A research data management method and device based on SOA architecture

Abstract

This invention discloses a research data management method and apparatus based on SOA architecture. The apparatus includes: a portal website module, a research plan budget management module, a project process management module, a project completion and acceptance module, and a research review module. The portal website module is connected to the other modules and is used to display research information. The research plan budget management module is used for budget management of research projects. The project process management module is used for process management of research projects. The research review module is used to provide review services for research projects. The comprehensive analysis management module is connected to the other modules and is used to perform statistical analysis on various types of research data to obtain research project classification information.

Description

Technical Field

[0001] This invention relates to the fields of scientific research and industrial data mining and processing, SOA programming, and data mining technology, specifically to a scientific research data management method and device based on SOA architecture. Background Technology

[0002] In the current field of scientific research management, research projects often involve multiple stages, including planning, budget allocation, process tracking, completion acceptance, and review and evaluation, with complex and scattered data types. Traditional scientific research data management systems mostly adopt a closed architecture, with each functional module operating independently, forming "information silos." This leads to poor flow of scientific research information, significant difficulties in cross-module data statistics and analysis, and makes it difficult for managers to have a comprehensive grasp of the overall situation of research projects. Furthermore, existing systems lack sufficient resource scalability. When research projects are submitted or reviewed in concentrated periods, the system is prone to lag or even crashes due to excessive load. Deployment and data synchronization in offline environments present obstacles, affecting the efficiency of the review process. Moreover, the lack of a robust disaster recovery and backup mechanism means that system failures or data loss can severely impact scientific research management, potentially even rendering critical research data unrecoverable. In addition, existing systems have limited data analysis capabilities, making it difficult to provide accurate statistical, query, and clustering results based on diverse user needs, thus failing to provide effective support for research project initiation decisions. Summary of the Invention

[0003] This invention primarily addresses the problem of how to effectively mine, classify, query, and cluster scientific research data for precise management. This invention discloses a scientific research data management method and apparatus based on SOA architecture.

[0004] In a first aspect, the present invention discloses a scientific research data management device based on SOA architecture, comprising: a portal website module, a scientific research plan budget management module, a project process management module, a project completion and acceptance module, a scientific research review module, and a comprehensive analysis management module;

[0005] The portal module is connected to the other modules respectively and is used to display scientific research information;

[0006] The research project budget management module is used for budget management of research projects;

[0007] The project process management module is used for process management of scientific research projects;

[0008] The project completion and acceptance module is used to manage the completion and acceptance of scientific research projects.

[0009] The research review module is used to provide review services for research projects;

[0010] The comprehensive analysis and management module is connected to other modules to perform statistical analysis on various types of scientific research data and obtain corresponding analysis results.

[0011] The portal website module is a website for various departments and managers to read and browse information related to scientific research projects;

[0012] The research project budget management module is used to receive and save research project plans and research funding budget information.

[0013] The project process management module is used to manage and store research project process data for various departments and research institutions.

[0014] The project completion and acceptance module is used to manage and store the completion and acceptance data of scientific research projects and scientific and technological achievements.

[0015] The research review module is used to provide review services for various stages of research projects.

[0016] The comprehensive analysis and management module is used to statistically analyze scientific research project data. Based on the conditions input by the user, it performs statistical analysis, query analysis, and cluster analysis on various types of scientific research data to obtain corresponding analysis results.

[0017] The comprehensive analysis and management module performs statistical analysis, query analysis, and cluster analysis on various types of scientific research data based on the conditions input by the user, and obtains corresponding analysis results, including:

[0018] The user-inputted condition information includes category information and content information;

[0019] The comprehensive analysis and management module extracts categories from the conditional information input by the user to obtain category information;

[0020] When the category information is statistical analysis, statistical analysis is performed on various types of scientific research data based on the content information in the condition information to obtain statistical analysis information;

[0021] When the category information is for query analysis, the various types of scientific research data are queried and analyzed based on the content information in the condition information to obtain query analysis information;

[0022] When the category information is cluster analysis, cluster analysis is performed on various types of scientific research data based on the content information in the condition information to obtain cluster analysis information.

[0023] The process of querying and analyzing various types of scientific research data based on the content information in the conditional information to obtain query and analysis information includes:

[0024] Using a keyword matching retrieval method, the content information in the condition information is queried and analyzed in various scientific research data to obtain the first query result information;

[0025] A semantic retrieval method based on vector embedding is used to query and analyze the content information in the condition information in various scientific research data to obtain the second query result information;

[0026] A structured retrieval method based on knowledge graphs is used to query and analyze the content information in the conditional information in various scientific research data to obtain the third query result information;

[0027] A structured query method based on semantic parsing is used to query and analyze the content information in the condition information in various scientific research data to obtain the fourth query result information;

[0028] All query results are clustered and fused to obtain query analysis information.

