Multi-dimensional after-sales service evaluation scoring system

Abstract

The invention relates to the technical field of service quality evaluation, in particular to a multi-dimensional after-sales service evaluation scoring system, which comprises a responsibility description comparison module, a node process propulsion module, a scoring dimension generation module, a scoring data acquisition module and a responsibility attribution judgment module. According to the invention, through the description and comparison of responsible persons of audio and text dual channels, the verification of node sequence consistency and conclusion content consistency, the linkage judgment of the fusion task state and personnel identity, and the linkage circulation logic of on-site sign-in data, and the dynamic mapping of the scoring dimension in combination with multivariate parameters such as the service type and the emergency degree. Scoring data are collected in a layered mode, score differences and time sequences are fused, responsibility traceability labels are established, multi-dimensional data synchronous verification, flow node collaborative propulsion, dynamic scoring scene matching and responsibility affiliation transparent management are achieved, the dispute recognition capacity in a complex service scene is enhanced, and the responsibility traceability is improved. And the adaptability of the whole-process tracing fineness of the work order and the service link evaluation is improved.

Description

Technical Field

[0001] This invention relates to the field of service quality evaluation technology, and in particular to a multi-dimensional after-sales service evaluation and scoring system. Background Technology

[0002] Service quality evaluation mainly involves monitoring, providing feedback, and evaluating the service process and results provided by enterprises through digital means, in order to achieve the goals of service quantification, traceability, and optimization. This includes service process tracking, customer feedback collection, employee performance quantification, service response timeliness management, and evaluation data visualization. It is widely used in various business scenarios such as enterprise after-sales service, customer service management, and maintenance processes. Among them, the traditional after-sales service evaluation system refers to a model that collects scoring information through a single user feedback entry and performs simple weighted summaries. Such systems usually use a form submission method where users fill in a service rating after the repair is completed, and then the platform calculates the total score through fixed weights.

[0003] Existing technologies rely on single-channel user feedback to collect ratings, resulting in a simplistic form data structure. The ratings only reflect the static experience of end users, and the dynamic information of multiple roles in the work order process is not systematically integrated. There is a lack of adaptability to business scenarios such as inconsistent responsibility descriptions, overlapping process divisions, and multiple confirmations at each node. When faced with conflicting node descriptions and multi-node collaboration scenarios, the attribution of responsibility is unclear, and the evaluation of service processes and results cannot cover the entire process chain. This leads to incomplete traceability information for disputed work orders, making it difficult for management to achieve multi-dimensional collaboration and real-time decision support. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a multi-dimensional after-sales service evaluation and scoring system.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a multi-dimensional after-sales service evaluation and scoring system, the system comprising:

[0006] The responsibility description comparison module analyzes the names of responsible persons and the order of event nodes based on the audio and text descriptions of work orders uploaded by users, verifies the consistency of the conclusion content, identifies nodes with inconsistent descriptions as abnormal nodes, and obtains a responsibility conflict feature set.

[0007] Based on the aforementioned responsibility conflict feature set, the node process advancement module screens work order task numbers, compares check-in time with the end time of the previous node, determines whether identity information and status identifier correspond, advances process flow, and obtains process flow sequence.

[0008] The scoring dimension generation module extracts service type and urgency based on the process flow sequence, constructs parameter structure, matches scoring rule library indicators, and adjusts scoring items in combination with the identity of the responsible party and node status to obtain multi-scenario scoring indicators.

[0009] The scoring data acquisition module analyzes the differences between the scoring fields and the standard scores based on the multi-scenario scoring indicators, determines the correspondence between the actual scores and the standards, records the differences in node scores, and organizes the scoring time to obtain the scoring data offset sequence.

[0010] The present invention improves upon this invention by including the following features: the responsibility conflict feature set includes conflict node identifiers, responsibility attribution types, and content consistency identifiers; the process flow sequence includes node sequence codes, process advancement identifiers, and node processing statuses; the multi-scenario scoring indicators include evaluation dimension markers, scoring item categories, and applicable identity labels; and the scoring data offset sequence includes score fluctuation identifiers, scoring anomalies, and data time indexes.

[0011] The present invention is improved in that the liability description comparison module includes:

[0012] The node sequence analysis submodule compares the order of each node in the audio with the time sequence of the nodes in the text based on the audio and text descriptions of the work orders uploaded by the user. It integrates the name, processing stage and occurrence order of each node, calculates the correspondence between node time and order, and determines the distribution of each node number and tag to obtain node sequence comparison data.

[0013] The responsibility difference analysis submodule analyzes the spelling of names, job numbers and organization numbers of nodes in the same order based on the node sequence comparison data, calculates the number of characters with name differences, counts the correspondence between job numbers and organization numbers, determines the offset status of account data, and obtains the degree of difference in responsibility fields.

[0014] The consistency analysis submodule filters keywords and sentence order in the conclusion description based on the degree of difference in the responsibility field, judges the keyword overlap rate and description completeness, identifies nodes with missing or incorrect keyword order, and integrates the differences between node number and responsibility field to obtain a responsibility conflict feature set.

[0015] The present invention is improved in that the node process advancement module includes:

[0016] The task identification submodule retrieves the work order database based on the responsibility conflict feature set, matches the work order number and node responsibility type, performs a set encoding difference comparison between each conflict node data and the registered task number in the work order database, filters out unregistered task numbers, and obtains a set of unrecorded task numbers.

[0017] The time series comparison submodule, based on the uncollected task number set, obtains the sign-in time of the dispatching engineer and the end time of the previous node for each number, compares the sign-in time and end time, and calculates the node time difference by combining the responsible person's identity code and sign-in location data, and marks each node to obtain the node time identifier sequence.

[0018] The workflow confirmation submodule extracts the node responsible person's account and the operator's identity based on the node time identifier sequence, compares the two types of identity data item by item, obtains the dispatch status field and the sign-in confirmation field in sequence, compares their field contents, gradually filters nodes with completely consistent contents, and integrates the node sequence code and processing status to obtain the workflow sequence.

[0019] The present invention is improved in that the rating dimension generation module includes:

[0020] The service parameter extraction submodule analyzes the node sequence code and node processing status based on the process flow sequence, compares the service type and urgency of each node, constructs joint parameters by integrating service type tags and urgency levels, and determines the service type and urgency combination data to obtain the service combination parameter set.

[0021] The indicator matching calculation submodule determines the correspondence between the categories of scoring items and identity tags in the scoring rule library based on the service combination parameter set, calculates the fitting range of scoring items, filters the categories of scoring items, and obtains the scoring matching indicators.

