AI-powered intelligent CRM supports dynamic configuration and rendering of heterogeneous data fields across multiple business scenarios.

Abstract

This invention discloses a method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM that supports multiple business scenarios. It relates to the field of data processing technology, and includes: collecting heterogeneous fields and access contexts from multiple business scenarios; constructing an original field event set and generating field semantic fingerprints; retrieving a standard field semantic library based on the semantic fingerprints to generate a unified standard field set; then combining business scenarios to complete field filtering, permission constraints, and conflict resolution, forming an effective field state set; further constructing a field dependency graph to complete page arrangement, component mapping, and initial interface rendering; finally, in runtime, performing local propagation based on field changes to achieve incremental reconfiguration and local re-rendering of fields. This method achieves semantic-level unified normalization and local incremental dynamic configuration rendering of heterogeneous CRM fields in multiple business scenarios, improving the accuracy of field configuration and the efficiency of page linkage response.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for dynamically configuring and rendering heterogeneous data fields in AI-powered intelligent CRM that supports multiple business scenarios. Background Technology

[0002] As enterprise digital operations continue to evolve towards refinement and end-to-end collaboration, CRM systems have gradually evolved from early single-business support tools primarily focused on customer profile maintenance and sales process recording to comprehensive business platforms covering multiple business scenarios such as lead generation, opportunity development, contract signing, fulfillment and delivery, after-sales service, and renewal management. In this process, the data sources accessed by CRM have expanded from single business tables to customer master data, business process data, channel collaboration data, and data synchronized from external platforms. Fields from different sources exhibit significant differences in naming conventions, data types, value structures, and usage stages. To adapt to the need for rapid multi-scenario page construction, existing technologies typically employ field mapping tables, template form configurations, conditional display rules, and front-end dynamic form engines to extract, categorize, and render fields, thereby enabling field display and interactive control across different pages, roles, and business stages. Especially in AI-assisted CRM scenarios, some systems have attempted to introduce semantic matching, rule recommendation, and page linkage mechanisms to improve field configuration efficiency and interface adaptability.

[0003] However, existing methods still have two main shortcomings: First, for the normalization process of heterogeneous fields from multiple sources, most solutions still rely on static field mapping, manual alias maintenance, or local rule matching. It is difficult to establish a stable field semantic attribution mechanism by simultaneously utilizing multi-dimensional information such as field name semantics, field value distribution, business stage, co-occurrence relationship, and manual correction trajectory. This leads to problems such as difficulty in unifying synonymous fields, mismatch of fields with the same name, and insufficient scalability of the standard field system when new data sources are accessed or reused across business scenarios. Second, for field linkage and dynamic rendering during page runtime, existing solutions mostly use decentralized condition judgment or whole-page recalculation to handle field display, validation, candidate values, and changes in business status. They lack a structured mechanism for local propagation and local re-rendering around field dependencies. As a result, in scenarios with complex field linkage relationships and frequent business status switching, it is easy to cause problems such as excessively large configuration calculation scope, decreased response efficiency, and coarse page update granularity. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a dynamic configuration and rendering method for heterogeneous data fields in AI-powered intelligent CRM that supports multiple business scenarios, solving the problems of difficulty in achieving high-precision unification and normalization of heterogeneous fields, and coarse page update granularity and insufficient response efficiency in complex linkage scenarios.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a method for dynamically configuring and rendering heterogeneous data fields in AI-powered intelligent CRM that supports multiple business scenarios, comprising:

[0008] Collect heterogeneous data field information and access context information from multiple business scenarios, construct an original field event set, extract multidimensional features of each heterogeneous field based on the original field event set, and construct a field semantic fingerprint set;

[0009] By using the field semantic fingerprint set to perform similarity retrieval in a pre-established standard field semantic library, the standard semantic slots corresponding to each heterogeneous field are determined, and a unified standard field set is generated.

[0010] Based on the current business scenario, page task objectives, and customer lifecycle stage, the unified standard field set is narrowed down to generate a task candidate field set. Permission constraints and data conflict resolution are performed on the task candidate field set to generate an effective field status set.

[0011] A field dependency graph is constructed based on the effective field status set to obtain a field relationship network. Page configuration orchestration is performed based on the field relationship network to generate the first-round page configuration scheme. Component mapping and initial business interface rendering are completed based on the first-round page configuration scheme.

[0012] The initial business interface is bound to the field dependency graph to form a runtime configuration context;

[0013] In the runtime configuration context, when a field change event is detected, local propagation is performed in the field dependency graph starting from the triggering field, and incremental reconfiguration and local re-rendering are performed only on the fields in the affected subgraph.

[0014] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the steps for constructing the original field event set are as follows:

[0015] Collect the access context information corresponding to the current request, and extract heterogeneous field records related to the target business object from multiple business data sources; after standardizing the heterogeneous field records, bind and aggregate them with the access context information to construct the original field event set.

[0016] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the steps of extracting multi-dimensional features of each heterogeneous field based on the original field event set and constructing a field semantic fingerprint set are as follows:

[0017] Based on the original field event set, extract the semantic features of field name text, sample value distribution features, upstream and downstream call relationship features, business stage features, historical co-occurrence field set features, and manual correction trajectory features for each heterogeneous field; normalize and uniformly encode the extracted features to construct a field semantic fingerprint set that represents the comprehensive semantic attributes of heterogeneous fields.

[0018] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the steps for generating a unified standard field set are as follows:

[0019] Based on the field semantic fingerprint set, similarity retrieval is performed in the pre-established standard field semantic library, and the standard semantic slots corresponding to each heterogeneous field are determined according to the consistency of business stage, historical manual correction, and master data field priority.

[0020] Extract standard field codes and field values ​​from heterogeneous fields that have completed the standard semantic slot assignment, generate standard field objects containing field source priority, normalized confidence and applicable scope markers, and summarize them to form a unified standard field set.

[0021] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the steps for generating the task candidate field set are as follows:

[0022] Read the current business scenario identifier, page task objective, and customer lifecycle stage, and establish the field filtering benchmark corresponding to this round of page processing based on the applicable scope marker in the unified standard field object;

[0023] Based on the field filtering benchmark, the applicable scope filtering and task direct relevance convergence are performed on the unified standard field set, retaining the standard fields required for this round of page processing, and generating a task candidate field set.

[0024] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the steps for generating the effective field status set are as follows:

[0025] Read the current user role, organizational position, data access security rules and page function permissions, and determine the visibility, editability, exportability and linkage trigger eligibility of each field in the task candidate field set accordingly;

[0026] The system identifies multiple source value conflicts for the same standard field in the candidate field set of the task, and determines the unique effective value of the field in the order of priority of the main data source, the most recent update time, and the manual confirmation result. It generates an effective field status set that includes field code, effective value, display status, editing status, linkage trigger status, and verification participation flag.

[0027] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the steps of constructing a field dependency graph based on the effective field status set to obtain a field relationship network, and performing page configuration orchestration based on the field relationship network to generate the first-round page configuration scheme are as follows:

[0028] Field nodes are constructed based on each field in the effective field status set, and the display dependencies, validation dependencies, candidate value dependencies, and business logic dependencies between fields are identified to generate a field dependency graph. The propagation direction, triggering condition, priority, and termination condition are written for each dependency edge in the field dependency graph to generate a field relationship network.

[0029] Based on the field relationship network and the effective field status set, combined with the current page task objectives, field usage frequency and business completion path, the page configuration orchestration is performed on the effective fields to generate the first-round page configuration scheme, which includes the field arrangement order, display area, collapse level and preset linkage markup.

