Small and medium-sized enterprise information management method and system based on multilayer architecture

Abstract

The invention discloses a small and medium-sized enterprise information management method and system based on a multi-layer architecture, and belongs to the technical field of information management, and the method comprises the steps: constructing a decoupling architecture of a user interface layer, a business logic layer and a data storage layer; analyzing the evaluation standard through a dynamic rule engine of a business logic layer, and generating a configurable rule unit comprising a logic expression and a dynamic weight parameter; an offline priority strategy is adopted at the mobile terminal, and a vector clock identifier containing a client ID and a logic clock sequence is added to a data record; executing dynamic weighted evaluation calculation to generate a total evaluation score; and performing industry reference threshold verification on the numeric field. The technical bottleneck of tight coupling of multiple layers of modules in a traditional single framework is broken through by constructing a decoupling framework of the user interface layer, the service logic layer and the data storage layer and an asynchronous message bus communication mechanism, so that cross-layer codes do not need to be synchronously modified when the system adds or adjusts the evaluation dimension, and the system reliability is improved. And the framework flexibility and the expansion efficiency are obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of information management, and in particular to a method and system for managing information of small and medium-sized enterprises based on a multi-layer architecture. Background Art

[0002] In the digital management of small and medium-sized enterprises, the architectural flexibility and adaptability of information systems and their evaluation functions are core technical challenges that hinder efficient enterprise operations. Existing solutions generally face the following technical bottlenecks when dealing with diverse evaluation scenarios: Traditional enterprise information management systems often employ a monolithic architecture or a traditional three-tier architecture (such as the MVC model), deeply integrating the user interface, business logic, and data storage. While some systems incorporate APIs for inter-layer communication, they lack the dynamic decoupling mechanisms of asynchronous message buses. Adding or adjusting evaluation dimensions requires simultaneous modifications to the front-end interaction logic, back-end business code, and database structure.

[0003] Existing systems often implement evaluation logic through hard-coding. While some solutions utilize rule engines (such as Drools) for basic rule configuration, rule parameters (such as indicator thresholds and scoring weights) lack a mechanism to link with dynamic industry data. For example, when industry regulators adjust evaluation standards, traditional systems require manual modification of rule files and restarting services. However, small and medium-sized enterprises often lag behind in rule updates due to insufficient technical capabilities.

[0004] While some mainstream client-server architecture systems use local caches (such as SQLite) to temporarily store data when the mobile network is interrupted, they lack distributed version control mechanisms (such as vector clocks). This results in data conflict rates as high as 28% when offline modifications are synchronized with online data. For example, if a network failure occurs during field data collection, already entered information may be lost due to version confusion.

[0005] Existing systems often display assessment data in static tables. While some tools support basic chart generation, they lack the ability to dynamically correlate indicators (e.g., linking them to real-time operational data). Business managers must manually integrate data from multiple sources, with a single analysis taking approximately three hours. Furthermore, they are unable to pinpoint the real-time correlation between assessment indicators and operational weaknesses. Summary of the Invention

[0006] In view of the above shortcomings of the prior art, the present invention provides a method and system for managing information of small and medium-sized enterprises based on a multi-layer architecture.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for information management of small and medium-sized enterprises based on a multi-layer architecture, comprising: A decoupled architecture is constructed for the user interface layer, business logic layer, and data storage layer, with each layer communicating via an asynchronous message bus. The dynamic rule engine in the business logic layer parses evaluation criteria and generates configurable rule units containing logical expressions and dynamic weight parameters. An offline-first strategy is adopted on mobile devices, with vector clock identifiers containing the client ID and logical clock sequence attached to data records. Perform dynamic weighted evaluation calculations, adjust indicator weights based on the technology lifecycle data in the industry knowledge graph, and generate a total evaluation score. When data conflicts are detected, perform industry benchmark threshold verification on numeric fields and perform semantic similarity merging on text fields.

[0008] Preferably, the parsing process of the dynamic rule engine is to convert the evaluation criteria into a structured rule file, which includes the unique code of the indicator, the logical expression and the weight calculation rule; construct an abstract syntax tree through the general syntax parsing framework to generate an executable verification instruction set; the dynamic weight parameter is calculated according to the formula weight = benchmark weight × (1 + technology iteration rate) (当前年度-基准年) Calculation, where the technology iteration rate is obtained by weighting the industry technology growth rate and the R&D investment growth rate.

