CN119809529A - A management system for enterprise scientific and technological achievements - Google Patents
A management system for enterprise scientific and technological achievements Download PDFInfo
- Publication number
- CN119809529A CN119809529A CN202411616366.1A CN202411616366A CN119809529A CN 119809529 A CN119809529 A CN 119809529A CN 202411616366 A CN202411616366 A CN 202411616366A CN 119809529 A CN119809529 A CN 119809529A
- Authority
- CN
- China
- Prior art keywords
- text
- user
- unbrowsed
- vector
- word
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
本发明涉及科技成果信息管理技术领域，具体公开了一种用于企业科技成果管理系统，具体包括：基于当前的检索场景在标定周期内获取用户当前检索后的浏览文本集合和未浏览文本集合，通过行为数据将用户浏览文本划分为目标文本和非目标文本，通过目标文本与未浏览文本进行余弦相似度计算，确定用户对未浏览文本对应的特征值，根据兴趣特征值排序推荐给用户，通过对用户感兴趣文本的筛选提高用户的阅读效率，从而提高科技成果管理和转化的效率。
The present invention relates to the technical field of scientific and technological achievement information management, and specifically discloses a scientific and technological achievement management system for an enterprise, which specifically comprises: obtaining a browsed text set and an unbrowsed text set after a user's current search within a calibration period based on a current search scenario, dividing the user's browsed text into target text and non-target text through behavior data, calculating cosine similarity between the target text and the unbrowsed text, determining feature values corresponding to the unbrowsed text by the user, sorting and recommending the unbrowsed text to the user according to the feature values of interest, improving the user's reading efficiency by screening the texts of interest to the user, thereby improving the efficiency of scientific and technological achievement management and conversion.
Description
Technical Field
The invention relates to the field of scientific and technological achievement information management, in particular to a scientific and technological achievement management system for enterprises.
Background
Scientific achievements refer to knowledge products that people have gained through complex intellectual work in scientific and technical activities and that have some accepted academic or economic value. The technological achievement can obtain good economic, social or ecological environmental benefits, the connotation of the technological achievement is basically consistent with intellectual property and proprietary technology, and the technological achievement is an indispensable important component in intangible assets.
Currently, information technology is commonly employed by businesses and research institutions to assist in the management of technological achievements, such as database systems, intellectual property management software, and the like. A user can conduct data mining and text analysis through keyword searching, and relevant texts can be screened out from a large number of scientific and technological achievement documents.
Along with rapid development of science and technology, the number of scientific and technological achievements shows a remarkable growing trend, and users still face a large amount of texts after searching through keywords, wherein the texts have high similarity in content but have different key contents, so that the users need to spend a large amount of time to compare one by one, and the management and conversion efficiency of the scientific and technological achievements are affected.
Disclosure of Invention
The invention aims to provide a scientific and technological achievement management system for enterprises, which solves the technical problems that:
The aim of the invention can be achieved by the following technical scheme:
A system for managing a scientific achievement of an enterprise, comprising:
The data acquisition module is used for acquiring a browsing text set and an unbrown text set which are currently searched by a user in a calibration period T, and acquiring the interaction times J and the watching time T of any browsing text of the user, wherein the user interaction comprises clicking and sliding;
The text recommendation module is used for determining an interest value G corresponding to any browsing text through the watching time T and the user interaction times J of the user in the browsing text, and calculating the average value of the watching time And the average of the number of user interactionsCalculating a threshold G ́, when G is more than or equal to G ́, marking the text as a target text, when G is less than G ́, marking the text as a non-target text, generating a comprehensive text by all target texts, respectively preprocessing the comprehensive text and any one of the non-browsed texts to obtain a preprocessed comprehensive text word vector set and a preprocessed any one of the non-browsed text word vector sets, respectively aggregating according to the comprehensive text word vector set and the any one of the non-browsed text word vector sets to obtain a comprehensive text overall vector A and a non-browsed text overall vector D n, wherein n is any one of the non-browsed texts;
and obtaining an included angle cosine value Q n by calculating the integrated text overall vector A and any one of the unbrown text overall vectors D n, obtaining all unbrown text characteristic values X according to the included angle cosine value Q n, sorting all unbrown texts according to the corresponding characteristic values X from large to small, generating a sorting result, and recommending the unbrown texts to a user according to the sorting result.
The data acquisition module further comprises a step of screening a text database according to the search words input by the user to obtain search data, and a step of sorting the search data according to the publication sequence of the text and the relevance of the article titles and the search words.
In the data processing module, the preprocessing comprises sentence division of the text, word segmentation, stop word removal and normalization processing of the divided text sentences.