[0030] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the steps for completing component mapping and initial business interface rendering based on the first-round page configuration scheme are as follows:

[0031] Based on the standard data type of the corresponding field in the unified standard field set, and combined with the display and editing status of the corresponding field in the effective field status set and the preset linkage markers in the first round of page configuration scheme, each field is mapped to the corresponding interface component;

[0032] Based on the display area, field arrangement order, and collapse hierarchy determined in the first round of page configuration scheme, the mapped interface components are organized into regions and arranged into hierarchical levels, and the initial business interface is rendered.

[0033] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the steps of binding the initial business interface with the field dependency graph to form a runtime configuration context are as follows:

[0034] Bind each component instance in the initial business interface to the corresponding field node in the field relationship network, and establish a field status cache table to record the latest display status, editing status and value status of each field;

[0035] Register event listeners for interactive components to capture field input, selection, deletion, toggle, and submit actions, and establish a local update queue to store the set of field nodes to be propagated, forming a runtime configuration context.

[0036] As a preferred embodiment of the AI-powered intelligent CRM heterogeneous data field dynamic configuration and rendering method supporting multiple business scenarios described in this invention, the method involves: when a field change event is detected, local propagation is performed in the field dependency graph starting from the triggering field, and incremental reconfiguration and local re-rendering are performed only on the fields in the affected subgraphs. The steps are as follows:

[0037] In the runtime configuration context, when the event listener detects a field change event, it writes the corresponding triggering field node into the local update queue and determines the affected subgraph in the field relationship network starting from the triggering field node.

[0038] Based on the affected subgraph, local propagation is performed according to the priority of dependent edges. Only the display state, candidate value set, validation rules, required state and component display parameters of the affected fields are recalculated. The field state cache table is written back and local re-rendering is performed to complete the incremental reconfiguration of the fields.

[0039] The beneficial effects of this invention are as follows: This invention can utilize multi-dimensional information such as field name semantics, field value distribution, business stage, historical co-occurrence relationship and manual correction trajectory to achieve more accurate normalization of heterogeneous fields to standard fields, and improve the reuse and expansion capabilities of multi-source fields; at the same time, through field dependency graph and local update mechanism, incremental reconfiguration and local re-rendering are only performed on the affected subgraph, reducing the repeated calculation of irrelevant fields and overall interface redrawing, and improving the response efficiency and configuration stability of complex business pages under multi-scenario switching and high-frequency linkage. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0041] Figure 1 A flowchart illustrating the dynamic configuration and rendering methods for heterogeneous data fields in AI-powered intelligent CRM to support multiple business scenarios.

[0042] Figure 2This is a diagram illustrating field dependencies.

[0043] Figure 3 This is a timing diagram for local updates in runtime.

[0044] Figure 4 Configure a relationship diagram for the page. Detailed Implementation

[0045] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0046] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0048] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for dynamically configuring and rendering heterogeneous data fields in AI-powered intelligent CRM that supports multiple business scenarios, including the following steps:

[0049] S1. Collect heterogeneous data field information and access context information under multiple business scenarios, construct the original field event set, extract the multi-dimensional features of each heterogeneous field based on the original field event set, and construct the field semantic fingerprint set;

[0050] S1.1: Collect the access context information corresponding to the current request, and extract heterogeneous field records related to the target business object from multiple business data sources;

[0051] Specifically, in the AI-powered CRM system, the customer master identifier serves as the unified anchoring center for business objects. Business records corresponding to the current request from sales leads, opportunity promotion, contract orders, after-sales service, work order processing, channel collaboration, and external platform synchronized data are uniformly associated with the same customer object to avoid fragmented data segments for the same customer at different business stages. Access context information corresponding to the current request is collected synchronously, and an access context vector is constructed. Using the current request timestamp as the endpoint, within a preset fixed time window, field records related to the target business object are extracted from the customer master data table, lead table, opportunity table, contract table, work order table, after-sales table, channel collaboration table, and external synchronized cache table to form a source field candidate pool. To ensure that field records entering the current processing cycle are time-related to the current request, a time validity coefficient is calculated for each candidate field record. The larger the time validity coefficient, the higher the time correlation between the field record and the current request; the smaller the time validity coefficient, the further the field record deviates from the current processing cycle. The calculated time validity coefficient is compared with a preset time validity threshold W to filter out valid candidate records and summarize them to form a heterogeneous field candidate record set.

[0052] The time efficiency coefficient is calculated using the following expression:

[0053]

[0054] in, This indicates the time validity coefficient of the field record. This represents the timestamp in seconds for the current request. This indicates the timestamp in seconds of the last update time recorded in the field. This indicates the time offset of the field record relative to the current request. This represents the time decay coefficient.

[0055] It should be noted that the access context information includes at least the business scenario identifier, user role identifier, page entry identifier, terminal type, customer lifecycle stage, trigger event type, and request timestamp;

[0056] Valid candidate records were filtered out, specifically including: when When the corresponding field is marked as a valid candidate record, when... At that time, the corresponding field record is marked as a historical weakly correlated record and removed from the construction process of the candidate record set of heterogeneous fields in this round;

[0057] The range of values ​​can be set to Preferred selection The value is determined by using an exponential decay modeling method to ensure that the field records maintain distinguishable time sensitivity within a preset 30-minute business processing time window, and to ensure that the time validity coefficient decreases monotonically as the time difference increases and the calculation result is stable.

[0058] S1.2: After standardizing heterogeneous field records, bind and aggregate them with access context information to construct the original field event set.

[0059] Specifically, a unified standardization process is performed on each field record in the heterogeneous candidate record set to obtain a standardized field record set. Then, each standardized field record is bound to an access context vector, so that each field record carries both the field's own information and the business environment information when the field occurred. After completing the context binding, the field records are further encapsulated into events, including generating a unique event identifier for each field record based on the target business object identifier, business scenario identifier, field name, and field effective time, to obtain an event record set. The field event records are then aggregated according to the rule of "same target business object + same business scenario + same processing cycle" to form the original field event set.

[0060] It should be noted that the standardization process includes at least: field name normalization, unified expression of null values, original data type normalization, and field value format cleaning. Specifically, field name normalization eliminates differences in case sensitivity, separator differences, and naming redundancy; unified expression of null values ​​standardizes NULL, empty strings, and default invalid placeholder values ​​into standard null value markers; original data type normalization maps field types from different source systems to text, numeric, enumeration, time, boolean, and relational types; and field value format cleaning standardizes the format of common fields such as dates, amounts, and phone numbers.

[0061] The bound field record should include at least the original field name, the standardized field value, the field source system, the module where the field is located, the basic data type, the type of the business object to which it belongs, the effective timestamp, and the business scenario identifier, user role identifier, page entry identifier, terminal type, customer lifecycle stage, and trigger event type in the access context vector.

[0062] S1.3: Based on the original field event set, extract the field name text semantic features, sample value distribution features, upstream and downstream call relationship features, business stage features, historical co-occurrence field set features, and manual correction trajectory features for each heterogeneous field;

[0063] Specifically, the original field event set is read, and the field event records are merged according to the original field name, the field source system, and the business object type to which they belong. Multiple event records belonging to the same heterogeneous field are organized into a field sample sequence to be analyzed.

