A platform for dynamic monitoring and visual analysis of scientific and technological achievement transformation benefits

Abstract

This application relates to the field of technology transfer management, and discloses a dynamic monitoring and visualization analysis platform for the benefits of technology transfer, comprising: a data acquisition module for collecting technology transfer-related data through multiple channels; the data includes invention patent data, market demand data, and transfer benefit evaluation data; and a data preprocessing module connected to the data acquisition module for standardizing the technology transfer-related data collected by the data acquisition module. Through a combination of fast classification algorithms and fine-tuning algorithms, it can achieve rapid initial classification in the field of invention patent technology using the TF-IDF keyword weight algorithm, solving the problems of low efficiency and cumbersome operation of traditional manual classification. Furthermore, it can perform secondary optimization of fuzzy samples using the cosine similarity algorithm, reducing the classification error rate and significantly improving the accuracy of patent technology segmentation, thus laying a precise data foundation for subsequent transfer benefit analysis.

Description

Technical Field

[0001] This invention relates to the field of technology transfer management, specifically to a platform for dynamic monitoring and visualization analysis of the benefits of technology transfer. Background Technology

[0002] The Dynamic Monitoring and Visual Analysis Platform for the Benefits of Technology Transfer is a digital tool that integrates multi-source data collection, intelligent analysis, and visualization. It is mainly used to track and quantify the benefit data of scientific and technological achievements (especially invention patents) throughout the entire process from R&D to market transfer in real time, and present the analysis results through intuitive charts to provide transfer decision support for universities, enterprises, research institutions, and other entities.

[0003] In existing technologies, such platforms mostly rely on manual input or simple interface connections to obtain information such as patent data and market demand. The classification of patent technology fields often adopts static matching based on IPC classification numbers or manual annotation. During the analysis process, only basic quantitative statistics and trend plotting can be achieved. Moreover, the evaluation of transformation benefits is mostly limited to the simple summary of a single indicator (such as transformation revenue), making it difficult to achieve dynamic correlation analysis between technology fields and market demand.

[0004] However, the classification efficiency of patent technology fields in the existing technology is low. Manual annotation is time-consuming and difficult to cope with massive patent data. Although simple algorithms (such as single keyword matching) are faster, they have a high error rate in classifying patents with ambiguous technical characteristics (such as cross-domain technologies and patents with atypical keyword descriptions), and misclassification and omission often occur. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a dynamic monitoring and visualization analysis platform for the benefits of technology transfer, which solves the problem of low classification efficiency.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a dynamic monitoring and visualization analysis platform for the benefits of technology transfer, comprising:

[0007] The data acquisition module is used to collect data related to the transformation of scientific and technological achievements through multiple channels;

[0008] The data includes invention patent data, market demand data, and conversion benefit evaluation data;

[0009] The data preprocessing module, connected to the data acquisition module, is used to standardize the data related to the transformation of scientific and technological achievements collected by the data acquisition module.

[0010] The intelligent analysis module is connected to the data preprocessing module and has a built-in fast classification algorithm for performing initial technical field classification of invention patents in the standardized data output by the data preprocessing module.

[0011] The visualization module, connected to the intelligent analysis module, is used to dynamically display the analysis results output by the intelligent analysis module and the conversion benefit evaluation data collected by the data acquisition module in multi-dimensional charts; the user interaction module, connected to the data acquisition module, intelligent analysis module and visualization module respectively, is used to receive user operation commands for each module and provide corresponding feedback results.

[0012] The data update module is connected to the data acquisition module, intelligent analysis module, and visualization module. It is used to automatically update the full / incremental data in the data acquisition module according to a preset cycle, and trigger the intelligent analysis module to re-execute the analysis process, and synchronize the update results to the visualization module.

[0013] The analysis process includes a fast classification algorithm and, if necessary, a fine-tuning algorithm.

[0014] Preferably, the multi-channel data acquisition method of the data acquisition module includes:

[0015] By connecting to the patent office database and third-party search systems through the patent data interface unit, basic data such as application number, abstract, and IPC classification number of invention patents can be obtained;

[0016] The market data crawling unit collects dynamic market demand data from corporate websites and industry databases.