[0009] Preferably, the method for generating the vector clock identifier includes: creating a version chain for each data record when the mobile terminal is offline; incrementing the logical clock sequence value each time a local modification is made; binding the client ID to the server time base to generate a unique space-time coordinate in the format of client ID: logical sequence@timestamp.

[0010] Preferably, the industry benchmark threshold verification is to call the industry data API to obtain the industry standard value; if the conflicting value deviates from the standard value by ±15%, a manual review process is triggered; and the maximum optimization strategy is adopted for the compliance conflicting value.

[0011] Preferably, the semantic similarity merging includes: using the BERT model to generate a word embedding vector for the text field; calculating the cosine similarity and retaining semantic segments with a comprehensive score greater than 0.7; and performing keyword deduplication on duplicate content.

[0012] The system of the small and medium-sized enterprise information management method based on the multi-layer architecture includes: The user interface layer integrates a dynamic form engine and a visual analysis module, supporting multi-terminal data entry and real-time chart linkage; The business logic layer includes a rule parsing module, a conflict resolution module, and a dynamic evaluation module. Each module implements asynchronous communication through a message queue. The data storage layer uses a relational database to store enterprise information, an in-memory database to cache evaluation rules, and a version log library to record data change trajectories; The containerized deployment module adapts to the container management platform through an abstract orchestration interface.

[0013] Preferably, the rule parsing module includes: a syntax analysis unit, which defines the rule expression structure based on context-free grammar; a weight calculation unit, which dynamically associates the technology evolution data in the industry knowledge graph; and a real-time loading unit, which makes the rule modification take effect within 300ms without the need for service restart.

[0014] Preferably, the conflict resolution module includes: a vector clock management unit, which generates a version identifier containing a client ID and a logical clock sequence; a numerical arbitration unit, which integrates an industry database interface to perform threshold verification; a text fusion unit, which uses a word embedding model to calculate semantic similarity and perform keyword extraction; and a containerized deployment unit, which connects to a container management platform through an abstract orchestration interface.

[0015] Preferably, the containerized deployment module includes: an elastic scaling controller that dynamically adjusts the number of service instances according to the CPU / memory load; a configuration management center that implements hot updates of environment variables and version rollbacks; and a service mesh agent that manages communication links and traffic distribution between microservices.

[0016] Preferably, the visualization analysis module includes: an indicator correlation analysis unit that dynamically binds the evaluation results with the business data; an event response unit that displays the corresponding original voucher and improvement suggestions when the user selects a chart element; and an adaptive rendering engine that automatically switches between Canvas and SVG rendering modes according to the terminal type.

[0017] Compared with the prior art, the present invention has at least the following benefits: 1. The present invention breaks through the technical bottleneck of the tight coupling of multi-layer modules in the traditional monolithic architecture by constructing a decoupled architecture of the user interface layer, business logic layer and data storage layer and an asynchronous message bus communication mechanism. This allows the system to add or adjust evaluation dimensions without the need to synchronously modify cross-layer code, significantly improving the flexibility and expansion efficiency of the architecture. The dynamic rule engine parses the evaluation criteria into configurable rule units, dynamically adjusts the indicator weights in combination with the industry knowledge graph, changes the static defects of traditional hard-coded rules, and realizes the real-time linkage between the evaluation logic and industry policies and technological evolution, solving the problem of disconnection between evaluation results and standards caused by lagging rule updates. The mobile terminal offline priority strategy implements data version chain management through vector clock identification, ensuring the continuity of data entry during network interruption and conflict traceability during synchronization, avoiding the pain points of offline data loss and duplicate entry in the traditional C / S architecture. The industry benchmark threshold verification of numerical fields and the semantic similarity merging mechanism of text fields build a conflict resolution system covering structured and unstructured data, improving the accuracy and consistency of enterprise information management.

