AI customer portrait automatic generation system, electronic equipment and storage medium

Abstract

The invention provides an AI client portrait automatic generation system, an electronic device and a storage medium, and the system comprises a multi-mode dialogue collection module which is used for real-time access and normalization processing of client interaction data; the mixed semantic analysis unit is connected with the multi-modal dialogue acquisition module and is configured to be used for performing context association understanding and metaphor expression recognition on the normalized data by adopting a neural network model and combining with a specific industry knowledge base; the multi-dimensional emotion calculation engine is connected with the multi-modal dialogue acquisition module and is configured to be used for performing multi-signal fusion emotion classification based on voiceprint features, text emotion word weights and dialogue rhythm features; and the dynamic graph builder is connected with the mixed semantic analysis unit and the multi-dimensional emotion calculation engine and is configured to be used for building and updating a customer portrait model in real time based on an analysis result. Real-time acquisition and fusion of multi-source heterogeneous data are realized, and real-time and comprehensive decision support is provided for client layering, accurate service and risk management and control.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology, specifically relating to an AI customer profile automatic generation system, electronic device, and storage medium. Background Technology

[0002] In the fields of Customer Relationship Management (CRM), talent services, and key account marketing, building accurate customer profiles is the core of achieving personalized services and precise resource matching. Traditional customer profiling methods mainly rely on manually entered information, static questionnaires, and simple transaction behavior data, which are slow to update and have limited dimensions.

[0003] With the diversification of interaction channels (such as telephone, WeChat, email, etc.), a large amount of dynamic and real-time customer needs and characteristics are contained in multimodal dialogue content. However, existing technologies struggle to effectively extract deep-value information from this unstructured data. Summary of the Invention

[0004] This invention provides an AI-powered customer profile automatic generation system, electronic device, and storage medium, offering real-time and comprehensive decision support for customer segmentation, precise service, and risk management.

[0005] To achieve the above objectives, one or more embodiments of this application provide an AI customer profile automatic generation system, an electronic device, and a storage medium, the system comprising:

[0006] The multimodal dialogue acquisition module is configured to access and normalize customer interaction data from multiple communication channels in real time, including text, structured data, and voice.

[0007] The hybrid semantic analysis unit is connected to the multimodal dialogue acquisition module and is configured to use a neural network model and a specific industry knowledge base to perform contextual understanding and metaphorical expression recognition on the normalized data.

[0008] A multi-dimensional emotion computing engine is connected to the multimodal dialogue acquisition module and is configured to perform multi-signal fusion emotion classification based on voiceprint features, text emotion word weights, and dialogue rhythm features.

[0009] A dynamic graph builder, connected to the hybrid semantic analysis unit and the multidimensional sentiment computing engine, is configured to build and update customer profile models in real time based on the analysis results.

[0010] Based on the above technical solution of the present invention, the following improvements can also be made:

[0011] Optionally, the multimodal dialogue acquisition module includes:

[0012] A smart call center interface equipped with an audio noise reduction chip for processing telephone voice data;

[0013] Email content parsing middleware, used to extract structured data from emails;

[0014] An API adaptation layer that supports semantic emojis for handling instant messaging text and emojis.

[0015] Optionally, the hybrid semantic analysis unit adopts a BERT-GRU neural network architecture and includes a knowledge base of specific industry terms and a metaphorical expression mapping rule engine.

[0016] Optionally, the multidimensional emotion computing engine is configured to extend the basic emotion model to detect multiple complex emotional states; its classifier integrates the fundamental frequency and formant features of the voiceprint, the text emotion word features weighted by the TF-IDF algorithm, and the dialogue rhythm features based on speech rate and pause patterns.

[0017] Optionally, the dynamic graph builder is implemented based on a graph database, and the node attributes in the customer profile model include at least one quantitative indicator among the consumption capacity index and risk preference coefficient.

