A channel merchant automated settlement processing method and system

Abstract

The present invention provides a channel merchant automated settlement processing method and system, relating to the field of data processing technology. The method comprises: obtaining standard sales data; obtaining rebate rule parameters, configuring corresponding weight values ​​for each field, and performing field structure adjustment and weight binding on the rebate rule parameters; obtaining, based on each sales record in the standard sales data, rule items whose matching relationship with the current sales record is not empty as an evaluable set, and performing field matching score processing on each rule item; screening the target rule item with the highest score, extracting its rebate coefficient, and multiplying the rebate coefficient by the sales quantity of the corresponding sales record to obtain a rebate amount; performing normalization calculation based on the distribution range of the maximum score value and other score values ​​to generate a confidence score; and generating, based on all sales records, a settlement item set with rebate amounts and confidence scores. The present invention improves the accuracy of channel merchant automated settlement processing.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method and system for automated settlement processing for channel merchants. Background Art

[0002] In existing technology, settlement processing for distributors typically relies on a company's internal financial system or ERP system. The system aggregates basic data such as distributor sales orders, delivery notes, and receipt records, and generates the payable amount according to preset settlement cycles and calculation rules. During this process, companies typically set fixed rebate percentages or rebate conditions, using automated scripts to categorize and aggregate data. For example, the system applies calculation templates based on factors such as the distributor's tier and product type, outputting settlement details and summary reports. After the final results are confirmed by financial personnel, a settlement document is generated and payment is executed.

[0003] During quarterly promotions for home appliances, distributors often need to upload their terminal sales data so that companies can calculate rebates based on actual sales volume. Promotional policies often include cross-rules for product models and time periods, such as "8% rebate for Model A sold between January and March 2025, 5% for the remainder of the year," while existing systems often only support single rule matching. When the data uploaded by distributors includes sales records for multiple models and different time periods, the system cannot accurately distinguish the applicable rules, resulting in miscalculation of certain rebate amounts. For example, the same product sold in January and April might receive a uniform 5% rebate rate, causing distributors to question the accuracy of the settlement and requiring manual financial intervention to correct the situation, reducing settlement efficiency and data credibility. Summary of the Invention

[0004] The purpose of the present invention is to provide a channel merchant automated settlement processing method and system, aiming to solve the problems mentioned in the background technology.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0006] In a first aspect, a method for automated settlement processing for a channel merchant is provided, the method comprising:

[0007] Obtain the original sales data, perform field detection and time sorting on the original sales data to obtain standard sales data;

[0008] Obtain rebate rule parameters, configure corresponding weight values ​​for each field based on the field information in the rebate rule parameters, and adjust the field structure and bind the weights of the rebate rule parameters to obtain a weighted rule dataset. The rebate rule parameters include the rebate rule number, applicable time range, applicable product model, applicable channel information, and rebate coefficient.

[0009] According to each sales record in the standard sales data, obtain the rule items with a non-empty matching relationship with the current sales record from the weighted rule dataset as an evaluable set, and perform field matching score processing on each rule item to obtain a rule adaptation score set;

[0010] Based on the rule adaptation score set, the target rule item with the highest score is selected, the rebate coefficient in the target rule item is extracted, and the rebate coefficient is multiplied by the sales quantity of the corresponding sales record to obtain the rebate amount corresponding to the sales record;

[0011] According to the rule adaptation score set, the maximum score value and the distribution range of other score values ​​are normalized and calculated to generate the corresponding confidence score;

[0012] Based on all sales records, a set of settlement items with rebate amounts and confidence scores is generated, and the set of settlement items is merged to obtain complete settlement detail data.

[0013] Preferably, according to the field information in the rebate rule parameters, a corresponding weight value is configured for each field, including:

[0014] Extract the applicable time range, applicable product model, and applicable channel information fields from the rebate rule parameters to form a field information set;

[0015] Calculate the field importance index based on the field stability, hit frequency, and distribution fluctuation of each field in the historical sales records of the field information set;

[0016] According to the field importance index and the distribution characteristics of the field in different product types or sales channels, set the weight adjustment factor corresponding to each field;

[0017] Multiply the field importance index and the corresponding weight adjustment factor to obtain the field score value set;

[0018] The field score value set is normalized to generate a field weight value set, including the time field weight value, the model field weight value, and the channel field weight value.

[0019] Preferably, the rebate rule parameters are subjected to field structure adjustment and weight binding to obtain a weight rule data set, including:

[0020] Based on the field weight value set, each field weight value is matched one by one with the corresponding field in the rebate rule parameters, and the matching results are marked and enhanced to generate weighted field records;

[0021] Combine and bind the weighted field record with its corresponding rebate rule number and rebate coefficient to construct a unified rule structure entry;

[0022] All rule structure entries are constructed into rule key-value combinations in the preset field order for index identification to generate a weighted rule dataset.

[0023] Preferably, according to each sales record in the standard sales data, rule items whose matching relationship with the current sales record is not empty are obtained from the weighted rule data set as an evaluable set, and field matching score processing is performed on each rule item to obtain a rule adaptation score set, including:

[0024] Extract the time field, model field, and channel field of each sales record in the standard sales data to form the current sales field set;

[0025] Traverse and filter all rule items in the weighted rule data set, identify rule items whose field values ​​have a non-empty intersection or fuzzy similarity relationship with the current sales field set, and generate an evaluable rule set;

[0026] For each rule item in the evaluable rule set, calculate the time coincidence score of the current sales field set with its time field, the exact match score with the model field, and the classification similarity score with the channel field;

[0027] Multiply the time overlap score, exact match score, and classification similarity score by the time field weight, model field weight, and channel field weight, respectively, and add up the product results to obtain the weighted score of the sales record and the rule item;

[0028] The weighted scores calculated for all rule items are combined into a set to generate a rule adaptation score set corresponding to the sales record.

[0029] Preferably, the field importance index is calculated based on the field stability, hit frequency, and distribution fluctuation of each field in the field information set in the historical sales records, including:

[0030] Based on historical sales record data, we perform time-dimension segmentation and statistics on each field in the field information set, analyze the magnitude of field value fluctuations within consecutive cycles, and obtain the stability score of the field.

[0031] Based on historical sales record data, calculate the frequency of each field value in the sales record and obtain the hit frequency score;

[0032] Based on historical sales record data, the standard deviation or information entropy of the distribution of each field value is calculated to obtain the volatility score of the field;

[0033] The stability score, hit frequency score, and volatility score are weighted and summed according to a preset ratio to generate the field importance index of the corresponding field.

[0034] Preferably, based on the field importance index and the distribution characteristics of the field in different product types or sales channels, a weight adjustment factor corresponding to each field is set, including:

[0035] Based on historical sales record data, extract the value set of each field in the field information set under different product types, sales regions or channel types to form field-business dimension mapping relationship data;

[0036] Based on the field-business dimension mapping relationship data, the value distribution density, concentration, and distinctiveness of each field in different business dimensions are counted to obtain a set of field distribution characteristic indicators;

[0037] Based on each indicator in the field distribution characteristic indicator set and combined with the field importance indicator, a preference factor evaluation calculation is performed to output the weight adjustment factor of each field.