In the text recommendation module, the specific process of aggregating the word vector sets comprises the following steps:
Embedding word vectors of all different dimensions into the highest dimensional representation in the word vector of all target texts and the word vector of any unbrown text, and according to a calculation formula ,Wherein, C a is any a word vector in all target texts, b is the total number of all target texts, C f is any f word vector in any unbrown text, and r is the total number of any unbrown text.
In the text recommendation module, the specific acquisition process of the target text comprises the following steps:
The interest value G corresponding to any browsing text is determined through the watching time T and the user interaction times J of the user in any browsing text, and the average value of the watching time is used for determining And the average of the number of user interactionsThe threshold G ́ is calculated as follows:
;;
G=Ti+Ji/y; G ́=+/y;
Wherein i is any browsing text, y is a preset coefficient, when G is more than or equal to G ́, the text is marked as a target text, and when G < G ́, the text is marked as a non-target text.
In the text recommendation module, the text cosine similarity has the following calculation formula:
;
Where a x D n represents the dot product of vectors a and D n, and |a| and |d n | represent the modular length of vector a and vector D n.
In the text recommendation module, the cosine value Q n of the passing included angle is obtained according to a calculation formula X n=k*Qn to obtain a characteristic value X n, wherein k is a preset coefficient, k is greater than 0, and the characteristic value X n represents a characteristic value of a user on any nth unbrown text.
In the data acquisition module, if the initial time of the calibration period is not first searched, acquiring all historical integrated text overall vectors in the historical calibration period, calculating cosine similarity Q ́ of the current integrated text overall vector A and all historical integrated text overall vectors, screening out historical integrated text overall vectors with Q ́ being more than or equal to 0.7 and marking the historical integrated text overall vectors as similar vectors, acquiring all historical integrated text word vector sets and the current integrated text word vector sets corresponding to the similar text vectors, comparing the two word vector sets, screening out repeated word vectors and generating a repeated word vector set, polymerizing the repeated word vector sets to obtain an overall correction vector, calculating the overall correction vector and any unbeared text overall vector D n to obtain an included angle cosine value Q 1, calculating a new feature value X xn through the included angle cosine value Q 1 and a feature value X n of a current user, sorting the unbeared texts according to the new feature value X, generating a sorting formula, recommending corresponding texts to the user according to the sorting result, and calculating as follows:
;
wherein p and m are preset coefficients, and p > m >0;
if the starting time of the calibration period is the first search, no correction is performed.
The method has the advantages that a browsing text set and an unbrown text set of a user after current retrieval are obtained in a calibration period based on a current retrieval scene, browsing texts of the user are divided into target texts and unbrown texts through behavior data, cosine similarity calculation is conducted on the target texts and the unbrown texts, characteristic values corresponding to the unbrown texts by the user are determined, the unbrown texts are recommended to the user according to the characteristic value sequence, if the initial time of the calibration period is not the first retrieval, a correction module is started, the current characteristic values are corrected, new characteristic values are calculated, the unbrown texts are recommended to the user according to the characteristic value sequence, reading efficiency of the user is improved through screening of texts of interest of the user, and therefore efficiency of scientific and technological achievement management and conversion is improved.
Drawings
The invention is further described below with reference to the accompanying drawings.
FIG. 1 is a schematic diagram of an architecture for an enterprise scientific and technological process management system according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention is based on the scene after the user searches through the key word, through the second screening of the user to the text which is left unbrown and is obtained after the search, the browsing text set and unbrown text set after the user searches are obtained in the calibration period based on the current search scene, the browsing text of the user is divided into the target text and the unbrown text through the behavior data, the cosine similarity calculation is carried out on the target text and the unbrown text, the characteristic value corresponding to the unbrown text of the user is determined, the characteristic value is ordered and recommended to the user according to the characteristic value, if the initial moment of the calibration period is not the first search, the correction module is started to correct the current characteristic value, the new characteristic value is calculated, the unbrown text is recommended to the user according to the characteristic value ordering, the reading efficiency of the user is improved through the screening of the text which is interested by the user, and the efficiency of scientific and technological achievement management and conversion is improved.