[0018] Optionally, the system further includes:

[0019] The policy sensitivity monitoring module is configured to dynamically load a policy keyword library and identify policy-sensitive words in conversations through a neural network model, and output the compliance risk level.

[0020] The strategy library management module is configured to dynamically adjust customer service strategies based on the compliance risk level and automatically trigger compliance prompts when high-risk expressions are identified.

[0021] Optionally, the policy sensitivity monitoring module calculates the policy impact score using the following formula:

[0022]

[0023] in, To influence scores due to policy changes As policy keywords, For its corresponding weight, For conversation duration, This represents the policy urgency coefficient.

[0024] Optionally, the system also includes a scoring calculation module configured to calculate a comprehensive score for the customer profile using the following formula:

[0025] ProfileScore = α (0.6A + 0.3B) + βC

[0026] Here, ProfileScore is the sum score of the customer profile, A is the qualification value, B is the behavioral characteristics, C is the risk level, and α and β are weight coefficients that are dynamically adjusted through reinforcement learning algorithms.

[0027] According to another aspect of the present invention, a computer program is provided that includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the functions of any of the systems described above.

[0028] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions that, when executed by a processor, implement the functions of any of the systems described above.

[0029] The beneficial effects of this invention are that it provides an AI-powered customer profile automatic generation system, electronic device, and storage medium. This system can achieve real-time acquisition and fusion of multi-source heterogeneous data, extract key features through deep semantic analysis and sentiment computing, and ultimately construct a dynamically updatable customer profile that includes multiple dimensions such as qualification value, behavioral characteristics, and risk level, thereby providing real-time and comprehensive decision support for customer segmentation, precise service, and risk management. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the system architecture of an embodiment of the present invention.

[0031] Figure 2 This is a schematic diagram of the processing flow of an embodiment of the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0033] It should be noted that, unless otherwise defined, the technical or scientific terms used in one or more embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms "first," "second," and similar terms used in one or more embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0034] like Figures 1-2 As shown, one or more embodiments of this application disclose an AI customer profile automatic generation system, electronic device, and storage medium. The system includes:

[0035] The multimodal dialogue acquisition module is configured to access and normalize customer interaction data from multiple communication channels in real time, including text, structured data, and voice.

[0036] The hybrid semantic analysis unit is connected to the multimodal dialogue acquisition module and is configured to use a neural network model and a specific industry knowledge base to perform contextual understanding and metaphorical expression recognition on the normalized data.

[0037] A multi-dimensional emotion computing engine is connected to the multimodal dialogue acquisition module and is configured to perform multi-signal fusion emotion classification based on voiceprint features, text emotion word weights, and dialogue rhythm features.

[0038] A dynamic graph builder, connected to the hybrid semantic analysis unit and the multidimensional sentiment computing engine, is configured to build and update customer profile models in real time based on the analysis results.

[0039] Based on the above technical solution of the present invention, the following improvements can also be made:

[0040] Optionally, the multimodal dialogue acquisition module includes:

[0041] A smart call center interface equipped with an audio noise reduction chip for processing telephone voice data;

[0042] Email content parsing middleware, used to extract structured data from emails;

[0043] An API adaptation layer that supports semantic emojis for handling instant messaging text and emojis.

[0044] Optionally, the hybrid semantic analysis unit adopts a BERT-GRU neural network architecture and includes a knowledge base of specific industry terms and a metaphorical expression mapping rule engine.

[0045] Optionally, the multidimensional emotion computing engine is configured to extend the basic emotion model to detect multiple complex emotional states; its classifier integrates the fundamental frequency and formant features of the voiceprint, the text emotion word features weighted by the TF-IDF algorithm, and the dialogue rhythm features based on speech rate and pause patterns.

[0046] Optionally, the dynamic graph builder is implemented based on a graph database, and the node attributes in the customer profile model include at least one quantitative indicator among the consumption capacity index and risk preference coefficient.