[0029] The process of clustering and fusion of all query results to obtain query analysis information includes:

[0030] Each text element in each query result is converted into a text vector to obtain the corresponding text vector.

[0031] For each query result, construct the corresponding set of text vectors for all text vectors.

[0032] For each set of text vectors representing query results, a matrix representation is performed to obtain the corresponding text matrix;

[0033] Perform a first clustering process on the text matrix to obtain a clustering matrix of the query result information;

[0034] A second clustering process is performed on the clustering matrix of all query results to obtain the final clustering matrix;

[0035] For each row vector of the final clustering matrix, perform vector-to-text conversion to obtain the corresponding text information;

[0036] By utilizing all the text information, query analysis information is constructed.

[0037] The first clustering process includes:

[0038] Principal component analysis is performed on the text matrix to obtain a coefficient matrix and a principal component matrix; each row vector of the principal component matrix is ​​a sample value of the extracted principal component index at each time step.

[0039] Perform cross-correlation calculations on the principal component matrices to obtain a cross-correlation matrix;

[0040] The eigenvalues ​​of the cross-correlation coefficient matrix are calculated to obtain the eigenvalue vector;

[0041] Using the eigenvalue vectors as the diagonal elements of the matrix, and setting the off-diagonal elements of the matrix to 0, we can construct the eigenma matrix.

[0042] Multiplying the feature matrix by the coefficient matrix yields the factor matrix;

[0043] The rotation factor matrix is ​​obtained by multiplying the factor matrix with a preset orthogonal matrix.

[0044] The evaluation vector is obtained by performing evaluation vector calculation on the rotation factor matrix, eigenvalue vector, and orthogonal matrix.

[0045] Find the index of all elements in the evaluation vector that are greater than the preset evaluation threshold;

[0046] Using all the found serial numbers, determine the row vectors of all corresponding serial numbers in the text matrix, and construct a clustering matrix using all the determined row vectors.

[0047] A second aspect of this invention discloses a research data management method based on SOA architecture, implemented using the aforementioned research data management device based on SOA architecture, comprising:

[0048] S1, collect the condition information input by the user;

[0049] S2, Based on the condition information, obtain the corresponding analysis results from the comprehensive analysis management module;

[0050] S3 uses a portal website module to display the analysis results.

[0051] A third aspect of this invention discloses a scientific research data management device based on SOA architecture, the device comprising:

[0052] Memory containing executable program code;

[0053] A processor coupled to the memory;

[0054] The processor calls the executable program code stored in the memory to execute the research data management method based on SOA architecture.

[0055] In a fourth aspect of this invention, a computer-storable medium is disclosed, wherein the computer-storable medium stores computer instructions, and when the computer instructions are invoked by a computer, they are used to execute the aforementioned research data management method based on SOA architecture.

[0056] A fifth aspect of the present invention discloses an information data processing terminal, which is used to implement the aforementioned research data management method based on SOA architecture.

[0057] The beneficial effects of this invention are as follows:

[0058] This invention connects the portal module with other modules, enabling centralized display and cross-departmental sharing of scientific research information. It eliminates the drawbacks of "information silos" in traditional systems, allowing departments and managers to easily access information such as notices, announcements, and research results and handle related matters, significantly improving the efficiency of scientific research information flow and the level of collaborative management.

[0059] This invention constructs a management system for the entire life cycle of scientific research projects through the professional division of labor in modules such as scientific research plan budget management, project process management, project completion acceptance, and scientific research review. From the preparation of project guidelines, project initiation, and budget preparation to the initiation, scheme management, and information change during the process, and then to the completion acceptance and review at each stage, standardized control is achieved, ensuring the systematicness and rigor of scientific research project management.

[0060] The comprehensive analysis and management module of this invention employs multiple retrieval methods and clustering fusion processing technology. It can perform accurate statistical analysis, query analysis, and clustering analysis based on the conditions and information input by the user. It can not only comprehensively present the overall deployment and progress of scientific research projects, but also provide data support for scientific research project decision-making, thereby improving the scientific nature and accuracy of decision-making.

[0061] This invention achieves elastic scaling of services based on cloud computing architecture. By dynamically adjusting application resources, it ensures stable operation when the system load is high and enables resource reuse when resource consumption is low, thereby improving the utilization efficiency of system resources.

[0062] This invention features rapid deployment capabilities on standalone computer terminals, meeting the needs for synchronizing review materials and transmitting review data in network-free environments, ensuring efficient paperless review processes. It also supports rapid deployment when terminals are replaced or added, significantly improving the flexibility and efficiency of review work. The business disaster recovery backup mechanism of this invention enables the business system to autonomously recover from software bugs, avoiding the tedious manual uninstallation and installation, and prevents data loss under abnormal conditions through database backup, ensuring the continuity of scientific research management and the security of scientific research data. Attached Figure Description

[0063] Figure 1 This is a flowchart illustrating the implementation of the method of the present invention;

[0064] Figure 2 This is a design diagram of the SOA-based architecture of the present invention. Detailed Implementation

[0065] To better understand the content of this invention, an embodiment is provided here.