[0064] The original names of the corresponding fields in the sample sequence are read, and the field names are processed by word segmentation, normalization and stop modifier removal. Different naming forms such as "customer contact phone", "contact person phone", "mobile phone number" and "customer mobile phone number" are uniformly converted into the normalized field name "contact phone" according to the preset field alias dictionary and naming specifications. Then, based on the preset field name word vector encoding model, the normalized field name is mapped to the field name text semantic feature vector.

[0065] Statistical analysis is performed on all field values ​​in the sample sequence to extract the mean length, standard deviation of length, proportion of empty values, proportion of unique values, proportion of numeric characters, date format matching rate, and enumeration repetition rate of field values. The stability coefficient of field values ​​is calculated, and the above statistics and the stability coefficient of field values ​​are combined to form the sample value distribution feature vector, which is used to characterize the value pattern and data type stability of the field.

[0066] Based on the original field event records, which contain information such as the module where the field is located, the page entry identifier, the trigger event type, and the field event source marker, the field's call position in the current business chain is reconstructed. Specifically, if the field event originates from a page input action, the field is recorded in the input node; if the field event originates from a submit and save action, the field is recorded in the save node; if the field event originates from a process state switch or process approval action, the field is recorded in the flow node; and if the field event originates from a statistical display or report output action, the field is recorded in the display node. Furthermore, the frequency of field occurrence in various nodes is statistically analyzed to construct a feature vector of upstream and downstream call relationships.

[0067] Based on the business scenario identifier and customer lifecycle stage identifier carried in the original field event records, the frequency of occurrence of each heterogeneous field in each stage of customer acquisition, follow-up, contract signing, delivery, service and renewal is counted, and a feature vector of the business stage is formed.

[0068] When constructing the original field event set by steps, the co-occurrence window is "same target business object + same business scenario + same processing cycle". The co-occurrence correlation degree of each heterogeneous field with other fields in the same event window is counted. Based on the co-occurrence correlation degree results, the top k fields with the highest co-occurrence strength with the current field are selected, and the identifiers of these fields and their corresponding co-occurrence correlation degree values ​​are used to form the historical co-occurrence field set feature of the i-th heterogeneous field.

[0069] Read the historical field normalization correction log, count the number of times the i-th heterogeneous field has been manually corrected to the slots of each target standard field in history, and calculate the correction consistency coefficient; combine the historical target standard slots, correction counts, most recent correction time, and correction consistency coefficient to form the manual correction trajectory feature.

[0070] The expression for calculating the stability coefficient of a field value is:

[0071]

[0072] In the formula, This represents the stability coefficient of the field value of the i-th heterogeneous field. This represents the standard deviation of the character length of the sample values ​​in this field. This represents the average character length of the sample values ​​in this field. This represents a smoothing constant to prevent the denominator from being zero;

[0073] Co-occurrence correlation, expressed as:

[0074]

[0075] In the formula, This indicates the co-occurrence correlation between field i and field j. This indicates the number of times field i and field j co-occur in the same co-occurrence window. and These represent the total number of times field i and field j appear in the co-occurrence window, respectively.

[0076] The corrected consistency coefficient is calculated using the following expression:

[0077]

[0078] In the formula, Let represent the corrected consistency coefficient for the i-th heterogeneous field. This indicates the number of times the field has been corrected to the target standard field slot that has been corrected the most times. This indicates the total number of times the field has been manually corrected.

[0079] It should be noted that the merge key is "same field name + same source system + same object type" instead of merging by field name alone. The reason is that in the AI ​​intelligent CRM system, even if the field names are the same in different source systems, the semantics may be different due to different object types. For example, the business meaning of the "status" field in the customer master is not the same as that in the work order module. Therefore, it is necessary to include both the source system and the object type in the merge basis.

[0080] The higher the stability coefficient of a field value, the more stable the field value structure is, which is more helpful in determining whether the field belongs to the number, enumeration, date, or free text category.

[0081] The correction log should at least record the original field identifier, the standard field slot before correction, the standard field slot after correction, the correction time, and the correction operation identifier.

[0082] S1.4: Normalize and uniformly encode the extracted features to construct a field semantic fingerprint set that represents the comprehensive semantic attributes of heterogeneous fields.

[0083] Specifically, after obtaining the semantic features of field name text, sample value distribution features, upstream and downstream call relationship features, business stage features, historical co-occurrence field set features, and manually corrected trajectory features, the above six types of features are uniformly normalized. The six types of features corresponding to the same heterogeneous field are concatenated in a fixed order to form a comprehensive feature vector. A uniform encoding transformation is performed on each comprehensive feature vector to generate the corresponding field semantic fingerprint. After the field semantic fingerprints of all heterogeneous fields have been generated, the field semantic fingerprints corresponding to each field are summarized according to the field number or field identifier to form a field semantic fingerprint set.

[0084] It should be noted that the normalization process includes scaling numerical features to a uniform numerical range, converting categorical features into index codes, and aligning vector features according to a preset dimension, thereby forming a normalized feature representation. The field semantic fingerprint is used to represent the comprehensive semantic attributes of the heterogeneous field in a unified semantic space. It not only reflects the semantic information of the field name itself, but also retains information such as field value patterns, business call locations, business stage distributions, stable co-occurrence relationships, and manual correction paths.

[0085] S2. Utilize the field semantic fingerprint set to perform similarity retrieval in the pre-established standard field semantic library to determine the standard semantic slots corresponding to each heterogeneous field and generate a unified standard field set;

[0086] S2.1: Based on the field semantic fingerprint set, perform similarity retrieval in the pre-established standard field semantic library, and determine the standard semantic slots corresponding to each heterogeneous field according to the consistency of business stage, historical manual correction, and master data field priority;

[0087] Specifically, a standard field semantic library is pre-established, storing multiple standard semantic slots, each corresponding to a standard field template. After the standard field semantic library is established, template encoding is performed on the standard field templates corresponding to each standard semantic slot to generate standard semantic slot template fingerprints. Each field semantic fingerprint in the field semantic fingerprint set is used as a search key and compared with the standard semantic slot template fingerprints to obtain the similarity between the i-th heterogeneous field and the j-th standard semantic slot. Based on the similarity results, the top N standard semantic slots with the highest similarity ranking are selected as the candidate standard semantic slot set for the heterogeneous field. The first round of screening is performed by comparing the heterogeneous field's business stage features with the applicable business stages in each candidate standard semantic slot template. After completing the business stage consistency screening, if multiple candidate standard semantic slots still exist, a second round of screening is performed by manually correcting the trajectory features. After screening for consistency at each stage and historical manual corrections, if multiple candidate standard semantic slots still exist, the final unique attribution determination is further completed according to the priority of the master data fields. After the unique attribution determination is completed, it is determined whether the highest matching similarity corresponding to the current heterogeneous field reaches the preset matching threshold. When the highest matching similarity is higher than or equal to the preset threshold, and a unique candidate standard semantic slot has been obtained after the aforementioned screening for consistency at each business stage, screening for historical manual corrections, and determination of the priority of the master data fields, it is confirmed that the heterogeneous field has successfully been assigned to the unique candidate standard semantic slot. If the highest matching similarity is lower than the matching threshold, it means that there is not a sufficiently reliable standard semantic slot in the existing standard field semantic library to match the heterogeneous field. The heterogeneous field is then marked as a candidate field and will not participate in the generation of the unified standard field set in this round. At the same time, the field semantic fingerprint and source information of the field are retained for subsequent manual confirmation or expansion of the standard field semantic library.