[0047] Optionally, the system further includes:

[0048] The policy sensitivity monitoring module is configured to dynamically load a policy keyword library and identify policy-sensitive words in conversations through a neural network model, and output the compliance risk level.

[0049] The strategy library management module is configured to dynamically adjust customer service strategies based on the compliance risk level and automatically trigger compliance prompts when high-risk expressions are identified.

[0050] Optionally, the policy sensitivity monitoring module calculates the policy impact score using the following formula:

[0051]

[0052] in, To influence scores due to policy changes As policy keywords, For its corresponding weight, For conversation duration, This represents the policy urgency coefficient.

[0053] Optionally, the system also includes a scoring calculation module configured to calculate a comprehensive score for the customer profile using the following formula:

[0054] ProfileScore = α (0.6A + 0.3B) + βC

[0055] Here, ProfileScore is the sum score of the customer profile, A is the qualification value, B is the behavioral characteristics, C is the risk level, and α and β are weight coefficients that are dynamically adjusted through reinforcement learning algorithms.

[0056] In another embodiment, a computer program is provided that includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to perform the functions of any of the systems described above.

[0057] In another embodiment, a computer-readable storage medium is provided that stores computer instructions which, when executed by a processor, implement the functions of any of the systems described above.

[0058] In another embodiment, a customer profile generation method based on the above system is also provided, including the following steps:

[0059] S1: Real-time collection of multimodal customer dialogue data from telephone, WeChat, and email channels, and normalization preprocessing;

[0060] S2: Perform deep semantic analysis on text data to identify intent, extract key information, and analyze industry metaphors;

[0061] S3: Perform voiceprint analysis and sentiment calculation on speech data, perform sentiment analysis on text data, and integrate dialogue rhythm features to output a complex emotional state;

[0062] S4: Integrate semantic analysis results, sentiment tags, real-time policy information, and historical behavioral data;

[0063] S5: Based on integrated data, build and update a customer profile model in real time in a graph database, which includes three dimensions: qualifications, behavior, and risk.

[0064] S6: Based on the profile scoring results, trigger the corresponding service strategy or early warning mechanism.

[0065] As an example, in the above embodiments:

[0066] The WeChat semantic analysis layer supports 128-dimensional vectorized representation of emojis (including multiple industry-specific emoji libraries).

[0067] The telephone voice processing engine is based on the Kaldi framework and includes 13 dialect acoustic models.

[0068] Policy-sensitive word monitoring: dynamically loads the policy keyword database of policy-publishing websites (automatically updated daily).

[0069] The dynamic graph database, Neo4j Enterprise Edition, includes 3 types of node attributes (qualification / behavior / risk) and 12 types of relationship edges.

[0070] Overall information processing can be based on the appendix Figure 1 and Figure 2In summary, the process involves inputting relevant information via audio or text through WeChat, phone calls, or other means. This information originates from direct communication between customer service representatives and customers via WeChat or phone. Voiceprint analysis (emotional analysis) and terminology decoding (intent classification) transform the raw input into analyzable information. Furthermore, relevant policy information (risk indicators) is synchronously retrieved via a policy API. Finally, a three-dimensional scoring model is used to apply a scoring algorithm to this information, resulting in a customer profile score.

[0071] Scenario 1: Talent Matching Service

[0072] Initial profile: Private equity fund practitioner qualification, score 80.

[0073] Initial request: "I only have a fund practitioner certificate, and my qualification is about to expire! Can you help me see if there are any companies hiring?" (Speech speed increased by 23%).

[0074] Terminology Decoding: Documents, incomplete follow-up vocational training (triggers warning flag).

[0075] Voiceprint analysis: Anxiety was detected by fundamental frequency fluctuations (intensity 72%).

[0076] Real-time query: Matched 1 asset management company requiring fund practitioner qualifications. The Asset Management Association of China (AMAC) is currently verifying continuing education records.

