Intelligent office information pushing method and system
By collecting data from multiple channels in the smart office information push system and calculating the intelligent push index using machine learning algorithms, the accurate push of information and dynamic adaptation of user needs are achieved, and the problems of flooding information and low correlation in traditional systems are solved. Improve user experience and work efficiency.
Patent Information
- Application Number
- CN202410961594.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional information push systems are difficult to personalize push according to users' real-time needs and preferences, resulting in information flooding and low correlation, poor user experience, and lack of effective analysis and utilization of user feedback, making it difficult to continuously optimize push strategies.
A smart office information push system was designed to collect data from multiple channels, including office software, project management tools, conference software, HR management system, etc., and perform data preprocessing and feature extraction, and use machine learning algorithms to calculate intelligent push index. Generate personalized prediction results, push relevant information in real time, and optimize push strategies through user feedback.
It realizes accurate push of information, improves the efficiency of users to obtain information, reduces irrelevant information interference, dynamically adapts to user needs, improves user satisfaction and work efficiency, provides intelligent decision-making support, promotes the digital transformation and intelligent office of enterprises .
Smart Images

Figure CN118921402A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of information push, and in particular to a smart office information push method and system. Background Art
[0002] In large and medium-sized enterprises, office information push is an important means to improve employee work efficiency and communication and collaboration. However, traditional information push methods often rely on manual operations or systems based on simple rules, which makes it difficult to accurately grasp the actual needs and preferences of each employee. With the development of informatization and digitalization, smart office information push has become a new trend. Through intelligent technology and data analysis, more accurate and efficient information push can be achieved.
[0003] Traditional push systems have problems such as excessive information, low relevance of push content, and poor user experience. First, it is difficult for traditional systems to perform personalized push based on users' real-time needs and preferences, which often results in a large amount of irrelevant information being pushed, interfering with users' normal work. Second, traditional systems lack effective analysis and utilization of user feedback, making it difficult to continuously optimize push strategies.
[0004] In the Chinese invention patent with application publication number CN116562814A, a smart office information push system and method are disclosed, which involves the field of information push technology, obtains the identity information and push preferences of employees, and establishes an employee information set after aggregation. According to whether the employee is a resident employee, a resident employee evaluation value and a non-resident employee coefficient are generated respectively; according to the relationship between the resident employee evaluation value and the non-resident employee coefficient and the corresponding threshold, the employees are divided into gears, and according to the generated different gears; according to the selected push strategy, combined with the employee's push preferences, notification information is sent to the employees in an orderly manner; within a preset period, the average time for employees to confirm the received notification information is counted, and the employee's push preference is corrected according to the length of the confirmation time. By judging the push priority of the employee, the sending and obtaining of the notification information can be carried out in an orderly manner, and there will be no omissions due to blockage, which guarantees the orderly development of work.
[0005] In the technical solutions recorded in the above applications, data collection is limited to the identity information of employees, push preferences, and the frequency of employees entering the office area, etc. The data source is relatively single and the information dimension is limited. When generating the evaluation value Zz of resident employees and the coefficient Fs of non-resident employees, only a limited number of parameters (such as department profit ratio, years of service, job salary, frequency of entry and confirmation time) are considered, and there is a lack of extraction and analysis of broader features. The push strategy is mainly based on the evaluation value and gear of employees, lacking a more detailed and personalized push strategy. The push preference correction is also limited to the length of the confirmation time, the adjustment is relatively simple, and the feedback mechanism is relatively simple. The push preference is mainly adjusted by the confirmation time. There is a lack of analysis of details such as user click behavior and reading time. The system optimization ability is limited, mainly relying on the evaluation value Zz of employees and the coefficient Fs of non-resident employees, dividing the gear according to the threshold, and pushing based on the gear. This system collects and processes data more comprehensively, extracts richer features, uses advanced machine learning algorithms for modeling and prediction, provides more personalized and accurate information push, and optimizes the system through continuous user feedback, significantly improving the effect of smart office information push. Summary of the invention
[0006] 1. Technical issues to be resolved
[0007] In view of the deficiencies in the prior art, the present invention provides a smart office information push method and system, which solves the problems mentioned in the background technology.