[0017] The user data upload unit receives patent conversion contracts, revenue reports, and other conversion benefit evaluation data uploaded by users, and supports one-click import of PDF and Excel formats.

[0018] Preferably, the standardization processing of the data preprocessing module includes:

[0019] The units for the conversion benefit evaluation data are standardized, conversion revenue is converted to RMB, and the cost-benefit ratio is rounded to two decimal places.

[0020] The 3σ criterion is used to eliminate outliers in patent implementation rate and cost-benefit ratio;

[0021] The missing monthly conversion revenue data is filled in using linear interpolation, with the following formula:

[0022]

[0023] in, For the missing conversion revenue, , Given the known income before and after the missing point, , For the time corresponding to known income

[0024] Interval point, For time points with missing data.

[0025] Preferably, the fast classification algorithm of the intelligent analysis module is a keyword weight algorithm based on TF-IDF, and the formula is as follows:

[0026]

[0027] in, Keywords In the invention patent text The weighted score in the middle, Keywords In the text Number of times it appears in For text Total number of occurrences of all keywords in the text This represents the total number of invention patents in the platform's database. For keywords The number of invention patents;

[0028] When the keyword set of a certain technical field accumulates The sum is greater than the threshold At that time, the invention patent will be classified into this field; the cumulative score is in When the sample is fuzzy, it is determined to be a fuzzy sample.

[0029] Preferably, the intelligent analysis module also incorporates a fine-tuning algorithm, which is a text matching algorithm based on cosine similarity. This fine-tuning algorithm is used to optimize fuzzy samples obtained by the fast classification algorithm, wherein the fuzzy samples are those whose cumulative scores in the fast classification are within a certain range. The formula for the invention patent within the interval is as follows:

[0030]

[0031] in, For invention patent text Technical Field Standard Texts similarity, , Text respectively , In the Word vector values ​​in the feature dimension This represents the total number of feature dimensions.

[0032] when > If the invention patent is confirmed to belong to the original technical field, it will be reassigned to another technical field with the highest similarity.

[0033] Preferably, the multi-dimensional charts in the visualization module include:

[0034] A heatmap showing the distribution of technical fields in statistics, using color depth to indicate the cost-benefit ratio.

[0035] A matching network diagram of related categories, where node size represents patent implementation rate and line thickness represents degree of matching with market demand;

[0036] The conversion trend line chart shows the relationship between annual application volume and conversion revenue in the technology field, and supports switching time dimensions.

[0037] Preferably, the user interaction module includes:

[0038] The role recognition unit identifies roles such as university researchers, corporate technical specialists, and patent agents.

[0039] The interface adaptation unit loads corresponding functional modules for different roles, such as displaying a patent application trend analysis module for university personnel.

[0040] The operation feedback unit displays the processing time, number of results, and error messages in a pop-up window after the user completes data upload and query.

[0041] Preferably, the data update module includes:

[0042] The trigger control unit supports both preset periodic updates and user-manually triggered updates.

[0043] The incremental update unit filters incremental data from various channels in the data collection module based on timestamps.

[0044] The linkage analysis unit calls the intelligent analysis module to perform classification algorithms and benefit evaluation analysis on incremental data;

[0045] The version management unit records data update batches, times, and change summaries, and pushes update notifications to the user interaction module.

[0046] This invention provides a platform for dynamic monitoring and visualization analysis of the benefits of technology transfer. It has the following advantages:

[0047] 1. This invention combines a fast classification algorithm and a fine-tuning algorithm. It can achieve rapid initial classification in the field of invention patent technology by using the TF-IDF keyword weight algorithm, solving the problems of low efficiency and cumbersome operation of traditional manual classification. It can also perform secondary optimization on fuzzy samples through the cosine similarity algorithm, reducing the classification error rate and significantly improving the accuracy of the classification in the field of patent technology, laying a precise data foundation for subsequent conversion benefit analysis.

