A method and system for developing a software development platform

Abstract

The application belongs to the technical field of software development, and discloses a development method and system of a software development platform, which comprises receiving at least two candidate technical schemes for the same development task in the design stage of software development. The scheme constructs a closed-loop method and system of "quantitative prediction-mapping tracking-aggregated analysis-tracing attribution-optimization feedback", breaks through the barrier between technical decision and business value, predicts the long-term business impact of the scheme in the development stage, accurately attributes cost, performance and other data to specific code and decision in the running stage, and automatically generates data-driven optimization suggestions, so that software development is changed from periodic optimization depending on experience to continuous dynamic optimization based on real-time data, effectively solves the traditional pain points of disconnection between decision and effect, feedback lag and problem tracing difficulty, and significantly improves the research and development accuracy and resource efficiency.

Description

Technical Field

[0001] This invention relates to the field of software development technology, and in particular to a method and system for developing a software development platform. Background Technology

[0002] In current software engineering practices, the common development process follows a linear, phased model. Typically, the development team first conducts requirements analysis and technical solution design, then moves into the coding and testing phase, and finally deploys the software system online. After the system goes live, the operations team is responsible for monitoring its technical indicators (such as response time and system availability), while the business operations team obtains data such as user behavior and business conversion from independent business analysis tools. After a period of data accumulation (such as an iteration cycle or quarter), all parties analyze the data through retrospective meetings, summarize problems, and schedule new optimization requirements into subsequent development plans, thus forming a cyclical process in which development, operations, and maintenance are relatively separated.

[0003] However, the aforementioned existing technological models have obvious limitations: 1. Long and delayed feedback cycle: From the time a technical decision is made to the time its real business impact is quantified and verified, it often takes a complete development, deployment and operation data accumulation cycle, resulting in a serious delay in optimization decisions; 2. Disconnect between technical decisions and business results: Technical selections during the development phase (such as frameworks, algorithms, and architectures) are often based on local factors such as technical performance and team experience, lacking a direct and quantitative correlation with final business metrics (such as user retention and cost per unit of revenue), resulting in insufficient decision support. 3. Difficulty in tracing problems: When business metrics fluctuate or costs are abnormal online, it is difficult to quickly and accurately trace the root cause of the problem to a specific historical technical decision or code module. Attribution analysis relies on expert experience, which is inefficient.

[0004] Therefore, the industry urgently needs a new method to overcome the above-mentioned shortcomings. The ideal method needs to establish a direct, quantitative, and traceable feedback mechanism in the software development process, so that technical decisions can be made based on predictions of their long-term business impact and can be dynamically adjusted according to real-time online results, thereby fundamentally improving the accuracy of software development and the efficiency of business value output. Summary of the Invention

[0005] This invention aims to provide a software development platform development method and system to solve the problems mentioned in the background. This solution constructs a closed-loop method and system of "quantitative prediction - mapping tracing - aggregation analysis - source attribution - optimization feedback", breaking down the barriers between technical decisions and business value. By predicting the long-term business impact of the solution during the development phase, and accurately attributing cost, performance and other data to specific code and decisions during the operation phase, it automatically generates data-driven optimization suggestions. This transforms software development from experience-based periodic optimization to continuous dynamic optimization based on real-time data, effectively solving the traditional pain points of disconnect between decision-making and results, delayed feedback, and difficulty in tracing problems, and significantly improving the accuracy of R&D and resource efficiency.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for developing a software development platform includes the following steps: S1, during the design phase of software development, receive input of at least two candidate technical solutions for the same development task, and based on a pre-built multi-dimensional prediction model, perform quantitative business impact prediction on each candidate technical solution to generate prediction results that include at least two prediction indicators among performance, cost, and maintainability. S2, for the target technical solution selected during development based on the prediction results, establish a traceable association between the code module generated by implementing the target technical solution and the hardware resources called or deployed by the code module and the target technical solution, and generate a storage decision mapping relationship; S3. After the application based on the target technical solution is launched and running, collect its running performance data, business indicator data and resource cost data, and aggregate the collected data to the corresponding target technical solution and code module dimensions according to the decision mapping relationship. S4. Based on the aggregated dimensional data, monitor the performance of business metrics. When the performance deviates from expectations, trace back to the relevant target technical solutions and code modules according to the decision mapping relationship and perform attribution analysis. S5. Based on the results of the attribution analysis, generate optimization suggestions containing specific quantitative improvement expectations, and feed these optimization suggestions back to step S1 as new development decision inputs.