[0038] Preferably, for each rule item in the evaluable rule set, the time coincidence score of the current sales field set and its time field, the exact match score with the model field, and the classification similarity score with the channel field are calculated respectively, including:

[0039] Calculate the interval intersection of the time field in the current sales field set and the rule item time interval, divide the number of intersection days by the total number of days in the rule time interval to obtain the time overlap score;

[0040] Perform a character-level comparison of the product model field in the sales field set with the model field in the rule item. If they are completely consistent, a full score is assigned. If they are partially matched, the score is reduced based on the similarity to obtain an exact match score.

[0041] Perform category mapping or label similarity calculation on the channel field in the sales field set and the rule item channel field, match the similarity value with the scoring criteria of the channel field, and output the classification similarity score.

[0042] In a second aspect, a channel merchant automated settlement processing system is provided, the system comprising:

[0043] The sales data acquisition module is used to obtain the original sales data, perform field detection and time sorting on the original sales data, and obtain standard sales data;

[0044] The rebate rule management module is used to obtain rebate rule parameters, configure corresponding weight values ​​for each field according to the field information in the rebate rule parameters, adjust the field structure and bind the weights of the rebate rule parameters, and generate a weight rule data set;

[0045] The rule matching and scoring module is used to obtain, from each sales record in the standard sales data, rule items that have a non-empty matching relationship with the current sales record from the weighted rule dataset as an evaluable set, and perform field matching scoring on each rule item to generate a rule adaptation score set;

[0046] The rebate calculation module is used to filter the target rule item with the highest score based on the rule adaptation score set, extract the rebate coefficient in the target rule item, and multiply the rebate coefficient by the sales quantity of the corresponding sales record to generate the rebate amount;

[0047] The confidence calculation module is used to perform normalization calculation based on the maximum score value and the distribution range of other score values ​​in the rule adaptation score set to generate the corresponding confidence score;

[0048] The settlement generation module is used to generate a set of settlement items with rebate amounts and confidence scores based on all sales records, and merge the settlement item sets to generate complete settlement detail data.

[0049] The above solution of the present invention includes at least the following beneficial effects:

[0050] By building a multi-field weighted matching mechanism, the problem of inaccurate adaptation caused by the rebate rules in the existing technology only supporting single condition matching can be solved. The system builds a multi-dimensional rule adaptation path based on the combination of time field, model field and channel field, and introduces field matching scoring and field weighting mechanism to make the matching process of rebate rules more refined and logically transparent. Unlike the existing system that can only choose the "hit is match" processing logic, the present invention can realize the pros and cons scoring between different rule items, so as to select the rule item with the best score for rebate calculation, significantly improving the rationality and accuracy of matching.

[0051] Furthermore, for the "model-time crossover" rebate strategy commonly used in home appliance promotions, this invention quantifies the matching strength between sales records and rules through a field overlap scoring and confidence generation mechanism. This not only automatically identifies the optimal rebate rule item corresponding to the sales record, but also uses confidence scoring to determine the credibility of the matching result. This scoring model avoids the hidden risk of mismatched rebate amounts. When low confidence scores occur, they can be flagged in advance or a manual review mechanism can be triggered, fundamentally reducing settlement errors and subsequent correction costs.

[0052] In addition, combined with the weight configuration mechanism and the field importance analysis model, the present invention can dynamically generate field weights based on historical sales behavior data, so that the rebate decision-making process not only relies on rule settings, but also has data-driven capabilities. Especially when faced with complex rebate policies for different channels and different product types, the system can automatically adjust the field influence, realize "same rules, multiple perspectives" intelligent judgment, and effectively deal with the rebate calculation complexity problems brought about by flexible and changeable promotion strategies. Overall, the automated settlement processing solution provided by the present invention can reduce manual intervention, improve the intelligence level of rebate accounting and the maintainability of the system, and is suitable for efficient settlement needs in large-scale channel merchant rebate scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 The present invention provides a flowchart of a method for processing automated settlement for channel merchants. DETAILED DESCRIPTION

[0054] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0055] like Figure 1 As shown, an embodiment of the present invention provides a channel merchant automated settlement processing method, the method comprising:

[0056] S100, obtaining original sales data, and performing field detection and time sorting on the original sales data to obtain standard sales data;

[0057] S200: Obtain rebate rule parameters, configure corresponding weight values ​​for each field based on field information in the rebate rule parameters, adjust the field structure of the rebate rule parameters and bind the weights to obtain a weighted rule dataset. The rebate rule parameters include a rebate rule number, applicable time range, applicable product model, applicable channel information, and a rebate coefficient.

[0058] S300: Based on each sales record in the standard sales data, obtain rule items whose matching relationship with the current sales record is not empty from the weighted rule data set as an evaluable set, and perform field matching score processing on each rule item to obtain a rule adaptation score set;

[0059] S400: Filter the target rule item with the highest score based on the rule adaptation score set, extract the rebate coefficient in the target rule item, and multiply the rebate coefficient by the sales quantity of the corresponding sales record to obtain the rebate amount corresponding to the sales record;

[0060] S500: performing normalization calculation on the maximum score value and the distribution range of other score values ​​in the rule adaptation score set to generate a corresponding confidence score;

[0061] S600: Generate a settlement item set with rebate amounts and confidence scores based on all sales records, and merge the settlement item set to obtain complete settlement detail data.

[0062] In an embodiment of the present invention, by constructing a complete set of channel merchant automated settlement processing methods, it is possible to achieve automatic matching and rebate calculation between large-scale sales data and rebate rules, significantly reduce the intensity of manual review, and improve the accuracy and traceability of rebate accounting. First, by obtaining the original sales data and performing field detection and time sorting operations, all sales records are ensured to have integrity and time series consistency, avoiding rule matching deviations caused by missing or disordered data. Based on the pre-processed standard sales data, the system introduces rebate rule parameters as the basis for settlement, including rebate rule number, applicable time range, applicable product model, applicable channel information and rebate coefficient. By structurally standardizing such parameters and configuring field weights, a structured and evaluation-usable weight rule data set is formed, providing a data basis for the subsequent scoring mechanism.

[0063] During the actual calculation process, for each sales record, the system selects all rule items with matching field values ​​from the weighted rule dataset as the evaluable set, avoiding a full data set traversal and improving matching efficiency. Based on this, the system performs a weighted summation operation based on the matching between the time, model, and channel fields, combining the weights of each field. This outputs a set of rule adaptation scores between the sales record and each rule item. This score set is not only used to select the target rule items with the highest scores for final settlement, but also normalizes the scores to generate a confidence score, which measures the reliability of the matching results. This confidence score provides an audit reference for the subsequent settlement process. If the confidence level is low, it can guide the system or manual retroactive correction of the results. Finally, the system combines the rebate amounts corresponding to all sales records with the confidence scores to generate a set of settlement items. This is then merged to output standardized settlement details, providing a clear and structured data output for subsequent payment, financial reconciliation, and system integration.