[0088] Furthermore, generating the standard semantic slot template fingerprint involves constructing a template feature vector of the same dimension as the field semantic fingerprint based on the standard field name, standard data type, applicable business stage, typical value pattern, stable co-occurrence field group and historical normalization case in the standard field template, and generating the template fingerprint of the j-th standard semantic slot through a unified encoding function.

[0089] The first round of screening involves reading the distribution of the current heterogeneous field across various business stages, including customer acquisition, follow-up, contract signing, delivery, service, and renewal. It also involves reading the pre-configured distribution of applicable business stages in each candidate standard semantic slot template. A vector similarity comparison method is used to calculate the degree of consistency between the current heterogeneous field and each candidate standard semantic slot across business stages. Candidate standard semantic slots with the highest degree of consistency across business stages are prioritized for retention. Other candidate standard semantic slots with a significantly lower degree of consistency across business stages than the optimal candidate slot are directly eliminated.

[0090] The second round of screening includes reading the historical manual correction records corresponding to the current heterogeneous field, counting the number of times the heterogeneous field was manually corrected to each candidate standard semantic slot, and using the historical correction hit ratio comparison method to determine which candidate standard semantic slot the heterogeneous field more stably belongs to during the historical manual intervention process, and prioritizing the retention of the candidate standard semantic slot with the highest historical correction hit ratio.

[0091] The final unique attribution determination includes pre-establishing a field source priority table according to the data system to which the field belongs, where the priority of fields in the customer master data system and the basic master file system is higher than that of extended fields used only in local business modules; when multiple candidate standard semantic slots can be matched, the source level of each candidate standard semantic slot is compared according to the field source priority table, and standard semantic slots belonging to the customer master data system or the basic master file system are retained first, while edge field slots used only in local business modules are placed in later positions.

[0092] Similarity between the i-th heterogeneous field and the j-th standard semantic slot The expression is:

[0093]

[0094] In the formula, It is the semantic fingerprint vector of the i-th heterogeneous field. The template fingerprint vector representing the j-th standard semantic slot;

[0095] It should be noted that each standard field template includes at least the standard field code, standard field name, standard data type, applicable business stage, typical value pattern, stable co-occurrence field group, and historical normalization case. The standard field code uniquely identifies the standard field; the standard field name characterizes the canonical semantics of the standard field; the standard data type defines the basic data type of the standard field; the applicable business stage identifies the business stage in which the standard field typically appears; the typical value pattern describes the common value structure of the standard field in historical data; the stable co-occurrence field group describes the high-frequency co-occurrence relationship between the standard field and other standard fields; and the historical normalization case records heterogeneous field samples that historically belonged to the slot of this standard field.

[0096] The matching threshold is preferably set to 0.75 to 0.90, with 0.80 as an example. After constructing a field attribution verification sample set based on historical manual normalization results, a tiered threshold test method is used to compare and analyze the automatic attribution accuracy, misattribution rate, and proportion of fields to be confirmed under different similarity thresholds to determine the threshold. The final value is determined by the threshold range that ensures the automatic attribution accuracy meets the preset requirements, controls the misattribution rate within an acceptable range, and avoids an excessive number of fields to be confirmed.

[0097] S2.2: Extract standard field codes and field values ​​from heterogeneous fields that have completed the standard semantic slot assignment, generate standard field objects containing field source priority, normalized confidence and applicable scope markers, and summarize them to form a unified standard field set.

[0098] Specifically, the standard field templates in the standard semantic slots corresponding to heterogeneous fields are read, and the standard field codes are extracted. Combined with the field event records retained in the original field event set, the current value of the corresponding heterogeneous field is extracted as the standard field value. A source priority flag is written to the standard field object based on whether the field originates from the customer's main data source, business process source, external synchronization source, or cache source. A higher source priority indicates that the field value should be retained more preferentially when there are multiple source conflicts for the same standard field in the future. Furthermore, a normalized confidence score is generated for each successfully assigned heterogeneous field. An applicable scope flag is written for each successfully assigned heterogeneous field based on the applicable business stage, applicable business scenario, and applicable object type in the standard field template. After obtaining the standard field code, standard field value, field source priority, normalized confidence score, and applicable scope flag, each successfully assigned heterogeneous field is converted into a standard field object with a unified structure. All standard field objects are then aggregated to form a unified standard field set.

[0099] It should be noted that the scope of application flag is used to characterize in which business tasks, page scenarios and lifecycle stages the standard field object can be called, so that it can be directly referenced when the task scope convergence is performed on the unified standard field set in the next step.

[0100] The normalized confidence score is obtained by reading the similarity retrieval results between the current heterogeneous field and each candidate standard semantic slot (calculated by cosine similarity) and taking the highest matching degree of the corresponding final assigned standard semantic slot as the normalized confidence score of the heterogeneous field.

[0101] Each standard field object includes at least: standard field code, standard field value, field source priority, normalized confidence level, and scope flag.

[0102] S3. Based on the current business scenario, page task objectives and customer lifecycle stage, the unified standard field set is narrowed down to generate a task candidate field set. Permission constraints and data conflict resolution are performed on the task candidate field set to generate an effective field status set.

[0103] S3.1: Read the current business scenario identifier, page task objective, and customer lifecycle stage, and establish the field filtering benchmark corresponding to this round of page processing based on the applicable scope marker in the unified standard field object;

[0104] Specifically, the system reads the business scenario identifier corresponding to the current request to determine the business processing category of the current page; it then reads the task objective of the current page to determine the specific processing task that the current page request intends to complete; it continues to read the lifecycle stage corresponding to the current customer; and it reads the scope of application markers of each standard field object in the unified standard field set, and maps the scope of application markers to the current business scenario identifier, page task objective, and customer lifecycle stage to establish the field filtering benchmark for this round of page processing.

[0105] It should be noted that the business scenario identifier is used to distinguish which business scenario the current page belongs to: lead creation, customer follow-up, opportunity promotion, contract signing, performance delivery, after-sales processing, work order flow, or renewal management; the page task objective should include at least one of the following: information viewing, information supplementation, status flow, business approval, result submission, and service processing; the customer lifecycle stage is used to characterize which stage the current customer is in: customer acquisition, follow-up, contract signing, delivery, service, or renewal; the scope of application marker should include at least three types of information: applicable business scenario, applicable page task, and applicable lifecycle stage.

[0106] S3.2: Based on the field filtering benchmark, perform applicable scope filtering and task direct relevance convergence on the unified standard field set, retain the standard fields required for this round of page processing, and generate a task candidate field set.

[0107] Specifically, based on the scope of application markers in each standard field object, the scope of application is filtered for the unified standard field set to form a preliminary converged field subset; the task-direct relevance convergence is then performed on the standard field subset based on the current page task objective to obtain the task-directly related standard field set; and a task candidate field set is generated according to the principle of minimum processing scope.

[0108] It should be noted that the scope of application screening includes determining whether the applicable business scenario, applicable page task, and applicable lifecycle stage of each standard field object are consistent with the field screening benchmark; if they are consistent, the standard field object is retained and enters the next round of processing; if they are inconsistent, the standard field object is excluded from the scope of this round of page processing.

[0109] The convergence of direct relevance of tasks includes determining whether each standard field object belongs to the fields that the current page task must participate in processing: if the current page task goal is information supplementation, then the standard fields used for entry, completion and confirmation are retained first; if the current page task goal is status transition, then the standard fields involved in status update, node switching and result recording are retained first; if the current page task goal is business approval, then the standard fields directly related to approval conditions, approval opinions and approval results are retained first.