[0007] Preferably, the multi-dimensional prediction model in step S1 is constructed based on: data on the effects of technical decisions in historical projects, attribute data of components constituting the technology stack, cloud resource cost pricing data, and business growth assumption parameters.

[0008] Preferably, the generation of decision mapping relationships in step S2 specifically involves: constructing a decision knowledge graph, wherein the target technical solution is used as an entity node, and it is linked to entity nodes representing code files, API interfaces, database tables, or cloud resource instances through relational edges.

[0009] Preferably, step S3 specifically includes: Obtain raw data streams from application performance management tools, business databases, and cloud service provider billing interfaces; Using the aforementioned decision knowledge graph, the original data stream is parsed and classified, and resource consumption metrics, performance metrics, and business event metrics are collected and accumulated under their corresponding target technical solution entity nodes and code-level entity nodes.

[0010] Preferably, the attribution analysis in step S4 specifically includes: Calculate the deviation between the actual operating data corresponding to the target technical solution and its prediction result in step S1; Analyze the correlation between the deviation value and the fluctuation of business indicators, and locate at least one code module that causes the main deviation.

[0011] Preferably, the generation of optimization suggestions in step S5 specifically includes: Based on the attribution analysis results, candidate optimization schemes are matched from the pre-set optimization strategy library; The candidate optimization scheme is input into the multi-dimensional prediction model to perform simulation prediction and obtain simulation results; By comparing the simulation results with the current actual operating effects, an optimization suggestion text containing the predicted improvement range and investment return cycle is generated.

[0012] A software development platform system, comprising: The decision prediction module is used to perform the above S1 step and provides a quantitative prediction function for technical solutions; The mapping management module is used to perform the above S2 step, creating and maintaining decision mapping relationships; The data aggregation engine is used to execute the above S3 steps to classify and aggregate multi-source data according to mapping relationships; The intelligent analysis module is used to execute the above S4 steps to complete the effect tracing and attribution analysis; The feedback loop module is used to execute the S5 steps mentioned above, and is responsible for generating optimization suggestions and injecting feedback into the development process.

[0013] Preferably, the decision prediction module includes: A multi-dimensional knowledge base is used to store technical attributes, cost models, and historical pattern data; The prediction model unit calls upon the data from the multi-dimensional knowledge base to perform multi-objective quantitative prediction on the input technical solution.

[0014] Preferably, the data aggregation engine is connected to the mapping management module, and the data aggregation engine includes: An interface for collecting data from multiple data sources; The association calculation unit is used to perform real-time association calculation and aggregation storage on the collected data according to the decision mapping relationship provided by the mapping management module.

[0015] Preferably, the system further includes a visual interaction platform, which is connected to the intelligent analysis module and the feedback loop module. The visual interaction platform is used for: Using the target technical solution as a dimension, a comparison view of its predicted indicators and actual operating indicators is presented; The cost-benefit view of resource consumption and business contribution is presented at the code module level. Publish and track the processing status of the optimization recommendations.

[0016] The beneficial effects of this technical solution compared to existing technologies are as follows: (1) By introducing a multi-dimensional prediction model in the development and design phase, the long-term impact of candidate technical solutions on performance, cost, maintainability and key business indicators can be quantitatively evaluated, providing objective data support for decision-making and transforming technology selection from relying on local experience to scientific decision-making based on global business value.

[0017] (2) By constructing and maintaining a mapping graph of “decision-code-resource” and performing multi-source data aggregation in accordance with it during operation, the system can automatically and quickly locate the root cause of the problem to specific historical technical decisions and specific code modules when business indicators fluctuate, which greatly improves the efficiency and accuracy of problem diagnosis and attribution analysis.