[0064] Among them, the original sales data is obtained, and the original sales data is subjected to field detection and time sorting processing, specifically including: collecting original sales data uploaded by channel dealers or from the company's internal sales system, and the original sales data includes at least multiple fields such as sales time, product model, channel code, sales quantity, etc. Taking into account the diverse sources and different formats of the original data, the system needs to perform preliminary verification processing on the data integrity and field consistency. The field detection process is mainly used to identify records with missing fields, empty field values, or abnormal data formats, and to remove, repair, or mark them. After the detection is passed, the system sorts all records in ascending or descending order according to the sales time field to ensure that the time sequence can be accurately referenced in the subsequent rule matching process and to avoid cross-matching errors. After completing the above operations, the sorted data is saved as standard sales data for subsequent rule adaptation and rebate calculation.

[0065] Among them, obtaining the rebate rule parameters specifically includes: extracting the set rebate rule content from the rebate policy management system, and the rebate rule parameters include fields such as rule number, applicable time range, applicable product model, applicable channel information, and rebate coefficient. These parameters are usually formulated in advance by the marketing or finance department and configured according to the promotion cycle, product category or channel level. During the extraction process, the system will perform a rule validity check to ensure that expired or ineffective rules are not mistakenly used in the current calculation process. In the specific implementation, the system can also establish an index according to the rule number to facilitate subsequent rapid retrieval in the sales record and rule matching link. This step ensures that the logical basic data for the rebate calculation is timely and complete, and can support subsequent field comparison and rebate amount calculation operations.

[0066] Based on the rule adaptation score set, the system selects the target rule item with the highest score, extracts the rebate coefficient within the target rule item, and multiplies the rebate coefficient by the sales quantity of the corresponding sales record to calculate the rebate amount corresponding to the sales record. Specifically, the system first sorts the rule adaptation score set generated during the scoring process for each sales record and selects the rule item with the highest score as the optimal adaptation rule for that record. The rebate coefficient within the target rule item is the rebate ratio coefficient to be applied to that sales record under the current rule structure. The system then multiplies the rebate coefficient by the sales quantity recorded in the sales record to calculate the rebate amount corresponding to that sales record. For example, if the rebate coefficient for the target rule item is 7% and the sales quantity is 150 units, the system calculates the rebate amount as 150 multiplied by 7%, which is the rebate amount for that record. This step ensures that the rebate amount is calculated based on the optimal rule option determined by the scoring mechanism while maintaining consistency and logical closure between the numerical result and the original sales data.

[0067] The system generates a confidence score based on the maximum score in the rule adaptation score set and the distribution range of other scores. Specifically, after obtaining all the rule adaptation scores for a sales record, the system uses the highest score in the set as the core reference value. It also calculates the minimum score in the set and the overall range. The system then calculates the confidence level of this maximum score based on its position in the overall score distribution and outputs a confidence score. This confidence score measures the plausibility of the match between the target rule item and the sales record. For example, if a sales record shows a moderate level of match under multiple rule items and the scores are similar, the system will generate a low confidence score, indicating a low degree of certainty in the subsequent rebate calculation process. However, if a score significantly exceeds that of other items, the confidence score will be close to full, indicating a high degree of confidence in the rule adaptation result. The confidence score can be used for flagging, grading, or manual intervention in subsequent settlement review.

[0068] Based on all sales records, a set of settlement items with rebate amounts and confidence scores is generated. These items are then merged to produce complete settlement details. Specifically, the system binds the rebate amount and confidence score corresponding to each sales record to generate a settlement item with structured fields. A settlement item may include fields such as the sales record number, rebate amount, matching rule number, confidence score, and settlement status. The system aggregates all settlement items, categorizing and merging them by channel ID, product, or cycle, and outputs the final settlement details. For example, multiple sales records generated by a channel within a quarter will be merged by sales time to produce a unified rebate detail for that channel. This detail can be directly accessed by the financial system for payment applications or exported as a reconciliation report, audit interface, or channel external settlement document. This approach not only improves the integrity and output efficiency of the rebate detail structure but also provides a unified settlement interface for multi-dimensional business integration.

[0069] In a preferred embodiment of the present invention, according to the field information in the rebate rule parameters, a corresponding weight value is configured for each field, including:

[0070] Extract the applicable time range, applicable product model, and applicable channel information fields from the rebate rule parameters to form a field information set;

[0071] Calculate the field importance index based on the field stability, hit frequency, and distribution fluctuation of each field in the historical sales records of the field information set;

[0072] According to the field importance index and the distribution characteristics of the field in different product types or sales channels, set the weight adjustment factor corresponding to each field;

[0073] Multiply the field importance index and the corresponding weight adjustment factor to obtain the field score value set;

[0074] The field score value set is normalized to generate a field weight value set, including the time field weight value, the model field weight value, and the channel field weight value.

[0075] In an embodiment of the present invention, in order to make the role of the fields of the rebate rules in the matching process have distinctiveness and weight differences, the flexibility and intelligence of the rule screening are effectively improved by introducing a field weight configuration mechanism. Specifically, the system extracts fields such as applicable time range, applicable product model and applicable channel information from the rebate rule parameters to form a unified field information set. By combining historical sales records, the performance characteristics of each field in the actual sales data are statistically analyzed to generate a field importance index. The analysis content includes: the frequency of value changes of the field in different time periods, which is used to evaluate the stability of the field; the frequency of the field being cited in all sales records, which is used to judge the hit rate; the distribution breadth and offset degree of the field value, which reflects its volatility. The above three indicators are quantified with numerical values, and preset proportions are set to perform weighted fusion to form a scoring result that reflects the overall importance of the field.

[0076] Furthermore, considering the differences in the importance of different fields under different product types or channel strategies, the system uses the value characteristics of the fields in each business dimension as the source of adjustment factors. By extracting the distribution characteristics of the fields in different product categories and channel scenarios, a weight adjustment factor related to the field matching strategy is generated, and it is multiplied by the field importance index. The product result represents the score value that the field should have in the current business environment, that is, the field score value set. In order to facilitate the subsequent scoring module to perform weighted calculations, the system normalizes all score values ​​and finally generates a field weight value set for actual rule matching calculations. This processing mechanism avoids the rigidity of fixed weight configuration, realizes the coupling of field importance and business context, and effectively enhances the accuracy of rebate calculations and the intelligence of rule adaptation.

[0077] In a preferred embodiment of the present invention, the rebate rule parameters are subjected to field structure adjustment and weight binding to obtain a weight rule data set, including:

[0078] Based on the field weight value set, each field weight value is matched one by one with the corresponding field in the rebate rule parameters, and the matching results are marked and enhanced to generate weighted field records;

[0079] Combine and bind the weighted field record with its corresponding rebate rule number and rebate coefficient to construct a unified rule structure entry;

[0080] All rule structure entries are constructed into rule key-value combinations in the preset field order for index identification to generate a weighted rule dataset.