[0110] The principle of minimum processing scope means that only standard fields that need to participate in subsequent status determination, permission constraints, field validation, or linkage configuration in the current round of page processing are retained, while standard fields that are related to the current business context but will not be actually called in the current round of page processing are not retained.

[0111] S3.3: Read the current user role, organization position, data access security rules and page function permissions, and determine the visibility, editability, exportability and linkage trigger eligibility of each field in the task candidate field set accordingly;

[0112] Specifically, the system reads the user role information and organizational position information corresponding to the current request; it continues to read the data access security rules and page function permissions corresponding to the current request; it determines the visibility status of each field in the task candidate field set; it continues to determine the editability status of the visible fields; after determining the editability status of the fields, it continues to determine the exportability status of each field; and it further determines whether each field is allowed to be used as a trigger source for linkage.

[0113] It should be noted that user role information is used to identify which role the current operator belongs to among sales, customer service, operations, business development supervisor, department head, or system administrator; organizational position information is used to identify the current operator's department level, job responsibilities, and data ownership scope; data access security rules are used to limit the boundaries of viewing, editing, and exporting different fields and data domains for different user roles and positions; page function permissions are used to limit whether the current page is allowed to perform page-level operations such as viewing, editing, exporting, submitting, approving, or triggering linkage.

[0114] The visibility determination specifically includes reading the pre-configured field security level markers in the standard field template. These markers are pre-written based on the data domain and business sensitivity of the field when the standard field semantic library is established. Then, combined with the current user's viewing permissions, the data access scope corresponding to the organizational position, and the viewing qualifications in the page function permissions, it is determined whether the field is allowed to be displayed on the current page. If it is allowed to be displayed, the field is marked as a visible field; if it is not allowed to be displayed, it is marked as an invisible field.

[0115] The business sensitivity level is pre-defined based on the confidentiality requirements, leakage risks, and impact on customer privacy, transaction security, or business management results of the information carried by the field in the existing CRM business. For example, it is divided into three levels: ordinary, restricted, and sensitive. The value is based on the common practices in the existing data classification and management, which can achieve stable and executable access control based on the field's business attributes.

[0116] The editability determination specifically includes, based on the field being visible, considering the current user's editing permissions, the scope of responsibilities of the organizational position, the editing qualifications in the page function permissions, and the modification restrictions of the field's current business stage, determining whether the field can be modified by the current user; if modification is allowed, the field is marked as an editable field; if only viewing is allowed and modification is not allowed, it is marked as a read-only field.

[0117] The exportability determination specifically includes determining whether the current field is allowed to be taken out of the system through data export, report download, or list output based on the pre-configured field security level flag in the standard field template, the export restrictions in the data access security rules, and the export eligibility in the page function permissions. If allowed, the field is marked as an exportable field; if not allowed, it is marked as a non-exportable field.

[0118] The determination of the linkage trigger source specifically includes reading the pre-configured field linkage attribute markers in the standard field template. These markers indicate whether the field belongs to the linkage category that allows triggering subsequent field state adjustments. Then, combining page function permissions, field business attributes, and whether the field currently meets both visibility and editability conditions, it is determined whether the field possesses the basic qualifications to be a linkage trigger source. Based on this, a valid change determination is performed on the field value change. That is, the field is considered to have undergone a linkage-triggering change only when the current value has actually changed relative to the previous valid value. Specifically, for enumeration, Boolean, and status fields, the difference between the previous and subsequent values ​​is used as the basis for the change; for numeric fields, the difference between the previous and subsequent values ​​reaches a preset change threshold; for text fields, the change is based on whether the standardized text content has changed. Finally, only when the field simultaneously meets the conditions of visibility, editability, page-allowed linkage, and a valid value change, and is marked as a linkage category field that allows triggering subsequent field state adjustments, is it marked as a permitted linkage trigger source; otherwise, it is not marked as a linkage trigger source.

[0119] S3.4: Identify multi-source value conflicts for the same standard field in the task candidate field set, and determine the unique effective value of the field in the order of priority of main data source, most recent update time, and manual confirmation result, and generate an effective field status set containing field code, effective value, display status, editing status, linkage trigger status, and verification participation mark.

[0120] Specifically, after determining the permission status, the candidate field objects for the task are further grouped according to the standard field code to identify whether there are multiple source values ​​for the same standard field and whether the field values ​​are inconsistent. For the identified multi-source value conflict fields, the first round of conflict resolution is performed according to the main data source priority rule. After the main data source priority processing is completed, if there are still multiple candidate field values ​​of the same level, the second round of conflict resolution is performed according to the most recent update time priority rule. If there are still multiple candidate field values, the final conflict resolution is completed according to the manual confirmation result priority rule. After obtaining the unique effective value of the field, a verification participation mark is generated for each field. Each task candidate field is organized into an effective field status record. After all task candidate fields have completed the status organization, all effective field status records are summarized to form an effective field status set.

[0121] It should be noted that grouping candidate field objects for tasks refers to comparing field objects with the same standard field code. If field objects in the same group are found to come from different source systems and their field values ​​are inconsistent, it is determined that the standard field has multiple source value conflicts.

[0122] The first round of conflict resolution includes: if among multiple conflicting field values ​​there are field values ​​that come from the customer's master data source, the basic master file source, or those written after unified master data governance, then the field value that comes from the customer's master data source, the basic master file source, or those written after unified master data governance will be retained as the candidate effective value, and other field values ​​that come from the business process table, the external synchronization table, or the cache table will be temporarily placed in the secondary candidate position.

[0123] The second round of conflict resolution refers to reading the last update timestamp corresponding to each candidate field value and prioritizing the field value with the most recent update time as the new candidate effective value;

[0124] The final conflict resolution process includes reading the manual confirmation record corresponding to the standard field. If a field value has been explicitly confirmed by business personnel in the past, the manual confirmation field value will be used as the final effective value.

[0125] The specific steps for generating validation participation markers include combining the field's visibility status, editability status, and the field's role in the current page task to determine whether the field needs to participate in subsequent field validity validation. If the field needs to participate in mandatory field validation, format validation, or range validation in subsequent steps, it is marked as a field participating in validation; otherwise, it is marked as a field not participating in validation.

[0126] Each effective field status record includes at least: field code, effective value, display status, edit status, linkage trigger status, and verification participation flag. Among them, the field code is used to uniquely identify the current field, the effective value is used to represent the unique valid value of the field in this round of page processing, the display status is used to represent whether the field is displayed on the page, the edit status is used to represent whether the field is allowed to be modified, the linkage trigger status is used to represent whether the field is allowed to be used as a linkage trigger source, and the verification participation flag is used to represent whether the field enters the subsequent verification chain.

[0127] S4. Construct a field dependency graph based on the effective field status set to obtain a field relationship network. Perform page configuration orchestration according to the field relationship network to generate the first round of page configuration scheme. Complete component mapping and initial business interface rendering according to the first round of page configuration scheme.

[0128] S4.1: Construct field nodes based on each field in the effective field status set, and identify the display dependencies, validation dependencies, candidate value dependencies, and business logic dependencies between fields to generate a field dependency graph;

[0129] Specifically, the system reads each effective field status record from the effective field status set and generates a field node for each field corresponding to that record, thus obtaining a set of field nodes. It then identifies the display dependencies, validation dependencies, candidate value dependencies, and business logic dependencies between fields. After generating the field nodes and identifying the display dependencies, validation dependencies, candidate value dependencies, and business logic dependencies, the system summarizes the set of field nodes and the sets of various dependencies to construct a field dependency graph.