[0008] (II) Technical solution
[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: a smart office information push system, including a data collection module, a data preprocessing module, a feature extraction module, a prediction module, an information push module and a user feedback module;
[0010] The data collection module is used to collect data from office software, project management tools, conference software, HR management system, browser plug-in, internal enterprise social platform, time management software, email client and collaboration tool to obtain: a first data group, a second data group and a third data group;
[0011] The data preprocessing module is used to clean, remove duplicates, convert formats and standardize the collected first data group, second data group and third data group, and then transmit them to the feature extraction module;
[0012] The feature extraction module is used to extract key features, user behavior patterns, keywords and usage frequencies from the preprocessed data, establish a training model, and use machine learning algorithms, collaborative filtering, content recommendation and deep learning to calculate and obtain: intelligent push index Znts;
[0013] The prediction module is used to generate personalized prediction results and predict user needs and preferences by comparing the smart push index Znts with the preset push threshold N and the preset push threshold M;
[0014] The information push module is used to push information related to user needs in real time according to the results of the prediction model, including documents, meeting notices and industry trends, to achieve accurate information push and improve the efficiency of user acquisition of information;
[0015] The user feedback module is used to collect user feedback on the pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy based on the feedback results.
[0016] Preferably, the data collection module includes a work situation collection unit, a personal interest situation collection unit and a behavior habit collection unit;
[0017] The work situation collection unit is used to collect task completion time, task priority, project progress, number of times a file is used, and employee activity in the meeting through project management tools, document management systems, and conference software, and obtain: expected completion time Yjsj, actual completion time Sjsj, access rate of a specific document Wdfw, editing frequency of a specific document Wdbj, progress percentage of the current project Jdbf, and conference participation activity Hyhy as a first data group;
[0018] The personal interest situation collection unit is used to collect document reading time Ydsj, web page stay time Tlsj, topic click rate Ztdj and skill application frequency Jnyy as the second data group through the document management system, browser plug-in and internal enterprise social platform;
[0019] The behavior habit collection unit is used to collect: daily working time Rcgz, application use time Yysy, work and rest ratio Sjbl, email response time Xysj and collaboration tool use frequency Gjsy as the third data group through time management software, email client and collaboration tools.
[0020] Preferably, the data preprocessing module includes a data preprocessing unit;
[0021] The data preprocessing unit is responsible for performing comprehensive data preprocessing on the first data group, the second data group and the third data group. The preprocessing steps include: removing noise data, processing missing values, correcting outliers, identifying and deleting duplicate data, and converting data from different sources into a unified format, and transmitting the preprocessed data to the feature extraction module.
[0022] Preferably, the feature extraction module includes a feature extraction unit and a model training unit;
[0023] The feature extraction unit is responsible for extracting key features from the cleaned and standardized data, including user behavior patterns, keywords and usage frequency, and removing redundant or irrelevant features using a feature selection algorithm;
[0024] The user behavior pattern is used to analyze the user's operation log and extract the behavior pattern, including the distribution of working time, application usage time and document reading time;
[0025] The keyword extraction is used to extract keywords from documents and communication content, using natural language processing (NLP) technology, TF-IDF and word vectors;
[0026] The usage frequency is used to calculate the usage frequency of various documents, applications and tools, and to count the number of visits and edits and meeting participation;
[0027] The model training unit is used to use the extracted and selected features, and to train the model using a variety of machine learning algorithms. The trained model is used to calculate: the smart push index Znts, the work content relevance coefficient Gzxg, the interest matching coefficient Xqpp, and the behavior habit coefficient Xwxg;
[0028] The smart push index Znts is calculated by the following formula:
[0029] Znts=Gzxg*α+Xqpp*β+Xwxg*χ;
[0030] In the formula, Znts represents the smart push index, Gzxg represents the work content relevance coefficient, Xqpp represents the interest matching coefficient, Xwxg represents the behavior habit coefficient, α, β and χ represent the proportional coefficients of the work content relevance coefficient Gzxg, the interest matching coefficient Xqpp and the behavior habit coefficient Xwxg respectively;
[0031] Among them, 0≤α≤1, 0≤β≤1, 0≤χ≤1, and α+β+χ≤1.0;
[0032] The work content relevance coefficient Gzxg is calculated by the following formula:
[0033]
[0034] In the formula, Yjsj represents the expected completion time, Sjsj represents the actual completion time, Wdfw represents the access rate of a specific document, Wdbj represents the editing frequency of a specific document, Jdbf represents the progress percentage of the current project, and Hyhy represents the meeting participation activity. It represents the ratio of the expected completion time to the actual completion time. represents the inverse of the access rate of a specific document, Represents the inverse of how often a particular document has been edited.
[0035] Preferably, the interest matching coefficient Xqpp is calculated by the following formula:
[0036]
[0037] Where Ydsj represents the document reading time, Tlsj represents the webpage dwelling time, Ztdj represents the topic click-through rate, Jnyy represents the skill application frequency, the logarithm of the document reading time log(Ydsj+1) represents taking the logarithm of the document reading time to reduce the impact of extreme values, and the logarithm of the webpage dwelling time log(Tlsj+1) represents taking the logarithm of the webpage dwelling time to reduce the impact of extreme values.