[0081] In an embodiment of the present invention, in order to address the problems of complex structure and diverse data formats of rebate rule parameters, a method for field structure adjustment and weight binding is provided, so that the rule parameters have a unified data structure and field weight information, which is convenient for subsequent scoring and rule call processing. In this embodiment, the system first reads the generated field weight value set, and matches the weight values ​​in the set with the corresponding time field, model field and channel field in the rebate rule parameters one by one. After the matching is completed, the weight value is added to each field, and the tag enhancement processing is performed, so that the original field not only has the original semantics, but also has the importance weight information when participating in the scoring, thereby forming a weighted field record.

[0082] The system then combines the weighted field records with their corresponding rebate rule numbers and rebate coefficients to generate uniformly structured rule structure entries. Each entry simultaneously contains field values, field weights, and settlement elements, with complete semantics and scoring capabilities. After the entire rule item structure is completed, the system combines the field values ​​according to the set field order (e.g., time → model → channel) to generate key values ​​for subsequent rule indexing and invocation, thereby forming a weighted rule dataset. This dataset has a standardized structure that efficiently supports screening, scoring, and decision logic execution during the matching process, while also providing a unified data interface for system maintenance.

[0083] Based on the set of field weight values, each field weight is matched one-to-one with the corresponding field in the rebate rule parameters. The matching results are then marked and enhanced to generate weighted field records. Specifically, the system reads the weights of the time field, product model field, and channel field from the generated field weight value set and matches these weights to the corresponding fields in the rebate rule parameters. For example, if a rebate rule applies to the first quarter of 2025, Class A products, and the first-level distribution channel, the system will append the corresponding weights of the time field, model field, and channel field to the parameter field of this rule, making it no longer static text but a structured field with weight attributes.

[0084] During the tagging enhancement process, the system explicitly expresses the relationship between a field and its corresponding weight through structural identifiers or data tags, ensuring that the numerical contribution of the field can be correctly identified during subsequent scoring calculations. For example, field weights can be encoded into structure entries as field attributes, or a "field name-field value-field weight" triple can be defined through a key-value pair structure to support subsequent weighted matching processing. This enhancement ensures that the rebate rule field has the complete semantic expression required to participate in the scoring calculation, effectively eliminating operational interruptions or deviations caused by missing field weights in the scoring process.

[0085] The system combines weighted field records with their corresponding rebate rule numbers and rebate coefficients to create a unified rule structure entry. Specifically, after field tagging enhancement, the system combines the field values ​​and field weights within each rebate rule with the rebate rule number and rebate coefficient for that rule item to generate a complete and unified data entry. This rule structure entry contains both field matching information and rebate calculation parameters, serving as a key reference in subsequent sales record scoring and rebate amount calculation.

[0086] To ensure structural consistency and ease of indexing, the system arranges fields in a predefined order and encapsulates them using a unified data model or structured object. For example, a rule structure entry might be organized in the following order: "Rule number → Time field (including weight) → Model field (including weight) → Channel field (including weight) → Rebate coefficient." This is stored in the system as a standard object for easy access and comparison. This step not only ensures the consistency and integrity of rebate rule data but also helps improve the computational efficiency and traceability of rule matching and evaluation.

[0087] The system constructs a rule key-value combination based on a preset field order for all rule structure entries, which is used for index identification and generates a weighted rule dataset. Specifically, the system concatenates or combines the field values ​​of all generated unified rule structure entries according to the set field priority order to form a unique rule key-value identifier. This key-value identifier can be generated through string concatenation, hash calculation, or code mapping, and is used to quickly retrieve matching rules during the sales record scoring phase, avoiding repeated traversal of the entire rule dataset.

[0088] For example, if the field order is set to "time → model → channel," the system concatenates the time period identifier, product model code, and channel category identifier for a rule to form a rule key, such as "2025Q1_A001_D1." This key is then used as the primary index field for the rule structure entry. By mapping this key to the rule structure entry, the system constructs a weighted rule dataset that supports fast queries.

[0089] This rule dataset can be indexed by key value, supporting advanced matching operations such as fuzzy queries, prefix matching, and similarity calculations. In practical applications, this structure significantly improves retrieval efficiency during large-scale rule matching, and is particularly suitable for complex business scenarios with hundreds or even thousands of rebate rules, effectively reducing latency and resource consumption during the scoring process.

[0090] In a preferred embodiment of the present invention, based on each sales record in the standard sales data, rule items that have a non-empty matching relationship with the current sales record are obtained from the weighted rule data set as an evaluable set, and field matching score processing is performed on each rule item to obtain a rule adaptation score set, including:

[0091] Extract the time field, model field, and channel field of each sales record in the standard sales data to form the current sales field set;

[0092] Traverse and filter all rule items in the weighted rule data set, identify rule items whose field values ​​have a non-empty intersection or fuzzy similarity relationship with the current sales field set, and generate an evaluable rule set;

[0093] For each rule item in the evaluable rule set, calculate the time coincidence score of the current sales field set with its time field, the exact match score with the model field, and the classification similarity score with the channel field;

[0094] Multiply the time overlap score, exact match score, and classification similarity score by the time field weight, model field weight, and channel field weight, respectively, and add up the product results to obtain the weighted score of the sales record and the rule item;

[0095] The weighted scores calculated for all rule items are combined into a set to generate a rule adaptation score set corresponding to the sales record.

[0096] In this embodiment of the present invention, to improve the matching accuracy and scoring efficiency between sales records and rebate rules, a step-by-step field scoring process is constructed. This allows the system to calculate the strength of the association between sales records and rule items in a structured manner, providing quantitative support for rebate calculations. For each standard sales data record, the system first extracts the time, model, and channel fields contained within it to form the current sales field set. This field set serves as the input source for subsequent matching and scoring.

[0097] The system analyzes each rule item in the weighted rule dataset, comparing the time, model, and channel field values ​​for each rule to identify those that have a non-empty intersection, partial overlap, or fuzzy match with the current sales field set. Eligible rule items are identified as the evaluable rule set. This limited set effectively reduces computing resource consumption and prevents invalid rules from interfering with the scoring results.

[0098] During the scoring phase, the system performs a scoring operation on each rule item in the evaluable set. The time field is processed using the interval intersection method, and the time overlap score is output by calculating the proportion of overlapping days between the sales time and the rule application time range. The model field uses an exact matching strategy to compare the field string character by character. If there is a complete match, the full score is assigned, and if there is a partial match, the score is proportionally reduced based on the similarity. The channel field uses a preset channel label system or mapping rules to evaluate the category proximity between fields and output a classification similarity score. The above three scoring values ​​respectively represent the degree of proximity between the sales record and the rule item in three key dimensions.

[0099] During the score calculation phase, the system multiplies the temporal overlap score, exact match score, and classification similarity score by the field weights, reflecting the contribution of each field score to the overall score. All these multiplication results are summed to form a weighted score for the sales record and the current rule item. The weighted scores corresponding to all evaluable rule items are combined to form the rule adaptation score set for the sales record. This set not only reflects the quality of the match between multiple rules and sales data but also provides the raw scoring basis for subsequent rebate calculations and confidence level generation, enhancing the accuracy, transparency, and adjustability of the system during the rule adaptation process.