[0022] The dimension expansion construction submodule, based on the scoring matching index and combined with the current node processing status code, retrieves the additional scoring dimension items associated with the scoring rule library, determines the matching situation between the identity label and the node status, and integrates the additional dimension parameters into the existing index to obtain multi-scenario scoring indexes.

[0023] The present invention is improved in that the scoring data acquisition module includes:

[0024] The field difference identification submodule, based on the multi-scenario scoring indicators, extracts the actual entered score and the standard score of the scoring rule according to each scoring field under the work order node, calculates the difference between the two under the corresponding scoring item, and identifies the directionality of the difference to obtain the field difference sequence.

[0025] The offset point extraction submodule compares the number of differences with the identification benchmark content configured in the scoring rules based on the field difference sequence, filters the scoring fields whose number of differences exceeds the benchmark content, marks the node where the scoring field is located, the scoring item category and its corresponding offset performance, and obtains the scoring offset node group.

[0026] The time index integration submodule collects scoring time information based on the scoring offset node group, extracts the time arrangement of the node where the scoring item is located, judges the fluctuation of adjacent offset scoring fields in time sorting, calculates the time dispersion of scoring fields, judges the distribution change of data offset in the time series, and obtains the scoring data offset sequence.

[0027] The present invention has an improvement, wherein the system further includes:

[0028] The responsibility attribution determination module determines the identity of the responsible party at each node, the maintenance report, and the acceptance status based on the offset sequence of the scoring data. It then summarizes the scoring items with attribution characteristics, integrates the associated numbers and accounts, and obtains the responsibility attribution mapping result.

[0029] The responsibility attribution mapping result includes the attribution node code, associated responsible person information, and traceability judgment label.

[0030] The present invention is improved in that the responsibility attribution determination module includes:

[0031] The identity verification submodule analyzes the relationship between the identity of the responsible party at each node and the maintenance report filling time and acceptance status based on the scoring data offset sequence, compares the chronological relationship between the scoring time and the report time, filters nodes whose scoring time is earlier than the report filling time, and determines whether the node corresponding to the scoring record has been accepted, thus obtaining the responsibility time mismatch record.

[0032] The item filtering submodule compares the scoring items corresponding to the field numbers with the node numbers and the responsible personnel accounts based on the responsibility time mismatch records. It analyzes the correspondence between the responsible person's identity in the scoring field and the responsible party's identity in the node, determines whether the role identity data is complete, and obtains the responsibility anomaly screening results.

[0033] Based on the results of the responsibility anomaly screening, the attribution mapping submodule integrates the data corresponding to the field number, node number, and responsible personnel account, optimizes the correlation between the scoring time, scoring field content, and node responsible person information, and sorts and aggregates the node and account information to obtain the responsibility attribution mapping result.

[0034] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0035] In this invention, the responsible person descriptions are compared through both audio and text channels. This is combined with the consistency verification of node order and conclusion content. The linkage judgment of task status and personnel identity is integrated. The on-site sign-in data is linked to the flow logic. The scoring dimensions are dynamically mapped by combining multiple parameters such as service type and urgency. Scoring data is collected in layers and the score differences and time series are integrated to establish responsibility traceability tags. This enables multi-dimensional data synchronous verification, collaborative advancement of process nodes, dynamic scoring scenario matching, and transparent management of responsibility attribution. It strengthens the ability to identify disputes in complex business scenarios and improves the granularity of work order traceability and the adaptability of service link evaluation. Attached Figure Description

[0036] Figure 1 This is a system flowchart of the present invention;

[0037] Figure 2 This is a flowchart of the liability description comparison module in this invention;

[0038] Figure 3 This is a flowchart of the node process advancement module in this invention;

[0039] Figure 4 This is a flowchart of the scoring dimension generation module in this invention;

[0040] Figure 5 This is a flowchart of the scoring data acquisition module in this invention;

[0041] Figure 6 This is a flowchart of the responsibility attribution determination module in this invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0044] Example

[0045] Please see Figure 1 This invention provides a technical solution: a multi-dimensional after-sales service evaluation and scoring system, comprising:

[0046] The responsibility description comparison module analyzes the responsible person's name in the audio description and the responsible person's information in the text description based on the audio and text descriptions of the work order uploaded by the user. It directly compares the order of the two sets of event nodes in the audio and text, checks the correspondence between the audio conclusion and the text conclusion, and judges the actual differences between the responsible person, the order of events and the content of the conclusion. When a description discrepancy occurs, it is marked as an abnormal node, and a responsibility conflict feature set is obtained.

[0047] The node process advancement module, based on the responsibility conflict feature set, determines the uncollected work order task number, analyzes the relationship between the work order dispatch engineer's sign-in time and the previous node's end time, matches the responsible person and the operating user according to identity data, and filters the matching of dispatch status identifier and sign-in confirmation content. If all conditions are met, the process is advanced to the next work order node, resulting in a process flow sequence.

[0048] The scoring dimension generation module is based on the process flow sequence, filters the service type and urgency of nodes, constructs combined parameters, analyzes the parameters and the indicator items in the scoring rule library, compares the identity of the responsible party, adjusts the scoring items for the purchaser or supplier, and adds related scoring dimensions according to the current node status to obtain multi-scenario scoring indicators.

[0049] The scoring data acquisition module analyzes the difference between the actual score and the standard score of each scoring field based on multi-scenario scoring indicators, determines the correspondence between the score entered by the scorer and the standard score, records the real difference of the scores of the scoring fields under the work order node, and organizes the scoring items with reference to the scoring time to obtain the scoring data offset sequence.

[0050] The responsibility attribution determination module determines the correspondence between the identity of the responsible party at each node and the time of completion of the maintenance report and the acceptance status based on the offset sequence of the scoring data. The scoring items with responsibility attribution characteristics are merged and summarized by field number, node number and responsible personnel account to obtain the responsibility attribution mapping result.

[0051] The responsibility conflict feature set includes conflict node identifier, responsibility attribution type, and content consistency identifier; the process flow sequence includes node sequence code, process progress identifier, and node processing status; the multi-scenario scoring indicators include evaluation dimension marker, scoring item category, and applicable identity label; the scoring data offset sequence includes score fluctuation identifier, scoring anomaly point, and data time index; and the responsibility attribution mapping result includes attribution node code, associated responsible person information, and source tracing judgment label.