[0038] Preferably, the behavior habit coefficient Xwxg is calculated by the following formula:
[0039]
[0040] In the formula, Rcgz represents daily working time, Yysy represents application usage time, Sjbl represents the ratio of work to rest, Xysj represents email response time, Gjsy represents the frequency of collaboration tool use, and the ratio of daily working time to application usage time is Reflects the ratio of working time to application usage time, balances work efficiency, and is inversely proportional to email response time Represents the inverse of email response time.
[0041] Preferably, the prediction module includes a prediction evaluation unit;
[0042] The prediction evaluation unit is used to compare the intelligent push index Znts with the preset push threshold N and the preset push threshold M to generate a personalized prediction result:
[0043] When the smart push index Znts> the preset push threshold N, the first evaluation level is obtained, and the system prioritizes pushing information related to user needs. Through feature extraction and model training, it ensures that the smart push index Znts reflects the accuracy of user needs.
[0044] When the preset push threshold N ≥ smart push index Znts ≥ preset push threshold M, the second evaluation level is obtained. The system pushes a balanced variety of information content based on user interests and historical behaviors, adds personalized setting options, allows users to adjust push preferences, including topic preferences and frequency preferences, introduces a multi-level push strategy, and flexibly adjusts push content based on users' different interest areas and behavior patterns;
[0045] When the smart push index Znts is less than the preset push threshold M, the third evaluation level is obtained, the user's browsing history, search preferences and click patterns are re-analyzed, and user-defined push preference settings are provided, including topic selection, push frequency and time preferences. A rapid response mechanism is established, and the push strategy is iterated according to real-time feedback to ensure timely adjustment and optimization of push content.
[0046] Preferably, the information push module includes a real-time push unit;
[0047] The real-time push unit is used to analyze the user's current needs and preferences based on the results of the prediction model, accurately push documents, meeting notifications and industry dynamics to the user, collect user feedback on the pushed information, and transmit the feedback information to the user feedback module.
[0048] Preferably, the user feedback module includes a feedback data collection unit and a feedback analysis unit;
[0049] The feedback data collection unit is used to monitor and record the user's click behavior on the pushed information, track the user's stay time on the pushed content, provide a user rating function, record the user's evaluation of the pushed content, and count the number of clicks on each piece of information and the user's reading time for each piece of information;
[0050] The feedback analysis unit is used to collect and analyze click-through rate, reading time and user rating data, identify user preferences and needs, and adjust the push strategy according to the feedback data analysis results, including push frequency, push time and push content, and optimize the personalized settings of push content.
[0051] The smart office information push method includes the following steps:
[0052] Step 1: Collect data from office software, project management tools, conference software, HR management system, browser plug-ins, internal enterprise social platforms, time management software, email clients, and collaboration tools to obtain: the first data group, the second data group, and the third data group;
[0053] Step 2: The collected first data group, second data group and third data group are cleaned, deduplicated, format converted and standardized, and then transmitted to the feature extraction module;
[0054] Step 3: Extract key features, user behavior patterns, keywords and usage frequency from the preprocessed data, build a training model, and use machine learning algorithms, collaborative filtering, content recommendation and deep learning to calculate and obtain: smart push index Znts;
[0055] Step 4: Generate personalized prediction results by comparing the smart push index Znts with the preset push threshold N and the preset push threshold M to predict the user's needs and preferences;
[0056] Step 5: Based on the results of the prediction model, push information related to user needs in real time, including documents, meeting notices, and industry trends, to achieve accurate information push and improve the efficiency of user information acquisition;
[0057] Step 6: The user feedback module is used to collect user feedback on the pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy based on the feedback results.
[0058] (III) Beneficial effects
[0059] The present invention provides a smart office information push method and system, which has the following beneficial effects:
[0060] (1) When the system is running, data is collected from multiple channels to obtain: a first data group, a second data group and a third data group. The collected first data group, second data group and third data group are pre-processed and transmitted to the feature extraction module to establish a training model, and the intelligent push index Znts is calculated and obtained. The intelligent push index Znts is compared with the preset push threshold N and the preset push threshold M to generate personalized prediction results, predict user needs and preferences, and according to the results of the prediction model, push information related to user needs in real time, including documents, meeting notices and industry trends, to achieve accurate information push and improve the efficiency of users in obtaining information. The user feedback module is used to collect user feedback on the pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy according to the feedback results.
