A business adaptation degree intelligent analysis method and system combined with a customer portrait

Abstract

This invention proposes an intelligent analysis method and system for business adaptability based on customer profiles, belonging to the field of data processing and large-scale model technology. The method includes steps for data tag construction, customer data acquisition, customer profile construction, business model adaptation, and model interaction. Dialogue data from potential customers is continuously acquired through a data acquisition middleware, and the target business model communicates with the middleware via a one-way data pipeline. The system connects multiple benchmark business models, which obtain input or update data from the middleware to implement the method. This technical solution, based on the data acquisition middleware, achieves precise interaction by dynamically constructing customer profiles and adapting business models. The one-way data pipeline ensures data controllability, the iteration mechanism improves response accuracy, and the collaborative operation of all system units results in high adaptability, efficiently meeting the customer service needs of the home decoration industry.

Description

Technical Field

[0001] This invention relates to large-scale modeling, data processing and analysis technology, belonging to the field of data processing and large-scale modeling technology, and particularly to intelligent analysis methods and systems for business adaptability based on customer profiles, and human-computer interaction devices. Background Technology

[0002] Customer profiles are virtual user models built from multi-dimensional data, extracting information such as customer characteristics, needs, preferences, and behavioral habits. Through tagging, customer profiles accurately present core customer attributes, providing concrete customer references for business decisions and serving as a key data carrier connecting businesses and customers.

[0003] Customer profiling involves collecting and analyzing data on customer behavior, needs, and feedback. Through tag extraction and weight adjustment, a dynamically updated customer characteristic model is constructed, with the core objective of accurately representing customer information. The business marketing model, on the other hand, integrates industry data with customer profiles based on algorithms. Through demand matching and rule optimization, it generates targeted marketing interaction plans. When the two work together, the customer profile provides precise input to the large model, and the large model, through real-time interaction, feeds back into the profile for updates, forming a closed loop of "data-profile-model-interaction" to improve the accuracy of marketing response. In related technologies, Chinese invention patent application CN202510639920 proposes a precise financial customer profile analysis platform based on big data, while Chinese invention patent application CN202410998315 discloses a method for constructing data service products based on supplier profiles. These technologies can accurately depict customer profiles, empower financial businesses, and enhance institutional competitiveness.

[0004] The home improvement industry is characterized by high average order value, long decision-making cycles, and highly personalized needs. Customers focus on design style, material quality, construction techniques, and after-sales guarantees, and their decisions are easily influenced by factors such as word-of-mouth and case studies. Data dimensions cover multiple scenarios including promotion, transactions, repairs, and reviews. Customer profiling and large-scale models are still relatively uncommon in the home improvement industry's marketing applications. Traditional technologies lack targeted data integration mechanisms for the home improvement sector, resulting in fragmented customer data with low utilization rates, making it difficult to create accurate and dynamic customer profiles. Business models have poor adaptability to customer needs, often using generic configurations that fail to respond to the personalized and dynamic demands of home improvement customers. Data transmission lacks secure and controllable channels, leading to chaotic data flow, and data from model-customer interactions cannot effectively update customer profiles, resulting in low marketing efficiency and a poor customer experience. Summary of the Invention

[0005] To address the aforementioned technical issues, this invention proposes a business adaptability intelligent analysis method and system, as well as a human-computer interaction device, that combines customer profiles.

[0006] In a first aspect of the invention, a business fit analysis method combining customer profiles is proposed, the method being implemented based on data acquisition middleware.

[0007] The method includes the following steps:

[0008] Data tag construction steps: Based on the categories of historical statistical data in the home decoration industry, construct multiple data tags, including home decoration promotion data, home decoration sales data, home decoration repair data, and home decoration evaluation data;

[0009] Customer data acquisition steps: Continuously acquire conversation data of potential customers through data acquisition middleware. When the conversation data is identified to contain statistical data of the corresponding category, add at least one data tag for the potential customer or adjust the weight of at least one existing data tag.

[0010] The dialogue data includes dialogue content actively entered by the customer and the customer's interactive operations on the currently displayed content.

[0011] Customer profile building steps: Based on at least one data tag for each potential customer, build or update a customer profile for each potential customer;

[0012] Business model adaptation steps: Based on the customer profile corresponding to each potential customer, analyze the customer's target business needs, and call or update the corresponding target business model based on the target business needs;

[0013] Model interaction steps: Engage in dialogue and interaction with the customer based on the target business model;

[0014] The target business model communicates with the data acquisition middleware through a one-way (business) data pipeline.

[0015] After the model interaction step, return to the customer data acquisition step.

[0016] In the method described above, when updating a customer profile, the proportion of profile dimensions is dynamically optimized based on the adjusted weights of existing data tags. Tags with increased weights have higher priority for their corresponding profile dimensions, while tags with decreased weights have lower priority for their corresponding dimensions, so that the profile reflects changes in customer characteristics in real time.

[0017] In the method described above, when updating the target business model, the weight coefficients of the matching rules in the model are adjusted in accordance with the trend of the adjusted label weight changes and the target business needs of the customer, thereby enhancing the model's response accuracy to the dynamic needs of the customer.