[0052] In the responsibility description comparison module, event nodes refer to key steps in the work order processing flow, including reporting a repair, dispatching a work order, accepting a work order, checking in, filling out a maintenance report, acceptance, and scoring. Each step is an independent node in the system, used to track the work order's progress. Direct comparison refers to the system comparing the user-uploaded audio with the text description one-to-one according to structured information such as nodes and responsible persons to check for content conflicts without requiring secondary manual screening. Correspondence check refers to comparing data fields of the same type (such as responsible persons and conclusion content) in the audio and text descriptions item by item to verify whether their data content is consistent in the two channels. Actual difference refers to inconsistencies identified by the system between the audio and text descriptions, including differences or omissions in responsible person information, node order, or conclusion descriptions. Abnormal nodes refer to specific nodes in the work order processing flow where conflicts, omissions, or inconsistencies in descriptions are found after comparison. These nodes are individually identified by the system for subsequent tracking and processing.

[0053] In the node process advancement module, the work order task number refers to the unique identifier of each after-sales service work order and its task flow node in the system, used to distinguish and track the processing progress of each work order; the corresponding judgment refers to the one-to-one verification of the information of each participant in the work order flow process (such as responsible person, operator user) according to the identity data in the system to ensure that the responsibility at the process node is correct; the dispatch status identifier refers to the status attribute of the work order in the dispatch process, reflecting whether the work order has been assigned to a specific engineer by the supplier, such as "pending dispatch" or "dispatched"; the sign-in confirmation content refers to the sign-in information uploaded by the after-sales engineer after arriving at the site through on-site photos or clock-in operation. The system automatically records the sign-in time, location and other data to confirm actual attendance; all conditions are met means that when all judgment items (such as time, identity, status) involved in the work order node advancement are verified to be correct, the system automatically advances the process to the next node; the next work order node refers to the next processing link automatically pointed to by the system according to the process after the current node is completed, such as entering the engineer order acceptance node after dispatch.

[0054] In the scoring dimension generation module, the node service type refers to the specific service category corresponding to the work order node, such as "equipment repair," "fault consultation," and "inspection and maintenance," which is used to distinguish the evaluation indicator content in the scoring system; urgency level refers to the priority division of the work order or task, generally divided into "urgent" and "normal," which determines the weight of subsequent dispatch response and indicator activation; combined parameters refer to the structured data formed by the system combining business parameters such as service type, urgency level, and responsible party role, which facilitates matching and searching in the scoring rule library; the scoring rule library refers to the system's built-in service evaluation standards and scoring item data set, which dynamically loads corresponding evaluation dimensions and scoring standards according to different business scenarios; indicator item content refers to the specific scoring items, such as "response speed," "service attitude," and "repair effect," which are dynamically selected by the system as parameters when generating scoring dimensions; scoring items refer to each evaluation item that the user actually needs to complete after work order acceptance, corresponding to each row or item in the system scoring table; associated scoring dimensions refer to the scoring indicators dynamically added according to the actual business attributes of the current node when generating scoring dimensions, so that the scoring results are consistent with the actual business.

[0055] In the rating data acquisition module, the rating rule standard score refers to the basic reference score set by the rating system for each rating item, such as a full score of "5 points" or "10 points", which is used to compare actual scoring data and analyze service deviations; the correspondence relationship refers to the item-by-item mapping between user rating data and rating rule standard scores, which identifies the gap between user ratings and standards by comparison; the actual difference of rating field scores refers to the direct difference data between each score actually entered by the user and the standard score for that item, which is used for subsequent statistics and offset analysis; the rating time refers to the system timestamp corresponding to each evaluation submitted by the rater, which is used for work order rating traceability and statistical analysis.

[0056] In the responsibility attribution determination module, the node responsible party identity refers to the user identity who actually performs the duties in each business node of the work order, including purchasers, suppliers, after-sales engineers, etc., for responsibility tracking; responsibility attribution characteristics refer to key data fields or behavioral characteristics used to clarify the attribution of responsibility in traceability scenarios such as scoring deviation and process anomalies, for example, scoring anomalies are directly bound to the account of the responsible person of a certain node; merging and summarizing refers to the system integrating and classifying multi-dimensional data such as scoring items, node numbers, and responsible person accounts to form structured responsibility attribution information, which is convenient for subsequent traceability and statistics.

[0057] Please see Figure 2 The responsibility description comparison module includes:

[0058] The node sequence analysis submodule compares the order of each node in the audio with the time sequence of the nodes in the text based on the audio and text descriptions of the work orders uploaded by the user. It integrates the name, processing stage and occurrence order of each node, calculates the correspondence between node time and order, and determines the distribution of each node number and tag to obtain node sequence comparison data.

[0059] The audio content is transcribed into structured text sentence by sentence using speech recognition. Key nouns such as "repair request," "order dispatch," "sign-in," and "acceptance" are identified, and the order of these nodes in the audio is determined by combining the timestamp field. For example, if "engineer arrives and signs in" appears after "order dispatch completed" in the audio, its sequence number is recorded as 2, while "order dispatch" is recorded as 1. Simultaneously, the text description is parsed in paragraphs or sentence groups to identify keywords for nodes of the same type. The sentence "engineer confirms arrival" is identified as the "sign-in" node, and its sequence number is recorded based on its context. The responsible person, job label, and node type are extracted simultaneously to construct a structured information set for each node. Based on this, the order of each node name in the audio and text is compared to retrieve the "repair request" node. Does it always appear before "Dispatch"? Does "Check-in" always appear after "Accept Order"? If any set of nodes is found to have an inconsistent order in the two descriptions, for example, "Check-in" appears before "Dispatch" in audio but the opposite in text, then it is marked as a node with an inconsistent order. For each node's processing stage, it is classified into three processing stages according to keywords: early response, processing, and closing stage. "Report Repair" and "Dispatch" are mapped to the response stage, "Check-in" and "Maintenance Report" are mapped to the processing stage, and "Rating" and "Acceptance" are classified as the closing stage. Then, the names, processing stages, and sequence numbers of each node in the two channels are integrated into a table record, and the comparison relationship is output. For example, if the node name is "Dispatch", the audio sequence number is 1, the text sequence number is 2, and the processing stage is the response stage, a comparison list of all nodes is compiled as node sequence comparison data.

[0060] The responsibility difference analysis submodule analyzes the spelling of names, job numbers and organization numbers of nodes in the same order based on the node sequence comparison data, calculates the number of characters with name differences, counts the correspondence between job numbers and organization numbers, determines the offset status of account data, and obtains the degree of difference in responsibility fields.