[0130] It should be noted that each field node inherits at least the field code, effective value, display status, editing status, linkage trigger status, and verification participation flag from the corresponding effective field status record, which are used to characterize the initial state of the field in the current page processing cycle;

[0131] Identifying display dependencies involves reading the page rule configuration, the display condition description in the standard field template, and the current effective value of the field to determine whether the display status of a certain field depends on the current value of another field. If so, a display dependency edge is established between the corresponding two field nodes. The display dependency relationship is used to characterize whether the display of a certain field depends on the current value of another field. For example, when the "Customer Type" is "Enterprise Customer", the "Unified Social Credit Code" field is displayed, while when the "Customer Type" is "Individual Customer", the field is hidden.

[0132] Identifying validation dependencies involves determining whether the required conditions, value range, or validity validation rules of a field will change as the current value of another field changes. If they will change, a validation dependency edge is established between the corresponding field nodes. Validation dependencies are used to characterize whether the validation logic of a field is affected by another field. For example, when "Document Type" is "ID Card", the "Document Number" field performs ID card format validation; when "Document Type" is "Passport", passport format validation is performed.

[0133] Identifying candidate value dependencies involves determining whether the candidate set, enumeration set, or recommended value set of a certain field needs to be dynamically updated based on the current value of another field. If so, candidate value dependency edges are established between the corresponding field nodes. The candidate value dependency relationship is used to characterize whether the range of possible values ​​for a certain field is affected by another field. For example, when "Industry" is "Manufacturing", the "Sub-industry" field loads the candidate set for Manufacturing; when "Industry" is "Education", the candidate set for Education is loaded.

[0134] Identifying business logic dependencies involves determining whether a field will trigger the addition of a new field, hiding of a field, or switching of a field state when a specific business state changes, based on business state transition rules, page task rules, and business logic descriptions in standard field templates. If it will trigger, a business logic dependency edge is established between the corresponding field nodes. Here, business logic dependencies are used to characterize the state linkage relationship between fields driven by business rules. For example, when the "contract status" changes to "signed", fields such as "signing time" and "performance start time" are triggered to enter the processing state or switch to the editable state.

[0135] S4.2: Write the propagation direction, triggering condition, priority and termination condition for each dependency edge in the field dependency graph to generate the field relationship network;

[0136] Specifically, after constructing the initial field dependency graph, a propagation direction is written for each dependency edge; the propagation direction is used to represent which field node the dependency relationship is propagated from to which field node, that is, to clarify the "triggering field" and the "affected field";

[0137] For example, in scenarios where "Customer Type" affects the display status of the "Unified Social Credit Code for Enterprises" field, the propagation direction shifts from the "Customer Type" field to the "Unified Social Credit Code for Enterprises" field.

[0138] Write trigger conditions for each dependency edge; where trigger conditions are used to characterize the field value change or field state change under which dependency propagation is activated, such as when a field value is equal to a preset value, belongs to a preset value set, or the state is switched to a specific business state, the corresponding dependency edge is triggered to take effect;

[0139] Continue to write a priority for each dependency edge; where the priority is used to characterize the execution order of each dependency edge when multiple dependency edges meet the triggering conditions at the same time; usually, business logic dependency edges are given higher priority, followed by display dependency edges and validation dependency edges, and then candidate value dependency edges, or sorted according to preset business rules.

[0140] Furthermore, a termination condition is written for each dependency edge; the termination condition is used to characterize the conditions under which dependency propagation stops, such as when the state of the affected field has reached the target state, the candidate value set has been updated, or the subsequent field no longer meets the conditions for continued propagation.

[0141] S4.3: Based on the field relationship network and the effective field status set, combined with the current page task objective, field usage frequency and business completion path, perform page configuration orchestration on the effective fields to generate the first-round page configuration scheme including field arrangement order, display area, folding level and preset linkage markup.

[0142] Specifically, the system reads the current page task objective, historical field usage frequency, and business completion path under the corresponding business scenario to obtain a set of page configuration orchestration parameters. Combining field usage frequency, business completion path, and the strength of the field's relationship in the field relationship network, it determines the order of each active field on the page. This includes prioritizing fields that are frequently used, located at the beginning of the business completion path, and directly support the current page task; and prioritizing fields that are used less frequently, located in the middle or later stages of the business completion path, or mainly serve an auxiliary explanatory role. The system also considers the field's business attributes and its position in the business completion path. The function is to determine the display area of ​​each field; further, based on the field's usage frequency, business necessity, and relevance to the current task, determine the field's folding level; write pre-set linkage markers for the corresponding fields according to the candidate value dependency edges and business logic dependency edges in the field relationship network. The pre-set linkage markers are used to indicate that once the value of the corresponding trigger field changes after the initial rendering of the page, the corresponding dependency edge in the field relationship network needs to be called to perform subsequent linkage updates; after completing the field arrangement order, display area, folding level, and writing of pre-set linkage markers, the relevant results are summarized to generate the first round of page configuration scheme.

[0143] It should be noted that the page task objective is used to characterize which type of task the current page intends to perform, such as information viewing, information supplementation, status transition, business approval, result submission, or service processing; the field usage frequency is used to characterize the historical access frequency and editing frequency of each field in the same type of page; and the business completion path is used to characterize the field processing order that is usually followed when completing the page task in the current business scenario.

[0144] Determining the display area for each field includes dividing the field into different areas such as basic information area, business processing area, approval result area, extended information area, or auxiliary information area; among them, fields that play a key supporting role in the current page task are prioritized to enter the basic information area and business processing area, while auxiliary fields are entered into the extended information area or auxiliary information area;

[0145] Determining the collapsibility level of fields includes identifying fields that must be directly displayed on the first screen as first-screen fields, fields that need to participate in the current round of business processing but can be viewed later as extended fields, and fields that only need to be viewed or operated on under special conditions as collapsed fields.

[0146] The initial page configuration scheme should include at least the field arrangement order, display area, collapse level, and preset linkage markers; if necessary, it may also include field grouping identifiers and sorting sequence numbers within the area.

[0147] S4.4: Based on the standard data type of the corresponding field in the unified standard field set, and combined with the display and editing status of the corresponding field in the effective field status set and the preset linkage marker in the first round of page configuration scheme, map each field to the corresponding interface component;

[0148] Specifically, cross-step associations are established according to field codes. Standard data types for each field are read from a unified standard field set, and display and editing states for each field are read from an effective field status set. Pre-set linkage markers for each field are read from the initial page configuration scheme, generating a component mapping parameter set. The component category for each field is determined based on its standard data type: text fields are mapped to text components, enumeration fields to selection components, date fields to date components, associated object fields to search selection components, and read-only statistical fields to summary display components. The specific presentation form of the component is determined by combining the display and editing states of each field. Pre-set linkage markers from the initial page configuration scheme are written into the corresponding component instances. If a field has a pre-set linkage marker in the initial page configuration scheme, the linkage trigger attribute for that field is registered in the corresponding component instance to indicate whether the component needs to respond to dependency propagation in the field relationship network in subsequent runtime states. If a field does not have a pre-set linkage marker, a regular component instance is generated. The component instances corresponding to each field are summarized to form a complete set of mapped field components.