[0100] Among them, all rule items in the weighted rule data set are traversed and screened to identify rule items whose field values ​​have a non-empty intersection or fuzzy similarity relationship with the current sales field set, and generate an evaluable rule set. Specifically, when the system processes each standard sales record, it first extracts the time field, product model field, and channel field from the record to form the current sales field set. The system uses this field set as input and compares the field values ​​of all rule items in the weighted rule data set one by one. During the comparison process, the system first determines whether there is a clear intersection between the current sales field set and the rule item field value, that is, whether the field values ​​are the same or included in the applicable scope of the rule item, such as overlapping time periods or models belonging to the same category.

[0101] If the exact match criteria aren't met, the system further performs fuzzy matching. For example, if the prefix or suffix in the product model field is similar, or if the channel field belongs to the same channel branch in the category mapping table, the system will consider the fields to be similar. During this process, the system sets matching strategies based on the field type, such as interval overlap determination for time fields, string similarity comparison for model fields, and channel classification mapping rules for channel fields.

[0102] Any rule items identified in the above process as having at least one field with a valid intersection or a similarity exceeding a preset threshold are added to the set of evaluable rules for the current sales record. This set serves as input for subsequent scoring calculations and rebate selection, ensuring that the scoring mechanism focuses on relevant rule items, improving computational efficiency and reducing system resource waste caused by irrelevant rules participating in the scoring process.

[0103] In a preferred embodiment of the present invention, the field importance index is calculated based on the field stability, hit frequency, and distribution fluctuation of each field in the field information set in the historical sales records, including:

[0104] Based on historical sales record data, we perform time-dimension segmentation and statistics on each field in the field information set, analyze the magnitude of field value fluctuations within consecutive cycles, and obtain the stability score of the field.

[0105] Based on historical sales record data, calculate the frequency of each field value in the sales record and obtain the hit frequency score;

[0106] Based on historical sales record data, the standard deviation or information entropy of the distribution of each field value is calculated to obtain the volatility score of the field;

[0107] The stability score, hit frequency score, and volatility score are weighted and summed according to a preset ratio to generate the field importance index of the corresponding field.

[0108] To improve the scientific and discriminatory nature of field weights during rule matching, this embodiment of the present invention introduces a method for calculating field importance indicators based on historical sales records. This method analyzes the stability, hit frequency, and volatility of the time, model, and channel fields extracted from the rebate rule parameters, ensuring that field weights are fully supported by historical data.

[0109] During the data analysis phase, the system first segments historical sales records by time period, such as monthly or quarterly, and calculates the value changes of each field. For each field, the system calculates the magnitude of its value fluctuation over consecutive periods. Small fluctuations indicate high field value stability, and the system assigns a higher score. Stable fields generally more reliably reflect channel characteristics or product sales patterns, and are therefore more valuable in rebate determination.

[0110] During the frequency scoring process, the system counts the frequency of each field value appearing in the historical records and uses the ratio of that frequency to the total number of records as the scoring basis. A high frequency indicates that the field is common in sales behavior and is therefore given a higher score. The system also identifies field values ​​that appear in very few records and lowers their scores to reduce the impact of noise.

[0111] To assess volatility, the system uses standard deviation or information entropy algorithms to analyze the dispersion of field values ​​within the sample distribution. The more concentrated and regular the field values, the higher the score; the wider and more discrete the distribution, the lower the score. These three scores represent the stability, coverage, and predictability of the field in historical data, respectively.

[0112] The system weights the stability score, hit frequency score, and volatility score in a preset ratio (e.g., 3:2:1) to generate a field importance index for each field. This index is directly used in the subsequent weight adjustment factor setting process, laying the data foundation for the reasonable determination of the field's final weight value.

[0113] For each rule item in the evaluable rule set, the time overlap score between the current sales field set and its time field is calculated separately. Specifically, the system extracts the sales time field of the sales record and compares it with the applicable time range defined in the rule item. If the sales time is completely within the effective time period of the rule, the system considers it a complete time match and assigns a high score. If the sales time only partially overlaps with the rule time interval, the system sets a score level based on the proportion of the overlapping days in the total rule time period. If the sales time is completely earlier or later than the rule time period, the score is the lowest or no score is given.

[0114] For example, if a sales record date is April 5, 2025, and the rule's time range is April 1, 2025, to April 30, 2025, the system will determine that it fully covers the period and give it a full score. However, if the sales date is March 31, which is adjacent to but not overlapping with the rule's time range, no score will be given. This scoring ensures that the time field plays a fundamental role in rule adaptation, and is particularly suitable for scenarios where rebate policies are tightly coupled with promotional periods.

[0115] For each rule item in the evaluable rule set, the system calculates an exact match score between the current sales field set and its product model field. Specifically, the system first performs a character-by-character comparison of the model field in the sales record with the model field in the rule item. If the two match exactly, it's considered a perfect match and awarded a full score. If there are some character differences but structural similarities, such as belonging to the same product series or having the same prefix, the system assigns a lower score based on the predefined similarity rules.

[0116] In practice, the system can determine whether product models have a derivative relationship. For example, if "A123-B" and "A123-C" are classified as products in the same series, even if a perfect match isn't possible, an intermediate matching score can still be assigned based on the same series principle. This strategy is suitable for scenarios where product model naming is inconsistent or data reported by sales terminals is simplified, maintaining scoring accuracy while ensuring matching flexibility.

[0117] For each rule item in the evaluable rule set, a classification similarity score is calculated between the current sales field set and its channel field. Specifically, the system first extracts the channel field value from the sales record and maps it to a standard channel category label based on a predefined channel classification table. The system then compares this label with the label of the channel field in the rule item to determine the degree of similarity within the channel hierarchy tree or classification mapping structure.

[0118] If the sales channel and the regular channel have identical labels, for example, both are "direct-operated first-level distribution," the highest score is assigned. If there is a superior-subordinate relationship or they belong to the same general category (such as "direct-operated channel" and "direct-operated second-level distribution"), a higher score, but not the full score, is assigned based on the similarity level. If they do not belong to the same general category, a lower score is assigned or the channel is considered mismatched. This approach, through structured processing of channel labels, supports flexible adaptation within complex channel systems, effectively addressing the limitations of traditional systems in handling cross-level channel similarity.

[0119] The time overlap score, exact match score, and category similarity score are multiplied by the time field weight, model field weight, and channel field weight, respectively. These products are then accumulated to obtain the weighted score for the sales record and the rule item. Specifically, after obtaining the score for each field, the system calls the weight value of the corresponding field in the field weight value set and multiplies each score by the corresponding weight value. The product represents the contribution of the field to the overall score.

[0120] After calculation, the contribution values ​​of all fields are summed to form a final weighted score between the current sales record and a specific rule item. This score reflects the overall compatibility between the current rule item and the sales record and serves as the core decision-making basis for subsequent selection of the optimal rebate rule item. This step, by introducing field weight control, not only reflects the varying importance of different fields to the scoring model but also makes the scoring mechanism adjustable and interpretable, preventing the final result from being skewed by scores from different dimensions.