[0061] Under the same node number, the corresponding responsible person's name field is read from both the audio and text. By comparing the differences in character positions word by word, each different character is counted. For example, if the responsible person in the audio is "Li XX" and the text is "Li XZ", "X" and "Z" are identified as different characters and counted as 1 difference. Then, the job number field to which both belong is read. For example, if the responsible person in the audio is "A2001" and the text is "A2003", the numbers are checked to see if they match. If they do not match, it is marked as a mismatch in job. At the same time, the organization number field is read. If the record in the audio is "B301" and the text is "B302", it is determined that the organization numbers do not match. Therefore, the job field in the record is marked as inconsistent, and the organization number is marked as inconsistent. If the fields are inconsistent, the account data fields are analyzed further. The position number of the account in the audio appears in the node sequence is checked to see if it is consistent with the sequence number in the text. For example, if the account "engineer001" appears in node number 3 in the audio but in node number 4 in the text, the account offset is recorded as +1, indicating that the account has been offset. Then, the four basic difference parameters, namely the number of differences in name characters, the consistency status of job number, the consistency status of organization number, and the account offset, are integrated and summarized into the responsibility field difference value of this node. This difference value is used to identify the consistency of the current node in the identity field in subsequent consistency analysis and judgment, and the responsibility field difference degree of each node is output.

[0062] The consistency analysis submodule filters keywords and sentence order in the conclusion description based on the degree of difference in the responsibility field, judges the keyword overlap rate and description completeness, identifies nodes with missing or incorrect keywords, and integrates the differences between node numbers and responsibility fields to obtain a responsibility conflict feature set.

[0063] The conclusion description field under each node is processed sequentially. First, the complete sentence in the description text is segmented into several phrases to locate key operation verbs and result noun groups, such as "installation complete," "submit report," and "verification complete." Then, a predefined set of keyword phrases is loaded from the built-in standard library to identify whether the phrases appearing in the current node description contain the keyword. If "submit report" exists in the standard library, it is determined to be a keyword overlap. The number of these keywords is counted one by one to obtain the actual number of overlaps, and the total number of phrases in the node description is recorded. Based on this, the keyword overlap rate is calculated, and the sentence order of the conclusion description is further checked. The conclusion statement of the previous node is compared with the node order in the standard process. For example, if the process stipulates that "dispatch" should be earlier than "check-in", if the current conclusion is "dispatch completed after check-in", it is judged as a sequence mismatch and marked as an incorrect sequence. Then, the keyword overlap rate, statement order correctness, and responsibility field difference obtained in the previous step are integrated. If the overlap rate is extremely low, the statement order is incorrect, and the field difference is large, the node is marked as a responsibility conflict node. Finally, the node number, number of keyword overlaps, total number of phrases, statement order consistency, responsibility field difference value, and other elements are recorded in the table and summarized into a complete responsibility conflict feature set.

[0064] Please see Figure 3 The node process advancement module includes:

[0065] The task identification submodule retrieves the work order database based on the responsibility conflict feature set, matches the work order number and node responsibility type, performs a set encoding difference comparison between each conflict node data and the registered task number in the work order database, filters out unregistered task numbers, and obtains a set of uncollected task numbers.

[0066] Extract the node number and responsibility attribution type fields from each conflict node. Perform primary key matching of these field values ​​with all work order number fields registered in the work order database. If a conflict node record is "Work Order Number: GD20250718012", check if this number exists in the work order database. If it exists, extract all registered node information under this work order. Then, retrieve the responsibility type information field from the conflict node, such as "dispatching party" or "sign-in party". Search the work order database to see if this type of responsibility field is registered under the corresponding task item. If no corresponding record is found, mark it as a missing task item. Finally, combine the node numbers in the conflict node data with the set of task nodes extracted from the work order database. The comparison operation performs a set difference analysis between the conflict node ID set and the task node set in the work order database. Specifically, it extracts ID items that appear only in the conflict node set but not in the work order database set. These IDs are considered unregistered IDs. For example, if the conflict node ID set is {3, 5, 6, 8} and the corresponding task node set in the work order database is {1, 2, 3, 5, 7}, then IDs 6 and 8 are marked as unregistered task nodes. The set difference results are then validated for ID format to ensure they conform to the work order coding rules, such as being presented in the form of "GD + year, month, day + three-digit serial number". IDs that do not meet the format are not included in the analysis. The selected IDs are then integrated into a set of unrecorded task IDs.

[0067] The time series comparison submodule, based on the uncollected task ID set, retrieves the dispatch engineer's check-in time and the previous node's end time for each ID, compares the check-in and end times, and combines this with the responsible person's identification code and check-in location data, using the following formula:

[0068]

[0069] Calculate the node time difference T s Each node is marked to obtain a node time identifier sequence, where n represents the number of check-in comparison nodes, and S i E represents the check-in time of the engineer at the i-th node. i This indicates the end time of the i-th preceding node, and m represents the number of check-in locations. This represents the code of the j-th check-in location. This represents the code for the j-th reference location;

[0070] The node time difference refers to the absolute time difference between the check-in time of the dispatching engineer and the end time of the previous node for each unrecorded work order task. The system calculates the absolute time difference between the two, takes the average of the differences, and accumulates the differences between the check-in location code and the reference location code. The differences in both time and location are reflected as a numerical value, which is used to quantify the temporal continuity and location consistency of the task in the node processing flow.

[0071] Extract the engineer check-in time S for each task. i Compared to the end time E of the previous node i A time series comparison sample set was constructed. The check-in times for task numbers WM2025041203, WM2025041204, and WM2025041205 were 120 min, 245 min, and 390 min, respectively, and the previous node's end times were 100 min, 250 min, and 360 min, respectively. The differences between the three sets were 20, 5, and 30 min. To avoid inconsistencies in dimensionality due to differences in the time spans of different nodes, the time differences were normalized before calculation. The maximum value of 30 min was taken as the normalization reference. The above time differences were converted into normalized results: 20 ÷ 30 = 0.667, 5 ÷ 30 = 0.167, 30 ÷ 30 = 1.000, resulting in a normalized time difference set of {0.667, 0.167, 1}. The calculation yielded:

[0072]

[0073] Next, the location code differences are processed. Let the check-in location code for the first task be 5.2, the reference location code be 5.0, and the difference be 0.2; for the second task, the check-in location code be 7.8, the reference code be 8.0, and the difference be 0.2. Assuming the maximum reference difference for location codes is 0.5, after normalization, the differences are 0.2 ÷ 0.5 = 0.4 and 0.2 ÷ 0.5 = 0.4 respectively. The sum of the code differences is then calculated.

[0074] 0.4 + 0.4 = 0.8;

[0075] Substitute the normalized value:

[0076] T s =0.611 + 0.800 = 1.411;

[0077] This result indicates that the node time difference T s =1.411 has exceeded the set upper limit range of the joint temporal and spatial offset reference [0.000, 1.200]. According to the built-in process rules, if the T of a certain node... s If the value exceeds this range, the node is determined to have inconsistencies in the check-in time sequence or check-in location information. Therefore, the node needs to be set to "unmatched" status in the subsequent node marking process, and its time matching identifier value is assigned to "0" otherwise "1". This value directly determines whether the node is included in the valid process flow chain. After the above calculation, the system outputs the node number and the matching status together to construct a standardized process time matching identifier sequence for the next stage of process flow confirmation and judgment.