[0048] 2. This invention achieves not only refined classification of patent technologies by dynamically calculating the matching degree between patent texts and technical standard texts, but also indirectly uncovers potential connections between cross-domain technologies, providing users with a reference direction for technology integration. At the same time, its output classification results are linked with conversion benefit evaluation data, which can directly support the visualization module to generate charts, providing users with full-chain data support from technology classification to conversion decisions. Attached Figure Description

[0049] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Example:

[0052] Please see the appendix Figure 1 This invention provides a platform for dynamic monitoring and visualization analysis of the benefits of technology transfer, comprising:

[0053] The data acquisition module is used to collect data related to the transformation of scientific and technological achievements through multiple channels;

[0054] The data includes invention patent data, market demand data, and commercialization benefit evaluation data;

[0055] The data preprocessing module, connected to the data acquisition module, is used to standardize the data related to the transformation of scientific and technological achievements collected by the data acquisition module.

[0056] The intelligent analysis module is connected to the data preprocessing module and has a built-in fast classification algorithm for performing initial technical field classification of invention patents in the standardized data output by the data preprocessing module.

[0057] The visualization module, connected to the intelligent analysis module, is used to dynamically display the analysis results output by the intelligent analysis module and the conversion benefit evaluation data collected by the data acquisition module in multi-dimensional charts; the user interaction module, connected to the data acquisition module, intelligent analysis module and visualization module respectively, is used to receive user operation commands for each module and provide corresponding feedback results.

[0058] The data update module, connected to the data acquisition module and the intelligent analysis module, is used to automatically update the full / incremental data in the data acquisition module according to a preset cycle, and trigger the intelligent analysis module to re-execute the analysis process, and synchronize the update results to the visualization display module.

[0059] The analysis process includes a fast classification algorithm and, if necessary, a fine-tuning algorithm.

[0060] In this embodiment, the multi-channel data acquisition method of the data acquisition module includes:

[0061] By connecting to the patent office database and third-party search systems through the patent data interface unit, basic data such as application number, abstract, and IPC classification number of invention patents can be obtained;

[0062] The market data crawling unit collects dynamic market demand data from corporate websites and industry databases.

[0063] The user data upload unit receives patent conversion contracts, revenue reports, and other conversion benefit evaluation data uploaded by users, and supports one-click import of PDF and Excel formats.

[0064] The patent data interface unit connects to the patent office database, which specifically includes the patent publication and announcement system and third-party search systems, including but not limited to SooPAT and IncoPat. The interface is called once per hour to ensure that the latest published invention patent data is obtained.

[0065] The market data crawling unit adopts a distributed crawler architecture to crawl the "Technical Requirements" section of the company's official website and the "Industry Technology Report" section of industry databases (such as iResearch and Analysys). The crawling fields include the time of requirement release, technical field tags, budget range, etc., and the IP proxy pool is used to avoid access restrictions by the target website.

[0066] The user data upload unit supports OCR to automatically recognize key information in PDF / Excel files (such as the amount and signing time in the conversion contract, and the monthly income in the revenue report).

[0067] In this embodiment, the standardization process of the data preprocessing module includes:

[0068] The units for the conversion benefit evaluation data are standardized, conversion revenue is converted to RMB, and the cost-benefit ratio is rounded to two decimal places.

[0069] The 3σ criterion is used to eliminate outliers in patent implementation rate and cost-benefit ratio;

[0070] The missing monthly conversion revenue data is filled in using linear interpolation, with the following formula:

[0071]

[0072] in, For the missing conversion revenue, , Given the known income before and after the missing point, , For the time corresponding to known income

[0073] Interval point, For time points with missing data.

[0074] In this embodiment, the fast classification algorithm of the intelligent analysis module is a keyword weight algorithm based on TF-IDF, and the formula is as follows:

[0075]

[0076] in, Keywords In the invention patent text The weighted score in the middle, Keywords In the text Number of times it appears in For text Total number of occurrences of all keywords in the text This represents the total number of invention patents in the platform's database. For keywords The number of invention patents;

[0077] When the keyword set of a certain technical field accumulates The sum is greater than the threshold At that time, the invention patent will be classified into this field; the cumulative score is in When the sample is fuzzy, it is determined to be a fuzzy sample.