[0149] It should be noted that the standard data type is used to determine which basic interface component a field should be mapped to; the display state is used to determine whether the field enters the initial interface rendering of this round; the edit state is used to determine whether the field is presented in input state or display state after mapping; and the preset linkage flag is used to determine whether the component reserves a linkage trigger interface after rendering.

[0150] Specifically, for fields that are not visible, no visible component instance is generated; for text fields that are visible and editable, they are mapped to a text input component; for text fields that are visible and read-only, they are mapped to a text display component; for enumerated fields that are visible and editable, they are mapped to an optional selection component; and for statistical fields that are visible and read-only, they are mapped to a summary display component.

[0151] S4.5: Based on the display area, field arrangement order, and collapse level determined by the first round of page configuration scheme, organize the mapped interface components into regions and arrange them into layers, and complete the initial business interface rendering.

[0152] Specifically, the system reads the display area, field order, and collapse level of each field in the initial page configuration scheme to generate a set of region loading parameters; it organizes the mapped field components into regions based on the display area corresponding to each field component; it loads components within each region container according to the field order, including loading corresponding field components from front to back within the same region according to the field order parameters, so that high-priority fields are located at the front of the region and low-priority fields are located at the back of the region; it performs hierarchical loading of components within the region according to the collapse level; and it calls the interface rendering engine to load and render each page region instance to generate the initial business interface for the current business scenario.

[0153] The initial business interface includes at least a core information area, a business operation area, an extended information area, and a historical trajectory area, as well as field components loaded in each area in a predetermined order and hierarchy.

[0154] It should be noted that the display area is used to indicate which area among the core information area, business operation area, extended information area, or historical trajectory area the field component should be loaded into; the field arrangement order is used to indicate the front-to-back arrangement relationship of the field components in the same area; the collapse hierarchy is used to indicate whether the field component belongs to the first screen direct presentation layer, the extended presentation layer, or the collapsed presentation layer.

[0155] The regional organization involves loading field components into corresponding regional containers for the core information area, business operation area, extended information area, and historical trajectory area. The core information area is used to carry the core information components of the current page's first screen, the business operation area is used to carry the operation-type components directly related to the current page task, the extended information area is used to carry low-frequency extended components, and the historical trajectory area is used to carry read-only historical information or process record components.

[0156] Hierarchical loading includes loading the first screen directly rendered field components into the area's direct display layer, loading extended field components into the extended display layer, and loading collapsed field components into the collapsed display layer; during the initial page rendering, only the first screen directly rendered layer and necessary extended entry points are directly displayed, without actively expanding the collapsed display layer.

[0157] S5. Bind the initial business interface to the field dependency graph to form a runtime configuration context;

[0158] S5.1: Bind each component instance in the initial business interface to the corresponding field node in the field relationship network, and establish a field status cache table to record the latest display status, editing status and value status of each field;

[0159] Specifically, the system reads each component instance from the initial business interface and matches and binds it one by one with the corresponding field nodes in the field relationship network based on the field codes carried in the component instances; and establishes a field status cache table.

[0160] It should be noted that each component instance is bound to at least one field node, enabling the component instance to inherit the node identifier and associated dependency edge information of that field node in the field relationship network;

[0161] The field state cache table is used to record the latest display state, editing state, and value state of each field in the current page. The display state and editing state can be directly applied through field state record inheritance, while the value state is initialized by the current display value or input value of the corresponding component instance when the initial business interface is rendered.

[0162] S5.2: Register event listeners for interactive components to capture field input, selection, deletion, toggle, and submit actions, and establish a local update queue to store the set of field nodes to be propagated, forming a runtime configuration context.

[0163] Specifically, interactive component instances are selected from the initial business interface; after establishing the field state cache table, event listeners are registered for the interactive components in the initial business interface; writing rules are established for field change events to the field nodes to be propagated; after completing the event listener registration, a partial update queue is further established; after completing the establishment of the component-field node binding mapping table, field state cache table, field event listeners, and partial update queue, the above structures are combined in a unified manner to form a runtime configuration context.

[0164] It should be noted that interactive component instances include at least text input components, selection components, date components, search selection components, and other components that allow users to actively modify field values; read-only display components and summary display components are not considered actively interactive components.

[0165] Event listeners are used to capture user actions such as inputting, selecting, deleting, toggling, and submitting fields. Input actions correspond to changes in the value of text input components, selection actions correspond to changes in the value of enumerated selection components or search selection components, deletion actions correspond to clearing field values, toggling actions correspond to switching values ​​of switch-type or state-type components, and submission actions correspond to the confirmation and submission of field values ​​on the current page or in the current area.

[0166] The writing rules include: when a component instance is detected to have a value change by an event listener, the field node corresponding to the component instance is determined through the component-field node binding mapping table, and the field node is written to the local update queue as a node to be propagated; if the same field node already exists in the local update queue in the current propagation cycle, it will not be written again.

[0167] The partial update queue is used to store a set of field nodes to be propagated by field change events.

[0168] S6. In the runtime configuration context, when a field change event is detected, local propagation is performed in the field dependency graph starting from the triggering field, and incremental reconfiguration and local re-rendering are performed only on the fields in the affected subgraph.

[0169] S6.1: In the runtime configuration context, when the event listener detects a field change event, it writes the corresponding triggering field node into the local update queue and determines the affected subgraph in the field relationship network starting from the triggering field node;

[0170] Specifically, a set of field event listeners continuously monitors the input, selection, deletion, switching, and submission actions of interactive components. When a field value changes in a component instance, the event listener captures the change event and determines the triggering field node corresponding to the component instance through the component-field node binding mapping table. After determining the triggering field node, it is written to the local update queue. If the triggering field node does not already exist in the local update queue during the current propagation cycle, it is directly enqueued; if it already exists, it is not written again to avoid the same triggering field triggering propagation repeatedly in the same propagation cycle.

[0171] Starting from the triggering field node, the search proceeds along the dependent edges that are directly or indirectly connected to the node and meet the triggering conditions in the field relationship network to determine the set of affected field nodes and form the affected subgraph accordingly. During the search, the search continues only along dependent edges that meet the current field value change condition and have not reached the termination condition. Dependency edges that do not meet the triggering conditions are not propagated.

[0172] S6.2: Based on the affected subgraph, perform local propagation according to the priority of dependent edges, recalculate the display state, candidate value set, validation rules, required state and component display parameters only for the affected fields, write back the field state cache table and perform local re-rendering to complete the incremental reconfiguration of fields.

[0173] Specifically, after identifying the affected subgraph, local propagation is performed on the propagation path within the affected subgraph based on the priority of the affected dependency edges. During local propagation, only the field states and component parameters in the affected subgraph are recalculated, and corresponding reconfigurations are performed according to the dependency types reached during propagation. The latest display state, editing state, value state, candidate value set, validation rules, and required state of the affected fields are written back to the field state cache table. After the field state cache table is written back, based on the incremental reconfiguration results of the affected fields, only the interface components or interface areas corresponding to the affected fields are refreshed. Specifically, if only the display state, candidate value set, or validation prompt of a single field component changes, only that field component is refreshed; if multiple affected fields are located in the same area and the hierarchical structure within that area changes, only the corresponding interface area is refreshed, without redrawing other unaffected areas; after the local re-rendering is completed, the trigger field nodes that have been processed in this round and the nodes to be propagated on their propagation chain are removed from the local update queue; if there are still new nodes to be propagated in the local update queue, the next round of local propagation is executed according to the above steps; if the local update queue is empty, the current round of local update ends.