[0078] The workflow confirmation submodule extracts the account of the person in charge of the node and the identity of the user based on the node time identifier sequence. It compares the two types of identity data item by item, obtains the dispatch status field and the sign-in confirmation field in turn, compares their field contents, and gradually filters nodes with completely consistent contents. Finally, it integrates the node sequence code and processing status to obtain the workflow sequence.

[0079] For each node's structured data item, extract the responsible person's account field and the operating user's identity field. The responsible person's account field is recorded as "eng_A078", and the operating user's identity field is recorded as "user_tech01". Compare the two types of account fields for consistency using a character-by-character comparison method. If the account characters are completely identical, it is recorded as a consistent identity; if the field difference exceeds two characters, it is marked as an inconsistent identity. Then, extract the dispatch status field and check-in confirmation field from the node data. For example, if a node's dispatch status field is "Dispatched" and the check-in confirmation field is "Attendance Confirmed", determine whether the dispatch field is "Dispatched" or "Dispatched Completed", and determine whether the check-in field is "Checked In" or "Attended". "Field Confirmation": If both fields meet one of the above specific values, the fields are considered to be consistent. The above four items of each node record are checked step by step. If any field value is empty in the record, the node is skipped and the field extraction and comparison are performed on the next node. After all nodes are screened, the node numbers of all nodes with consistent identities and matching field contents are extracted and summarized. At the same time, the sequence code field and processing status field corresponding to the node are extracted. The processing status field is recorded as text values ​​such as "processing", "completed", "pending acceptance", etc. The node sequence code is used as the horizontal sequence field, and the processing status is recorded as the current processing stage of the corresponding node. The node sequence number and processing status field are integrated in order and output as a process flow sequence.

[0080] Please see Figure 4 The rating dimension generation module includes:

[0081] The service parameter extraction submodule analyzes the node sequence code and node processing status based on the process flow sequence, compares the service type and urgency of each node, constructs joint parameters by integrating service type tags and urgency levels, and determines the service type and urgency combination data to obtain the service combination parameter set.

[0082] The process reads the node sequence code field and node processing status field of each record in the workflow sequence. The node sequence code is an integer value representing the sequence of workflow stages, such as 1 for "Report Repair", 2 for "Dispatch", 3 for "Check-in", and so on. The processing status field records the current service processing stage information of each node, such as "Pending Response", "Processing", "Accepted", etc. The node data is arranged in sequence number and a mapping table is built. Then, the service type label field is read from the node information field. This field records the service category content involved in the current node, such as "Equipment Repair", "Inspection and Maintenance", "Troubleshooting", etc. At the same time, the urgency level field is extracted. This field uses a fixed enumeration value to represent the urgency level, such as "0" for normal and "1" for urgent. "2" indicates urgent. The two fields mentioned above are extracted node by node, and a cross-combination operation is performed. That is, each service type is combined with its corresponding urgency level into a structured pair. For example, if the service type corresponding to node number 3 is "equipment repair" and the urgency level is "1", the combination result is "equipment repair-urgent". Following this logic, all nodes are traversed to construct multiple combinations of service types and urgency levels. After combination and deduplication, a unified service combination list is formed. The list records all service and urgency level pairings that actually appear in the process. Each combination record includes the node number of its first appearance and its corresponding sequence code. Finally, the above combination data is integrated to form a service combination parameter set, which is used for subsequent matching of the scoring dimensions associated with each combination parameter in the scoring rules.

[0083] The indicator matching calculation submodule, based on the service composition parameter set, determines the correspondence between the scoring item categories and identity tags in the scoring rule library, using the following formula:

[0084]

[0085] Calculate the SD of the rating item fit r The rating item categories are filtered to obtain rating matching indicators, among which PD s Represents the service type tag code, ED s Represents the emergency level code, RD s Represents responsibility identity mapping encoding, TD s Represents the scoring criteria sequence items, CD r Represents the category difference number of the rating item, ID s Represents identity compatibility number;

[0086] The fitting range of scoring items is mainly used to measure the degree of matching between the actual situation of the work order and the scoring rule items under various structural parameters. The larger the fitting range, the more obvious the difference between the business parameters and the scoring items; the smaller the fitting range, the more consistent the structural parameters of the two are. It is usually used to filter the scoring item categories in the future and select the scoring indicators that best match the attributes of the current node, thereby supporting the expansion of subsequent scoring dimensions.

[0087] Extract the normalized service type label code PD from work order node ND001 s =0.60, Emergency Level Code ED s =0.83, Responsibility Identity Mapping Code RD s =0.40, and simultaneously extract the scoring standard sequence item TD of the scoring rule item "Service Response Timeliness". s =0.95, rating item category difference number CD r =0.32, Identity Adaptability Number ID s =0.21, substitute the above parameters into the formula, and first perform the addition operation inside the parentheses:

[0088] PD s +ED s =0.60 + 0.83 = 1.43;

[0089] Then multiply the result by the responsibility identity mapping code:

[0090] 1.43·0.40=0.572;

[0091] Then perform the subtraction operation:

[0092] 0.572 - 0.95 = -0.378;

[0093] For the denominator, first add the category difference number of the scoring item to the identity suitability number:

[0094] 0.32 + 0.21 = 0.53;

[0095] Then take its square root:

[0096]

[0097] Finally, perform division and take the absolute value to obtain the fit range of the rating items:

[0098]

[0099] This result indicates that the SD of the rating item fit is... r=0.519 falls within the adaptation level range [0.4, 0.7]. According to the scoring rules, this range corresponds to the medium adaptation level, indicating that node ND001 has a moderate structural matching relationship with the service response time of the scoring item under the current service type, urgency level and responsible party identity combination. Its function is to quantify the degree of difference between the structural attributes of the scoring item and the node business parameters. By comparing with the set range, the classification and screening of the scoring item category can be further realized. Only the scoring items in the adaptation range are retained for the construction of scoring matching indicators. Therefore, this adaptation range not only completes the structural screening of the scoring item, but also directly determines whether the scoring item is included in the subsequent dimension construction process, and is finally used by the system for the screening conditions in the matching judgment logic.

[0100] The dimension expansion construction submodule is based on the scoring matching index, combined with the current node processing status code, to retrieve the additional scoring dimension items associated with the scoring rule library, determine the matching of identity tags and node status, and integrate the additional dimension parameters into the existing index to obtain multi-scenario scoring indexes.