[0121] In a preferred embodiment of the present invention, based on the field importance index and the distribution characteristics of the field in different product types or sales channels, a weight adjustment factor corresponding to each field is set, including:

[0122] Based on historical sales record data, extract the value set of each field in the field information set under different product types, sales regions or channel types to form field-business dimension mapping relationship data;

[0123] Based on the field-business dimension mapping relationship data, the value distribution density, concentration, and distinctiveness of each field in different business dimensions are counted to obtain a set of field distribution characteristic indicators;

[0124] Based on each indicator in the field distribution characteristic indicator set and combined with the field importance indicator, a preference factor evaluation calculation is performed to output the weight adjustment factor of each field.

[0125] In this embodiment of the present invention, to further enhance the adaptability of field weights to actual business strategies, the system introduces a mechanism for setting weight adjustment factors based on field importance indicators. This mechanism comprehensively considers the distribution characteristics of fields across different business dimensions, dynamically adjusts field scores, and enhances the scenario-wide generalization capabilities of the scoring model.

[0126] The system first extracts all values ​​for each field from historical sales records across different product types, sales regions, and channel types, and then analyzes these values ​​based on business dimensions. For example, the product model field may be more influential in high-end product lines, but less so in low-priced promotional items. By identifying the frequency of occurrence, distribution concentration, and ability to distinguish between categories of field values ​​across different dimensions, the system generates distribution characteristic parameters that describe the differences in the field's performance across various business contexts.

[0127] The system then performs a reconciliation analysis on the aforementioned distribution characteristic parameters, combining them with the field importance index. For example, if a field has a high historical importance score and a high distribution density in the current product line, the system assigns it a higher reconciliation factor. If a field has a high score but sparse or mixed values ​​in a particular business dimension, the system lowers its score impact to avoid matching bias.

[0128] The system weights or multiplies the adjustment factor with the field importance index to generate a weight adjustment factor for each field. This factor is then applied to the importance index to obtain a corrected score. These corrected values ​​form the field score set, which is then normalized to generate the final weight. This mechanism ensures that the system's scoring strategy is business-aware, inheriting its data-driven nature while also embodying business flexibility, thereby improving the practicality and accuracy of the rebate rule adaptation model.

[0129] Based on historical sales records, the system extracts the value sets for each field in the field information set for different product types, sales regions, or channel types to form field-to-business dimension mapping relationship data. Specifically, the system first cleans and clusters the historical sales records to extract specific value records for the time field, product model field, and channel field under different business attributes (such as product category, region code, or channel level). The system summarizes the frequency, value types, and relationships of these fields across different business dimensions to form a mapping between fields and business dimensions.

[0130] In the specific implementation, the system takes the product model field as an example, and records the specific values ​​of the field in different product lines such as high-end, basic, and entry-level, and counts its coverage ratio in each category. Similarly, the channel field will be mapped to multiple categories such as direct sales, distribution, and platforms. The above data is not used directly for scoring, but as the basic data structure for downstream analysis of field distribution characteristics, it is used to reveal the usage characteristics of the field under different business conditions. This mapping relationship not only helps the system understand in which business dimensions a field is active, but also provides data support for field scoring.

[0131] Among them, based on the field-business dimension mapping relationship data, the value distribution density, concentration and distinctiveness of each field in different business dimensions are counted to obtain a set of field distribution characteristic indicators. Specifically, the system performs three types of statistical analysis on each field from the field-dimension mapping data constructed in the first step.

[0132] First, distribution density reflects the prevalence of field values ​​across business dimensions. The system quantifies the field's coverage by calculating the frequency of occurrence of the field in each dimension and comparing it to the total number of sales records in that dimension. For example, if a field value is present in multiple product types, the density score will be high.

[0133] Secondly, concentration is used to assess whether field values ​​are concentrated within a few specific dimensions. The system analyzes the frequency distribution of field values ​​across all dimensions and determines the proportion of values ​​concentrated in a few dimensions to assign a concentration score. A higher concentration indicates a strong business focus on the field.

[0134] Finally, discriminability reflects the ability of a field value to discriminate business dimensions. The system determines whether a field helps distinguish different business scenarios by analyzing the field's distribution range across dimensions—that is, the difference in the field's proportion between the most and least frequently occurring dimensions. For example, if a channel field appears frequently in direct sales channels but rarely in third-party platforms, it will have a high discriminability score. These distribution characteristic scores will be used as parameters to subsequently generate weight adjustment factors.

[0135] Based on each indicator in the field distribution characteristic indicator set and the field's importance index, a preference factor evaluation calculation is performed to output a weight adjustment factor for each field. Specifically, the system takes the distribution density, concentration, and discriminability scores obtained for each field in the previous step, combines them with the importance index values ​​obtained for that field in historical sales records, and calculates the weight adjustment factor using a comprehensive scoring model. This calculation model can adopt a linear weighting structure, a segmented evaluation strategy, or a fusion function set according to empirical rules to reflect the actual importance of the field in the current business context.

[0136] During the comprehensive analysis, the system sets a corresponding adjustment weight for each distribution feature to control its contribution to the overall score. For example, in some businesses, the system will increase the weight of the distinguishing score if it places greater emphasis on the field's ability to distinguish; if it focuses more on whether the field is widely applicable to multiple dimensions, the influence of the distribution density score will be increased. Ultimately, the system combines the adjusted three score results with the field's importance index to generate a weight adjustment factor for the field. This factor will further act on the field score value in the weight generation module to achieve dynamic adjustment for different business characteristics.

[0137] For example, in one implementation scenario, the model field's importance score is high, but its distribution density across different product types is too low. The system then generates an adjustment factor slightly lower than the original score, thereby reducing the field's excessive influence on the scoring model. This strategy seamlessly integrates historical data-driven and business strategy-guided approaches, making field weight configuration highly flexible and adaptable to business needs.

[0138] In a preferred embodiment of the present invention, the calculation formula of the weight adjustment factor is:

[0139] ;

[0140] in:

[0141] For fields The weight adjustment factor of

[0142] For fields The importance index is derived from the weighted score mentioned above;

[0143] Score the density of the distribution;

[0144] Score the distribution concentration;

[0145] Scoring for discrimination;

[0146] 、 、 are adjustment coefficients corresponding to the three feature scores, which can be set by the system or learned through training;

[0147] For fields The number of values ​​in all business dimensions comes from the field and business mapping data;

[0148] For fields In the business dimension The number of occurrences of the following is obtained by counting;

[0149] For business dimension The total number of sales records under is used for normalization to solve the dimensionality problem;

[0150] It is the total constant of all business dimensions, the number of dimensions;

[0151] For fields The total number of occurrences in all business dimensions, used to calculate the relative frequency of the field;

[0152] is the entropy normalized denominator to ensure that the concentration score is between [0,1];

[0153] and For fields The maximum and minimum proportions in each dimension are used to distinguish the differences in field distribution;

[0154] The log base not shown in the above calculation formula is e, which means the natural logarithm is used by default.