[0061] (2) Through data preprocessing steps, noise data is removed, missing values are processed, and outliers are corrected to ensure data integrity and consistency, unify data formats, and enable data from different sources to be effectively combined and analyzed, thereby improving the efficiency of data processing and analysis. Through comprehensive data collection of work conditions, personal interests, and behavioral habits, the system can better understand user needs and preferences. Accurately push relevant work information, task reminders, and collaboration suggestions to help users complete work tasks more efficiently.
[0062] (3) Through personalized information push, users can obtain information related to their needs and interests in a timely manner, reducing the interference of irrelevant information. Through the real-time feedback mechanism, the push content and strategy are continuously optimized to ensure that the system can dynamically adapt to the changing needs of users and improve user satisfaction with the system. Provide accurate task reminders and collaboration suggestions to help users better manage work tasks and time and improve work efficiency. By pushing the latest industry trends and skill applications, help users continuously improve their professional knowledge and skills and enhance their professional competitiveness. Through big data analysis and machine learning models, the system can provide data-based intelligent decision support to help enterprises optimize resource allocation and management decisions. Realize intelligent information push and promote the development of enterprise digital transformation and smart office.
[0063] (4) By comparing the intelligent push index Znts with the preset threshold, the system can dynamically adjust the push strategy to ensure that the push content is highly matched with user needs and improve the accuracy and relevance of the push. Provide multi-level push strategies and personalized setting options, so that users can flexibly adjust the push content and frequency according to their own needs and preferences, and enhance user participation and satisfaction. Strengthen the user feedback mechanism, optimize the push content and strategy in a timely manner based on user feedback, and improve the user experience. Through feature extraction and model training, continuously optimize the recommendation algorithm to improve the system's ability to predict user needs and push efficiency. Utilize real-time feedback and rapid response mechanisms to ensure that the system can adjust and optimize push content in a timely manner and maintain efficient operation. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] Figure 1 This is a schematic diagram of the flow chart of the smart office information push system of the present invention;
[0065] Figure 2 Schematic diagram of the steps of the smart office information push method of the present invention. DETAILED DESCRIPTION
[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0067] Example 1
[0068] The present invention provides a smart office information push system, please refer to Figure 1 , including data collection module, data preprocessing module, feature extraction module, prediction module, information push module and user feedback module;
[0069] The data collection module is used to collect data from office software, project management tools, conference software, HR management system, browser plug-in, internal enterprise social platform, time management software, email client and collaboration tool to obtain: a first data group, a second data group and a third data group;
[0070] The data preprocessing module is used to clean, remove duplicates, convert formats and standardize the collected first data group, second data group and third data group, and then transmit them to the feature extraction module;
[0071] The feature extraction module is used to extract key features, user behavior patterns, keywords and usage frequencies from the preprocessed data, establish a training model, and use machine learning algorithms, collaborative filtering, content recommendation and deep learning to calculate and obtain: intelligent push index Znts;
[0072] The prediction module is used to generate personalized prediction results and predict user needs and preferences by comparing the smart push index Znts with the preset push threshold N and the preset push threshold M;
[0073] The information push module is used to push information related to user needs in real time according to the results of the prediction model, including documents, meeting notices and industry trends, to achieve accurate information push and improve the efficiency of user acquisition of information;
[0074] The user feedback module is used to collect user feedback on the pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy based on the feedback results.
[0075] In this embodiment, data is collected from multiple channels to obtain: a first data group, a second data group and a third data group; the collected first data group, second data group and third data group are pre-processed and transmitted to a feature extraction module to establish a training model, and the intelligent push index Znts is calculated and obtained; the intelligent push index Znts is compared with a preset push threshold N and a preset push threshold M to generate personalized prediction results, predict user needs and preferences, and according to the results of the prediction model, information related to user needs is pushed in real time, including documents, meeting notices and industry trends, to achieve accurate information push and improve the efficiency of users in obtaining information; the user feedback module is used to collect user feedback on pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy according to the feedback results.
[0076] Example 2
[0077] This embodiment is explained in Example 1, please refer to Figure 1,Specifically: the data collection module includes a work situation collection unit, a personal interest situation collection unit and a behavior habit collection unit;
[0078] The work situation collection unit is used to collect task completion time, task priority, project progress, number of times a file is used, and employee activity in the meeting through project management tools, document management systems, and conference software, and obtain: expected completion time Yjsj, actual completion time Sjsj, access rate of a specific document Wdfw, editing frequency of a specific document Wdbj, progress percentage of the current project Jdbf, and conference participation activity Hyhy as a first data group;
[0079] The personal interest situation collection unit is used to collect document reading time Ydsj, web page stay time Tlsj, topic click rate Ztdj and skill application frequency Jnyy as the second data group through the document management system, browser plug-in and internal enterprise social platform;
[0080] The behavior habit collection unit is used to collect: daily working time Rcgz, application use time Yysy, work and rest ratio Sjbl, email response time Xysj and collaboration tool use frequency Gjsy as the third data group through time management software, email client and collaboration tools.