[0101] For each indicator, the processing status code field of its corresponding node is read. This field represents the status value in the process flow sequence, such as "Processing," "Completed," or "Under Acceptance." Using this status value as a query condition, the system accesses the additional scoring dimension configuration items in the scoring rule base that match the status value. For example, the status "Completed" corresponds to the additional scoring dimensions "Repair Completion Rate" and "Customer Feedback Response Rate" in the rule base. After identifying a match, the additional scoring dimension is integrated with the basic dimension of the current scoring matching indicator. Before this, the identity tag field in the indicator item needs to be compared item by item with the applicable identity tag fields recorded in the rule base. For example, the identity tag field recorded in the current indicator item... If the label is "Supplier" and the additional dimension in the rule base is applicable to "Supplier" or "Engineer", it is considered compatible. If the current indicator's identity label is "Purchaser" but the additional dimension only applies to "Supplier", then the dimension expansion is skipped. The identity label matching rule uses text value equality or a subordinate mapping table for judgment. After confirming that the identity and node processing status match, the additional scoring dimension is written into the scoring indicator set of the current node. The integration method is to add a structured record to the original scoring item set. The record contains the dimension name, applicable identity label, and active node status. All integrated dimension indicator items constitute a complete multi-scenario scoring indicator for subsequent scoring data collection operations.

[0102] Please see Figure 5 The scoring data collection module includes:

[0103] The field difference recognition submodule is based on multi-scenario scoring indicators. According to each scoring field under the work order node, it extracts the actual entered score and the standard score of the scoring rule, calculates the difference between the two under the corresponding scoring item, and identifies the direction of the difference to obtain the field difference sequence.

[0104] First, read the work order node number and rating field name corresponding to each record in the rating indicators, and extract the actual entered user rating value. For example, in a record where the field is "service attitude", the node number is 5, and the user rating value is 7, then access the rating rule standard table to locate the benchmark score value corresponding to the rating field in the standard rule. For example, if the standard score for "service attitude" is 9, then the difference between the actual entered score and the standard score is -2. Record this difference value as the field difference value for this rating field. Continue processing the next rating field data, extract the node number and rating item name corresponding to the field, and obtain its actual entered score. Then, compare it with the benchmark value of the corresponding field in the rule by subtraction to obtain the difference value. If the actual score is higher than the standard value, it is considered a positive difference; if it is lower than the standard value, it is considered a negative difference. The directional field is recorded as "positive" or "negative". The above steps are performed on each scoring indicator record in turn to organize the differences and directions of all scoring fields into a structured field difference list. Each record contains the scoring field name, node number, scoring item category, actual score value, standard score value, difference value, and directional identifier. If the same scoring item appears repeatedly in different nodes, it is recorded separately according to the node number and the difference direction and absolute difference are analyzed independently. For example, "response speed" is -1 in node 3 and +2 in node 7. The two difference instances and their directions are recorded separately to generate a field difference series.

[0105] The offset point extraction submodule is based on the field difference sequence. It compares the number of differences with the recognition benchmark content configured in the scoring rules, filters the scoring fields whose number of differences exceeds the benchmark content, marks the node where the scoring field is located, the scoring item category and its corresponding offset performance, and obtains the scoring offset node group.

[0106] Each scoring field difference is read item by item and compared with the preset identification benchmark in the scoring rule library. The identification benchmark is the allowable deviation range set for the scoring item. For example, the identification benchmark value for the "service attitude" scoring item is 1 point, meaning that any difference exceeding ±1 is considered an offset item. The difference value in the scoring field is compared with this benchmark value. If the absolute value of the difference value exceeds 1, it is judged as a scoring offset field. For example, the score difference for "response speed" under a certain node is -2, which exceeds the benchmark value of 1 and is marked as an offset item. Then, the node number and scoring item category corresponding to the scoring field are read, and the above information is integrated with the current field difference direction. If the score difference is greater than 1, the score difference is considered as an offset field. If the sub-field has a negative difference, the offset is marked as "negative offset". If it has a positive difference, it is marked as "positive offset". Each scoring field is traversed and all records that meet the offset conditions are filtered according to this logic. The frequency of occurrence of the offset field is further counted. If the same scoring item repeatedly exceeds the benchmark threshold in multiple nodes, its offset in all nodes is recorded and the nodes are grouped. Each group of records contains the scoring item name, node number, offset direction, difference value, identification benchmark value, and exceedance range. The offset records filtered above are classified and integrated according to the node number to form scoring offset node groups, and each node group is marked with the scoring field name and offset direction.

[0107] The time index integration submodule, based on the score offset node group, collects score time information, extracts the time arrangement of the nodes where the score items are located, and determines the fluctuation of adjacent offset score fields in time sorting using the following formula:

[0108]

[0109] Calculate the time dispersion (DQ) of the rating field to determine the distribution change of the data offset in the time series, and obtain the rating data offset sequence, where n DQ Qs represents the total number of rating offset fields that are being counted. z Qr represents the actual score of the z-th rating field. z Qp represents the standard score of the rating rule for the z-th rating field. z This represents the category code of the rating item to which the z-th rating field belongs, in Qt. z This represents the time index value of the z-th rating field in the rating record table.

[0110] The time dispersion of the scoring field refers to the distribution and fluctuation of different scoring fields on the time index in the work order scoring process. It is a weighted average of the difference between the actual score and the standard score, the scoring item category and the corresponding scoring time, reflecting the dispersion and change intensity of the scoring data in the time dimension. The larger the value, the more dispersed the abnormal offset of the scoring field is in time and the more obvious the change is.

[0111] For all identified rating fields, the rating time information is extracted. This information is numbered using the recorded timestamp sequence, and each rating field has a corresponding time index in Qt. z Extract the actual score Qs based on the rating field number. z Scoring Rules Standard Score Qr z Scoring item category code Qp z and scoring time index Qt z Then, based on the distribution of the rating fields over time, the time dispersion DQ of the rating fields is calculated, where Qs z With Qr z The actual entered score and the rule-set score are respectively used to reflect the direction and degree of deviation of the scoring item. Qp z This indicates the service category code to which the rating item belongs, representing its classification weight attribute in the overall rating system. (Qt) z The time sequence number corresponds to the time of rating entry, representing the relative position of the rating field on the timeline. The calculation formula weights the rating content offset by category, harmonizes its influence using a time index, performs absolute value transformation and summation, and finally executes a square root transformation to extract the global offset trend. This structure integrates three factors of rating behavior: numerical deviation, weighted classification, and time distribution. Rating fields numbered 1, 3, and 4 participate in the calculation, and their parameters are normalized as follows:

[0112] Field 1: Qs1 = 6.5, Qr1 = 7.0, Qp1 = 2, Qt1 = 5, which are normalized to 0.65, 0.70, 0.20, and 0.55 respectively;

[0113] Field 3: Qs3 = 8.2, Qr3 = 8.0, Qp3 = 2, Qt3 = 4, which are normalized to 0.82, 0.80, 0.20, and 0.44 respectively;

[0114] Field 4: Qs4 = 5.9, Qr4 = 6.0, Qp4 = 1, Qt4 = 2, which are normalized to 0.59, 0.60, 0.10, and 0.22 respectively;

[0115] Substitute the above data into the calculation, and the steps are as follows:

[0116] Field 1:

[0117] Field 3:

[0118] Field 4:

[0119]

[0120] The results indicate that the time dispersion of the rating field, DQ≈0.2538, exceeds the upper limit of the preset interval for identifying abnormal rating fluctuations, which is 0.2. Based on the archiving of historical rating data, this upper limit is calculated from the 95% confidence interval of the rating behavior of a large-scale sample in the time axis offset distribution. The current calculated value is outside this interval, indicating that the rating offset field exhibits a relatively concentrated and abrupt arrangement structure in the time series, suggesting that there is a batch deviation in ratings or a phased tilt in rating behavior. This upward fluctuation in time dispersion is directly recorded as a fluctuation trigger in the rating data offset sequence. Furthermore, this value is associated and mapped with the rating item type, the identity of the responsible person, and the node status to form the behavioral tracing factor for the subsequent responsibility attribution mapping results.

[0121] Please see Figure 6 The responsibility attribution determination module includes:

[0122] The identity verification submodule analyzes the relationship between the identity of the responsible party at each node and the maintenance report filling time and acceptance status based on the scoring data offset sequence. It compares the chronological relationship between the scoring time and the report time, filters nodes whose scoring time is earlier than the report filling time, and determines whether the node corresponding to the scoring record has been accepted, thus obtaining the responsibility time mismatch record.

[0123] For each rating record, extract the node number, rating time, and the account of the responsible party. Then, access the maintenance report information of the corresponding node, read the report completion time field and acceptance status field of the record under that node, and perform a comparison operation between the two types of times for each node. That is, compare the rating time with the report time in chronological order. If the timestamp corresponding to the rating time is earlier than the timestamp corresponding to the report completion time, mark the node as a time mismatch record and record the mismatch direction as "early rating". If the rating time is equal to or later than the report completion time, mark it as "normal" or "delayed rating". On this basis, continue to read the acceptance status of the node. If the status field value is "not accepted" or "pending acceptance", it constitutes a logical conflict with the case where the rating time is early. Nodes whose scoring time is earlier than the report time and whose acceptance status is "not accepted" are extracted separately. The extracted records are then numbered and organized, and the corresponding responsible person's identity field is read to determine whether their identity is the work order's responsible party identifier. For example, if the responsible person's account is "eng093" and the role is "supplier", then the value is compared with the responsible role field registered in the node. If they match completely, the record is recorded as a normal identity consistent node. If they do not match or the role is missing, the record is recorded as an abnormal identity node. All scoring records are processed in sequence and the above time comparison and status judgment actions are performed. Finally, the node entries whose scoring time is earlier than the maintenance report filling time, whose acceptance status is not passed, or whose responsible person's identity is inconsistent are filtered out and output in a structured form as responsibility time mismatch records.

[0124] The item filtering submodule compares the scoring items corresponding to the field numbers with the node numbers and the responsible personnel accounts based on the records of mismatched responsibility time. It analyzes the correspondence between the identity of the person in charge of the scoring field and the identity of the party in charge of the node, determines whether the role identity data is complete, and obtains the responsibility anomaly screening results.

[0125] Extract the field number, node number, and responsible person account for each rating entry. Then, compare the rating field name with the responsible party identity field of each node. First, read the rating field entries associated with the field number, and locate the responsible person identity field within each entry. For example, the responsible person account for the rating field "Service Quality" is "sup_015," corresponding to the role of "Supplier." Next, read the responsible party account and identity information registered in the work order node information for the node number containing that field. For example, the responsible person for node number 7 is "sup_015," and their identity is "Supplier." Compare the two identity data. If the account and identity type are completely identical, the identity data is considered complete. If the account is inconsistent or the identity is missing, the data is considered incomplete. Finally, check if the field number exists. If a field number is found to correspond to multiple node numbers, and the field number "F008" corresponds to both node 7 and node 9, with the responsible accounts for the two nodes being "sup_015" and "eng_012" respectively, it indicates that there is a conflict of identity attribution for this scoring field across different nodes. This situation is recorded as a multi-node responsibility conflict entry, and the field number is marked. Continue to filter all records where the identity of the responsible person for the scoring field does not match the identity of the responsible party at the node, where the identity is missing, or where the record appears repeatedly across nodes. Compile the field number, node number, scorer account, node responsible account, identity matching status, and role integrity judgment result for each entry, and mark the screening status as "abnormal" or "normal". Construct a responsibility anomaly screening result for all entries identified as abnormal.

[0126] The attribution mapping submodule integrates the data corresponding to field numbers, node numbers, and responsible personnel accounts based on the results of the responsibility anomaly screening. It optimizes the correlation between scoring time, scoring field content, and node responsible person information, and sorts and aggregates node and account information to obtain the responsibility attribution mapping results.

[0127] First, the field number, node number, and responsible personnel account are read from each exception record. A number mapping index table is created, combining the field number and node number into a composite primary key. A relational mapping structure is established using the responsible personnel account as the index field. Then, the rating time and rating field name are extracted from each rating record. The rating time is compared with the service completion time field in the node record to confirm whether the rating action falls at the end of the process. The rating time is then formatted and used as the time base field for sorting. Finally, the rating field content is cross-matched with the node responsible party record to ensure a correspondence between field items and responsible identities. For example, if the field is... "Response speed" should be attributed to "dispatcher". If the current record of the responsible account role is "customer", it is marked as a mismatch. After the matching verification is completed, the node number and the responsible account field are aggregated. The responsible account associated with multiple field numbers is merged into the account set under the same node. The nodes are sorted in ascending order according to the node order field to form a node responsible person mapping sequence. The mapping result of each node records the total number of rating fields, the number of abnormal fields and the set of responsible accounts. Finally, the above information is integrated into the responsibility attribution mapping result, and the corresponding node number, responsible account, associated field number and time order position of each record are output.