[0018] (3) Based on accurate source analysis, the system can automatically generate optimization suggestions containing quantitative improvement expectations and directly feed them back to the development process, forming a continuous "monitoring-analysis-optimization" closed loop. This completely changes the situation of serious lag in optimization decision-making under the traditional model, enabling the software development process to be dynamically adjusted according to real-time business feedback, and continuously improving the system's cost-effectiveness and business adaptability. Attached Figure Description

[0019] Figure 1 A flowchart illustrating the development method of the software development platform provided by this invention; Figure 2 A schematic diagram of the composition of the software development platform system provided by the present invention. Detailed Implementation

[0020] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments: This invention provides a software development platform development method and system, aiming to solve core problems in traditional software development processes such as the disconnect between technical decisions and business results, delayed feedback, and difficulty in problem tracing. Its core idea is to construct a quantifiable and traceable dynamic optimization closed loop driven by business metrics throughout the entire software development lifecycle. The following will combine... Figure 1 (Flowchart of software development platform development methodology) and Figure 2 (A schematic diagram of the software development platform system) provides a detailed description of the specific implementation of the present invention.

[0021] Example 1: A method for developing a software development platform refer to Figure 1 The development method provided in this embodiment of the invention mainly includes the following five core steps (S1 to S5): S1: Development Decisions and Quantitative Forecasting During the design or architecture review phase of a software project, developers or architects can use the platform interface to input at least two candidate technical solutions for comparison and selection for a specific development task (e.g., "implementing intelligent image cropping functionality for users").

[0022] The platform's built-in multi-dimensional prediction model will automatically perform quantitative analysis on this. The model is built on a continuously updated multi-dimensional knowledge base, which integrates historical decision-making effect data, technical component attribute data, cloud resource cost pricing data, and business growth assumption parameters.

[0023] The prediction model integrates the above information and outputs a set of quantitative prediction metrics for each candidate solution, for example: Option A (using commercial image processing interfaces): The development cycle is predicted to be shortened by 70%, the monthly interface call cost will increase by a certain number of units, and the image processing success rate (business metric) is expected to be 98.5%.

[0024] Option B (using self-developed image processing algorithm): The development cycle is predicted to increase by 40%, the monthly cloud computing resource cost will increase by a certain amount, and the image processing success rate is expected to be 97.2%.

[0025] These forecasts provide data-driven, objective evidence for decision-making, directly linking technology selection to future business costs and outcomes.

[0026] S2: Code Association and Mapping Graph Construction After the development team selects the target technical solution (e.g., solution B) based on the prediction results of S1, it enters the development and implementation phase. In this invention, the development process needs to establish a traceable connection.

[0027] Association Establishment: When developers submit code, the platform will prompt them to associate this submission with the previously decided "Target Technology Solution ID". Alternatively, the platform can automatically identify and establish associations by analyzing the code repository's submission information, dependency files, and infrastructure configuration files.

[0028] Mapping Graph Generation: The core component of the platform—the mapping management module—will dynamically build and maintain a decision knowledge graph. In this graph, "image cropping using self-developed algorithms" is a decision entity node. This node is linked to specific image processing program module entity nodes through the "implemented in" relation edge, and to cloud server resource entity nodes through the "deployed in" relation edge. This graph constitutes the "skeleton" for all subsequent data tracing.

[0029] S3: Multi-dimensional aggregation of runtime data After the application goes live, it enters the operation monitoring and data collection stage. Traditional monitoring data is scattered, while the key to this invention is to aggregate it according to the decision-making dimension.

[0030] Data Acquisition: The data aggregation engine obtains raw data streams from various sources through its rich collection interfaces. 1. Obtain the number of calls and the time taken for the image processing module from application performance monitoring tools; 2. Obtain the utilization rate and cost details of relevant cloud server resources from the cloud service provider's billing interface; 3. Retrieve the event log (business metric) of "user successfully saved image after uploading" from the business database.

[0031] Data Association and Aggregation: The association calculation unit in the data aggregation engine calls the decision knowledge graph built in S2. Based on the graph, it classifies and accumulates all the scattered data such as "costs generated by cloud server resources", "performance data of image processing program module calls", and "successfully processed image business data" into the decision node "image cropping using self-developed algorithm" and its associated code module nodes. Finally, it forms an analytical dataset that integrates cost, performance and business results with "decision" and "code module" as dimensions.

[0032] S4: Effect Source Tracing and Attribution Analysis The platform continuously monitors key business metrics. For example, if operations staff find that the "success rate of saving user uploaded images" has dropped from the expected 97.2% to 95.5%.

[0033] Source tracing trigger: When the intelligent analysis module detects that the business indicator deviates from the preset threshold, it automatically starts source tracing analysis.