[0018] In the method described above, a one-way (business) data pipeline is used. The one-way data pipeline only allows the data acquisition middleware to transmit customer profile data to the target business model. The interactive feedback (business) data of the target business model is not transmitted in reverse, ensuring the controllability of the (business) data flow between the data acquisition end and the model end.

[0019] In a second aspect of the invention, to implement the method described in the first aspect, a business adaptability intelligent analysis system combining customer profiles is also proposed. The system connects multiple benchmark business models, which obtain input data or update data from data acquisition middleware.

[0020] The system also includes:

[0021] Data tag construction unit: Based on the categories of historical statistical data in the home decoration industry, multiple data tags are constructed. The statistical data includes home decoration promotion data, home decoration sales data, home decoration repair data, and home decoration evaluation data.

[0022] Customer data acquisition unit: continuously acquires conversation data of potential customers through data acquisition middleware. When the conversation data is identified to contain statistical data of the corresponding category, at least one data tag is added for the potential customer or the weight of at least one existing data tag is adjusted. The conversation data includes conversation content actively entered by the customer and the customer's interactive operations on the currently displayed content.

[0023] Customer profile building unit: Based on at least one data tag for each potential customer, build or update a customer profile for each potential customer;

[0024] Business model adaptation unit: Based on the customer profile corresponding to each potential customer, analyze the customer's target business needs, and call or update the corresponding target business big model based on the target business needs;

[0025] Model Interaction Unit: Engages in dialogue with the customer based on the target business big model, and notifies the customer data acquisition unit after the dialogue ends.

[0026] After the dialogue ends, the target business big model generates a notification message containing a dialogue end identifier, customer ID, and interaction summary through the message interface of the data acquisition middleware. After verification by the middleware, the notification message is pushed to the customer data acquisition unit, triggering the subsequent data update process.

[0027] The business model adaptation unit obtains real-time tag weight changes of customer profiles through data acquisition middleware, and dynamically adjusts the model call priority or updates the model's requirement matching algorithm by combining the target business requirements with the adaptation parameter library of the benchmark business model.

[0028] The technical solution of this invention can be automatically executed by computer program instructions through an electronic device including a processor and a memory, and the computer program instructions can be loaded into a computer-readable storage medium. Therefore, in a third aspect of this invention, a computer-readable storage medium is also proposed, wherein the computer-readable storage medium includes computer program instructions, which, when executed by a processor, implement the intelligent analysis method for business adaptability based on customer profiles described in the first aspect.

[0029] The electronic device is preferably a human-computer interaction device, which includes a human-computer interaction interface and is equipped with at least one customer service application. The customer service application is connected to a large model server to implement the intelligent analysis method for business adaptability based on customer profiles as described in the first aspect.

[0030] The technical solution proposed in this application focuses on the pain points of marketing in the home decoration industry and has multiple core advantages.

[0031] In terms of accuracy, the solution builds a tagging system based on the unique data of the home decoration industry. By continuously collecting customer dialogue data and dynamically adjusting the tag weights, the customer profile can reflect changes in demand in real time, providing an accurate basis for business adaptation and solving the adaptation problem of highly personalized customer needs in the home decoration industry.

[0032] In terms of dynamic response capabilities, a closed-loop mechanism of "data collection - profile update - model adaptation - interactive feedback" is formed. The business model can be dynamically optimized according to changes in tag weights and customer needs, which greatly improves the accuracy of responding to dynamic customer needs and avoids the rigidity problem of generalized models.

[0033] Regarding data security and controllability, a unidirectional data pipeline is used to achieve communication between the data acquisition middleware and the business model, strictly controlling the data flow to prevent data chaos and ensure customer data security. In terms of system collaboration efficiency, the methods, systems, and equipment form a complete technical link, with each unit and step working efficiently together. The unified scheduling of the data acquisition middleware makes data transmission and processing more efficient, significantly improving marketing interaction efficiency and customer experience, and helping home decoration companies achieve precise marketing and cost reduction.

[0034] Further advantages of the present invention will be further detailed in the Specific Embodiments section in conjunction with the accompanying drawings. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a schematic diagram of the main steps of a business adaptability intelligent analysis method combining customer profiles according to an embodiment of the present invention.

[0037] Figure 2 yes Figure 1 A further preferred embodiment of the method is illustrated in the diagram;

[0038] Figure 3 yes Figure 1 The method described uses data acquisition middleware to obtain data for one-way transmission to a large model.

[0039] Figure 4 It is to achieve Figure 1 The functional (module) unit architecture diagram of the business adaptability intelligent analysis system combining customer profiles in the embodiment of the method. Detailed Implementation

[0040] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments identical to those described in this application. Rather, they are merely examples of apparatuses and methods identical to some aspects of this application as detailed in the appended claims.