[0018] 2. The dynamic rule engine achieves refined configuration of rule parameters and deep integration of dynamic industry data through structured rule file parsing, abstract syntax tree construction, and dynamic weight formulas. The unique spatiotemporal coordinate identification method of the mobile vector clock provides accurate version traceability for data changes. The application of the BERT model and cosine similarity algorithm in semantic similarity merging enhances the intelligent level of text data conflict resolution. The containerized deployment module enhances the system's adaptability to high-concurrency scenarios and operational efficiency through elastic scaling and configuration hot update mechanisms. The synergistic effect of these technical features forms a highly efficient, flexible, and intelligent information management solution for small and medium-sized enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0020] Figure 1 This is a multi-index combination diagram of the performance comparison experiment of the present invention; Figure 2 A dual-axis line graph for threshold verification of the present invention; Figure 3 It is a schematic diagram of the decoupling of the multi-layer architecture of the present invention; Figure 4 This is a workflow framework diagram of the dynamic rule engine of the present invention; Figure 5 This is a timing diagram of mobile terminal offline data processing according to the present invention; Figure 6 This is a module composition diagram of the information management system for small and medium-sized enterprises based on a multi-layer architecture of the present invention. DETAILED DESCRIPTION

[0021] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0022] Example 1: A method for managing information of small and medium-sized enterprises based on a multi-layer architecture, comprising: Build a decoupled architecture for the user interface layer, business logic layer, and data storage layer. Each layer communicates through an asynchronous message bus based on a message queue (such as RabbitMQ), achieving asynchronous processing of requests and state decoupling between layers. The dynamic rule engine of the business logic layer parses the evaluation criteria and generates configurable rule units containing logical expressions, dynamic weight parameters, and unique indicator codes. The dynamic weight parameters are linked in real time with the technology lifecycle data in the industry knowledge graph. On the mobile side, an offline-first strategy is adopted, and a vector clock identifier is added to each data record. The identification format is client ID: logical clock sequence@timestamp; The logical clock sequence is generated by an atomic increment algorithm, and the timestamp is synchronized to the server time reference based on the NTP protocol; Perform dynamic weighted evaluation calculations based on the technology iteration rate in the industry knowledge graph (obtained by weighting the industry patent growth rate and R&D investment growth rate). When data conflicts are detected, call the industry data API for numeric fields to obtain industry standard values ​​and perform a ±15% deviation check. For text fields, use the BERT model to generate word embedding vectors, calculate cosine similarity, and retain semantic segments with a comprehensive score greater than 0.7.

[0023] The parsing process of the dynamic rule engine is to convert the evaluation criteria into a structured rule file, which contains the unique code of the indicator, the logical expression and the weight calculation rules; construct an abstract syntax tree through the general syntax parsing framework to generate an executable verification instruction set; the dynamic weight parameter is calculated according to the formula weight = benchmark weight × (1 + technology iteration rate) (当前年度-基准年) Calculation, where the technology iteration rate is obtained by weighting the industry technology growth rate and the R&D investment growth rate.

[0024] The method for generating a vector clock identifier includes: creating a version chain for each data record when the mobile terminal is offline; incrementing the logical clock sequence value each time a local modification is made; and binding the client ID to the server time base to generate a unique spatiotemporal coordinate in the format of client ID: logical sequence@timestamp.

[0025] The industry benchmark threshold verification is to call the industry data API to obtain the industry standard value; if the conflicting value deviates from the standard value by ±15%, a manual review process is triggered; and the maximum optimization strategy is adopted for the compliance conflicting value.

[0026] The semantic similarity merging includes: using the BERT model to generate word embedding vectors for text fields; calculating cosine similarity and retaining semantic segments with a comprehensive score greater than 0.7; and performing keyword deduplication on duplicate content; The semantic similarity merging includes: Use the BERT model fine-tuned based on industry corpora (corporate annual reports, technical texts) to generate 768-dimensional word embedding vectors for text fields; The semantic similarity of text segments is calculated using the cosine similarity formula, and segments with a similarity greater than 0.7 are retained. This threshold is tested and verified by 50,000 sample data, achieving an optimal balance between semantic retention rate (82.1%) and false positive rate (7.2%) (e.g. Figure 2 shown); The top 5 keywords are extracted from duplicate content to generate a combined summary. Keyword extraction uses the TF-IDF algorithm combined with the industry terminology library. Embodiment 2: A system for the small and medium-sized enterprise information management method based on a multi-layer architecture according to embodiment 1, comprising: The user interface layer integrates a dynamic form engine and a visual analysis module, supports data entry from multiple terminals (Web / Android / iOS), and implements real-time chart linkage via WebSocket. The chart rendering engine automatically switches between Canvas (mobile) and SVG (PC) modes based on the terminal type. The business logic layer includes a rule parsing module, a conflict resolution module, and a dynamic evaluation module. Each module implements asynchronous communication through the Kafka message queue. The rule parsing module supports hot rule updates within 300ms without requiring a service restart. The data storage layer uses a MySQL relational database to store enterprise information, a Redis in-memory database cache evaluation rule (cache validity period is 1 hour), and a version log library based on MongoDB's vector clock trajectory to record data changes. The containerized deployment module adapts to the Kubernetes platform through the DockerCompose abstract orchestration interface, integrates the elastic scaling controller (based on HPA automatic expansion and contraction), the configuration management center (supports SpringCloudConfig hot updates) and the service mesh agent (Istio manages microservice communication) to achieve dynamic scheduling of system resources.

