic coordinates of vector similarity are determined based on the results of the vectorization process. 2) Based on the customer database (or knowledge base) and vectorization representation technology based on large models, the complaint description data of the complaint business is transformed into a high-bit vector form and then the key information of the complaint business is extracted. The key information of the complaint business may include at least one of the following: customer name, business type, product instance, and fault address. 3) Based on the extracted key information of the complaint business (such as city, district / county, business type, etc.), use Structured Query Language (SQL) to perform query filtering to narrow down the matching search range; 4) Based on the results obtained after query filtering, perform vector similarity calculation relative to the structured text of the customer information (e.g., customer name, business type, product instance, running status, etc.), and obtain a candidate list of product instances that are semantically matched or close to the complaint description data, which may include one or more product instances, based on the magnitude of the vector similarity. 5) Select one or more suitable product instances from the product instance candidate list as a reference for processing or responding to the complaint.

[0044] Figure 1 This is a flowchart illustrating a large-model-based intelligent identification and processing method for complaint handling according to an embodiment of the present invention.

[0045] In such Figure 1 The illustrated embodiment demonstrates an intelligent identification and processing method for complaint handling based on a large model, comprising the following steps: 110: Based on the received customer complaint, the customer information related to the complaint is vectorized to form a structured text of the customer information, and the semantic coordinates of vector similarity are determined based on the result of the vectorization. 120: Based on the customer database and vectorization representation technology based on large models, the complaint description data of the complaint business is transformed into a high-bit vector form and then the key information of the complaint business is extracted. The key information of the complaint business includes at least one of the following: customer name, business type, product instance, and fault address. 130: Based on the extracted key information of the complaint business, use Structured Query Language (SQL) to perform query filtering to narrow down the matching search scope; 140: Based on the results obtained after query filtering, perform vector similarity calculation relative to the structured text of the customer information, and obtain a candidate list of product instances that are semantically matched or close to the complaint description data based on the magnitude of the vector similarity. 150: Select a suitable product instance from the candidate list of product instances as a reference for processing or responding to the complaint.

[0046] According to a second aspect of the present invention, a large-model-based intelligent identification and analysis system for complaint handling is provided, for implementing the large-model-based intelligent identification and processing method for complaint handling as described above, including: The complaint business vectorization processing module is used to vectorize customer information related to the complaint business received from the customer to form structured text of the customer information, and determine the semantic coordinates of vector similarity based on the result of the vectorization processing. The complaint business key information extraction module is used to convert the complaint description data of the complaint business into a high-bit vector form based on the customer database and vectorization representation technology based on a large model, and then extract the key information of the complaint business. The key information of the complaint business includes at least one of the following: customer name, business type, product instance, and fault address. The complaint business key information filtering module is used to perform query filtering using structured query language SQL based on the extracted key information of the complaint business in order to narrow the matching search range. The complaint business similarity calculation and matching module is used to perform vector similarity calculation relative to the structured text of the customer information based on the results obtained after query filtering, and to obtain a candidate list of product instances that are semantically matched or close to the complaint description data based on the magnitude of the vector similarity. The complaint service instance selection module is used to select a suitable product instance from the product instance candidate list as a reference for processing or responding to the complaint service.

[0047] Thus, for received customer complaints, the relevant customer information (e.g., customer name, customer address, service type, complaint / fault description, etc.) is extracted and vectorized to form structured text of the customer information. Based on the vectorization result, semantic coordinates of vector similarity are determined for subsequent vector similarity calculations. For complaint description data from customer complaints, information recorded in the customer database is used, and vectorization representation technology based on a large model is applied to transform the complaint description data into high-order vector form. Key information from this vector form (e.g., customer name, service type, product instance, fault address, etc.) can then be extracted. This key information can be used for subsequent... Continuing vector similarity calculation, product instances with high similarity are matched for the complaint business. For efficiency considerations, based on the extracted key information of the complaint business, Structured Query Language (SQL) can be used for query filtering to narrow the matching search range, thereby effectively reducing the workload in subsequent similarity matching processes and improving efficiency. Based on the results obtained after query filtering, vector similarity is calculated relative to the structured text of the customer information. It should be understood that the higher the vector similarity, the higher the matching degree, and the more suitable it is as a reference for responding to or resolving the business complaint. Therefore, based on the magnitude of the vector similarity, a semantic match or... A candidate list of similar product instances is provided. It should be understood that this candidate list may contain multiple product instances with different high matching or similarity scores. However, in some cases, only one suitable product instance with a high matching or similarity score may exist. Selecting a suitable product instance with a high matching or similarity score for inclusion in the candidate list may depend on a pre-set matching or similarity score threshold range (e.g., only product instances with a matching or similarity score of 90% or higher are eligible for inclusion in the candidate list). Thus, one suitable product instance (e.g., the one with the highest matching or similarity score) or multiple product instances (e.g., matching or similarity scores ranked from highest to lowest) can be selected from the candidate list. The two product examples ranked first and second in the lower category can be used as a reference for processing or responding to the complaint. Of course, when necessary (e.g., after further communication with the customer or based on supplementary explanations provided by the customer), other more suitable product examples can be selected from the candidate list of product examples as a reference. Furthermore, the matching degree or similarity threshold range can be adjusted (e.g., product examples with a matching degree or similarity of more than 80% can be re-evaluated to qualify for inclusion in the candidate list) to obtain the product example that best matches the actual needs of the customer's business complaint as a reference for processing or responding to the complaint. This can significantly improve the efficiency of complaint processing and provide customers with strong technical support.