[0041] The specific implementation of this application obviously involves the acquisition and processing of user-related data. Therefore, it should be emphasized that this application can display a prompt interface, pop-up window, or output voice prompt information before and during the collection of user-related data. This prompt interface, pop-up window, or voice prompt information is used to inform the user that their relevant data is being collected. This ensures that this application only begins executing the steps related to acquiring user-related data after receiving confirmation from the user regarding the prompt interface or pop-up window; otherwise (i.e., without receiving confirmation from the user), the steps related to acquiring user-related data end, meaning that user-related data is not acquired. In other words, all user data collected in this application is collected with the user's consent and authorization, and the collection, use, and processing of relevant user data must comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0042] See Figure 1 , Figure 1 This is a schematic diagram of the main steps of a business adaptability intelligent analysis method that combines customer profiles according to an embodiment of the present invention.

[0043] Figure 1The method illustrates the steps of tag construction, customer data acquisition, customer profile construction, business model adaptation, and model interaction.

[0044] Preferably, in actual execution, the customer data acquisition step is returned after the model interaction step.

[0045] Preferably, the method is implemented based on data acquisition middleware. The target business model communicates with the data acquisition middleware through a one-way data pipeline.

[0046] Specifically, the unidirectional data pipeline only allows the data acquisition middleware to transmit customer profile data to the target business model, and the interactive feedback data of the target business model is not transmitted in reverse, ensuring the controllability of the data flow between the data acquisition end and the model end.

[0047] To further describe the steps of the above method, in Figure 1 Based on this, see Figure 2 , Figure 2 yes Figure 1 A further preferred embodiment of the method is illustrated in the schematic diagram, which elaborates on the following: Figure 1 The specific implementation of each step is as follows:

[0048] Data tag construction steps: Based on the categories of historical statistical data in the home decoration industry, construct multiple data tags, including home decoration promotion data, home decoration sales data, home decoration repair data, and home decoration evaluation data;

[0049] Customer data acquisition steps: Continuously acquire conversation data of potential customers through data acquisition middleware. When the conversation data is identified to contain statistical data of the corresponding category, add at least one data tag for the potential customer or adjust the weight of at least one existing data tag.

[0050] Customer profile building steps: Based on at least one data tag for each potential customer, build or update a customer profile for each potential customer;

[0051] Business model adaptation steps: Based on the customer profile corresponding to each potential customer, analyze the customer's target business needs, and call or update the corresponding target business model based on the target business needs;

[0052] Model interaction steps: Engage in dialogue and interaction with the customer based on the target business model;

[0053] Preferably, the dialogue data includes dialogue content actively entered by the customer and the customer's interactive operations on the currently displayed content.

[0054] Figure 2The feedback loop self-learning process, which returns to the customer data acquisition step after the model interaction step, is clearly shown.

[0055] Specifically, when updating customer profiles, the proportion of profile dimensions is dynamically optimized based on the adjusted weights of existing data tags. Tags with increased weights have higher priority for corresponding profile dimensions, while tags with decreased weights have lower priority for corresponding dimensions, so that the profile reflects changes in customer characteristics in real time.

[0056] When updating the target business model, the weight coefficients of the matching rules in the model are adjusted in accordance with the trend of the adjusted label weight changes and the customer's target business needs, thereby enhancing the model's response accuracy to the customer's dynamic needs.

[0057] The following sections will describe the specific implementation methods of each of the above steps in more concrete scenarios. It is understood that the scenario parameter limitations and data types included in the specific implementation methods presented here are merely illustrative and not exhaustive.

[0058] In the data tag construction step, multiple data tags are mainly constructed based on the categories of historical statistical data in the home decoration industry. The statistical data includes home decoration promotion data, home decoration sales data, home decoration repair data, and home decoration evaluation data.

[0059] The tagging system for the multi-scenario characteristics of data in the home decoration industry is implemented as follows:

[0060] Based on the four core categories of historical statistical data in the home decoration industry (home decoration promotion data, home decoration sales data, home decoration repair data, and home decoration evaluation data), a hierarchical construction technique of "first-level category - second-level dimension - third-level tag" is adopted.

[0061] The primary categories correspond to the four major data categories mentioned above. The secondary dimensions are subdivided according to the characteristics of the home decoration industry. For example, home decoration promotion data includes dimensions such as "Promotion Channels (Online Platforms / Offline Exhibitions)" and "Promotion Content (Design Cases / Material Discounts)". Home decoration sales data includes dimensions such as "Housing Type (Essential Housing / Upgraded Apartments / Villas)", "Renovation Budget (100,000-200,000 / 200,000-500,000 / Over 500,000)" and "Decoration Style (Modern Minimalist / New Chinese / Nordic)". Home decoration repair data includes dimensions such as "Repair Area (Walls / Plumbing / Carpentry)" and "Reason for Repair (Material Quality / Construction Process)". Home decoration evaluation data includes "Evaluation Dimensions (Design Satisfaction / Construction Satisfaction / After-Sales Response Speed)". The tertiary tags are specific feature identifiers, such as "Decoration Style - New Chinese", "Budget - 200,000-500,000", and "Repair Area - Plumbing".

[0062] This hierarchical tagging system accurately covers all business data scenarios in the home decoration industry, from customer acquisition to after-sales service, solving the shortcomings of traditional tags that only cover a single business link and cannot adapt to long-cycle home decoration services.