[0078] The construction method of the keyword set in the technical field is as follows: Based on the mapping relationship between IPC classification number and "National Economic Industry Classification", 12 primary technical fields are preset (such as artificial intelligence, biomedicine, new energy, etc.). Each field contains 50-200 core keywords (for example, the field of "artificial intelligence" includes "machine learning", "neural network", "natural language processing", etc.). The keyword weight is determined by combining expert review and training with historical conversion data.

[0079] threshold and The range of values ​​for: Set to 0.6-0.8 (dynamically adjusted based on the number of keywords in the technical field; the more keywords, the lower the value). (The higher the value) = -0.2, for example when When =0.7, =0.5, and invention patents with a cumulative score between 0.5 and 0.7 are judged as fuzzy samples.

[0080] threshold Through training with historical samples, the following criteria were determined: 1000 classified patents from each technical field were selected as the training set, and adjustments were made with a classification accuracy rate of ≥95% as the target. Values; Fixed as -0.2, tested and found to cover 90% of fuzzy samples.

[0081] In this embodiment, the intelligent analysis module also incorporates a fine-tuning algorithm, which is a text matching algorithm based on cosine similarity. This algorithm is used to optimize fuzzy samples obtained by the fast classification algorithm. Fuzzy samples are those whose cumulative scores in the fast classification are within a certain range. The formula for the invention patent within the interval is as follows:

[0082]

[0083] in, For invention patent text Technical Field Standard Texts similarity, , Text respectively , In the Word vector values ​​in the feature dimension This represents the total number of feature dimensions.

[0084] when > If the invention patent is confirmed to belong to the original technical field, it will be reassigned to another technical field with the highest similarity.

[0085] Sources of technical standard texts: 10-20 standard texts correspond to each primary technical field, selected from the abstracts of invention patent specifications that have achieved high conversion efficiency (top 10% of conversion revenue) in that field, and are updated regularly (quarterly) by the platform;

[0086] Word vector generation method: The invention patent text d and standard text s are segmented using the BERT pre-trained model to generate 768-dimensional word vectors, with a total feature dimension m=768;

[0087] threshold The value is fixed at 0.7. That is, when the similarity Sim(d,s) > 0.7, the original classification is confirmed; otherwise, the field with the highest Sim(d,s) is selected from all technical fields and reassigned. If the highest similarity is still ≤ 0.5, it is marked as "cross-domain technology" and pushed to the user interaction module for manual confirmation of classification.

[0088] In the conversion benefit evaluation data of this platform, the cost-benefit ratio is calculated using the following formula: Cost-benefit ratio = (Conversion revenue - Conversion cost) / Conversion cost × 100%, where conversion cost includes patent application fees, maintenance fees, industrialization investment, etc.; the patent implementation rate is calculated using the following formula: Patent implementation rate = Number of patents that have been converted / Total number of patents in this technical field collected by the platform × 100%. The above indicators are automatically calculated by the intelligent analysis module based on the pre-processed conversion benefit data and serve as the core data source for the visualization module.

[0089] In this embodiment, the multi-dimensional charts in the visualization module include:

[0090] A heatmap showing the distribution of technical fields in statistics, using color depth to indicate the cost-benefit ratio.

[0091] A matching network diagram of related categories, where node size represents patent implementation rate and line thickness represents degree of matching with market demand;

[0092] The conversion trend line chart shows the relationship between annual application volume and conversion revenue in the technology field, and supports switching time dimensions.

[0093] The coordinate system of the heatmap for the distribution of technical fields is as follows: the horizontal axis represents the primary technical fields, the vertical axis represents the provinces, and the cost-benefit range corresponding to the color depth is [0, 50%] (light red), [50%, 100%] (red), [100%, 200%] (dark red), and >200% (purple red). Clicking on a cell in the heatmap allows you to view the specific patent list for that region and field.

[0094] Dynamic interactive features of the matching relationship network diagram: nodes can be dragged and dropped, and clicking on a node displays detailed information about the patent / market demand (such as patent abstract and demand description). Connections can be clicked to view the basis for the matching degree calculation (such as common keywords and technology complementarity score).