[0066] Figure 1 This is a flowchart illustrating the implementation of the method of the present invention. Figure 2 This is a design diagram of the SOA-based architecture of the present invention.

[0067] In a first aspect, the present invention discloses a scientific research data management device based on SOA architecture, comprising: a portal website module, a scientific research plan budget management module, a project process management module, a project completion and acceptance module, a scientific research review module, and a comprehensive analysis management module;

[0068] The portal module is connected to other modules and is used to display scientific research information.

[0069] The research project budget management module is used for budget management of research projects;

[0070] The project process management module is used for process management of scientific research projects;

[0071] The project completion and acceptance module is used to manage the completion and acceptance of scientific research projects.

[0072] The research review module is used to provide review services for research projects;

[0073] The comprehensive analysis and management module is connected to other modules and is used to perform statistical analysis on various types of scientific research data to obtain scientific research project classification information.

[0074] The portal website module is a website for various departments and managers to read and browse information related to scientific research projects and to handle matters related to scientific research projects, including notices and announcements, management updates, research results, information review and release, etc.

[0075] The research project budget management module is used to receive and save research project plans and research funding budget information, mainly including the management of research project guidelines, project initiation, research funding budget preparation, and research plan formulation.

[0076] The project process management module is used to manage and store research project process data for various departments and research units, including the management of data such as project initiation, overall technical solutions, engineering design schemes, and project information changes.

[0077] The project completion and acceptance module is used to manage and store the completion and acceptance data of scientific research projects and scientific and technological achievements, including the management of project acceptance applications, project acceptance reviews, and project acceptance archive data.

[0078] The research review module is used to provide review services for various stages of research projects. The review services include functions such as expert registration, expert information management, expert selection, project information acquisition, project review, submission of project review opinions, and statistical printing of review results.

[0079] The comprehensive analysis and management module is used to statistically analyze scientific research project data. Based on the conditions and information input by the user, it performs statistical analysis, query analysis, and cluster analysis to obtain corresponding analysis results. This facilitates an effective understanding of the overall deployment and progress of scientific research projects and assists in decision-making regarding scientific research project approval.

[0080] The comprehensive analysis and management module performs statistical analysis, query analysis, and cluster analysis based on the conditions input by the user, and obtains the corresponding analysis results, including:

[0081] The user-inputted condition information includes category information and content information;

[0082] The comprehensive analysis and management module extracts categories from the conditional information input by the user to obtain category information;

[0083] When the category information is statistical analysis, statistical analysis is performed on various types of scientific research data based on the content information in the condition information to obtain statistical analysis information;

[0084] When the category information is for query analysis, the various types of scientific research data are queried and analyzed based on the content information in the condition information to obtain query analysis information;

[0085] When the category information is cluster analysis, cluster analysis is performed on various types of scientific research data based on the content information in the condition information to obtain cluster analysis information;

[0086] The process of querying and analyzing various types of scientific research data based on the content information in the conditional information to obtain query and analysis information includes:

[0087] Using a keyword matching retrieval method, the content information in the condition information is queried and analyzed in various scientific research data to obtain the first query result information;

[0088] A semantic retrieval method based on vector embedding is used to query and analyze the content information in the condition information in various scientific research data to obtain the second query result information;

[0089] A structured retrieval method based on knowledge graphs is used to query and analyze the content information in the conditional information in various scientific research data to obtain the third query result information;

[0090] A structured query generation based on semantic parsing is used to query and analyze the content information in the conditional information in various scientific research data to obtain the fourth query result information;

[0091] All query results are clustered and fused to obtain query analysis information.

[0092] The process of clustering and fusion of all query results to obtain query analysis information includes:

[0093] Each text element in each query result is converted into a text vector to obtain the corresponding text vector.

[0094] For each query result, construct the corresponding set of text vectors for all text vectors.

[0095] For each set of text vectors in the query results, a matrix representation is performed to obtain the corresponding text matrix; the row vectors of the text matrix are the text vectors in the set of text vectors in the query results.

[0096] Perform a first clustering process on the text matrix to obtain a clustering matrix of the query result information;

[0097] A second clustering process is performed on the clustering matrix of all query results to obtain the final clustering matrix;

[0098] For each row vector of the final clustering matrix, perform vector-to-text conversion to obtain the corresponding text information;

[0099] By utilizing all the text information, query analysis information is constructed.

[0100] The first clustering process includes:

[0101] Principal component analysis is performed on the text matrix to obtain the coefficient matrix and the principal component matrix Y;

[0102] Perform cross-correlation calculations on the principal component matrices to obtain a cross-correlation matrix;

[0103] The cross-correlation coefficient matrix is ​​subjected to eigenvector calculation to obtain the eigenma matrix U; the eigenma matrix is ​​a matrix constructed from all eigenvectors.