[0063] In the customer data acquisition step, conversation data of potential customers is continuously acquired through data acquisition middleware. When the conversation data is identified to contain statistical data of the corresponding category, at least one data tag is added for the potential customer or the weight of at least one existing data tag is adjusted.

[0064] In this invention, the improvement in this step mainly focuses on the multi-form collection and real-time tag adjustment of customer dialogue data in the home decoration industry, and is specifically implemented as follows:

[0065] An integrated processing module is built based on data acquisition middleware, which integrates "real-time access to multi-source dialogue data, home decoration feature recognition, and dynamic tag updates".

[0066] The data acquisition middleware connects with home decoration companies' intelligent customer service systems, online consultation platforms, and offline store interaction terminals through API interfaces to continuously acquire two types of dialogue data from potential customers: first, dialogue content actively entered by customers (such as "I want to decorate in a new Chinese style, with a budget of around 300,000 yuan" or "There is a water leakage problem after the water and electricity renovation"); second, customer interaction with the displayed content (such as clicking on the "Villa Decoration Case" page, long-pressing the "Environmentally Friendly Coating" introduction pop-up window, staying on the "After-sales Repair Application" interface for more than 30 seconds, etc., this invention does not impose specific restrictions on this.

[0067] Meanwhile, the data acquisition middleware incorporates a keyword and operation feature recognition model for the home decoration industry. When the dialogue data is identified to contain corresponding category statistical data features, a tag update mechanism is triggered: if a customer mentions "New Chinese Style" for the first time, a new tag "Decoration Style - New Chinese Style" is added for them; if a customer inquires about "Environmentally Friendly Paint" multiple times (e.g., mentioning it 3 times or more), the weight of the "Material Preference - Environmentally Friendly Paint" tag is increased from the initial 0.3 to 0.8; if the customer does not subsequently pay attention to "Nordic Style Cases," the weight of the "Decoration Style - Nordic Style" tag is decreased from 0.5 to 0.1. This allows the tag weight to be dynamically adjusted according to the customer's home decoration needs and attention, adapting to the industry characteristics where the needs of home decoration customers are gradually becoming clearer.

[0068] Building upon the steps outlined above, the next step is to construct a customer profile. This step involves building or updating a customer profile for each potential customer based on at least one data tag for each customer.

[0069] Compared to existing technologies, the improvement of this step for the home decoration industry lies in the dynamic optimization of home decoration customer profiles, specifically as follows:

[0070] Based on the three-level tags mentioned above for each potential customer, a dynamic construction technology of "tag weight driving profile dimension priority" is adopted.

[0071] First, we set three core dimensions for home decoration customer profiles: "demand preferences (decoration style / materials / budget)," "service concerns (design / construction / after-sales service)," and "house characteristics (type / area / floor layout)." Each dimension is associated with a corresponding three-level tag.

[0072] When customer tag weights are adjusted, the proportion of profile dimensions is optimized in real time: If the weight of the "Repair Area - Plumbing and Electrical" tag increases from 0.2 to 0.7, it indicates that the customer is concerned about the quality of plumbing and electrical construction. The priority of the corresponding "Service Focus - Construction" dimension increases from the 3rd to the 1st position, and the information proportion of this dimension in the profile display is increased (e.g., from 20% to 40%). If the weight of the "Renovation Budget - 100,000 to 200,000" tag decreases from 0.6 to 0.2, it indicates that the customer's budget may increase. The priority of the corresponding "Needs Preference - Budget" dimension decreases from the 2nd to the 3rd position, and its information proportion is reduced (e.g., from 30% to 15%).

[0073] This dynamic optimization technology enables customer profiles to reflect the changes in characteristics of home improvement customers in real time from initial consultation to the determination of their needs, avoiding the shortcomings of traditional static profiles that cannot adapt to the long decision-making cycle and dynamic adjustment of needs in the home improvement industry.

[0074] The next step is to adapt the business model, which involves analyzing the customer's target business needs based on the customer profile corresponding to each potential customer, and calling or updating the corresponding target business model based on the target business needs.

[0075] Compared to existing technologies, this step focuses on improving the technology for the home improvement industry by precisely matching the dynamic needs of customers with business models. The specific implementation is as follows:

[0076] A benchmark business model library specifically for the home decoration industry is pre-built, including "design scheme recommendation model", "material selection matching model", "construction team assignment model", "after-sales problem response model", etc. Each model has built-in matching rules adapted to home decoration scenarios (such as the matching rules of the "design scheme recommendation model" including "style-house type-budget" association rules).

[0077] The business model adaptation unit acquires real-time data on changes in the tag weights of customer profiles through data acquisition middleware. Combined with the customer's target business needs (e.g., "obtain design solutions" during the consultation phase, and "resolve repair issues" during the after-sales phase), it updates the target business model using a "dynamic adjustment of rule weight coefficients" technology.