[0034] Attribution localization: The module first queries the aggregated dataset generated by S3 and quickly locates the decision node related to the "image saving" process - "using self-developed algorithm to implement image cropping". Then, it drills down to analyze the detailed data under this node and finds that the number of failed calls to the image processing program module has increased abnormally, and its utilization of computing resources remains high.

[0035] Deviation Comparison: The module further calculates the deviation between the actual data and the predicted value in the S1 stage: the actual success rate (95.5%) is lower than the predicted value (97.2%), and the actual resource consumption rate is higher than the estimation of the prediction model. The analysis shows that the main reason for the deviation may be that the image complexity in the actual production environment is higher than the assumptions made during model training, which causes the self-developed algorithm to fail or take too long when processing some images.

[0036] S5: Optimization suggestion generation and feedback closed loop Based on S4 attribution analysis, the system generates actionable optimization suggestions.

[0037] Suggested solutions: The feedback loop module matches candidate solutions from the optimization strategy library based on the root cause of "insufficient ability of the algorithm to process complex images", such as: "1. Optimize the existing algorithm; 2. Switch to a higher precision commercial image processing interface (i.e., solution A)".

[0038] Simulation prediction: The module takes the candidate solution (especially solution A) as new input, calls the multi-dimensional prediction model of S1 again, and performs simulation prediction in combination with the current actual traffic data. The prediction result shows: "Using solution A, the success rate is expected to be increased to 98.8%, but the monthly cost will increase by a certain amount." Feedback Generation: The platform generates a structured optimization suggestion: "To improve image processing success rate, it is recommended to evaluate switching to a commercial image processing interface. Predicted data: The success rate is expected to increase from 95.5% to 98.8%, with a corresponding increase in monthly costs and a return on investment period of approximately 4 months (due to reduced user churn). Related Decision: This suggestion stems from a review of 'adopting a self-developed image processing algorithm' (Decision ID: DP002)." This suggestion is published through a visual interactive platform and can be directly created as a new development task.

[0039] Closed-loop feedback: When the development team accepts this task and designs a technical solution for it, the task carries historical data and optimization goals and re-enters step S1, starting a new data-driven optimization cycle.

[0040] Example 2: A software development platform system refer to Figure 2 The system used to implement the above method mainly includes the following modules: 1. Decision Prediction Module: Responsible for executing S1, which includes a multi-dimensional knowledge base and prediction model units, and provides an interactive interface for users to input solutions and view quantitative prediction reports.

[0041] 2. Mapping Management Module: Responsible for executing S2. Its core is to build and maintain the decision knowledge graph database and provide APIs or plugins to integrate with development toolchains (such as code version management systems and integrated development environments) to capture and establish relationships.

[0042] 3. Data Aggregation Engine: Responsible for executing S3, including data acquisition interfaces (adapting to various performance monitoring, cloud platforms, and business data sources) and related computing units. This unit is the system's "connector," using real-time stream processing or batch processing to associate and aggregate raw data according to a knowledge graph.

[0043] 4. Intelligent Analysis Module: Responsible for executing S4, with a built-in attribution analysis engine, capable of configuring business indicator alarm rules, and automatically executing drill-down, correlation, and deviation analysis processes when triggered.

[0044] 5. Feedback Loop Module: Responsible for executing S5, including an optimization strategy library and suggestion generator, which can transform analysis results into specific development suggestions with quantitative evaluation.

[0045] 6. Visual Interactive Platform: As the unified portal of the system, it provides a decision-making benefit dashboard (comparing predictions with actual results), a module cost heatmap (identifying high-cost and low-efficiency modules), and an optimization suggestion workbench, serving as an interface for collaboration among development, maintenance, and operation roles.

[0046] In summary, this invention constructs a complete closed loop from "predictive decision-making" to "retrospective optimization" through five clearly defined steps in the method and six corresponding modules in the system. It tightly connects the originally separate development, operation and maintenance, and operational activities through data and models, so that every code commit can be associated with business value and every business fluctuation can be traced back to the technical root cause, thereby realizing a fundamental transformation of the software development process from "experience-driven" to "data-driven".