[0104] The feature matrix and principal component matrix are transformed to obtain the clustering matrix;

[0105] The expression for the transformation process is:

[0106] J1=(U T U) -1 U T Y,

[0107] Where J1 is the clustering matrix;

[0108] The principal component analysis process can be implemented using the PCA algorithm, and both the principal component matrix and the coefficient matrix are determined through the principal component analysis process.

[0109] The second clustering process is performed on the clustering matrix of all query results to obtain the final clustering matrix, including:

[0110] Perform matrix clustering processing on the clustering matrices of all query results to obtain the clustering information for each clustering matrix;

[0111] The clustering matrices belonging to the same clustering information are weighted and summed to obtain the fused clustering matrix;

[0112] All the fused clustering matrices are concatenated along the column dimensions to obtain the final clustering matrix.

[0113] The weight values ​​for the weighted summation process can be the reciprocals of the largest eigenvalue of each matrix;

[0114] The matrix clustering process can employ either spectral clustering or hierarchical clustering methods.

[0115] The first through fourth query results come from different retrieval methods, including keyword matching, semantic retrieval, knowledge graph structured retrieval, and semantic parsing structured query. Each method has its own focus and advantages (e.g., keyword matching emphasizes literal matching, semantic retrieval emphasizes contextual relationships, and knowledge graph retrieval emphasizes entity relationships). Although the corresponding clustering matrix has achieved internal clustering of single-source results through the first clustering process, there may be information differences or complementarities between different source matrices.

[0116] The second clustering process uses matrix clustering (such as spectral clustering and hierarchical clustering) to group cluster matrices with high semantic relevance into one class, ensuring that the information being integrated is essentially related and avoiding the mixing of irrelevant information. Then, it uses weighted summation (using the reciprocal of the largest eigenvalue as the weight, which can reflect the "significance" or "reliability" of the matrix information) to merge matrices of the same class, and finally splices them together to form a unified final clustering matrix. This realizes the transformation of multi-source query results from "dispersed and independent" to "organic integration", overcoming the one-sidedness of single retrieval methods.

[0117] Different retrieval methods may produce slightly different results for the same query conditions (e.g., keyword matching may miss synonyms, and semantic retrieval may introduce weakly relevant information). In the second clustering process, the matrix clustering step identifies "core consistent" information (i.e., key content that multiple methods point to) in different source results by measuring the similarity between clustering matrices (e.g., based on matrix features, row vector distribution, etc.) and merging them into the same category. The weighted summation further strengthens the weight of high-reliability matrices (e.g., matrices with smaller eigenvalues ​​and more stable information), weakens the interference of low-reliability matrices, and makes the merged information closer to the true intent of the query, reducing noise and bias, and improving the accuracy of the overall results.

[0118] Multi-source search results may contain duplicate or highly similar content (such as different descriptions of the same research project retrieved by different methods). The second clustering process aggregates similar cluster matrices through matrix clustering, and then merges similar information through weighted summation. Essentially, this is a "deduplication and integration" of duplicate or similar content. The final clustering matrix formed by concatenating columns can arrange key information from different dimensions (such as basic project information, progress information, review information, etc.) in an orderly manner, avoiding information piling up and making core content (such as frequently occurring project characteristics and key indicators) more prominent. This makes it easier to convert the data into text information later, resulting in more concise and focused query and analysis results.

[0119] The final clustering matrix is ​​a structured data carrier resulting from multiple rounds of clustering and fusion. Its row vectors correspond to the integrated core text vectors, and its column dimensions cover the key dimensions of multi-source retrieval. This structured feature enables the subsequent "vector-to-text conversion" step to efficiently generate clear and comprehensive query analysis information. It includes the advantages of different retrieval methods (such as entity relationships in knowledge graphs and deep associations in semantic retrieval) and eliminates conflicting or redundant information through clustering and fusion. The resulting query analysis information can more comprehensively reflect the multi-dimensional characteristics of scientific research data (such as project type, progress status, and related results), providing users (such as research managers) with a more reliable and systematic basis for understanding the overall deployment of research projects and assisting in project initiation decisions.

[0120] In summary, the second clustering process, through clustering aggregation, weighted fusion, and structured splicing of multi-source matrices, has upgraded query results from "fragmented" to "systematic" and from "multi-source heterogeneous" to "homogeneous integration." This not only ensures the comprehensiveness of information but also improves accuracy and usability, laying a core foundation for efficient querying and analysis of scientific research data.