[0095] The time dimension of the conversion trend line chart can be switched: it supports switching by quarter (last 12 quarters) and year (last 5 years). The line chart includes a prediction line (based on linear regression prediction of data from the last 3 years), the prediction interval is marked with shade, and the prediction results are updated every quarter.

[0096] Responding to user interaction module commands: When a user selects a specific technical field, the heatmap automatically focuses on the data in that field; after dragging and dropping network graph nodes, the system updates the matching degree calculation of associated connections in real time.

[0097] In this embodiment, the user interaction module includes:

[0098] The role recognition unit identifies roles such as university researchers, corporate technical specialists, and patent agents.

[0099] The interface adaptation unit loads corresponding functional modules for different roles, such as displaying a patent application trend analysis module for university personnel.

[0100] The operation feedback unit displays the processing time, number of results, and error messages in a pop-up window after the user completes data upload and query.

[0101] The verification method for the role recognition unit is as follows: University researchers need to upload a photo of their employee ID card and verify it through the school's email address; enterprise technical specialists need to upload a photo of their enterprise's business license; and patent agents need to upload their practice certificate number. Role permissions will be automatically assigned after verification.

[0102] Examples of specific functional modules for the interface adaptation unit: The interface for enterprise technical specialists includes modules for "demand posting", "matching patent push", and "conversion benefit prediction"; the interface for patent agents includes modules for "patent tracking", "annual fee reminder" and "infringement monitoring".

[0103] Operation feedback unit error message handling: When there are data inconsistencies in the conversion revenue report uploaded by users (such as the monthly revenue total not matching the annual revenue), a pop-up window will display the specific inconsistency item (such as "March 2024 revenue + April 2024 revenue ≠ second quarter revenue") and provide two options: "automatic correction" and "manual modification". Automatic correction is based on the weighted average method to adjust the inconsistency data.

[0104] In this embodiment, the data update module includes:

[0105] The trigger control unit supports both preset periodic updates and user-manually triggered updates.

[0106] Full updates are only performed during initial deployment or when manually triggered by the user; preset periodic updates use incremental mode by default, and automatically switch to full updates only when the incremental data volume exceeds 20% of the total.

[0107] The incremental update unit filters incremental data from various channels in the data collection module based on timestamps.

[0108] The linkage analysis unit calls the intelligent analysis module to perform the analysis process on the incremental data (including classification algorithms and, when necessary, fine-tuning algorithms).

[0109] The version management unit records data update batches, times, and change summaries, and pushes update notifications to the user interaction module.

[0110] Preset cycle options for triggering the control unit: Users can select daily (2-4 AM), weekly (Sunday morning), or monthly (last day morning) updates in the platform settings. Manually triggering updates requires users to enter a verification code (to prevent accidental operation). The update process does not affect the normal use of the platform (using an incremental synchronization mechanism).

[0111] Incremental update unit timestamp rules: Based on the last update time of the data collection module, filter the patent data, market demand data and user-uploaded data added after that time point. The timestamp of patent data is based on the "publication date", and the timestamp of market demand data is based on the "release date".

[0112] The change summary of the version management unit includes: "Added XX invention patents (involving XX technical fields)", "Updated XX market demands (mainly from XX industry)", and "Corrected XX conversion benefit data (due to user completion)". Users can view all update records on the "Data Version History" page and can roll back to the 3 most recent historical versions.

