CN-121981677-A - Knowledge propaganda system based on big data

CN121981677ACN 121981677 ACN121981677 ACN 121981677ACN-121981677-A

Abstract

The invention relates to the technical field of knowledge propaganda and discloses a knowledge propaganda system based on big data, which comprises the following steps of S1, constructing a multi-source data acquisition system, acquiring full-dimension standardized data, acquiring audience data and resource data, S2, preprocessing the data, constructing a dynamic audience image, S3, pushing personalized knowledge, constructing an individual pushing engine and a group pushing engine based on the audience image and the resource label, S4, evaluating full-dimension effects, constructing an evaluation system combining four-level quantization indexes of a touch layer, an interaction layer, a conversion layer and a retention layer with NLP (non-line-of-sight) qualitative analysis, and S5, performing closed-loop optimization iteration, reversely optimizing contents, pushing strategies, audience images and the data acquisition system based on the comprehensive effect scores. In the invention, a knowledge propaganda system of a whole flow is constructed, a time sequence analysis algorithm is introduced into audience knowledge dynamic capture, the limitation of the traditional static image is broken through, and the knowledge propaganda and audience knowledge dynamic real-time adaptation is realized.

Inventors

CHEN CHEN
WANG SHANSHAN
LV RUI
LIU QIANSHU
XIE MENG

Assignees

山东理工职业学院

Dates

Publication Date: 20260505
Application Date: 20260122

Claims (9)

1. A big data based knowledge propaganda system comprising the steps of: S1, constructing a multi-source data acquisition system, acquiring full-dimension standardized data, and acquiring audience data and resource data, wherein the audience data comprises basic attribute data, behavior data, preference data and knowledge dynamic association data, and the resource data is subjected to multi-dimension labeling processing to construct a dynamically updated standardized knowledge resource base; S2, data preprocessing and dynamic audience portrait construction, namely cleaning, integrating, normalizing and desensitizing the audience data and the resource data, dividing an audience group by adopting an algorithm, extracting static characteristics, constructing an audience knowledge demand time sequence model by adopting the algorithm, capturing audience knowledge dynamic change through a timing update and event trigger update mechanism, dynamically updating the audience portrait, and reversely optimizing a resource tag system; S3, personalized knowledge pushing, namely constructing an individual pushing engine and a group pushing engine based on the audience portraits and the resource labels, screening the adaptive content through a cosine similarity algorithm, combining the audience pushing preference and the multi-channel characteristics, performing personalized and multi-channel accurate pushing, and simultaneously setting a repeated pushing control and recall mechanism; S4, full-dimensional effect evaluation, namely, an evaluation system combining four-level quantitative indexes of a reach layer, an interaction layer, a conversion layer and a retention layer with NLP (non-line-of-sight) mass analysis is constructed, a comprehensive effect evaluation model is constructed by adopting a weighted summation algorithm, and comprehensive effect scores of contents, channels and groups are generated; s5, closed-loop optimization iteration is performed, and reverse optimization content, pushing strategies, audience images and a data acquisition system are scored based on the comprehensive effect.
2. The knowledge propaganda system based on big data according to claim 1, wherein the multidimensional labeling in step S1 includes a topic label, a form label, a difficulty label, a failure label and an adaptation group label, and the knowledge resource base is provided with an audit mechanism for dynamically updating a current event hotspot.
3. The big data-based knowledge propaganda system according to claim 2, characterized in that in step S2, a K-Means clustering algorithm is adopted, a hierarchical clustering algorithm is combined to optimize a clustering result, audience groups are divided based on basic attribute data and preference data, a plurality of audience groups with clear audience characteristics are obtained, and each audience group is marked with a core demand label.
4. The knowledge propaganda system based on big data according to claim 3, wherein the capturing audience knowledge dynamic change in the step S2 is characterized in that a time sequence analysis algorithm is combined with a collaborative filtering algorithm and a decision tree algorithm to construct an audience knowledge demand time sequence model, and the audience knowledge demand time sequence model is updated by timing update and event triggering update, and meanwhile, a knowledge demand change early warning mechanism is established, when the attention of a certain audience group to a specific theme is improved in a short period, the audience knowledge demand dynamic change is automatically marked as an important attention group, and a group core demand label is updated.
5. The knowledge propaganda system based on big data according to claim 4, wherein the reverse resource label in step S2 adds a new sub-division label to the topic with high audience attention, merges or deletes the label with poor suitability, and prioritizes the content of the resource library by combining with the dynamically updated audience core demand label based on the audience image feature and the knowledge demand change.
6. The big data-based knowledge propaganda system according to claim 5 in which the individual pushing engine in step S3 calculates the matching degree of the audience and the resource by a cosine similarity algorithm in combination with the static characteristics, the dynamic demand labels and the historical behavior data, screens the content of the matching degree Top5-8 to be included in the pushing list, and simultaneously refers to the pushing time preference and the common platform of the audience to optimize the pushing time and the pushing format.
7. The big data based knowledge propaganda system according to claim 6 in which the group pushing engine in step S3 pushes content meeting group commonality requirements and dynamic preferences for the same audience group while integrating personalized elements.
8. The knowledge propaganda system based on big data according to claim 7, characterized in that in the step S3, a short video platform, a WeChat public number, a community APP, an enterprise OA system, an off-line terminal screen and a push channel are butted, a content format is optimized according to channel characteristics, push intervals of the same content to the same audience in the repeated push control mechanism are higher than 72 hours, and the audience not viewing push content in the recall mechanism pushes simplified content twice through other common platforms after 24 hours.
9. The knowledge propaganda system based on big data according to claim 8, characterized in that the reach layer comprises content reach rate and channel reach rate, the interaction layer comprises content completion rate, interaction rate and secondary propagation rate, the conversion layer comprises cognitive enhancement rate, participation enhancement rate and knowledge acceptance quantification value, the retention layer comprises audience retention rate and group liveness, the qualitative analysis adopts a natural language processing algorithm, combines an emotion analysis model and a theme extraction model to deeply mine audience comments, messages, questionnaire feedback and offline questionnaire contents, and the comprehensive effect scoring model assigns weights according to quantification index importance, integrates quantification index standardization scores and quality index quantification scores in a weighting mode to generate comprehensive effect scores of each propaganda content, each pushing channel and each audience group.