[0121] Based on the content information in the conditional information, cluster analysis is performed on various types of scientific research data to obtain cluster analysis information, including:

[0122] The content information in the conditional information includes clustering label values ​​and a set of clustering models;

[0123] Based on the clustering model set in the content information of the conditional information, clustering processing is performed on all categories of scientific research data to obtain a clustering information set for each scientific research data; the clustering information set consists of the clustering label value and the corresponding confidence value obtained by using each clustering model in the clustering model set for the scientific research data.

[0124] A reliable fusion calculation is performed on the clustering information set of each scientific research data to obtain the clustering analysis information of each scientific research data;

[0125] The expression for the trusted fusion computation is:

[0126]

[0127] Where L0 is the label value corresponding to the cluster analysis information of the scientific research data, L i and k i For each i-th clustering model, the clustering label value and corresponding confidence value obtained for the scientific research data are given, where k0 is a preset standard confidence value, which can be 0.8. The first to third clustering models are density clustering, spectral clustering, and fuzzy clustering, respectively.

[0128] Based on the label value of each scientific research data point, determine the corresponding clustering analysis information;

[0129] The clustering model set includes density clustering methods, spectral clustering methods, and fuzzy clustering methods;

[0130] For density clustering, the reliability can be measured by the sample's "local density" (number of surrounding neighbors) or "reachability distance" (distance from the core point). Fuzzy distance directly outputs the membership degree (0-1) of each sample to all clusters. The cluster with the highest membership degree is the cluster to which the sample belongs, and this membership degree value can be directly used as the reliability value (the closer the value is to 1, the more certain the affiliation).

[0131] For the confidence value of spectral clustering, the intermediate process of spectral clustering (such as similarity matrix, eigenvector) or the stability of clustering results can be used to indirectly calculate the confidence value of sample belonging (i.e. the reliability of a sample belonging to a certain cluster).

[0132] Scientific research data is characterized by its multi-source nature (e.g., project plans, budgets, process records, review comments), heterogeneity (text, numerical data, structured tables, etc.), and complexity (the same research field may involve interdisciplinary collaborations, and different projects may have implicit connections). Cluster analysis can overcome the subjectivity and limitations of manual classification, automatically identifying clusters of research projects with similar characteristics from massive datasets (e.g., by research direction, technical field, progress stage, funding scale, etc.). For example, density clustering can discover project clusters in the interdisciplinary field of "artificial intelligence + medicine," spectral clustering can identify highly similar research directions in key laboratories, and fuzzy clustering can distinguish between "basic research" and "applied research" projects with ambiguous boundaries. Uncovering these potential structures can help management departments discover the distribution patterns of research resources, the evolving trends of research hotspots, or potential opportunities for cross-disciplinary collaboration, transforming research data from "passive storage" into assets that "actively provide value."

[0133] This clustering analysis employs a "model set" consisting of three models: density clustering, spectral clustering, and fuzzy clustering. The final labels are generated through reliable fusion calculations, significantly improving the robustness of the results.

[0134] Single clustering models have inherent limitations (such as density clustering being sensitive to noise, spectral clustering relying on similarity matrix design, and fuzzy clustering being sensitive to membership thresholds), while parallel processing of multiple models can complement each other—density clustering excels at discovering clusters of arbitrary shapes, spectral clustering is suitable for clustering high-dimensional data after dimensionality reduction, and fuzzy clustering can handle samples with ambiguous boundaries. The combination of the three can cover diverse scenarios of scientific research data (such as clearly categorized "completed projects" and "projects under research," and "interdisciplinary projects" with ambiguous boundaries).

[0135] The credible fusion calculation (which strengthens the weight of high-credibility results through the tan function and filters reliable information with k0=0.8 as the standard) further filters noise: it assigns higher weights to model results with high credibility (such as samples with membership degree close to 1 in fuzzy clustering) and appropriately weakens low-credibility results (such as the classification of edge points in density clustering), so that the final label retains the advantages of multiple models and avoids errors caused by the bias of a single model, ensuring that the clustering results are highly matched with the true characteristics of scientific research data.

[0136] Cluster analysis information can be directly applied to multiple scenarios in scientific research management, addressing the pain points of "extensive classification" in traditional management:

[0137] In terms of resource allocation, clustering can be used to identify "project clusters with the same research direction" to avoid duplicate projects or resource dispersion (for example, if multiple similar projects are found in a certain field, the tendency to initiate subsequent projects can be adjusted).

[0138] In terms of process control, by "clustering by progress stage", we can strengthen supervision of "lagging project clusters" and summarize and promote the experience of "leading project clusters".

[0139] In terms of technology transfer, by "clustering by technology field", the results of similar projects can be quickly matched with industry needs, promoting the integration of industry, academia and research.

[0140] Furthermore, the standardization of clustering labels (such as unified "research field labels" and "progress status labels") can enable structured management of scientific research data, reduce the tediousness and errors of manual classification, and enable managers to quickly locate target data through labels, thereby improving the efficiency of management processes.