[0113] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A platform for dynamic monitoring and visualized analysis of transformation benefits of scientific and technological achievements, characterized in that, include: The data acquisition module is used to collect data related to the transformation of scientific and technological achievements through multiple channels; The data includes invention patent data, market demand data, and conversion benefit evaluation data; The data preprocessing module, connected to the data acquisition module, is used to standardize the data related to the transformation of scientific and technological achievements collected by the data acquisition module. The intelligent analysis module is connected to the data preprocessing module and has a built-in fast classification algorithm for performing initial technical field classification of invention patents in the standardized data output by the data preprocessing module. The visualization module, connected to the intelligent analysis module, is used to dynamically display the analysis results output by the intelligent analysis module and the conversion benefit evaluation data collected by the data acquisition module in multi-dimensional charts. The user interaction module is connected to the data acquisition module, intelligent analysis module, and visualization module respectively, and is used to receive user operation commands for each module and provide corresponding results. The data update module, connected to the data acquisition module, intelligent analysis module, and visualization module, is used to automatically update the full / incremental data in the data acquisition module according to a preset cycle, and trigger the intelligent analysis module to re-execute the analysis process, and synchronize the update results to the visualization module. The analysis process includes a fast classification algorithm and, if necessary, a fine-tuning algorithm. 2.The platform of claim 1, wherein, The data acquisition module employs multiple data acquisition methods, including: By connecting to the patent office database and third-party search systems through the patent data interface unit, basic data such as application number, abstract, and IPC classification number of invention patents can be obtained; The market data crawling unit collects dynamic market demand data from corporate websites and industry databases. The user data upload unit receives patent conversion contracts, revenue reports, and other conversion benefit evaluation data uploaded by users, and supports one-click import of PDF and Excel formats. 3.The platform of claim 1, wherein, The standardization processing of the data preprocessing module includes: The units for the conversion benefit evaluation data are standardized, conversion revenue is converted to RMB, and the cost-benefit ratio is rounded to two decimal places. The 3σ criterion is used to eliminate outliers in patent implementation rate and cost-benefit ratio; The missing monthly conversion revenue data is filled in using linear interpolation, with the following formula: wherein, is the missing conversion revenue, , are the known revenues before and after the missing point, , is the time Interval point, For time points with missing data.

4. The platform for dynamic monitoring and visualization analysis of the benefits of technology transfer according to claim 2, characterized in that, The fast classification algorithm of the intelligent analysis module is a keyword weight algorithm based on TF-IDF, and the formula is as follows: in, Keywords In the invention patent text The weighted score in the middle, Keywords In the text Number of times it appears in For text Total number of occurrences of all keywords in the text This represents the total number of invention patents in the platform's database. For keywords The number of invention patents; When the keyword set of a certain technical field accumulates The sum is greater than the threshold At that time, the invention patent will be classified into this field; the cumulative score is in When the sample is fuzzy, it is determined to be a fuzzy sample.

5. The platform for dynamic monitoring and visualization analysis of the benefits of technology transfer according to claim 4, characterized in that, The intelligent analysis module also incorporates a fine-tuning algorithm, a text matching algorithm based on cosine similarity, used to optimize fuzzy samples obtained by the fast classification algorithm. These fuzzy samples are those whose cumulative scores in the fast classification are within a certain range. The formula for the invention patent within the interval is as follows: in, For invention patent text Technical Field Standard Texts similarity, , Text respectively , In the Word vector values ​​in the feature dimension This represents the total number of feature dimensions. when > If the invention patent is confirmed to belong to the original technical field, it will be reassigned to another technical field with the highest similarity.

6. The platform for dynamic monitoring and visualization analysis of the benefits of technology transfer according to claim 1, characterized in that, The multi-dimensional charts in the visualization module include: A heatmap showing the distribution of technical fields in statistics, using color depth to indicate the cost-benefit ratio. A matching network diagram of related categories, where node size represents patent implementation rate and line thickness represents degree of matching with market demand; The conversion trend line chart shows the relationship between annual application volume and conversion revenue in the technology field, and supports switching time dimensions.

7. The platform for dynamic monitoring and visualization analysis of the benefits of technology transfer according to claim 1, characterized in that, The user interaction module includes: The role recognition unit identifies roles such as university researchers, corporate technical specialists, and patent agents. The interface adaptation unit loads corresponding functional modules for different roles, such as displaying a patent application trend analysis module for university personnel. The operation feedback unit displays the processing time, number of results, and error messages in a pop-up window after the user completes data upload and query.

8. The platform for dynamic monitoring and visualization analysis of the benefits of technology transfer according to claim 3, characterized in that, The data update module includes: The trigger control unit supports both preset periodic updates and user-manually triggered updates. The incremental update unit filters incremental data from various channels in the data collection module based on timestamps. The linkage analysis unit calls the intelligent analysis module to perform classification algorithms and benefit evaluation analysis on incremental data; The version management unit records data update batches, times, and change summaries, and pushes update notifications to the user interaction module.

Images