Description

Knowledge propaganda system based on big data Technical Field The invention relates to the technical field of knowledge propaganda, in particular to a knowledge propaganda system based on big data. Background Currently, the traditional knowledge propaganda method still uses 'unified pushing and wide coverage' as core logic, relies on offline lectures, poster display boards and themes class meeting, is developed in single or combined forms such as online public number pushing, short video matrix, community forwarding and the like, can realize basic propaganda coverage, but has obvious limitations in the aspects of accuracy, effectiveness, dynamic performance and the like, and is difficult to adapt to the core requirements of propaganda 'accurate, efficient, normalized and personalized', and the specific problems are as follows: in the prior art, although some schemes try to apply big data technology to the education field, the big data technology focuses on scenes such as academic score analysis, accurate pushing of teaching resources, student management early warning and the like, and the whole-flow big data application scheme aiming at knowledge propaganda is deficient. The few schemes related to the combination of knowledge propaganda and big data only stay at the level of simple content push optimization. Therefore, a knowledge propaganda system based on big data is needed to break the pain point and realize the accurate, dynamic and efficient upgrading of knowledge propaganda. Disclosure of Invention The invention aims to solve the defect that in the prior art, knowledge propaganda only stays at a simple content pushing optimization level, and provides a knowledge propaganda system based on big data. In order to achieve the above purpose, the invention adopts the following technical scheme that the method comprises the following steps: S1, constructing a multi-source data acquisition system, acquiring full-dimension standardized data, and acquiring audience data and resource data, wherein the audience data comprises basic attribute data, behavior data, preference data and knowledge dynamic association data, and the resource data is subjected to multi-dimension labeling processing to construct a dynamically updated standardized knowledge resource base; S2, data preprocessing and dynamic audience portrait construction, namely cleaning, integrating, normalizing and desensitizing the audience data and the resource data, dividing an audience group by adopting an algorithm, extracting static characteristics, constructing an audience knowledge demand time sequence model by adopting the algorithm, capturing audience knowledge dynamic change through a timing update and event trigger update mechanism, dynamically updating the audience portrait, and reversely optimizing a resource tag system; S3, personalized knowledge pushing, namely constructing an individual pushing engine and a group pushing engine based on the audience portraits and the resource labels, screening the adaptive content through a cosine similarity algorithm, combining the audience pushing preference and the multi-channel characteristics, performing personalized and multi-channel accurate pushing, and simultaneously setting a repeated pushing control and recall mechanism; S4, full-dimensional effect evaluation, namely, an evaluation system combining four-level quantitative indexes of a reach layer, an interaction layer, a conversion layer and a retention layer with NLP (non-line-of-sight) mass analysis is constructed, a comprehensive effect evaluation model is constructed by adopting a weighted summation algorithm, and comprehensive effect scores of contents, channels and groups are generated; s5, closed-loop optimization iteration is performed, and reverse optimization content, pushing strategies, audience images and a data acquisition system are scored based on the comprehensive effect. As a further description of the above technical solution: The multidimensional labeling in the step S1 comprises a theme label, a form label, a difficulty label, a failure label and an adaptation group label, wherein an auditing mechanism is arranged in the knowledge resource base, and a current event hot spot is dynamically updated. As a further description of the above technical solution: In the step S2, a K-Means clustering algorithm is adopted, a hierarchical clustering algorithm is combined to optimize a clustering result, audience groups are divided based on basic attribute data and preference data, a plurality of audience groups with clear audience characteristics are obtained, and each audience group is marked with a core demand label. As a further description of the above technical solution: The step S2 of capturing audience knowledge dynamic change adopts a time sequence analysis algorithm, a collaborative filtering algorithm and a decision tree algorithm to construct an audience knowledge demand time sequence model, and updates and event triggers the