[0141] Scientific research activities are dynamic (project progress and research hotspots change over time) and have personalized needs (different managers may focus on different dimensions of clustering results). This clustering analysis supports flexible configuration through "content information in the conditional information" (such as customizing cluster label values ​​and selecting specific model combinations):

[0142] When users focus on the "cluster distribution of Youth Fund projects in the past 3 years", they can limit the time range and project type by setting the conditional information, and the clustering model will process the data of that subset in a targeted manner.

[0143] When new directions emerge in the field of scientific research (such as "quantum computing"), the model can be quickly adapted to new classification requirements by updating the clustering label values, without having to reconstruct the entire analysis system.

[0144] This dynamic adaptability ensures that cluster analysis information can continuously meet the real-time needs of scientific research management, avoiding the problem of traditional static classification methods "lagging behind scientific research development".

[0145] In summary, this clustering analysis method, through multi-model fusion, reliable computing, and flexible configuration, not only achieves accurate mining of the deep structure of scientific research data, but also ensures the reliability and adaptability of the results. Ultimately, it provides a key bridge from "data to insight" for scientific research management, helping to improve the efficiency of scientific research resource allocation, the scientific nature of decision-making, and the level of refined management.

[0146] The statistical analysis can be used to obtain the quantity of various types of scientific research information, such as the number of ongoing projects and the number of completed projects.

[0147] The research data management device based on SOA architecture, also known as the management system, adopts the mainstream cloud computing architecture and is divided into three layers: infrastructure layer, platform layer, and application layer.

[0148] To meet the application service requirements of scientific research project management, the overall technical architecture should possess the following capabilities:

[0149] 1) Business Elastic Scaling: By setting different resource usage policies for applications, the system can automatically scale applications. After setting an automatic scaling policy, the system will automatically scale the application dynamically based on the resource policy and the application's resource utilization (e.g., increasing or decreasing application resources) to achieve the optimal resource utilization state. For example, when the system's CPU and memory resource utilization is high, the system will automatically expand resources for the application and automatically deploy the application software to reduce the application's resource load and ensure the healthy operation of the system; when the system's resource utilization is low, the system can automatically reduce the resources used by the application and release the corresponding resources to achieve effective reuse of system resources.

[0150] 2) Rapid Business Deployment: Since the review process for various solutions often takes place in environments without network connectivity, the application system should have the capability for rapid deployment on standalone computer terminals. This allows for the synchronization of relevant review materials to the review terminal, enabling paperless review work by experts. After the review, the review data should be synchronized to the data center system, thus enabling review work in network-free environments. Furthermore, the application system should be able to be rapidly deployed within ten minutes when computer terminals are replaced or added.

[0151] 3) Business Disaster Recovery and Backup: Business systems play a crucial role in scientific research management. Business interruptions can lead to significant economic losses and potentially the loss of critical data. Therefore, the construction of a platform disaster recovery and backup system is essential. Business disaster recovery should encompass business system disaster recovery and data disaster recovery. Business system disaster recovery refers to the ability of the business system to autonomously recover from software bugs within its own application, avoiding the need for manual uninstallation and reinstallation. Data disaster recovery refers to backing up data stored in the database to prevent data loss under abnormal conditions.

[0152] The infrastructure layer is located at the bottom of the scientific research project management platform system. It is the foundation of the entire platform, responsible for ensuring physical resources and providing resource support for platform services. The infrastructure layer uses virtualization technology to virtualize the lowest-level physical resources and allocate and schedule resources such as network bandwidth, CPU, storage and memory as needed.

[0153] The platform layer consists of services such as database services, web services, and caching services. For example, data management technology is used to aggregate and process scientific research project data to form an efficient comprehensive scientific research project database, providing the necessary support environment for application operation, building an integrated scientific research project service system, and providing scientific research project managers with comprehensive and accurate scientific research project information, as well as management and decision-making basis.

[0154] The application layer provides a variety of user-facing applications, including portal websites, review systems, business data systems, and more.

[0155] The system employs MDA to elevate the development process to a higher level of abstraction (analysis model level), automating the coding work specific to the computing platform. This successfully decouples business logic from implementation technology, resolving the problem in traditional software development methods where synchronizing UML models and code generated during the design phase becomes increasingly difficult as the project progresses; thus enhancing the system's robustness. The system uses the SOA model to connect different functional units of the application (called services) through well-defined interfaces and contracts between these services. These interfaces are defined in a neutral manner, independent of the hardware platform, operating system, and programming language used to implement the services. Services built into various subsystems can interact in a unified and universal way.

[0156] The system adopts the J2EE technology framework, which can fully utilize the existing system and achieve easy coupling. This architecture offers excellent stability, reducing severe losses caused by service downtime. Its efficient development performance also shortens the business system development cycle. It supports heterogeneous environments and is suitable for various mainstream operating systems, middleware, and hardware. Its scalability also meets the needs of future system expansion and improvements. The system uses a B / S architecture, which effectively achieves distributed computing, allowing various business operations to be performed anytime, anywhere, avoiding cumbersome client installations. Business expansion is simple and convenient, facilitating the future addition of new business systems and functions. The cross-platform and open nature of the architecture also allows the system to better adapt to various complex deployment scenarios.