[0128] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A multi-dimensional after-sales service evaluation and scoring system, characterized in that, The system includes: The responsibility description comparison module analyzes the names of responsible persons and the order of event nodes based on the audio and text descriptions of work orders uploaded by users, verifies the consistency of the conclusion content, identifies nodes with inconsistent descriptions as abnormal nodes, and obtains a responsibility conflict feature set. Based on the aforementioned responsibility conflict feature set, the node process advancement module screens work order task numbers, compares check-in time with the end time of the previous node, determines whether identity information and status identifier correspond, advances process flow, and obtains process flow sequence. The scoring dimension generation module extracts service type and urgency based on the process flow sequence, constructs parameter structure, matches scoring rule library indicators, and adjusts scoring items in combination with the identity of the responsible party and node status to obtain multi-scenario scoring indicators. The scoring data acquisition module analyzes the differences between the scoring fields and the standard scores based on the multi-scenario scoring indicators, determines the correspondence between the actual scores and the standards, records the differences in node scores, and organizes the scoring time to obtain the scoring data offset sequence.

2. The multi-dimensional after-sales service evaluation and scoring system according to claim 1, characterized in that, The responsibility conflict feature set includes conflict node identifier, responsibility attribution type, and content consistency identifier; the process flow sequence includes node sequence code, process advancement identifier, and node processing status; the multi-scenario scoring indicators include evaluation dimension marker, scoring item category, and applicable identity label; and the scoring data offset sequence includes score fluctuation identifier, scoring anomaly point, and data time index.

3. The multi-dimensional after-sales service evaluation and scoring system according to claim 1, characterized in that, The responsibility description comparison module includes: The node sequence analysis submodule compares the order of each node in the audio with the time sequence of the nodes in the text based on the audio and text descriptions of the work orders uploaded by the user. It integrates the name, processing stage and occurrence order of each node, calculates the correspondence between node time and order, and determines the distribution of each node number and tag to obtain node sequence comparison data. The responsibility difference analysis submodule analyzes the spelling of names, job numbers and organization numbers of nodes in the same order based on the node sequence comparison data, calculates the number of characters with name differences, counts the correspondence between job numbers and organization numbers, determines the offset status of account data, and obtains the degree of difference in responsibility fields. The consistency analysis submodule filters keywords and sentence order in the conclusion description based on the degree of difference in the responsibility field, judges the keyword overlap rate and description completeness, identifies nodes with missing or incorrect keyword order, and integrates the differences between node number and responsibility field to obtain a responsibility conflict feature set.

4. The multi-dimensional after-sales service evaluation and scoring system according to claim 1, characterized in that, The node process advancement module includes: The task identification submodule retrieves the work order database based on the responsibility conflict feature set, matches the work order number and node responsibility type, performs a set encoding difference comparison between each conflict node data and the registered task number in the work order database, filters out unregistered task numbers, and obtains a set of unrecorded task numbers. The time series comparison submodule, based on the uncollected task number set, obtains the sign-in time of the dispatching engineer and the end time of the previous node for each number, compares the sign-in time and end time, and calculates the node time difference by combining the responsible person's identity code and sign-in location data, and marks each node to obtain the node time identifier sequence. The workflow confirmation submodule extracts the node responsible person's account and the operator's identity based on the node time identifier sequence, compares the two types of identity data item by item, obtains the dispatch status field and the sign-in confirmation field in sequence, compares their field contents, gradually filters nodes with completely consistent contents, and integrates the node sequence code and processing status to obtain the workflow sequence.

5. The multi-dimensional after-sales service evaluation and scoring system according to claim 1, characterized in that, The scoring dimension generation module includes: The service parameter extraction submodule analyzes the node sequence code and node processing status based on the process flow sequence, compares the service type and urgency of each node, constructs joint parameters by integrating service type tags and urgency levels, and determines the service type and urgency combination data to obtain the service combination parameter set. The indicator matching calculation submodule determines the correspondence between the categories of scoring items and identity tags in the scoring rule library based on the service combination parameter set, calculates the fitting range of scoring items, filters the categories of scoring items, and obtains the scoring matching indicators. The dimension expansion construction submodule, based on the scoring matching index and combined with the current node processing status code, retrieves the additional scoring dimension items associated with the scoring rule library, determines the matching situation between the identity label and the node status, and integrates the additional dimension parameters into the existing index to obtain multi-scenario scoring indexes.

6. The multi-dimensional after-sales service evaluation and scoring system according to claim 1, characterized in that, The scoring data acquisition module includes: The field difference identification submodule, based on the multi-scenario scoring indicators, extracts the actual entered score and the standard score of the scoring rule according to each scoring field under the work order node, calculates the difference between the two under the corresponding scoring item, and identifies the directionality of the difference to obtain the field difference sequence. The offset point extraction submodule compares the number of differences with the identification benchmark content configured in the scoring rules based on the field difference sequence, filters the scoring fields whose number of differences exceeds the benchmark content, marks the node where the scoring field is located, the scoring item category and its corresponding offset performance, and obtains the scoring offset node group. The time index integration submodule collects scoring time information based on the scoring offset node group, extracts the time arrangement of the node where the scoring item is located, judges the fluctuation of adjacent offset scoring fields in time sorting, calculates the time dispersion of scoring fields, judges the distribution change of data offset in the time series, and obtains the scoring data offset sequence.

7. The multi-dimensional after-sales service evaluation and scoring system according to claim 1, characterized in that, The system also includes: The responsibility attribution determination module determines the identity of the responsible party at each node, the maintenance report, and the acceptance status based on the offset sequence of the scoring data. It then summarizes the scoring items with attribution characteristics, integrates the associated numbers and accounts, and obtains the responsibility attribution mapping result. The responsibility attribution mapping result includes the attribution node code, associated responsible person information, and traceability judgment label.

8. The multi-dimensional after-sales service evaluation and scoring system according to claim 7, characterized in that, The responsibility attribution determination module includes: The identity verification submodule analyzes the relationship between the identity of the responsible party at each node and the maintenance report filling time and acceptance status based on the scoring data offset sequence, compares the chronological relationship between the scoring time and the report time, filters nodes whose scoring time is earlier than the report filling time, and determines whether the node corresponding to the scoring record has been accepted, thus obtaining the responsibility time mismatch record. The item filtering submodule compares the scoring items corresponding to the field numbers with the node numbers and the responsible personnel accounts based on the responsibility time mismatch records. It analyzes the correspondence between the responsible person's identity in the scoring field and the responsible party's identity in the node, determines whether the role identity data is complete, and obtains the responsibility anomaly screening results. Based on the results of the responsibility anomaly screening, the attribution mapping submodule integrates the data corresponding to the field number, node number, and responsible personnel account, optimizes the correlation between the scoring time, scoring field content, and node responsible person information, and sorts and aggregates the node and account information to obtain the responsibility attribution mapping result.

Images