[0047] The above descriptions are merely embodiments of the present invention, and common knowledge such as specific technical solutions and / or characteristics are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the technical solutions of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for developing a software development platform, characterized in that, Includes the following steps: S1, during the design phase of software development, receive input of at least two candidate technical solutions for the same development task, and based on a pre-built multi-dimensional prediction model, perform quantitative business impact prediction on each candidate technical solution to generate prediction results that include at least two prediction indicators among performance, cost, and maintainability. S2, for the target technical solution selected during development based on the prediction results, establish a traceable association between the code module generated by implementing the target technical solution and the hardware resources called or deployed by the code module and the target technical solution, and generate a storage decision mapping relationship; S3. After the application based on the target technical solution is launched and running, collect its running performance data, business indicator data and resource cost data, and aggregate the collected data to the corresponding target technical solution and code module dimensions according to the decision mapping relationship. S4. Based on the aggregated dimensional data, monitor the performance of business metrics. When the performance deviates from expectations, trace back to the relevant target technical solutions and code modules according to the decision mapping relationship and perform attribution analysis. S5. Based on the results of the attribution analysis, generate optimization suggestions containing specific quantitative improvement expectations, and feed these optimization suggestions back to step S1 as new development decision inputs.

2. The development method of a software development platform according to claim 1, characterized in that, The multi-dimensional prediction model in step S1 is constructed based on: data on the effects of technical decisions in historical projects, attribute data of components constituting the technology stack, cloud resource cost pricing data, and business growth assumption parameters.

3. The development method of a software development platform according to claim 1, characterized in that, The specific steps in step S2 to generate the decision mapping relationship are as follows: construct a decision knowledge graph, in which the target technical solution is used as the entity node, and it is linked to entity nodes representing code files, API interfaces, database tables or cloud resource instances through relation edges.

4. The development method of a software development platform according to claim 3, characterized in that, The S3 step specifically includes: Obtain raw data streams from application performance management tools, business databases, and cloud service provider billing interfaces; Using the aforementioned decision knowledge graph, the original data stream is parsed and classified, and resource consumption metrics, performance metrics, and business event metrics are collected and accumulated under their corresponding target technical solution entity nodes and code-level entity nodes.

5. The development method of a software development platform according to claim 1, characterized in that, The attribution analysis in step S4 specifically includes: Calculate the deviation between the actual operating data corresponding to the target technical solution and its prediction result in step S1; Analyze the correlation between the deviation value and the fluctuation of business indicators, and locate at least one code module that causes the main deviation.

6. The development method of a software development platform according to claim 5, characterized in that, The generation of optimization suggestions in step S5 specifically includes: Based on the attribution analysis results, candidate optimization schemes are matched from the pre-set optimization strategy library; The candidate optimization scheme is input into the multi-dimensional prediction model to perform simulation prediction and obtain simulation results; By comparing the simulation results with the current actual operating effects, optimization suggestions are generated, including the predicted improvement range and investment return cycle.

7. A software development platform system for implementing the development method according to any one of claims 1 to 6, characterized in that, include: The decision prediction module is used to execute the S1 step and provides a quantitative prediction function for the technical solution; The mapping management module is used to execute step S2, creating and maintaining decision mapping relationships; A data aggregation engine is used to execute the S3 step to classify and aggregate multi-source data according to mapping relationships; The intelligent analysis module is used to execute the S4 step to complete the effect tracing and attribution analysis; The feedback loop module is used to execute the S5 step and is responsible for generating optimization suggestions and injecting feedback into the development process.

8. A software development platform system according to claim 7, characterized in that, The decision prediction module includes: A multi-dimensional knowledge base is used to store technical attributes, cost models, and historical pattern data; The prediction model unit calls upon the data from the multi-dimensional knowledge base to perform multi-objective quantitative prediction on the input technical solution.

9. A software development platform system according to claim 7, characterized in that, The data aggregation engine is connected to the mapping management module, and the data aggregation engine includes: An interface for collecting data from multiple data sources; The association calculation unit is used to perform real-time association calculation and aggregation storage on the collected data according to the decision mapping relationship provided by the mapping management module.

10. A software development platform system according to claim 7, characterized in that, The system also includes a visual interaction platform, which is connected to the intelligent analysis module and the feedback loop module. The visual interaction platform is used for: Using the target technical solution as a dimension, a comparison view of its predicted indicators and actual operating indicators is presented; The cost-benefit view of resource consumption and business contribution is presented at the code module level. Publish and track the processing status of the optimization recommendations.

Images