[0081] The data preprocessing module includes a data preprocessing unit;
[0082] The data preprocessing unit is responsible for performing comprehensive data preprocessing on the first data group, the second data group and the third data group. The preprocessing steps include: removing noise data, processing missing values, correcting outliers, identifying and deleting duplicate data, and converting data from different sources into a unified format, and transmitting the preprocessed data to the feature extraction module.
[0083] In this embodiment, through the data preprocessing step, noise data is removed, missing values are processed, and outliers are corrected to ensure the integrity and consistency of the data, unify the data format, and enable data from different sources to be effectively combined and analyzed, thereby improving the efficiency of data processing and analysis. Through comprehensive data collection of work conditions, personal interests, and behavioral habits, the system can better understand the needs and preferences of users. Accurately push relevant work information, task reminders, and collaboration suggestions to help users complete work tasks more efficiently.
[0084] Example 3
[0085] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the feature extraction module includes a feature extraction unit and a model training unit;
[0086] The feature extraction unit is responsible for extracting key features from the cleaned and standardized data, including user behavior patterns, keywords and usage frequency, and removing redundant or irrelevant features using a feature selection algorithm;
[0087] The user behavior pattern is used to analyze the user's operation log and extract the behavior pattern, including the distribution of working time, application usage time and document reading time;
[0088] The keyword extraction is used to extract keywords from documents and communication content, using natural language processing (NLP) technology, TF-IDF and word vectors;
[0089] The usage frequency is used to calculate the usage frequency of various documents, applications and tools, and to count the number of visits and edits and meeting participation;
[0090] The model training unit is used to use the extracted and selected features, and to train the model using a variety of machine learning algorithms. The trained model is used to calculate: the smart push index Znts, the work content relevance coefficient Gzxg, the interest matching coefficient Xqpp, and the behavior habit coefficient Xwxg;
[0091] The smart push index Znts is calculated by the following formula:
[0092] Znts=Gzxg*α+Xqpp*β+Xwxg*χ;
[0093] In the formula, Znts represents the smart push index, Gzxg represents the work content relevance coefficient, Xqpp represents the interest matching coefficient, Xwxg represents the behavior habit coefficient, α, β and χ represent the proportional coefficients of the work content relevance coefficient Gzxg, the interest matching coefficient Xqpp and the behavior habit coefficient Xwxg respectively;
[0094] Among them, 0≤α≤1, 0≤β≤1, 0≤χ≤1, and α+β+χ≤1.0;
[0095] The work content relevance coefficient Gzxg is calculated by the following formula:
[0096]
[0097] In the formula, Yjsj represents the expected completion time, Sjsj represents the actual completion time, Wdfw represents the access rate of a specific document, Wdbj represents the editing frequency of a specific document, Jdbf represents the progress percentage of the current project, and Hyhy represents the meeting participation activity. It represents the ratio of the expected completion time to the actual completion time. represents the inverse of the access rate of a specific document, Represents the inverse of how often a particular document has been edited.
[0098] The interest matching coefficient Xqpp is calculated by the following formula:
[0099]
[0100] Where Ydsj represents the document reading time, Tlsj represents the webpage dwelling time, Ztdj represents the topic click-through rate, Jnyy represents the skill application frequency, the logarithm of the document reading time log(Ydsj+1) represents taking the logarithm of the document reading time to reduce the impact of extreme values, and the logarithm of the webpage dwelling time log(Tlsj+1) represents taking the logarithm of the webpage dwelling time to reduce the impact of extreme values.
[0101] The behavior habit coefficient Xwxg is calculated by the following formula:
[0102]
[0103] In the formula, Rcgz represents daily working time, Yysy represents application usage time, Sjbl represents the ratio of work to rest, Xysj represents email response time, Gjsy represents the frequency of collaboration tool use, and the ratio of daily working time to application usage time is Reflects the ratio of working time to application usage time, balances work efficiency, and is inversely proportional to email response time Represents the inverse of email response time.
[0104] In this embodiment, through personalized information push, users can obtain information related to their own needs and interests in a timely manner, reducing the interference of irrelevant information. Through the real-time feedback mechanism, the push content and strategy are continuously optimized to ensure that the system can dynamically adapt to the changing needs of users and improve user satisfaction with the system. Provide accurate task reminders and collaboration suggestions to help users better manage work tasks and time and improve work efficiency. By pushing the latest industry trends and skill applications, help users continuously improve their professional knowledge and skills and enhance their professional competitiveness. Through big data analysis and machine learning models, the system can provide data-based intelligent decision support to help enterprises optimize resource allocation and management decisions. Realize the intelligent push of information and promote the development of enterprise digital transformation and smart office.