[0027] The rule parsing module includes: a syntax analysis unit that defines the rule expression structure based on context-free grammar; a weight calculation unit that dynamically associates the technology evolution data in the industry knowledge graph; and a real-time loading unit that makes rule modifications take effect within 300ms without the need for service restart.

[0028] The conflict resolution module includes: a vector clock management unit that generates a version identifier containing a client ID and a logical clock sequence; a numerical arbitration unit that integrates an industry database interface to perform threshold verification; a text fusion unit that uses a word embedding model to calculate semantic similarity and perform keyword extraction; and a containerized deployment unit that connects to a container management platform via an abstract orchestration interface.

[0029] The containerized deployment module includes: an elastic scaling controller that dynamically adjusts the number of service instances based on CPU / memory load; a configuration management center that implements hot updates of environment variables and version rollbacks; and a service grid agent that manages communication links and traffic distribution between microservices.

[0030] The visualization analysis module includes: an indicator correlation analysis unit that dynamically binds evaluation results with business data; an event response unit that displays corresponding original documents and improvement suggestions when a user selects a chart element; and an adaptive rendering engine that automatically switches between Canvas and SVG rendering modes according to the terminal type.

[0031] To address the problems of poor scalability and delayed rule updates in traditional monolithic architectures, this solution discloses a decoupled layered architecture. This architecture is divided into a user interaction layer, a business logic layer, and a data storage layer, and loosely coupled interaction between layers is achieved through an asynchronous message communication mechanism.

[0032] The user interaction layer includes a dynamic form generation module that automatically loads a set of fields based on the assessment type (e.g., professional assessment, innovation capability assessment, etc.), supporting numeric, text, and enumeration inputs. Field rendering rules are stored in the business logic layer, and configuration changes are synchronized in real time via the message bus.

[0033] Furthermore, the architecture incorporates a multi-terminal adaptation engine, employing a responsive design that automatically adjusts layout and interaction logic based on the terminal type (web, mobile). In offline mode, the engine enables local caching to ensure data entry continuity during network outages.

[0034] The dynamic rule processing engine is responsible for parsing structured rule files, generating executable instruction sets, and supporting hot updates of rules, thereby ensuring that evaluation criteria can take effect in real time after policy adjustments.

[0035] In terms of task processing, the assessment task is split into multiple subtasks and processed in parallel. The system allocates computing nodes through a consistent hashing algorithm and integrates a dynamic weighting model to achieve industry-aware calibration of assessment results, ensuring efficient and accurate assessments.

[0036] At the data storage layer, a multi-mode storage system is employed to meet diverse data structures and access requirements. The structured storage component is used to store basic enterprise information, evaluation rules, and historical records. The cache component employs a least-repeated (LRU) eviction strategy to cache frequently accessed rules and industry benchmark data, with a caching period set to one hour. The version control component records the spatiotemporal coordinates of data changes, supporting version rollback and conflict analysis to ensure data integrity and traceability.

[0037] The core of the dynamic rule processing mechanism lies in the structured representation of rules. This mechanism converts natural language evaluation criteria into machine-parseable structured rule units, which include the following elements: Indicator identification, clearly defining the evaluation indicators corresponding to the rules; Logical expressions define the conditions that the evaluation indicators must meet. They support a combination of relational operators (>, <, =) and logical operators (&&, ||), for example, value>=0.05; Base weight is the basic weight value set for the rule, used for initial evaluation calculation; Industry impact parameters include factors related to industry characteristics, such as technology iteration factors and base years. The technology iteration factors are calculated through industry data analysis, and the base year is used to calibrate the time validity of the parameters. For example, in a specific embodiment, for the evaluation indicator of the proportion of R&D investment, the corresponding structured rule can be expressed as the indicator identifier is the proportion of R&D investment, the logical expression is value>=0.05, the benchmark weight is 15, and the industry impact parameters include the technology iteration factor 0.12 (calculated through industry data analysis) and the benchmark year (for example, 2024).