[0078] If a customer's "Decoration Style - New Chinese Style" tag weight increases from 0.4 to 0.9 (the magnitude of the change can be determined based on the adjustment algorithm of the tag weight change rate), and the target requirement is "to obtain a design plan", then the "Design Plan Recommendation Model" is invoked, and the weight coefficient of the "New Chinese Style - Corresponding Apartment Type (e.g., three bedrooms and two living rooms)" matching rule in the model is increased from 0.5 to 0.9, while the weight coefficient of the "Modern Minimalist Style - Corresponding Apartment Type" rule is decreased from 0.4 to 0.1. If a customer's "After-sales Response Speed" evaluation tag weight increases from 0.3 to 0.8, and the target requirement is "to resolve repair issues", then the "After-sales Issue Response Model" is updated, and the weight coefficient of the "Dispatch within 2 hours" rule is increased from 0.6 to 0.9, enhancing the model's response accuracy to the home decoration customer's concern about after-sales efficiency and solving the problem that traditional general models cannot adapt to the segmented needs of the home decoration industry.

[0079] Finally, the model interaction step is executed, which involves dialogue and interaction with the customer based on the target business model.

[0080] The improved technical approach in this step is to achieve a closed loop of scenario-based model interaction and data feedback in the home decoration industry, specifically as follows:

[0081] Based on the target business model, we use a technical approach of "scenario-based interactive content generation + real-time feedback of interactive data" to engage in dialogue and interaction with customers.

[0082] If the "Design Scheme Recommendation Model" is invoked, the model combines the tags "House Type - Improved Villa", "Decoration Style - New Chinese Style", and "Budget - 300,000-500,000 RMB" from the customer profile to generate interactive content containing 3 New Chinese Style villa design schemes (including floor plans, material lists, and detailed quotations) that meet the customer's needs. This content is displayed through a human-computer interaction interface and allows customers to "favorite", "ask questions", and "modify requirements" for the schemes. If the "After-Sales Problem Response Model" is invoked, the model combines the tags "Repair Area - Water and Electricity" and "House Address - XX Community" to generate interactive content including "Water and Electricity Repair Team has been assigned and is expected to arrive within 1 hour" and "Repair Personnel Contact Information XXX", while simultaneously prompting the customer that "the repair team's location can be viewed in real time".

[0083] After the interaction ends, the target business model generates an interaction summary containing "customer ID, interaction model type, and customer operation feedback (such as saving the plan / confirming the repair time)" through the message interface of the data acquisition middleware. After being verified by the middleware, the summary is pushed to the customer data acquisition unit, triggering a new round of data collection and tag adjustment in the customer data acquisition steps, forming a closed loop of "interaction-feedback-update", which is suitable for the service characteristics of the home decoration industry that require continuous follow-up on customer needs.

[0084] As a further preferred embodiment, after the model interaction step is completed, the closed-loop process is triggered through the message notification mechanism of the data acquisition middleware, specifically implemented as follows:

[0085] After the target business model concludes its interaction with the customer (e.g., the customer terminates the consultation, confirms the solution, or completes the after-sales response), it immediately generates a structured interaction log containing the customer's unique identifier (ID), the interaction scenario type (e.g., design consultation / after-sales repair), key interaction behaviors (e.g., clicking the "modify budget" button, entering "prefer solid wood flooring" feedback), and the interaction duration. This log, after being formatted and validated on the model side, is pushed to the data acquisition middleware via a pre-defined one-way data pipeline notification channel.

[0086] After receiving the logs, the data acquisition middleware starts the home decoration industry-specific interactive feature extraction module to analyze the information in the logs related to the tag system: for example, if it recognizes "modify the budget to more than 500,000", it will trigger the weight increase logic of the "decoration budget" tag in the customer data acquisition step; if it captures "browsing solid wood flooring cases multiple times", it will activate the process of adding or strengthening the weight of the "material preference - solid wood" tag.

[0087] Meanwhile, the middleware generates a closed-loop trigger command, which associates the extracted interaction features with the customer's existing tag library, driving the customer data acquisition step into a new round of real-time collection—including listening to the customer's subsequent dialogue data in channels such as intelligent customer service and store terminals, realizing a full-link closed loop of "interaction feedback - tag update - profile iteration - model re-adaptation," adapting to the industry characteristics of home decoration customer needs becoming clearer as communication deepens, and ensuring that the system continuously learns customer dynamic preferences.

[0088] Based on the above improvements, see Figure 3 , Figure 3 yes Figure 1 The method uses data acquisition middleware to obtain data and transmit it unidirectionally to a large model.

[0089] As emphasized in the aforementioned method steps, the customer data acquisition step continuously acquires conversation data of potential customers through a data acquisition middleware, and the target business model communicates with the data acquisition middleware through a one-way data pipeline.

[0090] like Figure 3 As shown, the unidirectional data pipeline only allows the data acquisition middleware to transmit profile data / dialogue data of N customers to the target business model. The interactive feedback data of the target business model is not transmitted in reverse, ensuring the controllability of the data flow between the data acquisition end and the model end.

[0091] exist Figure 3 In this embodiment, the data acquisition middleware adopts a "distributed access + home decoration industry protocol adaptation" architecture, which is implemented as follows:

[0092] Access Layer: It connects with multiple data nodes such as the intelligent customer service system, store interaction terminal, and online consultation platform of home decoration enterprises through multi-protocol adaptation interfaces (HTTP / HTTPS, WebSocket, MQTT), supports concurrent access of 1000+ terminal devices, and meets the data collection needs of the home decoration industry in multiple online and offline scenarios.