[0157] The system incorporates a Business Process Management (BPM) engine, promoting integration between people, between people and systems, and between systems based on changes in the business environment. BPM includes processes such as modeling, implementation, monitoring, and management; it perfectly solves the shortcomings of traditional information systems, such as unclear requirement expression, low efficiency, and rigid business processes; it automates business process systems through BPM, and integrates and statistically analyzes business processes through process analysis and monitoring functions; and it achieves quantifiable improvements from multiple dimensions, including business, organizational, and system operation and maintenance perspectives.

[0158] Based on SOA architecture, cross-application system BPM processes can be provided by bundling services from multiple application systems, and new services can be generated by combining multiple business services. Furthermore, system administrators are provided with various management methods such as form configuration and graphical configuration, improving the user-friendliness of human-computer interaction.

[0159] This system has evolved from processing traditional tabular data to handling multiple data types. In practical applications, it is often necessary to perform statistics and analysis on data during the approval process. It needs to provide comprehensive form content management functions. In the process configuration, it can flexibly customize document-type data and business-type structured data, and automatically process the two types of data separately to ensure that the data can be easily searched, queried, and statistically analyzed later. Appropriate permissions can also be set.

[0160] To eliminate the "information silos" phenomenon in traditional systems, this system adheres to common technical standards and specifications, enabling seamless integration with different business systems in terms of business processing and data interaction. It facilitates cross-system data exchange, sharing, and fusion. Advanced text retrieval technology is employed, supporting full-text search. A relational database is used for centralized data import and export, facilitating data backup, recovery, and other maintenance tasks.

[0161] This system has flexible deployment capabilities, and can be divided into three deployment modes according to application scenarios: stand-alone, LAN, and network. The system mainly relies on the network to run, and can be uniformly deployed in environments with network connectivity. Stand-alone / LAN deployment modes are used for scenarios without network connectivity and temporary office work. The complete deployment mode improves the overall availability of the system.

[0162] The application system also has the capability for rapid containerization deployment. Using advanced container technology, some system modules are containerized, enabling unified deployment of subsystems across all environments. The deployment process is simple and fast. For systems such as review systems that require separate deployment, containerization deployment avoids cumbersome dependency environment installations and achieves one-click deployment.

[0163] Based on the user information and permission information database of the information platform, permissions are divided into physical, portal, and business categories according to user identity and role. Based on different permission divisions and configurations, the system provides users and roles with permission allocation and diverse view control for system functions, business operations, and data operations. In the system, when a user enters the corresponding interface, different application modules will be loaded according to their role permissions, while modules that do not have their own permissions will be blocked.

[0164] Physically, it supports standalone systems, local area networks (LANs), and network access control. A standalone system means there is only one business system, not multiple business systems, and the system platform needs to shield functions related to LANs and networks. A LAN system is a multi-functional integrated system within a region, such as the LAN system of a department that includes all news and business functions under that department. Each LAN system has its own resources, roles, and users.

[0165] The overall system is designed based on SOA architecture, such as Figure 2As shown, the structure is divided into a data service layer, a business logic layer, and a presentation layer. Considering practical application scenarios, advanced technologies such as dynamic workflow technology, database read / write separation technology, and the TRS full-text search system are used.

[0166] Service-Oriented Architecture (SOA) is a component model that connects different services of an application through well-defined interfaces and contracts, enabling repeated service calls, reducing costs, and improving efficiency. SOA architecture treats business logic as linked services or repetitive business tasks as integrated components, allowing access to these services and tasks over a network when needed.

[0167] Nested dynamic workflow technology involves dynamically judging the workflow based on the research project management process and the processing opinions from the business system, thereby achieving dynamic workflow management. The workflow definition clearly defines the research project management process, uses process modeling tools to define the workflow of business processing, defines different roles for each stage of processing, and implements the defined process through certain rules and constraints.

[0168] Dynamic workflows consist of predefined template processes and non-automated processes dynamically assigned based on professional knowledge in the research project field. By using workflow technology, administrative offices in research departments can be asynchronously coordinated, and information flow can be made faster and more convenient. This results in an efficient, stable, and transparent research project approval process, and dynamic situations that arise in real time.

[0169] In the calculation expressions of this invention, the variables involved have all been dimensionless before calculation.

[0170] A second aspect of this invention discloses a scientific research data management device based on an SOA architecture, the device comprising:

[0171] Memory containing executable program code;

[0172] A processor coupled to the memory;

[0173] The processor calls the executable program code stored in the memory to execute the research data management method based on SOA architecture.