[0038] The dynamic weight adjustment algorithm is an important component of the dynamic rule processing mechanism. It is used to dynamically adjust the weights of evaluation indicators based on industry characteristics, thereby ensuring that the evaluation results can more accurately reflect changes in the current industry environment. Its core calculation formula is:

[0039] Among them, W base is the benchmark weight of the indicator, usually determined by expert evaluation, F tech It is the technology iteration factor, which is used to quantify the impact of technology development on the indicator weight. It is calculated as follows:

[0040] Among them, ΔP and ΔR are the industry technology growth rate and R&D investment growth rate in the past three years respectively; P base and R base is the base value of the number of patents and R&D investment in the base year, Y curr and Ybase They are the current year and the base year; This dynamic weight adjustment algorithm allows the weights of evaluation indicators to be adjusted in real time based on dynamic factors in industry technological development. In actual testing within the semiconductor industry, applying this dynamic weight adjustment model significantly increased the correlation coefficient between evaluation results and industry competitiveness rankings, from 0.58 to 0.91 (p<0.01). This demonstrates that the dynamic weight adjustment algorithm can significantly improve the accuracy and timeliness of evaluation results, better adapting to the rapidly changing realities of the industry.

[0041] To further optimize the flexibility and stability of the dynamic rule processing mechanism, this solution exposes a rule hot update function to ensure rapid response and real-time effectiveness when the evaluation criteria change. The system uses an event-driven mechanism to implement rule hot updates. When a rule is modified, the system immediately triggers a rule change event and broadcasts the event to all relevant computing nodes via the message bus. After receiving the rule change event, each node must complete the update of the rule cache in memory within 300 milliseconds to ensure that the new rule can take effect in time and be applied to subsequent evaluation tasks. To reduce risks during rule updates, the system also introduces a grayscale release strategy. New rules are first tested online on 10% of computing nodes. After verifying the accuracy and stability of the new rules on this small number of nodes, the scope is gradually expanded, and eventually the full version is pushed to all nodes. This strategy can effectively avoid global problems caused by rule updates and ensure the stable operation of the system during the update process.

[0042] By combining an event-driven mechanism with a grayscale release strategy, the hot rule update function can not only quickly respond to rule changes, but also achieve real-time updates and effectiveness of evaluation criteria while ensuring system stability.

[0043] To effectively resolve data conflicts in distributed systems, this solution discloses a data conflict resolution strategy, covering distributed version control and conflict handling mechanisms for numeric and text data, ensuring data accuracy and consistency. The system uses a distributed version control mechanism to track the spatiotemporal coordinates of data changes. The specific identification method is as follows: Unique device identifier, a cryptographic hash value generated based on the hardware fingerprint (MAC address + CPU ID); Logical clock sequence, using AtomicLong to implement atomic increment, with an initial value of 0.

[0044] Timestamp, physical time synchronized through NTP, with millisecond accuracy.

[0045] Identification format, the device hash value is the logical clock@timestamp; For example, a1b2c3:42@1689234567890. This identification method can accurately record the source and time sequence of data changes, facilitating version backtracking and conflict analysis.

[0046] For numerical data conflicts, the following processing flow is adopted: Call the public API (HTTPS protocol) to obtain the industry standard value V industry , after removing outliers, take the average value of the past three years to ensure the reliability of the benchmark value; Calculate the conflict value V conflict Compared with the industry benchmark value V industry The deviation of:

[0047] According to the size of the deviation D, the following strategies are implemented: When D≤15%, select max(V conflict , V industry ), encourage data to meet standards; When D>15%, manual review is triggered, the interface automatically marks abnormal fields and pushes historical data comparison charts for further analysis and decision-making.

[0048] For text data conflicts, the following processing flow is adopted: We fine-tune the BERT model based on industry corpora (corporate annual reports and patent texts), using contrastive learning to improve the aggregation of similar texts. We use cosine similarity to quantify the semantic similarity between texts. Cosine similarity is calculated as:

[0049] Based on the semantic similarity, the following strategies are implemented: When sim≥0.7, the top 5 keywords are extracted to generate a summary, retaining the core semantics and ensuring the consistency of the text content.