[0105] Example 4
[0106] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the prediction module includes a prediction evaluation unit;
[0107] The prediction evaluation unit is used to compare the intelligent push index Znts with the preset push threshold N and the preset push threshold M to generate a personalized prediction result:
[0108] The prediction evaluation unit is used to compare the intelligent push index Znts with the preset push threshold N and the preset push threshold M to generate a personalized prediction result:
[0109] When the smart push index Znts> the preset push threshold N, the first evaluation level is obtained, and the system prioritizes pushing information related to user needs. Through feature extraction and model training, it ensures that the smart push index Znts reflects the accuracy of user needs.
[0110] When the preset push threshold N ≥ smart push index Znts ≥ preset push threshold M, the second evaluation level is obtained. The system pushes a balanced variety of information content based on user interests and historical behaviors, adds personalized setting options, allows users to adjust push preferences, including topic preferences and frequency preferences, introduces a multi-level push strategy, and flexibly adjusts push content based on users' different interest areas and behavior patterns;
[0111] When the smart push index Znts is less than the preset push threshold M, the third evaluation level is obtained, the user's browsing history, search preferences and click patterns are re-analyzed, and user-defined push preference settings are provided, including topic selection, push frequency and time preferences. A rapid response mechanism is established, and the push strategy is iterated according to real-time feedback to ensure timely adjustment and optimization of push content.
[0112] The information push module includes a real-time push unit;
[0113] The real-time push unit is used to analyze the user's current needs and preferences based on the results of the prediction model, accurately push documents, meeting notifications and industry dynamics to the user, collect user feedback on the pushed information, and transmit the feedback information to the user feedback module.
[0114] The user feedback module includes a feedback data collection unit and a feedback analysis unit;
[0115] The feedback data collection unit is used to monitor and record the user's click behavior on the pushed information, track the user's stay time on the pushed content, provide a user rating function, record the user's evaluation of the pushed content, and count the number of clicks on each piece of information and the user's reading time for each piece of information;
[0116] The feedback analysis unit is used to collect and analyze click-through rate, reading time and user rating data, identify user preferences and needs, and adjust the push strategy according to the feedback data analysis results, including push frequency, push time and push content, and optimize the personalized settings of push content.
[0117] In this embodiment, by comparing the intelligent push index Znts with the preset threshold, the system can dynamically adjust the push strategy to ensure that the push content is highly matched with user needs and improve the accuracy and relevance of the push. Provide multi-level push strategies and personalized setting options, so that users can flexibly adjust the push content and frequency according to their own needs and preferences, and enhance user participation and satisfaction. Strengthen the user feedback mechanism, optimize the push content and strategy in a timely manner based on user feedback, and improve the user experience. Through feature extraction and model training, continuously optimize the recommendation algorithm to improve the system's ability to predict user needs and push efficiency. Use real-time feedback and rapid response mechanisms to ensure that the system can adjust and optimize push content in a timely manner and maintain efficient operation.
[0118] Example 5
[0119] For information push methods of smart office, please refer to Figure 2 , specifically: including the following steps:
[0120] Step 1: Collect data from office software, project management tools, conference software, HR management system, browser plug-ins, internal enterprise social platforms, time management software, email clients, and collaboration tools to obtain: the first data group, the second data group, and the third data group;
[0121] Step 2: The collected first data group, second data group and third data group are cleaned, deduplicated, format converted and standardized, and then transmitted to the feature extraction module;
[0122] Step 3: Extract key features, user behavior patterns, keywords and usage frequency from the preprocessed data, build a training model, and use machine learning algorithms, collaborative filtering, content recommendation and deep learning to calculate and obtain: smart push index Znts;
[0123] Step 4: Generate personalized prediction results by comparing the smart push index Znts with the preset push threshold N and the preset push threshold M to predict the user's needs and preferences;
[0124] Step 5: Based on the results of the prediction model, push information related to user needs in real time, including documents, meeting notices, and industry trends, to achieve accurate information push and improve the efficiency of user information acquisition;
[0125] Step 6: The user feedback module is used to collect user feedback on the pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy based on the feedback results.