[0174] It should be noted that local propagation includes the priority of propagation when multiple dependent edges meet the triggering conditions at the same time. The dependent edge with higher priority updates its target field node state first, and the dependent edge with lower priority executes later, thereby avoiding state conflicts when multiple propagation paths act on the same field node at the same time.

[0175] Reconfiguration includes recalculating the display state of the target field for display-dependent edges; recalculating the candidate value set of the target field for candidate value-dependent edges; recalculating the validation rules and required fields for the target field for validation-dependent edges; recalculating whether to add, hide, or switch component display parameters for the target field for business logic-dependent edges; and keeping the display state, candidate value set, validation rules, required fields, and component display parameters unchanged for fields not in the affected subgraph.

[0176] In summary, this invention unifies heterogeneous fields scattered across customer master data, business process data, channel collaboration data, and externally synchronized data into a computable, searchable, and propagable field processing system. This not only improves the consistency, scalability, and adaptability of field configurations across multiple business scenarios but also enables page configurations to dynamically respond to changes in business scenarios, page tasks, and fields. Furthermore, this invention achieves field-level local propagation and local re-rendering through field dependency graphs and runtime configuration contexts, maintaining high page response efficiency and good interaction stability even in complex business scenarios. Therefore, it comprehensively enhances the intelligence level and engineering application value of AI-powered intelligent CRM systems in multi-scenario field organization, dynamic configuration, interface rendering, and runtime linkage control.

[0177] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for dynamically configuring and rendering heterogeneous data fields in AI-powered intelligent CRM that supports multiple business scenarios, characterized in that... include: Collect heterogeneous data field information and access context information from multiple business scenarios, construct an original field event set, extract multidimensional features of each heterogeneous field based on the original field event set, and construct a field semantic fingerprint set; By using the field semantic fingerprint set to perform similarity retrieval in a pre-established standard field semantic library, the standard semantic slots corresponding to each heterogeneous field are determined, and a unified standard field set is generated. Based on the current business scenario, page task objectives, and customer lifecycle stage, the unified standard field set is narrowed down to generate a task candidate field set. Permission constraints and data conflict resolution are performed on the task candidate field set to generate an effective field status set. A field dependency graph is constructed based on the effective field status set to obtain a field relationship network. Page configuration orchestration is performed based on the field relationship network to generate the first-round page configuration scheme. Component mapping and initial business interface rendering are completed based on the first-round page configuration scheme. The initial business interface is bound to the field dependency graph to form a runtime configuration context; In the runtime configuration context, when a field change event is detected, local propagation is performed in the field dependency graph starting from the triggering field, and incremental reconfiguration and local re-rendering are performed only on the fields in the affected subgraph.

2. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 1, characterized in that: The steps for constructing the original field event set are as follows: Collect the access context information corresponding to the current request, and extract heterogeneous field records related to the target business object from multiple business data sources; after standardizing the heterogeneous field records, bind and aggregate them with the access context information to construct the original field event set.

3. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 2, characterized in that: The steps for extracting multidimensional features from each heterogeneous field based on the original field event set and constructing a field semantic fingerprint set are as follows: Based on the original field event set, extract the semantic features of field name text, sample value distribution features, upstream and downstream call relationship features, business stage features, historical co-occurrence field set features, and manual correction trajectory features for each heterogeneous field; normalize and uniformly encode the extracted features to construct a field semantic fingerprint set that represents the comprehensive semantic attributes of heterogeneous fields.

4. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 3, characterized in that: The steps for generating a unified standard field set are as follows: Based on the field semantic fingerprint set, similarity retrieval is performed in the pre-established standard field semantic library, and the standard semantic slots corresponding to each heterogeneous field are determined according to the consistency of business stage, historical manual correction, and master data field priority. Extract standard field codes and field values ​​from heterogeneous fields that have completed the standard semantic slot assignment, generate standard field objects containing field source priority, normalized confidence and applicable scope markers, and summarize them to form a unified standard field set.

5. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 4, characterized in that: The steps for generating the candidate field set for the task are as follows: Read the current business scenario identifier, page task objective, and customer lifecycle stage, and establish the field filtering benchmark corresponding to this round of page processing based on the applicable scope marker in the unified standard field object; Based on the field filtering benchmark, the applicable scope filtering and task direct relevance convergence are performed on the unified standard field set, retaining the standard fields required for this round of page processing, and generating a task candidate field set.

6. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 5, characterized in that: The steps for generating the effective field status set are as follows: Read the current user role, organizational position, data access security rules and page function permissions, and determine the visibility, editability, exportability and linkage trigger eligibility of each field in the task candidate field set accordingly; The system identifies multiple source value conflicts for the same standard field in the candidate field set of the task, and determines the unique effective value of the field in the order of priority of the main data source, the most recent update time, and the manual confirmation result. It generates an effective field status set that includes field code, effective value, display status, editing status, linkage trigger status, and verification participation flag.

7. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 6, characterized in that: The process of constructing a field dependency graph based on the effective field state set to obtain a field relationship network, and then performing page configuration orchestration based on the field relationship network to generate the first round of page configuration scheme, is as follows: Field nodes are constructed based on each field in the effective field status set, and the display dependencies, validation dependencies, candidate value dependencies, and business logic dependencies between fields are identified to generate a field dependency graph. Write the propagation direction, triggering condition, priority, and termination condition for each dependency edge in the field dependency graph to generate a field relationship network; Based on the field relationship network and the effective field status set, combined with the current page task objectives, field usage frequency and business completion path, the page configuration orchestration is performed on the effective fields to generate the first-round page configuration scheme, which includes the field arrangement order, display area, collapse level and preset linkage markup.

8. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 7, characterized in that: The steps for completing component mapping and initial business interface rendering based on the initial page configuration scheme are as follows: Based on the standard data type of the corresponding field in the unified standard field set, and combined with the display and editing status of the corresponding field in the effective field status set and the preset linkage markers in the first round of page configuration scheme, each field is mapped to the corresponding interface component; Based on the display area, field arrangement order, and collapse hierarchy determined in the first round of page configuration scheme, the mapped interface components are organized into regions and arranged into hierarchical levels, and the initial business interface is rendered.

9. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 8, characterized in that: The steps for binding the initial business interface with the field dependency graph to form a runtime configuration context are as follows: Bind each component instance in the initial business interface to the corresponding field node in the field relationship network, and establish a field status cache table to record the latest display status, editing status and value status of each field; Register event listeners for interactive components to capture field input, selection, deletion, toggle, and submit actions, and establish a local update queue to store the set of field nodes to be propagated, forming a runtime configuration context.

10. The method for dynamic configuration and rendering of heterogeneous data fields in AI-powered intelligent CRM supporting multiple business scenarios as described in claim 9, characterized in that: When a field change event is detected, local propagation is performed in the field dependency graph starting from the triggering field. Incremental reconfiguration and local re-rendering are only performed on the fields in the affected subgraphs. The steps are as follows: In the runtime configuration context, when the event listener detects a field change event, it writes the corresponding triggering field node into the local update queue and determines the affected subgraph in the field relationship network starting from the triggering field node. Based on the affected subgraph, local propagation is performed according to the priority of dependent edges. Only the display state, candidate value set, validation rules, required state and component display parameters of the affected fields are recalculated. The field state cache table is written back and local re-rendering is performed to complete the incremental reconfiguration of the fields.

Images