[0050] When sim<0.7, the conflicting texts are displayed side by side, and the difference areas (such as changed content and operator information) are marked to facilitate user identification and conflict resolution.

[0051] Through the above-mentioned distributed version control mechanism and numerical and text data conflict handling strategies, the system can effectively solve data conflict problems in a distributed environment, ensure data accuracy and consistency, and improve system reliability and availability.

[0052] Experimental Example 1: Performance comparison test ( Figure 1 shown); The experimental hardware configuration is: CPU: Intel Xeon Gold 8370C, 8 cores, 2.60GHz main frequency, supports Turbo Boost technology, up to 3.50GHz; Memory: 32GB DDR4-3200 ECC server memory with error correction function to improve data processing reliability; Storage, 512GB NVMe SSD, sequential read speed up to 3500MB / s, sequential write speed up to 3000MB / s, providing high-performance storage solutions; Network, 1Gbps fiber optic network interface, low latency, and high bandwidth ensure efficient data transmission.

[0053] The commercial terminal is a Dell Latitude 7420 laptop equipped with an 11th generation Intel Core i7 processor and 16GB of memory, simulating multiple network environments.

[0054] The network environment uses software to simulate a mixed 4G (LTE), 5G (NR), and Wi-Fi (802.11ac / ax) network environment, simulating user access scenarios under different network conditions.

[0055] The software configuration is: On the server side, CentOS Linux release 7.9.2009 (Core), kernel version 3.10.0-1160.81.1.el7.x86_64, is stable and widely used in server environments.

[0056] The client, Windows 10 Enterprise LTSC 2021, ensures compatibility with the server and supports various testing tools.

[0057] The test tools are: The stress testing tool, JMeter 5.4.3, is used to simulate 200-1000 concurrent users to test the performance and stability of the system.

[0058] The network simulation tool, NetEm (Network Emulator), is integrated into the Linux kernel and is used to simulate different network conditions (latency, packet loss rate, etc.).

[0059] Monitoring tools, Prometheus + Grafana, are used to monitor the server's CPU, memory, disk I / O, and network usage in real time.

[0060] The test environment is: 1. Install the CentOS 7.9 operating system, configure basic network and storage services, deploy the application to be tested, ensure all services start normally, install the Prometheus monitoring agent, and configure the Grafana dashboard for real-time monitoring of server performance indicators; 2. Install JMeter 5.4.3 on the client, configure the test plan to simulate different numbers of concurrent users, use the NetEm tool to simulate 4G, 5G, and Wi-Fi network environments, set different network parameters (such as latency, bandwidth, and packet loss rate), and install the necessary test scripts and tools to ensure that the client can communicate normally with the server; 3. Configure multiple virtual network interfaces on the server side, connect them to different network simulation environments, and use routing tables and iptables rules to ensure that traffic passes through the correct network path.

[0061] The experimental steps are: 1. Start the application on the server, initialize the test data set, use JMeter to simulate 200-1000 concurrent users, perform data read and write operations, and record and count the data consistency rate achieved by each solution within 1 hour; 2. Generate a preset amount of conflict data, simulate multiple users concurrently writing the same data, and record the average time required for each solution to resolve the conflict; 3. Modify the evaluation rules, trigger a rule change event, and record the P99 delay from the event triggering to the completion of the rule update on all nodes; 4. In high-concurrency scenarios, continuously perform data read and write operations and count the number of transactions processed per second for each solution.

[0062] Experimental Example 2: Threshold Verification Experiment ( Figure 2 shown); Experimental steps: 1. Collect 50,000 text data samples, covering different fields and semantic features, and annotate the semantic core and conflicts of each data; 2. Set the text similarity thresholds to 0.65, 0.70, and 0.75, respectively, calculate the semantic similarity and resolve conflicts for each sample data, and record and calculate the semantic retention rate and false positive rate under different thresholds; 3. Analyze the balance between semantic retention rate and false positive rate under different thresholds, and select the threshold that achieves the best balance between semantic retention rate and false positive rate as the final value.

[0063] like Figure 1 and Figure 2 As shown, through detailed test environment configuration, performance comparison experiments and threshold verification data, the solution of the present invention shows significant advantages in data consistency, conflict resolution efficiency, rule hot update delay and system throughput, and can effectively meet the high concurrency and high reliability requirements in practical applications.