[0126] In this embodiment, data is collected from multiple channels to obtain: a first data group, a second data group and a third data group; the collected first data group, second data group and third data group are pre-processed and transmitted to a feature extraction module to establish a training model, and the intelligent push index Znts is calculated and obtained; the intelligent push index Znts is compared with a preset push threshold N and a preset push threshold M to generate personalized prediction results, predict user needs and preferences, and according to the results of the prediction model, information related to user needs is pushed in real time, including documents, meeting notices and industry trends, to achieve accurate information push and improve the efficiency of users in obtaining information; the user feedback module is used to collect user feedback on pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy according to the feedback results.
[0127] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. Smart office information push system, Features: It includes data collection module, data preprocessing module, feature extraction module, prediction module, information push module and user feedback module; The data collection module is used to collect data from office software, project management tools, conference software, HR management system, browser plug-in, internal enterprise social platform, time management software, email client and collaboration tool to obtain: a first data group, a second data group and a third data group; The data preprocessing module is used to clean, remove duplicates, convert formats and standardize the collected first data group, second data group and third data group, and then transmit them to the feature extraction module; The feature extraction module is used to extract key features, user behavior patterns, keywords and usage frequencies from the preprocessed data, establish a training model, and use machine learning algorithms, collaborative filtering, content recommendation and deep learning to calculate and obtain: intelligent push index Znts; The prediction module is used to generate personalized prediction results and predict user needs and preferences by comparing the smart push index Znts with the preset push threshold N and the preset push threshold M; The information push module is used to push information related to user needs in real time according to the results of the prediction model, including documents, meeting notices and industry trends, to achieve accurate information push and improve the efficiency of user acquisition of information; The user feedback module is used to collect user feedback on the pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy based on the feedback results.
2. According to claim 1, the smart office information push system, Features: The data collection module includes a work situation collection unit, a personal interest situation collection unit and a behavior habit collection unit; The work situation collection unit is used to collect task completion time, task priority, project progress, number of times a file is used, and employee activity in the meeting through project management tools, document management systems, and conference software, and obtain: expected completion time Yjsj, actual completion time Sjsj, access rate of a specific document Wdfw, editing frequency of a specific document Wdbj, progress percentage of the current project Jdbf, and conference participation activity Hyhy as a first data group; The personal interest situation collection unit is used to collect document reading time Ydsj, web page stay time Tlsj, topic click rate Ztdj and skill application frequency Jnyy as the second data group through the document management system, browser plug-in and internal enterprise social platform; The behavior habit collection unit is used to collect: daily working time Rcgz, application use time Yysy, work and rest ratio Sjbl, email response time Xysj and collaboration tool use frequency Gjsy as the third data group through time management software, email client and collaboration tools.
3. According to claim 1, the smart office information push system, Features: The data preprocessing module includes a data preprocessing unit; The data preprocessing unit is responsible for performing comprehensive data preprocessing on the first data group, the second data group and the third data group. The preprocessing steps include: removing noise data, processing missing values, correcting outliers, identifying and deleting duplicate data, and converting data from different sources into a unified format, and transmitting the preprocessed data to the feature extraction module.
4. The smart office information push system according to claim 1, Features: The feature extraction module includes a feature extraction unit and a model training unit; The feature extraction unit is responsible for extracting key features from the cleaned and standardized data, including user behavior patterns, keywords and usage frequency, and removing redundant or irrelevant features using a feature selection algorithm; The user behavior pattern is used to analyze the user's operation log and extract the behavior pattern, including the distribution of working time, application usage time and document reading time; The keyword extraction is used to extract keywords from documents and communication content, using natural language processing (NLP) technology, TF-IDF and word vectors; The usage frequency is used to calculate the usage frequency of various documents, applications and tools, and to count the number of visits and edits and meeting participation; The model training unit is used to use the extracted and selected features, and to train the model using a variety of machine learning algorithms. The trained model is used to calculate: the smart push index Znts, the work content relevance coefficient Gzxg, the interest matching coefficient Xqpp, and the behavior habit coefficient Xwxg; The smart push index Znts is calculated by the following formula: Znts=Gzxg*α+Xqpp*β+Xwxg*χ; In the formula, Znts represents the smart push index, Gzxg represents the work content relevance coefficient, Xqpp represents the interest matching coefficient, Xwxg represents the behavior habit coefficient, α, β and χ represent the proportional coefficients of the work content relevance coefficient Gzxg, the interest matching coefficient Xqpp and the behavior habit coefficient Xwxg respectively; Among them, 0≤α≤1, 0≤β≤1, 0≤χ≤1, and α+β+χ≤1.0; The work content relevance coefficient Gzxg is calculated by the following formula: In the formula, Yjsj represents the expected completion time, Sjsj represents the actual completion time, Wdfw represents the access rate of a specific document, Wdbj represents the editing frequency of a specific document, Jdbf represents the progress percentage of the current project, and Hyhy represents the meeting participation activity. It represents the ratio of the expected completion time to the actual completion time. represents the inverse of the access rate of a specific document, Represents the inverse of how often a particular document has been edited.