[0048] The intelligent complaint identification and processing system based on a large model, as provided in this invention, enables rapid and intelligent location and matching of customer complaints. In existing technologies, the average manual response time is 10-15 minutes. However, the intelligent complaint identification and processing method based on a large model, as described in this invention, significantly optimizes this process, enabling complaint location within one minute. This greatly improves complaint location efficiency, significantly reduces customer waiting time, and can even achieve a zero-wait experience with immediate response to requests. Furthermore, it can reduce repetitive manpower input (such as basic information queries and cross-system coordination) by more than 60%, effectively reducing operating costs while improving efficiency.

[0049] Therefore, the intelligent identification and processing system for complaint handling based on a large model provided by one or more embodiments of the present invention can significantly improve the efficiency of complaint handling.

[0050] Optionally, in any embodiment, the intelligent identification and processing system for complaint business based on a large model further includes: a noise reduction and compensation module, which is used to identify and filter non-standard information (i.e., so-called semantic noise) such as ambiguous expressions, colloquialisms, and typos in the complaint description of business complaints from users, and to enhance the expression of key information through semantic completion and synonym expansion, thereby improving matching accuracy.

[0051] Optionally, in any embodiment, the intelligent identification and processing system for complaint services based on a large model further includes: a correction module, which is used to verify and adjust the retrieval results after semantic noise reduction compensation processing by the noise reduction compensation module by combining communication industry business logic rules or training a dedicated classification model, correcting erroneous matches caused by semantic deviations or retrieval errors, and ensuring that the final output fault service type is accurate and reliable.

[0052] Figure 2 This is a schematic diagram of the structure of a large-model-based intelligent identification and processing system for complaint handling, according to an embodiment of the present invention.

[0053] In such Figure 2 The illustrated embodiment shows a complaint business intelligent identification and processing system based on a large model, including: The complaint business vectorization processing module 201 is used to vectorize customer information related to the complaint business received from the customer to form structured text of the customer information, and determine the semantic coordinates of vector similarity based on the result of the vectorization processing. The complaint business key information extraction module 202 is used to convert the complaint description data of the complaint business into a high-bit vector form based on the customer database and vectorization representation technology based on a large model, and then extract the key information of the complaint business. The key information of the complaint business includes at least one of the following: customer name, business type, product instance, and fault address. The complaint business key information filtering module 203 is used to perform query filtering using structured query language SQL based on the extracted key information of the complaint business in order to narrow the matching search range. The complaint business similarity calculation and matching module 204 is used to perform vector similarity calculation relative to the structured text of the customer information based on the results obtained after query filtering, and to obtain a candidate list of product instances that are semantically matched or close to the complaint description data based on the magnitude of the vector similarity. The complaint service instance selection module 205 is used to select a suitable product instance from the product instance candidate list as a reference for processing or responding to the complaint service.

[0054] According to one or more embodiments of the present invention, a big model-based intelligent identification and processing method and system for complaint business has been established, which builds a vectorized search model for the communications industry. It can accurately understand the user's ambiguous expressions, including colloquial expressions, elliptical sentences and other non-standard language forms. Moreover, it can clearly understand the user's true intention through semantic noise reduction, semantic compensation or semantic correction, thereby significantly improving the naturalness and accuracy of human-computer interaction.

[0055] The intelligent complaint identification and processing method and system based on a large model, provided by one or more embodiments of the present invention, adopts a three-layer architecture design of "vector retrieval + noise reduction compensation + bias correction": First, key information is captured through word segmentation and part-of-speech tagging; then, the results are finely sorted and filtered based on compensation rules (customer, business, address, etc.), thereby efficiently and accurately matching the product instance that is closest in semantics to the user's business complaint description. This architecture not only ensures search efficiency but also endows the system with dynamic reasoning capabilities, enabling it to automatically adjust the search strategy based on dialogue context and scene features. The final end-to-end intelligent search chain (retrieval-sorting-feedback) significantly reduces the complexity of user operations and provides an efficient solution for various application scenarios.

[0056] The intelligent identification and processing method and system for complaint services based on a large model, provided by one or more embodiments of the present invention, focuses on an intelligent scheduling strategy for rapidly analyzing, identifying, and locating customer complaints. Relying on a large-model natural language processing (NLP) model, it extracts keywords, performs sentiment analysis, and translates information such as customer complaint descriptions and fault addresses from received complaint work orders. Combined with context engineering, it quickly locates the fault service information that the customer is complaining about. By building this model, it can assist installation and maintenance personnel in quickly locating fault services, solving the dilemma of finding information and support through various channels, thereby greatly improving work efficiency.