[0064] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for information management of small and medium-sized enterprises based on a multi-layer architecture, characterized in that: include: Build a decoupled architecture for the user interface layer, business logic layer, and data storage layer, with each layer communicating through an asynchronous message bus; The dynamic rule engine of the business logic layer parses the evaluation criteria and generates configurable rule units containing logical expressions and dynamic weight parameters; On the mobile side, an offline-first strategy is adopted, and a vector clock identifier containing the client ID and the logical clock sequence is attached to the data record. Perform dynamic weighted evaluation calculations, adjust indicator weights based on technology lifecycle data in the industry knowledge graph, and generate a total evaluation score; When data conflicts are detected, industry benchmark threshold checks are performed on numeric fields, and semantic similarity merges are performed on text fields.

2. The method for managing information of small and medium-sized enterprises based on a multi-layer architecture according to claim 1, characterized in that: The parsing process of the dynamic rule engine is to convert the evaluation criteria into a structured rule file, which includes the unique code of the indicator, the logical expression and the weight calculation rules; Build an abstract syntax tree through a general syntax parsing framework to generate an executable verification instruction set; Dynamic weight parameter is calculated as follows: Weight = Baseline Weight × (1 + Technology Iteration Rate) (当前年度-基准年) Calculation, where the technology iteration rate is obtained by weighting the industry technology growth rate and the R&D investment growth rate.

3. The method for managing information of small and medium-sized enterprises based on a multi-layer architecture according to claim 1, characterized in that: The method for generating the vector clock identifier includes: When the mobile terminal is offline, a version chain is created for each data record; the logical clock sequence value is incremented each time a local modification is made; the client ID is bound to the server time base to generate a unique time and space coordinate.

4. The method for information management of small and medium-sized enterprises based on a multi-layer architecture according to claim 1, characterized in that: The industry benchmark threshold verification is to call the industry data API to obtain the industry standard value; If the conflict value deviates from the standard value, the manual review process will be triggered; the maximum optimization strategy will be adopted for the compliance conflict value.

5. The method for managing information of small and medium-sized enterprises based on a multi-layer architecture according to claim 1, characterized in that: The semantic similarity merging includes: using the BERT model to generate word embedding vectors for text fields; calculating cosine similarity, retaining semantic segments with a comprehensive score greater than 0.7; and performing keyword deduplication on duplicate content.

6. The system for the information management method for small and medium-sized enterprises based on a multi-layer architecture according to any one of claims 1 to 5, characterized in that: include: The user interface layer integrates a dynamic form engine and a visual analysis module, supporting multi-terminal data entry and real-time chart linkage; The business logic layer includes a rule parsing module, a conflict resolution module, and a dynamic evaluation module. Each module implements asynchronous communication through a message queue. The data storage layer uses a relational database to store enterprise information, an in-memory database to cache evaluation rules, and a version log library to record data change trajectories; The containerized deployment module adapts to the container management platform through an abstract orchestration interface.

7. The system for information management of small and medium-sized enterprises based on a multi-layer architecture according to claim 6, characterized in that: The rule parsing module includes: Syntax analysis unit, which defines the regular expression structure based on context-free grammar; Weight calculation unit, dynamically associates technology evolution data in the industry knowledge graph; Real-time loading unit, so that rule changes take effect within 300ms without service restart.

8. The system of the small and medium-sized enterprise information management method based on multi-layer architecture according to claim 6, characterized in that: The conflict resolution module includes: The vector clock management unit generates a version identifier containing the client ID and the logical clock sequence; Numerical arbitration unit, integrating industry database interface for threshold verification; The text fusion unit uses a word embedding model to calculate semantic similarity and perform keyword extraction; Containerized deployment unit, connected to the container management platform through an abstract orchestration interface.

9. The system for information management of small and medium-sized enterprises based on a multi-layer architecture according to claim 6, characterized in that: The containerized deployment module includes: An elastic scaling controller that dynamically adjusts the number of service instances based on CPU / memory load; Configuration management center to achieve hot update of environment variables and version rollback; Service mesh proxy manages communication links and traffic distribution between microservices.

10. The system for small and medium-sized enterprise information management method based on multi-layer architecture according to claim 6, characterized in that: The visual analysis module includes: Indicator correlation analysis unit dynamically binds assessment results with business data; Event response unit: When the user selects a chart element, the corresponding original document and improvement suggestions are displayed; Adaptive rendering engine, automatically switching between Canvas and SVG rendering modes according to the terminal type.

Images