[0155] In an embodiment of the present invention, the system first extracts the values ​​of each field in different product types, sales regions or channel types based on historical sales records, and forms a mapping relationship between the field and the business dimension. On this basis, the distribution density, distribution concentration and distinguishing characteristics of the field in each dimension are calculated respectively. The three types of features are quantified into scoring values ​​through normalization. The distribution density measures the extent of the field in each dimension; the concentration reflects whether the field value is concentrated in a few dimensions; and the distinguishingness measures the performance difference of the field between different dimensions. These statistical features are integrated through formulas and embedded in the field weight adjustment process, so that the field score is no longer determined by static rules, but dynamically fluctuates based on data performance.

[0156] By combining field importance indicators with performance in business dimensions, a multi-level, multi-factor weight generation mechanism is constructed. This not only reflects the statistical significance of the data but also allows for flexible adjustment of behavioral strategies. During the rebate rule matching process, when the importance of a field increases within a channel or product dimension, the adjustment factor is increased accordingly, thereby increasing the weight of the field in the matching score. Conversely, if the field is too sparsely distributed or noisy, its weight is suppressed, reducing its interference in rule screening. This dynamic adjustment strategy can effectively reduce matching error rates in complex rebate policies and improve the robustness and adaptability of the scoring model.

[0157] The entire weight adjustment factor generation process supports parameter training and business policy intervention, providing an algorithmic interface for future intelligent system upgrades and leaving room for iterative optimization of rebate policies. This approach demonstrates significant scalability and industrial application prospects. Compared to traditional methods that statically configure field weights based on empirical values, this scoring formula possesses greater generalization and dynamic responsiveness, making it an innovative computational mechanism that combines theoretical interpretability with practical application.

[0158] in, 、 、 The setting method is as follows:

[0159] First, static configuration mode (system settings):

[0160] System administrators or rule designers can manually set the weighting of each coefficient based on experience or industry characteristics. For example, if a company's sales strategy emphasizes the breadth of a field's distribution across various business dimensions, a larger α value, such as 0.6, might be used. If greater emphasis is placed on a field's ability to differentiate between different business categories, a γ value of 0.5 might be used, while other coefficients might be appropriately reduced. This model is suitable for system scenarios where rule changes are infrequent and rapid deployment is required.

[0161] Second, dynamic training mode (model optimization):

[0162] The system can use existing historical rebate data as training samples, establish an optimization objective function using target variables (such as settlement accuracy and rebate error backtesting rate), reversely adjust the values ​​of α, β, and γ in the scoring model, and iteratively learn the optimal coefficient combination through algorithms such as gradient descent, grid search, or Bayesian optimization. This method is suitable for rebate systems with large data volumes, complex models, and frequently changing rules, and can significantly improve matching quality and automation capabilities.

[0163] Suppose a company is implementing rebate rule scoring for its "Customized Channel Promotion" business. In this business scenario, the channel field is highly differentiated across regional branches, while product models are widely distributed across multiple channels but have limited impact. Based on this experience, the system administrator sets the following parameters:

[0164] =0.2 (indicating low importance of distribution density);

[0165] =0.3 (indicating a moderate concentration effect);

[0166] =0.5 (indicates the greatest discriminatory effect);

[0167] In this case, if a field has clear discriminatory power within a business dimension, even if its coverage is limited, its higher γ value will give it a greater weight in the weight adjustment, ultimately increasing its influence in the overall matching score. Conversely, in scenarios where the distribution across all channels is balanced and there are no clear dividing lines, the system can appropriately lower the γ value to suppress ineffective noise interference.

[0168] This adjustment coefficient mechanism has the following beneficial effects:

[0169] Supports dual modes of expert knowledge injection and data-driven training;

[0170] It enables the system to flexibly cope with differences in field weights in different business scenarios;

[0171] Avoid structural rigidity or weight bias in scoring models when facing heterogeneous field distribution;

[0172] Provide explainability and customization capabilities for scoring algorithms.

[0173] In a preferred embodiment of the present invention, for each rule item in the evaluable rule set, a time coincidence score between the current sales field set and its time field, an exact match score with the model field, and a classification similarity score with the channel field are calculated respectively, including:

[0174] Calculate the interval intersection of the time field in the current sales field set and the rule item time interval, divide the number of intersection days by the total number of days in the rule time interval to obtain the time overlap score;

[0175] Perform a character-level comparison of the product model field in the sales field set with the model field in the rule item. If they are completely consistent, a full score is assigned. If they are partially matched, the score is reduced based on the similarity to obtain an exact match score.

[0176] Perform category mapping or label similarity calculation on the channel field in the sales field set and the rule item channel field, match the similarity value with the scoring criteria of the channel field, and output the classification similarity score.

[0177] To improve the interpretability and computability of multi-field scoring between sales records and rule items, this embodiment of the present invention employs a scoring algorithm designed for each field, ensuring a logically rigorous and mathematically sound matching process. The time field, model field, and channel field are processed using three methods: interval scoring, character comparison, and categorical similarity assessment, respectively, forming a comprehensive field matching scoring system.

[0178] For time fields, the system calculates the interval intersection between the sales record's sales time and the rule's applicable time range. The system then counts the number of days of this intersection and divides it by the total number of days in the rule's time span. The system then outputs a time overlap score between 0 and 1. This score quantifies the degree of alignment between the sales time and the rule's effective time, balancing time granularity and adaptability.

[0179] For product model fields, the system performs a character-by-character comparison. If the model field in the sales record matches the rule entry exactly, a full score is assigned. If there is a partial character match or prefix / suffix overlap, the system uses the edit distance or similarity function to calculate the score, and the score is reduced based on the degree of similarity. This strategy maintains strict matching standards while allowing for a certain range of fuzzy recognition to accommodate model naming differences.

[0180] When processing channel fields, the system uses a predefined channel classification table to map sales and rule fields to standard category labels. It then generates a category similarity score based on the category hierarchy or label similarity metrics. For example, if two fields belong to the same channel category but different subcategories, the system may assign a higher score, but not a perfect score. All scores are normalized according to established scoring criteria and output as input for the next step, weighted scoring.

[0181] Through the above scoring method, the system provides clear, traceable, and adjustable matching scoring results for each rule item, effectively improving the accuracy of rule screening and the transparency of rebate calculation logic.

[0182] An embodiment of the present invention further provides a channel merchant automated settlement processing system, the system comprising:

[0183] The sales data acquisition module is used to obtain the original sales data, perform field detection and time sorting on the original sales data, and obtain standard sales data;

[0184] The rebate rule management module is used to obtain rebate rule parameters, configure corresponding weight values ​​for each field according to the field information in the rebate rule parameters, adjust the field structure and bind the weights of the rebate rule parameters, and generate a weight rule data set;

[0185] The rule matching and scoring module is used to obtain, from each sales record in the standard sales data, rule items that have a non-empty matching relationship with the current sales record from the weighted rule dataset as an evaluable set, and perform field matching scoring on each rule item to generate a rule adaptation score set;

[0186] The rebate calculation module is used to filter the target rule item with the highest score based on the rule adaptation score set, extract the rebate coefficient in the target rule item, and multiply the rebate coefficient by the sales quantity of the corresponding sales record to generate the rebate amount;

[0187] The confidence calculation module is used to perform normalization calculation based on the maximum score value and the distribution range of other score values ​​in the rule adaptation score set to generate the corresponding confidence score;

[0188] The settlement generation module is used to generate a set of settlement items with rebate amounts and confidence scores based on all sales records, and merge the settlement item sets to generate complete settlement detail data.