[0093] Processing layer: Built-in NLP semantic parsing module and interactive behavior recognition engine dedicated to the home decoration industry (such as adopting the Transformer architecture and pre-training strategy for home decoration terminology), which can analyze key information such as "decoration style, budget range, and material preferences" in customer dialogues in real time, as well as feature values ​​of interactive operations such as clicks, pauses, and favorites, and transform unstructured data into structured data that conforms to the tag system.

[0094] Storage layer: It adopts a distributed cache (Redis) + time-series database (InfluxDB) combined architecture to cache real-time customer tag data (such as temporary preferences in the current session), persist historical interaction data and tag weight change trajectory, and support fast traceability query by customer ID and timestamp, which is suitable for the characteristics of long decision-making cycle and historical traceability needs of customers in the home decoration industry.

[0095] The unidirectional data pipeline is built on the technology of "hardware-level isolation + transmission access control", which specifically includes:

[0096] Physical isolation mechanism: A one-way network gateway device is used to achieve physical link isolation between the data acquisition middleware and the target business model, retaining only the transmission channel from the middleware to the model and blocking the reverse data flow path, thus eliminating the reverse transmission of model interaction feedback data from the hardware level.

[0097] Transmission Protocol Design: A custom data transmission protocol for the home decoration industry is designed. The protocol fields only contain necessary data such as customer profile ID, tag weight set, and business scenario identifier. It does not contain any sensitive fields that can be used to reverse locate middleware or customer's original data. The transmitted content must be anonymized by the middleware (such as hiding customer's mobile phone number, detailed address, etc.).

[0098] Permission verification mechanism: Dynamic token verification is set at the pipeline entry point. The data collection middleware generates a temporary access token every 30 seconds. The target business model can only receive data with a valid token, and the token permission is strictly limited to "read only", and it cannot initiate any data write or query requests.

[0099] Improvements to home decoration data adaptation in the middleware: Add a "home decoration terminology dictionary" (containing industry terms such as "water and electricity renovation, whole-house customization, and soft furnishing matching") and a "scenario-based rule engine" to the data processing layer (such as automatically associating the "construction specification" tag with the identification of "whether a load-bearing wall can be demolished") to improve the parsing accuracy of home decoration professional dialogue data.

[0100] Batch transmission optimization for unidirectional pipelines: Supports batch transmission of profile data by customer groups (such as "new Chinese style prospective customers" and "after-sales repair customers"), improving transmission efficiency by 50% and adapting to the data needs of centralized marketing activities in the home decoration industry.

[0101] Visualized monitoring of data flow: The middleware has a built-in flow monitoring module that records the transmission time, customer ID, data volume, and other information of each piece of data in real time, generating a visual flow graph to facilitate managers in tracing the data transmission path.

[0102] Through the above-mentioned technological improvements, the home decoration industry can achieve the following significant improvements:

[0103] 1. Enhanced Data Security: The physical isolation and access control of the one-way channel ensure that the customer's original dialogue data and privacy information are stored only in the middleware. The target business model cannot obtain or tamper with the data, which complies with the requirements of the Personal Information Protection Law for data minimization and secure transmission, reducing the risk of data leakage by 80% compared with the traditional two-way transmission mode.

[0104] 2. Optimized transmission efficiency and accuracy: The middleware's adaptation to the home decoration industry improves the data parsing accuracy from 72% in the general model to 91%. The batch transmission function meets the high-concurrency data needs of home decoration companies during promotional periods, and the transmission latency is controlled within 100ms.

[0105] 3. Enhanced system stability: Through unified scheduling of middleware and one-way control of pipelines, the data acquisition process is prevented from being interfered with by abnormal feedback data from the target business model. The continuous operation stability of the system is improved to 99.9%, solving the system crash problem caused by chaotic data interaction between the model and the acquisition end in the traditional architecture.

[0106] 4. Enhanced Business Adaptability: The middleware's professional processing of home decoration data, combined with the precise data supply from one-way transmission, enables the target business model to focus on the in-depth utilization of customer profiles, rather than data cleaning and security control (which is completed by the middleware and one-way data pipeline). The model's response accuracy to home decoration customer needs is improved by 40%, significantly optimizing the customer interaction experience.

[0107] It is important to note that in the aforementioned embodiments, it is explicitly stated that the unidirectional data pipeline only allows the data acquisition middleware to transmit "customer profile data" to the target business model; the interactive feedback data of the target business model is not transmitted in reverse (i.e., the unidirectional data pipeline is essentially a unidirectional "business" data pipeline). However, after the target business model generates notification information through the message interface of the data acquisition middleware, it needs to be verified by the middleware before being pushed to the customer data acquisition unit. This "notification information" is essentially a business trigger notification from the target business model (a simple "control" signal), which is only used to trigger the subsequent data update process (and does not contain actual interactive feedback data). Therefore, this process does not contradict the statement that the unidirectional data pipeline does not transmit "feedback data / business data" in reverse.