Referring to fig. 1, the present invention is a system for managing scientific achievements of enterprises, comprising:
The data acquisition module is used for acquiring a browsing text set and an unbrown text set which are currently searched by a user in a calibration period T, and acquiring the interaction times J and the watching time T of any browsing text of the user, wherein the user interaction comprises clicking and sliding;
The text recommendation module is used for determining an interest value G corresponding to any browsing text through the watching time T and the user interaction times J of the user in the browsing text, and calculating the average value of the watching time And the average of the number of user interactionsCalculating a threshold G ́, when G is more than or equal to G ́, marking the text as a target text, when G is less than G ́, marking the text as a non-target text, generating a comprehensive text by all target texts, respectively preprocessing the comprehensive text and any one of the non-browsed texts to obtain a preprocessed comprehensive text word vector set and a preprocessed any one of the non-browsed text word vector sets, respectively aggregating according to the comprehensive text word vector set and the any one of the non-browsed text word vector sets to obtain a comprehensive text overall vector A and a non-browsed text overall vector D n, wherein n is any one of the non-browsed texts;
and obtaining an included angle cosine value Q n by calculating the integrated text overall vector A and any one of the unbrown text overall vectors D n, obtaining all unbrown text characteristic values X according to the included angle cosine value Q n, sorting all unbrown texts according to the corresponding characteristic values X from large to small, generating a sorting result, and recommending the unbrown texts to a user according to the sorting result.
It will be appreciated that the dwell time T is the logic that determines the user's interest in the text by the number J of user interactions, as the user's interest in a segment of text increases the interaction with the text and the dwell time.
It can be understood that the text of interest of the user is assembled into a comprehensive text, the whole vector A is used as a center vector, then cosine similarity between the rest of the unbrown text and the comprehensive text is calculated to obtain the similar text of interest of the user, when the search is not performed for the first time in the current calibration period, the historical comprehensive text vector is obtained, the whole vector A of the current comprehensive text is used as the center vector, the highly similar comprehensive text is screened out, then the repeated word vectors are screened out through the word vectors of all the similar comprehensive texts and the word vector of the current comprehensive text, the repeated word vectors are aggregated into corrected whole vectors, finally the cosine similarity between the rest of the unbrown text and the corrected whole vector is calculated again to correct the characteristic value of the current user.
In a preferred embodiment of the present invention, the data obtaining module further includes screening the text database according to the search terms input by the user to obtain search data, and sorting the search data according to the publication order of the text, and the relevance between the article titles and the search terms.
It can be understood that the invention is based on the scene of the user after the keyword is searched, and the text database is screened according to the search word input by the user, so that the search data highly related to the user requirement can be ensured to be obtained, thereby improving the relevance and the accuracy of the information.
In a preferred embodiment of the present invention, in the data processing module, the preprocessing includes sentence division for the text, word segmentation for the divided text sentence, stop word removal, and normalization.
It can be appreciated that by sentence division of text, text structure can be more accurately analyzed and understood, thereby improving accuracy of subsequent processing. The keyword segmentation and the stop word removal can effectively extract key information in the text, reduce interference of irrelevant contents, and enable subsequent data analysis to be more efficient. The normalization processing can unify words in different forms into a standard format, and reduces complexity caused by word diversity, so that subsequent comparison and analysis are facilitated.
In a preferred embodiment of the present invention, in the text recommendation module, the specific process of aggregating a set of word vectors:
Embedding word vectors of all different dimensions into the highest dimensional representation in the word vector of all target texts and the word vector of any unbrown text, and according to a calculation formula ,Wherein, C a is any a word vector in all target texts, b is the total number of all target texts, C f is any f word vector in any unbrown text, and r is the total number of any unbrown text.
When the method is used for understanding, word vectors with different dimensions are embedded into unified high-dimensional representation, so that the consistency of data is ensured, and the subsequent calculation and comparison are facilitated. The aggregation process may reduce computational complexity so that the system can generate recommendations more quickly when processing large-scale text.
In a preferred embodiment of the present invention, in the text recommendation module, a specific acquisition process of the target text:
The interest value G corresponding to any browsing text is determined through the watching time T and the user interaction times J of the user in any browsing text, and the average value of the watching time is used for determining And the average of the number of user interactionsThe threshold G ́ is calculated as follows:
;;
G=Ti+Ji/y; G ́=+/y;
Wherein i is any browsing text, y is a preset coefficient, when G is more than or equal to G ́, the text is marked as a target text, and when G < G ́, the text is marked as a non-target text.
It can be appreciated that the interest value is calculated through the watching time and the interaction times, so that the real interest of the user to the text can be reflected more accurately, and the individuation of the recommendation is enhanced. The dynamic threshold G' is calculated by using the mean value, so that the recommendation system can adapt to the behavior modes and preferences of different users, and has higher flexibility and adaptability. By focusing on the content which is really interested by the user, the interference of irrelevant information can be reduced, and the satisfaction and the use experience of the user are improved.
In a preferred embodiment of the present invention, in the text recommendation module, a calculation formula of the text cosine similarity is as follows:
;
Where a x D n represents the dot product of vectors a and D n, and |a| and |d n | represent the modular length of vector a and vector D n.