[0174] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A scientific research data management device based on SOA architecture, characterized in that, include: The system includes a portal website module, a research plan budget management module, a project process management module, a project completion and acceptance module, a research review module, and a comprehensive analysis management module. The portal module is connected to the other modules respectively and is used to display scientific research information; The research project budget management module is used for budget management of research projects; The project process management module is used for process management of scientific research projects; The project completion and acceptance module is used to manage the completion and acceptance of scientific research projects. The research review module is used to provide review services for research projects; The comprehensive analysis and management module is connected to other modules and is used to perform statistical analysis on various types of scientific research data to obtain corresponding analysis results. The comprehensive analysis and management module performs statistical analysis, query analysis, and cluster analysis on various types of scientific research data based on the conditions input by the user, and obtains corresponding analysis results, including: The user-inputted condition information includes category information and content information; The comprehensive analysis and management module extracts categories from the conditional information input by the user to obtain category information; When the category information is for query analysis, based on the content information in the condition information, various types of scientific research data are queried and analyzed to obtain query analysis information, including: The content information in the condition information is retrieved and analyzed in various scientific research data using keyword matching retrieval method, vector embedding-based semantic retrieval method, knowledge graph-based structured retrieval method, and semantic parsing-based structured query method, respectively, to obtain first query result information, second query result information, third query result information, and fourth query result information. All query results are clustered and fused to obtain query analysis information, including: Each text element in each query result is converted into a text vector to obtain the corresponding text vector. For each query result, construct the corresponding set of text vectors for all text vectors. For each set of text vectors representing query results, a matrix representation is performed to obtain the corresponding text matrix; Perform a first clustering process on the text matrix to obtain a clustering matrix of the query result information; A second clustering process is performed on the clustering matrix of all query results to obtain the final clustering matrix; For each row vector of the final clustering matrix, perform vector-to-text conversion to obtain the corresponding text information; By utilizing all the text information, query analysis information is constructed.

2. The research data management device based on SOA architecture as described in claim 1, characterized in that, The portal website module is a website for various departments and managers to read and browse information related to scientific research projects; The research project budget management module is used to receive and save research project plans and research funding budget information. The project process management module is used to manage and store research project process data for various departments and research institutions.

3. The research data management device based on SOA architecture as described in claim 1, characterized in that, The project completion and acceptance module is used to manage and store the completion and acceptance data of scientific research projects and scientific and technological achievements. The research review module is used to provide review services for various stages of research projects. The comprehensive analysis and management module is used to statistically analyze scientific research project data. Based on the conditions input by the user, it performs statistical analysis, query analysis, and cluster analysis on various types of scientific research data to obtain corresponding analysis results.

4. The research data management device based on SOA architecture as described in claim 1, characterized in that, The first clustering process includes: Principal component analysis is performed on the text matrix to obtain a coefficient matrix and a principal component matrix; each row vector of the principal component matrix is ​​a sample value of the extracted principal component index at each time step. Perform cross-correlation calculations on the principal component matrices to obtain the cross-correlation coefficient matrix; The eigenvalues ​​of the cross-correlation coefficient matrix are calculated to obtain the eigenvalue vector; Using the eigenvalue vectors as the diagonal elements of the matrix, and setting the off-diagonal elements of the matrix to 0, we can construct the eigenma matrix. Multiplying the feature matrix by the coefficient matrix yields the factor matrix; The rotation factor matrix is ​​obtained by multiplying the factor matrix with a preset orthogonal matrix. The evaluation vector is obtained by performing evaluation vector calculation on the rotation factor matrix, eigenvalue vector, and orthogonal matrix. Find the index of all elements in the evaluation vector that are greater than the preset evaluation threshold; Using all the found serial numbers, determine the row vectors of all corresponding serial numbers in the text matrix, and construct a clustering matrix using all the determined row vectors.

5. The research data management device based on SOA architecture as described in claim 1, characterized in that, The comprehensive analysis and management module performs statistical analysis, query analysis, and cluster analysis on various types of scientific research data based on the conditions input by the user, and obtains corresponding analysis results, including: The comprehensive analysis and management module extracts categories from the conditional information input by the user to obtain category information; When the category information is statistical analysis, statistical analysis is performed on various types of scientific research data based on the content information in the condition information to obtain statistical analysis information; When the category information is cluster analysis, cluster analysis is performed on various types of scientific research data based on the content information in the condition information to obtain cluster analysis information.

6. A research data management method based on SOA architecture, characterized in that, Implemented using a research data management device based on SOA architecture as described in any one of claims 1 to 5, comprising: S1, collect the condition information input by the user; S2, Based on the condition information, obtain the corresponding analysis results from the comprehensive analysis management module; S3 uses a portal website module to display the analysis results.

7. A computer-storable medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked by the computer, are used to execute the research data management method based on SOA architecture as described in claim 6.

Images