[0189] It should be noted that this system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0190] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the above-described method. All implementations in the above-described method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0191] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the above-described method. All implementations in the above-described method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0192] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A channel merchant automated settlement processing method, characterized in that: The method comprises: Obtain the original sales data, perform field detection and time sorting on the original sales data to obtain standard sales data; Obtain rebate rule parameters, configure corresponding weight values ​​for each field based on the field information in the rebate rule parameters, and adjust the field structure and bind the weights of the rebate rule parameters to obtain a weighted rule dataset. The rebate rule parameters include the rebate rule number, applicable time range, applicable product model, applicable channel information, and rebate coefficient. Based on each sales record in the standard sales data, obtain the rule items with a non-empty matching relationship with the current sales record from the weighted rule dataset as the evaluable set, and perform field matching score processing on each rule item to obtain the rule adaptation score set, which specifically includes: Extract the time field, model field, and channel field of each sales record in the standard sales data to form the current sales field set; Traverse and filter all rule items in the weighted rule data set, identify rule items whose field values ​​have a non-empty intersection or fuzzy similarity relationship with the current sales field set, and generate an evaluable rule set; For each rule item in the evaluable rule set, three scores are calculated between it and the current sales field set: the time overlap score of the time field, the exact match score of the model field, and the classification similarity score of the channel field; Multiply the time overlap score, exact match score, and category similarity score by the time field weight, model field weight, and channel field weight, respectively, and add up the product results to obtain the weighted score of the sales record and the rule item. The weighted scores calculated for all rule items are combined into a set to generate a rule adaptation score set corresponding to the sales record; Based on the rule adaptation score set, the target rule item with the highest score is selected, the rebate coefficient in the target rule item is extracted, and the rebate coefficient is multiplied by the sales quantity of the corresponding sales record to obtain the rebate amount corresponding to the sales record; According to the rule adaptation score set, the maximum score value and the distribution range of other score values ​​are normalized and calculated to generate the corresponding confidence score; Based on all sales records, a set of settlement items with rebate amounts and confidence scores is generated, and the set of settlement items is merged to obtain complete settlement detail data.

2. A channel merchant automated settlement processing method according to claim 1, characterized in that: According to the field information in the rebate rule parameters, configure the corresponding weight value for each field, including: Extract the applicable time range, applicable product model, and applicable channel information fields from the rebate rule parameters to form a field information set; Calculate the field importance index based on the field stability, hit frequency, and distribution fluctuation of each field in the historical sales records; According to the field importance index and the distribution characteristics of the field in different product types or sales channels, set the weight adjustment factor corresponding to each field; Multiply the field importance index and the corresponding weight adjustment factor to obtain the field score value set; The field score value set is normalized to generate a field weight value set, including the time field weight value, the model field weight value, and the channel field weight value.

3. The automated settlement processing method for channel merchants according to claim 2, characterized in that: Adjust the field structure and weight binding of the rebate rule parameters to obtain the weighted rule dataset, including: Based on the field weight value set, each field weight value is matched one by one with the corresponding field in the rebate rule parameters, and the matching results are marked and enhanced to generate weighted field records; Combine and bind the weighted field record with its corresponding rebate rule number and rebate coefficient to construct a unified rule structure entry; All rule structure entries are constructed into rule key-value combinations in the preset field order for index identification to generate a weighted rule dataset.

4. The automated settlement processing method for channel merchants according to claim 2, characterized in that: Calculate the field importance index based on the field stability, hit frequency, and distribution fluctuation of each field in the historical sales records, including: Based on historical sales record data, we perform time-dimension segmentation and statistics on each field in the field information set, analyze the magnitude of field value fluctuations within consecutive cycles, and obtain the stability score of the field. Based on historical sales record data, calculate the frequency of each field value in the sales record and obtain the hit frequency score; Based on historical sales record data, the standard deviation or information entropy of the distribution of each field value is calculated to obtain the volatility score of the field; The stability score, hit frequency score, and volatility score are weighted and summed according to a preset ratio to generate the field importance index of the corresponding field.

5. The automated settlement processing method for channel merchants according to claim 4, characterized in that: Based on the field importance index and the distribution characteristics of the field in different product types or sales channels, set the weight adjustment factor corresponding to each field, including: Based on historical sales record data, extract the value set of each field in the field information set under different product types, sales regions or channel types to form field-business dimension mapping relationship data; Based on the field-business dimension mapping relationship data, the value distribution density, concentration, and distinctiveness of each field in different business dimensions are counted to obtain a set of field distribution characteristic indicators; Based on each indicator in the field distribution characteristic indicator set and combined with the field importance indicator, a preference factor evaluation calculation is performed to output the weight adjustment factor of each field.

6. The automated settlement processing method for channel merchants according to claim 1, characterized in that: For each rule item in the evaluable rule set, three scores are calculated between it and the current sales field set, namely the time overlap score of the time field, the exact match score of the model field, and the classification similarity score of the channel field, including: Calculate the interval intersection of the time field in the current sales field set and the rule item time interval, divide the number of intersection days by the total number of days in the rule time interval to obtain the time overlap score; Perform a character-level comparison of the product model field in the sales field set with the model field in the rule item. If they are completely consistent, a full score is assigned. If they are partially matched, the score is reduced based on the similarity to obtain an exact match score. Perform category mapping or label similarity calculation on the channel field in the sales field set and the rule item channel field, match the similarity value with the scoring criteria of the channel field, and output the classification similarity score.

7. A channel dealer automated settlement processing system, characterized in that: Applied to the method according to any one of claims 1 to 6, the system comprises: The sales data acquisition module is used to obtain the original sales data, perform field detection and time sorting on the original sales data, and obtain standard sales data; The rebate rule management module is used to obtain rebate rule parameters, configure corresponding weight values ​​for each field according to the field information in the rebate rule parameters, adjust the field structure and bind the weights of the rebate rule parameters, and generate a weight rule data set; The rule matching and scoring module is used to obtain, from each sales record in the standard sales data, rule items that have a non-empty matching relationship with the current sales record from the weighted rule dataset as an evaluable set, and perform field matching scoring on each rule item to generate a rule adaptation score set; The rebate calculation module is used to filter the target rule item with the highest score based on the rule adaptation score set, extract the rebate coefficient in the target rule item, and multiply the rebate coefficient by the sales quantity of the corresponding sales record to generate the rebate amount; The confidence calculation module is used to perform normalization calculation based on the maximum score value and the distribution range of other score values ​​in the rule adaptation score set to generate the corresponding confidence score; The settlement generation module is used to generate a set of settlement items with rebate amounts and confidence scores based on all sales records, and merge the settlement item sets to generate complete settlement detail data.

8. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images