[0108] In other words, the one-way data pipeline is only responsible for transmitting "core business data", namely, structured data that directly supports business model decisions, such as customer profile data and tag weights. It is strictly limited to one-way transmission from the data acquisition middleware to the target business model, and the reverse transmission of any business data is prohibited.

[0109] The "notification information" after model interaction (including dialogue end identifier, customer ID, and interaction summary) belongs to "business trigger notification" ("control" signal / identifier, summary, ID, etc., which can be represented as simple open, activate, etc.). It does not involve the customer's original data or core profile content. Its transmission is achieved through a "control signal channel" set up separately by the data acquisition middleware (physically isolated from the one-way business data pipeline). This channel is only used to trigger the process loop and does not have the ability to transmit business data in reverse.

[0110] At this point, when the target business model pushes notification information through the "message interface" of the data acquisition middleware, the interface only grants "signal reception" permissions and does not allow the model to read any customer data stored in the middleware (such as original conversation records or complete profiles). The format of the notification information must be predefined by the middleware (e.g., containing only encrypted customer IDs and standardized interaction type encodings), and the transmission process does not carry any information that could restore customer privacy or business data.

[0111] exist Figures 1-3 Based on the detailed description of the method embodiments, see also Figure 4 , Figure 4 Implementation shown Figure 1 The functional (module) unit architecture diagram of the business adaptability intelligent analysis system combining customer profiles in the embodiment of the method.

[0112] Figure 4 The system connects multiple benchmark business models, which obtain input data or update data from the data acquisition middleware.

[0113] The system also includes:

[0114] Data tag construction unit: Based on the categories of historical statistical data in the home decoration industry, multiple data tags are constructed. The statistical data includes home decoration promotion data, home decoration sales data, home decoration repair data, and home decoration evaluation data.

[0115] Customer data acquisition unit: continuously acquires conversation data of potential customers through data acquisition middleware. When the conversation data is identified to contain statistical data of the corresponding category, at least one data tag is added for the potential customer or the weight of at least one existing data tag is adjusted. The conversation data includes conversation content actively entered by the customer and the customer's interactive operations on the currently displayed content.

[0116] Customer profile building unit: Based on at least one data tag for each potential customer, build or update a customer profile for each potential customer;

[0117] Business model adaptation unit: Based on the customer profile corresponding to each potential customer, analyze the customer's target business needs, and call or update the corresponding target business big model based on the target business needs;

[0118] Model Interaction Unit: Engages in dialogue with the customer based on the target business big model, and notifies the customer data acquisition unit after the dialogue ends.

[0119] After the dialogue ends, the target business big model generates a notification message containing a dialogue end identifier, customer ID, and interaction summary through the message interface of the data acquisition middleware. After verification by the middleware, the notification message is pushed to the customer data acquisition unit, triggering the subsequent data update process.

[0120] The business model adaptation unit obtains real-time tag weight changes of customer profiles through data acquisition middleware, and dynamically adjusts the model call priority or updates the model's requirement matching algorithm by combining the target business requirements with the adaptation parameter library of the benchmark business model.

[0121] Similar to the method implementation, this system offers significant advantages tailored to the characteristics of the home decoration industry. It establishes an industry-specific tagging system through a data tagging construction unit; accurately captures customer dynamics through a customer data acquisition unit; enables real-time updates to customer profiles through a customer profiling construction unit; dynamically adjusts the model through a business model adaptation unit to improve response accuracy; and ensures closed-loop optimization through a model interaction unit. The message interface and one-way transmission mechanism of the data acquisition middleware ensure data security and controllability, and the efficient collaboration among all units significantly improves the adaptation of home decoration business processes and customer interaction.

[0122] Since the system implementation and method implementation are essentially corresponding to each other, their implementation principles, improvement methods, and beneficial effects should also correspond to each other, and will not be elaborated further here. The slight difference is that the method implementation uses the concept of a "target business model," while the system implementation, to reflect the big data processing process, uses the concepts of "baseline business model" and "target business model," but both describe the same concept.

[0123] Other model principles and technical means not elaborated in this invention can be found in the prior art.

[0124] Compared to existing technologies, the improvements and related beneficial effects of this invention are progressive, mainly reflected in:

[0125] First, this invention addresses the multi-scenario characteristics of data in the home decoration industry (promotion, sales, repair, evaluation) by constructing a "three-level hierarchical tagging system." It also uses data collection middleware to analyze customer dialogue and interaction data, accurately adding or adjusting tag weights to solve the problem of traditional tags having limited coverage and being unable to adapt to long-term home decoration services, thus achieving preliminary and accurate capture of customer needs.

[0126] Secondly, based on the characteristics of long decision-making cycles and dynamic changes in needs of home decoration customers, this invention uses tag weights to drive dynamic optimization of customer profile dimensions, so that the profile can reflect the changes in customer characteristics from consultation to after-sales service in real time. At the same time, the business model adaptation unit combines tag weight trends and home decoration segmentation needs to adjust the model matching rule weights, avoid the rigidity of general models, and improve the accuracy of demand response.