[0077] Output profile: Fund practitioner qualification certificate holder, score 65, risk level 3, note that supplementary training hour proof is required.

[0078] Scenario 2: Talent Demand Services

[0079] Initial requirements: We urgently need 3 refrigerated truck drivers (A2 driver's license + refrigerated truck operator's certificate), who can complete the registration in the Ministry of Transport system. The qualification review deadline is next Tuesday. Salary can be increased by 15% (voiceprint analysis shows a 25% increase in speaking speed).

[0080] Terminology Decoding: Filing with the Ministry of Transport system requires providing special equipment operation records (triggering compliance review).

[0081] Policy relevance: The company's registered location, Zhejiang Province, is currently verifying the qualifications of its cold chain workers.

[0082] Output profile: Operation records are incomplete. Note: It is recommended to provide three consecutive months of transportation records and equipment operation logs.

[0083] Scenario 3: Demand for Overseas Warehouse Expansion by Cross-Border E-commerce Enterprises

[0084] Initial requirements: We urgently need 2 overseas warehouse operations managers (fluent in English and familiar with FBA / WMS systems), who can complete Amazon SPN service provider certification. The deadline for signing contracts is next Wednesday. Salary can be increased by 20%.

[0085] Terminology Decoding: SPN service provider certification requires providing official Amazon partnership qualifications (triggering platform compliance review).

[0086] Policy relevance: It was detected that Shenzhen, where the company is registered, is strictly investigating the qualifications of cross-border e-commerce service providers.

[0087] Resource matching: Automatically connects you with 5 suppliers who have the qualifications to produce technology-related videos.

[0088] Output profile: Incomplete certification qualifications, insufficient practical experience in overseas warehouses, score 60, risk level 3.

[0089] Scenario 4: New Product Promotion

[0090] Initial requirements: We urgently need 3 sets of product promotional videos (smart home + health monitoring + energy saving and environmental protection), which can highlight the advantages of our technological patents. The videos must be delivered before the launch event next Friday. The budget can be increased by 18%.

[0091] Terminology Decoding: Technical patent advantages require providing patent certificates and technical white papers (triggering technical review).

[0092] Policy Relevance: The system is currently verifying the authenticity of the company's advertising claims, as the company is registered in Beijing.

[0093] Resource matching: Automatically connects you with 5 suppliers who have the qualifications to produce technology-related videos.

[0094] Output profile: Patent verification needs improvement, technology visualization solution is lacking, score 80, risk level 2.

[0095] This invention constructs several industry-first multimodal customer profiling systems that integrate policy sensitivity analysis, breaking through the three technological barriers of traditional CRM:

[0096] Dynamic information tracking, based on standard data exchange protocols, establishes an information status response mechanism and deviation detection to achieve second-level updates of industry practice license status and early warning of anomalies.

[0097] Industry semantic decoding: Establish a mapping rule library containing 529 different industry slang terms.

[0098] The policy response engine identifies 126 policy-sensitive words using the BiLSTM-CRF model, improving risk warning response speed by 40 times.

[0099] The intelligent call center interface is equipped with a construction site-specific noise reduction chip (signal-to-noise ratio improved by 22dB) and supports voiceprint recognition of 13 dialects (accuracy rate of 96.8%).

[0100] The WeChat semantic extension module analyzes various industry-specific emojis and identifies 287 metaphorical expressions.

[0101] The hybrid semantic analysis unit adopts the BERT-GRU neural network architecture, and the industry knowledge base includes: full lifecycle rules for multiple types of certificates.

[0102] Dynamic graph builder, constructing 3D relational networks based on Neo4j:

[0103] In terms of qualifications, the scarcity of certificate combinations is scored.

[0104] Behavioral dimensions: negotiation patterns / service preferences.

[0105] Risk dimension, certification compliance level (0-5).

[0106] The policy impact dashboard quantifies the impact of policies such as "qualification reform" and "social security network," and automatically generates compliance script templates (response time <800ms).