In a preferred embodiment of the present invention, in the text recommendation module, the through-included angle cosine value Q n obtains a feature value X n according to a calculation formula X n=k*Qn, where k is a preset coefficient, and k >0, and the feature value X n represents a feature value of a user on any nth unviewed text.
In a preferred embodiment of the present invention, in the data acquisition module, if the initial time of the calibration period is not the first search, all the history integrated text overall vectors in the history calibration period are acquired, the cosine similarity Q ́ is calculated for the current integrated text overall vector a and all the history integrated text overall vectors, the history integrated text overall vectors with Q ́ equal to or greater than 0.7 are screened out and marked as similar vectors, all the history integrated text word vector sets and the current integrated text word vector sets corresponding to the similar text vectors are acquired, the two word vector sets are compared and screened out to generate a repeated word vector set, the repeated word vector sets are aggregated to obtain an overall correction vector, the overall correction vector and any one of the unbrown text overall vectors D n are calculated to obtain an included angle cosine value Q 1, a new feature value X xn is calculated through the included angle cosine value Q 1 and the feature value X n of the current user, all the unbrown texts are ordered from large to small according to the new feature value X, a sequencing result is generated, and the text corresponding to a user recommendation formula is calculated according to the sequencing result is calculated as:
;
wherein p and m are preset coefficients, and p > m >0;
if the starting time of the calibration period is the first search, no correction is performed.
It can be understood that by acquiring the historical comprehensive text vector, the accuracy of current recommendation can be improved by utilizing the existing data, and the learning ability of the system is enhanced. The cosine similarity Q' is used for screening similar texts, and historical target texts similar to the current target text in terms of semantics can be effectively identified, so that the recommendation correlation is improved. And comparing and screening the repeated word vectors and generating an overall correction vector, redundant information can be reduced, and the final recommendation result is more refined and effective.
The foregoing describes one embodiment of the present invention in detail, but the description is only a preferred embodiment of the present invention and should not be construed as limiting the scope of the invention. All equivalent changes and modifications within the scope of the present invention are intended to be covered by the present invention.
Claims (8)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411616366.1A CN119809529A (en) | 2024-11-13 | 2024-11-13 | A management system for enterprise scientific and technological achievements |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411616366.1A CN119809529A (en) | 2024-11-13 | 2024-11-13 | A management system for enterprise scientific and technological achievements |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN119809529A true CN119809529A (en) | 2025-04-11 |
Family
ID=95266960
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411616366.1A Withdrawn CN119809529A (en) | 2024-11-13 | 2024-11-13 | A management system for enterprise scientific and technological achievements |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119809529A (en) |
-
2024
- 2024-11-13 CN CN202411616366.1A patent/CN119809529A/en not_active Withdrawn
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220284234A1 (en) | Systems and methods for identifying semantically and visually related content | |
| Yuan et al. | Video summarization by learning deep side semantic embedding | |
| US20220035827A1 (en) | Tag selection and recommendation to a user of a content hosting service | |
| AU2017268661B2 (en) | Large-scale image search and tagging using image-to-topic embedding | |
| CN107220365B (en) | Precise recommendation system and method based on collaborative filtering and parallel processing of association rules | |
| US8027977B2 (en) | Recommending content using discriminatively trained document similarity | |
| CN112035658B (en) | Enterprise public opinion monitoring method based on deep learning | |
| US8433140B2 (en) | Image metadata propagation | |
| WO2020211566A1 (en) | Method and device for making recommendation to user, computing apparatus, and storage medium | |
| US8856114B2 (en) | Ranking results of multiple intent queries | |
| CN112434151A (en) | Patent recommendation method and device, computer equipment and storage medium | |
| US20110029561A1 (en) | Image similarity from disparate sources | |
| CN101727454A (en) | Method and system for automatic classification of objects | |
| CN111460251A (en) | Data content personalized push cold start method, device, equipment and storage medium | |
| CN117874347A (en) | Content recommendation technology based on business characteristics | |
| EP4579489A1 (en) | Query intent specificity | |
| CN119809529A (en) | A management system for enterprise scientific and technological achievements | |
| CN117056392A (en) | Big data retrieval service system and method based on dynamic hypergraph technology | |
| CN117725149A (en) | Query method, device and equipment for content feedback | |
| WO2023079703A1 (en) | Processing execution system, processing execution method, and program | |
| Chaudhuri et al. | Advanced feature identification towards research article recommendation: a machine learning based approach | |
| Sowbhagya et al. | User profiling for web personalization | |
| Liu | Implicit semantic-based personalized micro-videos recommendation | |
| CN118035549B (en) | Computer cloud data optimization method and system based on deep learning | |
| KR101137491B1 (en) | System and Method for Utilizing Personalized Tag Recommendation Model in Web Page Search |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20250411 |