5. According to claim 4, the smart office information push system, Features: The interest matching coefficient Xqpp is calculated by the following formula: Where Ydsj represents the document reading time, Tlsj represents the webpage dwelling time, Ztdj represents the topic click-through rate, Jnyy represents the skill application frequency, the logarithm of the document reading time log(Ydsj+1) represents taking the logarithm of the document reading time to reduce the impact of extreme values, and the logarithm of the webpage dwelling time log(Tlsj+1) represents taking the logarithm of the webpage dwelling time to reduce the impact of extreme values.
6. The smart office information push system according to claim 4, Features: The behavior habit coefficient Xwxg is calculated by the following formula: In the formula, Rcgz represents daily working time, Yysy represents application usage time, Sjbl represents the ratio of work to rest, Xysj represents email response time, Gjsy represents the frequency of collaboration tool use, and the ratio of daily working time to application usage time is Reflects the ratio of working time to application usage time, balances work efficiency, and is inversely proportional to email response time Represents the inverse of email response time.
7. The smart office information push system according to claim 1, Features: The prediction module includes a prediction evaluation unit; The prediction evaluation unit is used to compare the intelligent push index Znts with the preset push threshold N and the preset push threshold M to generate a personalized prediction result: When the smart push index Znts> the preset push threshold N, the first evaluation level is obtained, and the system prioritizes pushing information related to user needs. Through feature extraction and model training, it ensures that the smart push index Znts reflects the accuracy of user needs. When the preset push threshold N ≥ smart push index Znts ≥ preset push threshold M, the second evaluation level is obtained. The system pushes a balanced variety of information content based on user interests and historical behaviors, adds personalized setting options, allows users to adjust push preferences, including topic preferences and frequency preferences, introduces a multi-level push strategy, and flexibly adjusts push content based on users' different interest areas and behavior patterns; When the smart push index Znts is less than the preset push threshold M, the third evaluation level is obtained, the user's browsing history, search preferences and click patterns are re-analyzed, and user-defined push preference settings are provided, including topic selection, push frequency and time preferences. A rapid response mechanism is established, and the push strategy is iterated according to real-time feedback to ensure timely adjustment and optimization of push content.
8. The smart office information push system according to claim 1, Features: The information push module includes a real-time push unit; The real-time push unit is used to analyze the user's current needs and preferences based on the results of the prediction model, accurately push documents, meeting notifications and industry dynamics to the user, collect user feedback on the pushed information, and transmit the feedback information to the user feedback module.
9. The smart office information push system according to claim 1, Features: The user feedback module includes a feedback data collection unit and a feedback analysis unit; The feedback data collection unit is used to monitor and record the user's click behavior on the pushed information, track the user's stay time on the pushed content, provide a user rating function, record the user's evaluation of the pushed content, and count the number of clicks on each piece of information and the user's reading time for each piece of information; The feedback analysis unit is used to collect and analyze click-through rate, reading time and user rating data, identify user preferences and needs, and adjust the push strategy according to the feedback data analysis results, including push frequency, push time and push content, and optimize the personalized settings of push content.
10. A smart office information push method, applied to the smart office information push system according to any one of claims 1 to 9, Features: The following steps are involved: Step 1: Collect data from office software, project management tools, conference software, HR management system, browser plug-ins, internal enterprise social platforms, time management software, email clients, and collaboration tools to obtain: the first data group, the second data group, and the third data group; Step 2: The collected first data group, second data group and third data group are cleaned, deduplicated, format converted and standardized, and then transmitted to the feature extraction module; Step 3: Extract key features, user behavior patterns, keywords and usage frequency from the preprocessed data, build a training model, and use machine learning algorithms, collaborative filtering, content recommendation and deep learning to calculate and obtain: smart push index Znts; Step 4: Generate personalized prediction results by comparing the smart push index Znts with the preset push threshold N and the preset push threshold M to predict the user's needs and preferences; Step 5: Based on the results of the prediction model, push information related to user needs in real time, including documents, meeting notices, and industry trends, to achieve accurate information push and improve the efficiency of user information acquisition; Step 6: The user feedback module is used to collect user feedback on the pushed information, including click-through rate, reading time and user rating, and optimize the model and push strategy based on the feedback results.
Citation Information
Cited By
Intelligent office assistant system based on natural language processing
CN119721970A
Marketing scene intelligent training management method and device and storage medium
CN120146893A
User interaction method, storage medium and user interaction system
CN120707955A