[0057] In summary, the intelligent identification and processing method and system for complaint handling based on a large model provided by one or more embodiments of the present invention can significantly improve the efficiency of complaint handling.

[0058] One embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed in a computer, it causes the computer to perform the intelligent identification and processing method for complaint business based on a large model as described in any embodiment of the present invention.

[0059] One embodiment of the present invention provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the method for intelligent identification and processing of complaint business based on a large model as described in any embodiment of the present invention.

[0060] It should be noted that the terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” used in the embodiments of this invention and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0061] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0062] It is understood that the structures illustrated in the embodiments of the present invention do not constitute a specific limitation on the apparatus of the embodiments of the present invention. In other embodiments of the specification, the above-described apparatus may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0063] The various embodiments in this invention are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0064] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using hardware, software, widgets, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium.

[0065] The specific embodiments or implementation methods described above further illustrate the purpose, technical solution and beneficial effects of the present invention. It should be understood that the above descriptions are only specific implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent identification and processing of complaint procedures based on a large model, characterized in that, include: Based on the received customer complaints, the customer information related to the complaints is vectorized to form structured text of the customer information, and the semantic coordinates of vector similarity are determined based on the result of the vectorization process. Based on the customer database and vectorization representation technology based on large models, the complaint description data of the complaint business is transformed into a high-bit vector form, and then the key information of the complaint business is extracted. The key information of the complaint business includes at least one of the following: customer name, business type, product instance, and fault address. Based on the key information of the complaint business extracted, the structured query language SQL is used to filter and narrow down the matching search range. Based on the results obtained after query filtering, vector similarity is calculated relative to the structured text of the customer information, and a candidate list of product instances that are semantically matched or close to the complaint description data is obtained based on the magnitude of the vector similarity. Select a suitable product instance from the candidate product instance list as a reference for processing or responding to the complaint.

2. The intelligent identification and processing method for complaint handling based on a large model according to claim 1, characterized in that, The customer information related to the complaint service includes at least one of the following: customer name, customer abbreviation, A / Z address, and product instance.

3. The intelligent identification and processing method for complaint handling based on a large model according to claim 1, characterized in that, The key information of the complaint service includes at least one of the following: service type, city, district / county.

4. The intelligent identification and processing method for complaint handling based on a large model according to claim 1, characterized in that, The vector similarity calculation includes performing vector similarity calculations for at least one of customer name, business type, product instance, and running status.

5. The intelligent identification and processing method for complaint handling based on a large model according to claim 1, characterized in that, The product instance candidate list includes multiple product instances arranged in descending order of semantic vector similarity to the complaint description data.

6. The intelligent identification and processing method for complaint handling based on a large model according to claim 1, characterized in that, The customer database includes: a communication industry-specific vector knowledge base, which includes at least one of the following: communication industry terminology, commonly used customer expressions, history, and fault cases.

7. The intelligent identification and processing method for complaint handling based on a large model according to claim 6, characterized in that, The step of calculating vector similarity relative to the structured text of the customer information includes: using the communication industry-specific vector knowledge base, employing vector retrieval technology to perform semantic similarity calculation on the complaint description data to obtain a candidate list of matching or close product instances.

8. The intelligent identification and processing method for complaint handling based on a large model according to claim 1, characterized in that, Before or after the step of calculating vector similarity relative to the structured text of the customer information, the method includes: performing semantic noise reduction processing on the complaint description data, identifying and filtering errors, ambiguities, or non-standard information in the complaint description data from the customer, and improving the complaint description data through semantic compensation processing.

9. The intelligent identification and processing method for complaint handling based on a large model according to claim 8, characterized in that, Before the step of calculating vector similarity relative to the structured text of the customer information, the method includes: performing verification, adjustment and semantic correction processing on the results of the semantic noise reduction and semantic compensation processing according to the business logic rules of the communications industry or training a dedicated classification model.

10. A complaint business intelligent identification and processing system based on a large model, characterized in that, It is used to implement the intelligent identification and processing method for complaint business based on a large model according to any one of claims 1-9, comprising: The complaint business vectorization processing module is used to vectorize customer information related to the complaint business received from the customer to form structured text of the customer information, and determine the semantic coordinates of vector similarity based on the result of the vectorization processing. The complaint business key information extraction module is used to convert the complaint description data of the complaint business into a high-bit vector form based on the customer database and vectorization representation technology based on a large model, and then extract the key information of the complaint business. The key information of the complaint business includes at least one of the following: customer name, business type, product instance, and fault address. The complaint business key information filtering module is used to perform query filtering using structured query language SQL based on the extracted key information of the complaint business in order to narrow the matching search range. The complaint business similarity calculation and matching module is used to perform vector similarity calculation relative to the structured text of the customer information based on the results obtained after query filtering, and to obtain a candidate list of product instances that are semantically matched or close to the complaint description data based on the magnitude of the vector similarity. The complaint service instance selection module is used to select a suitable product instance from the product instance candidate list as a reference for processing or responding to the complaint service.

Images