[0127] Furthermore, this invention relies on a one-way data pipeline and middleware message interface to achieve one-way secure transmission of data collection middleware to the business model. After the interaction, the middleware pushes a notification to trigger a closed loop of data updates, which not only protects the privacy and security of home decoration customers' data, but also forms a continuous optimization link of "collection-profiling-adaptation-interaction-update", adapting to the characteristic of home decoration services that require continuous follow-up.

[0128] Finally, at the system level, all units work together in accordance with the characteristics of the home decoration industry. The middleware professionally processes home decoration data, the model adaptation unit focuses on specific home decoration scenarios, and the interaction unit generates scenario-based content. Overall, this significantly improves the efficiency of home decoration business adaptation and customer experience, helping home decoration companies achieve precise services and reduce costs and increase efficiency.

[0129] The foregoing has shown and described the method embodiments and systems of the present invention, but it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A business fit analysis method combining customer profiles, the method being implemented based on data acquisition middleware, characterized in that, The method includes the following steps: Data tag construction steps: Based on the categories of historical statistical data in the home decoration industry, construct multiple data tags. The historical statistical data includes home decoration promotion data, home decoration sales data, home decoration repair data, and home decoration evaluation data. Customer data acquisition steps: Continuously acquire conversation data of potential customers through data acquisition middleware. When the conversation data is identified to contain statistical data of the corresponding category, add at least one data tag for the potential customer or adjust the weight of at least one existing data tag. Customer profile building steps: Based on at least one data tag for each potential customer, build or update a customer profile for each potential customer; Business model adaptation steps: Based on the customer profile corresponding to each potential customer, analyze the customer's target business needs, and call or update the corresponding target business model based on the target business needs; Model interaction steps: Engage in dialogue and interaction with the customer based on the target business model; The target business model communicates with the data acquisition middleware through a one-way data pipeline.

2. The intelligent analysis method for business adaptability based on customer profiles as described in claim 1, characterized in that, The dialogue data includes dialogue content actively entered by the customer and the customer's interactive operations on the currently displayed content.

3. The intelligent analysis method for business adaptability based on customer profiles as described in claim 1, characterized in that, After the model interaction step, return to the customer data acquisition step.

4. The intelligent analysis method for business adaptability based on customer profiles as described in claim 1, characterized in that, When updating customer profiles, the proportion of profile dimensions is dynamically optimized based on the adjusted weights of existing data tags. Tags with increased weights have higher priority for their corresponding profile dimensions, while tags with decreased weights have lower priority for their corresponding dimensions, ensuring that the profile reflects changes in customer characteristics in real time.

5. The intelligent analysis method for business adaptability based on customer profiles as described in claim 1, characterized in that, When updating the target business model, the weight coefficients of the matching rules in the model are adjusted in accordance with the trend of the adjusted tag weight changes and the target business needs of the customer, so as to enhance the model's response accuracy to the dynamic needs of the customer.

6. The intelligent analysis method for business adaptability based on customer profiles as described in claim 1, characterized in that, A one-way data pipeline only allows the data acquisition middleware to transmit customer profile data to the target business model. The interactive feedback data of the target business model is not transmitted in reverse, ensuring the controllability of the data flow between the data acquisition end and the model end.

7. A business fit intelligent analysis system that combines customer profiles, wherein the system connects multiple benchmark business models, and the benchmark business models obtain input data or update data from data acquisition middleware, characterized in that, The system also includes: Data tag construction unit: Based on the categories of historical statistical data in the home decoration industry, multiple data tags are constructed. The statistical data includes home decoration promotion data, home decoration sales data, home decoration repair data, and home decoration evaluation data. Customer data acquisition unit: continuously acquires conversation data of potential customers through data acquisition middleware. When the conversation data is identified to contain statistical data of the corresponding category, at least one data tag is added for the potential customer or the weight of at least one existing data tag is adjusted. The conversation data includes conversation content actively entered by the customer and the customer's interactive operations on the currently displayed content. Customer profile building unit: Based on at least one data tag for each potential customer, build or update a customer profile for each potential customer; Business model adaptation unit: Based on the customer profile corresponding to each potential customer, analyze the customer's target business needs, and call or update the corresponding target business big model based on the target business needs; Model Interaction Unit: Engages in dialogue with the customer based on the target business big model, and notifies the customer data acquisition unit after the dialogue ends.

8. The intelligent analysis system for business adaptability based on customer profiles as described in claim 7, characterized in that, After the conversation ends, the target business big model generates a notification message containing a conversation end identifier, customer ID, and interaction summary through the message interface of the data acquisition middleware. After verification by the middleware, the message is pushed to the customer data acquisition unit, triggering the subsequent data update process.

9. The intelligent analysis system for business adaptability based on customer profiles as described in claim 7, characterized in that, The business model adaptation unit obtains real-time tag weight changes of customer profiles through data acquisition middleware, and dynamically adjusts the model call priority or updates the model's requirement matching algorithm by combining the target business requirements with the adaptation parameter library of the benchmark business model.

10. A human-computer interaction device, the human-computer interaction device comprising a human-computer interaction interface, characterized in that, The human-computer interaction device is equipped with at least one customer service application, which is connected to a large model server to implement the intelligent analysis method for business adaptability based on customer profiles as described in any one of claims 1-6.

Images