[0107] The intelligent recommendation engine recommends the optimal certificate allocation region based on the ratio of existing certificates to demand, and the dynamic pricing model has an error rate of <7.3%.

[0108] The real-time certificate status tracking algorithm achieves real-time tracking of certificate status by calculating a certificate score. Its core formula is: the sum of the products of the effective days for each dimension and their corresponding weights, divided by the policy impact factor, yields the certificate score. The numerator comprehensively considers the certificate's effective duration and the importance of each dimension, while the denominator incorporates policy factors to adjust the evaluation results. Ultimately, the quantified score directly reflects the current status of the certificate, enabling dynamic and comprehensive evaluation and tracking of certificate status.

[0109] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0110] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0111] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction set implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0112] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0113] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0114] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An AI-powered customer profile automatic generation system, characterized in that, include: The multimodal dialogue acquisition module is configured to access and normalize customer interaction data from multiple communication channels in real time, including text, structured data, and voice. The hybrid semantic analysis unit is connected to the multimodal dialogue acquisition module and is configured to use a neural network model and a specific industry knowledge base to perform contextual understanding and metaphorical expression recognition on the normalized data. A multi-dimensional emotion computing engine is connected to the multimodal dialogue acquisition module and is configured to perform multi-signal fusion emotion classification based on voiceprint features, text emotion word weights, and dialogue rhythm features. A dynamic graph builder, connected to the hybrid semantic analysis unit and the multidimensional sentiment computing engine, is configured to build and update customer profile models in real time based on the analysis results.

2. The AI ​​customer profile automatic generation system according to claim 1, characterized in that, The multimodal dialogue acquisition module includes: A smart call center interface equipped with an audio noise reduction chip for processing telephone voice data; Email content parsing middleware, used to extract structured data from emails; An API adaptation layer that supports semantic emojis for handling instant messaging text and emojis.

3. The AI ​​customer profile automatic generation system according to claim 1, characterized in that, The hybrid semantic analysis unit adopts a BERT-GRU neural network architecture and includes a knowledge base of specific industry terms and a metaphorical expression mapping rule engine.

4. The AI ​​customer profile automatic generation system according to claim 1, characterized in that, The multidimensional emotion computing engine is configured to extend the basic emotion model to detect multiple complex emotional states; its classifier integrates the fundamental frequency and formant features of the voiceprint, the text emotion word features weighted by the TF-IDF algorithm, and the dialogue rhythm features based on speech rate and pause patterns.

5. The AI ​​customer profile automatic generation system according to claim 1, characterized in that, The dynamic graph builder is implemented based on a graph database, and the node attributes in the customer profile model include at least one quantitative indicator among the consumption capacity index and risk preference coefficient.

6. The AI ​​customer profile automatic generation system according to claim 1, characterized in that, The system also includes: The policy sensitivity monitoring module is configured to dynamically load a policy keyword library and identify policy-sensitive words in conversations through a neural network model, and output the compliance risk level. The strategy library management module is configured to dynamically adjust customer service strategies based on the compliance risk level and automatically trigger compliance prompts when high-risk expressions are identified.

7. The AI ​​customer profile automatic generation system according to claim 6, characterized in that, The policy sensitivity monitoring module calculates the policy impact score using the following formula: in, To influence scores due to policy changes As policy keywords, For its corresponding weight, For conversation duration, This represents the policy urgency coefficient.

8. The AI ​​customer profile automatic generation system according to claim 1, characterized in that, The system also includes a scoring calculation module, configured to calculate a comprehensive score for the customer profile using the following formula: ProfileScore = α (0.6A + 0.3B) + βC Here, ProfileScore is the sum score of the customer profile, A is the qualification value, B is the behavioral characteristics, C is the risk level, and α and β are weight coefficients that are dynamically adjusted through reinforcement learning algorithms.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the functions of the system as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the functions of the system as claimed in any one of claims 1-8.

Images