Electronic archive recommendation method and system based on big data

Abstract

The invention discloses an electronic archive recommendation method and system based on big data, relates to the technical field of big data processing, and solves the technical problem that an archive recommendation method is often relatively single and cannot accurately provide personalized services for individual preference differences of users. According to the method, the user portrait is constructed by collecting the basic information of the user in multiple dimensions, the interest points and potential requirements of the user can be more accurately captured, more accurate guidance is provided for subsequent archive matching, the accuracy and reliability of primary matching are greatly improved, and the user experience is improved when the pre-selected archive is processed. By calculating the association degree between the archive content and the real-time data of the user and performing intelligent recommendation according to the archive type and the association degree threshold, the recommendation content can be dynamically adjusted according to the current browsing situation of the user to adapt to different real-time demand scenes, and the flexibility and adaptability of the recommendation system are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of big data processing, and in particular to a method and system for recommending electronic archives based on big data. Background Art

[0002] With the rapid development of information technology, the number of electronic archives has exploded. Among the numerous electronic archive resources, how to enable users to quickly and accurately obtain archives that meet their needs and interests has become an urgent problem to be solved.

[0003] According to publication number CN117493645B, a method and system for recommending electronic archives based on big data are disclosed. The method includes: splicing image text column vectors, objective column vectors and multiple text column vectors that meet the correlation conditions to obtain an electronic archive feature matrix; if the electronic archive feature matrix and the user portrait matrix generated according to the user browsing history and collection records are similar matrices, and the Hamming distance between the image ahash value and the user portrait ahash value is greater than a preset threshold, recommending the electronic archive to the user.

[0004] Traditional archival recommendation methods are often simplistic, relying solely on basic archival classification or simple user behavior statistics. These methods struggle to deeply explore users' latent interests and complex evolving needs. For example, some early electronic archival systems only crudely categorized and recommended archival materials based on their subject areas, failing to precisely tailor personalized services to individual user preferences. Furthermore, in a big data environment, despite the abundance of data, effectively integrating and leveraging this data for precise recommendations faces numerous challenges, such as the diversity and complexity of the data and the dynamic nature of user interests. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides an electronic archive recommendation method and system based on big data, which solves the problem that archive recommendation methods are often relatively simple and cannot accurately provide personalized services based on the preferences of individual users.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for recommending electronic archives based on big data, the method specifically comprising the following steps: Step S1: Collect basic user information and build a corresponding user profile based on the basic information; Step S2: Match the user profile with the electronic profile in the big data to obtain a matching profile, and perform a secondary match with the user's historical data to generate pre-selected profile information; Step S3: obtaining a pre-selected file and analyzing it in combination with the user's real-time data. By analyzing the file content of the pre-selected file, an existence analysis signal and an absence analysis signal are generated; Step S4, processing the obtained presence analysis signal, calculating the content relevance between the pre-selected archive and the user's real-time data, and generating recommendation information based on the relevance analysis; Step S5: Analyze the obtained non-existence analysis results, identify similar users by calculating the similarity between users, and generate recommendation information based on the recommendation profiles corresponding to the similar users.

[0007] As a further solution of the present invention, the specific method of obtaining the matching file in step S2 is: Obtain user portraits, extract portrait features corresponding to user portraits, and match them with electronic files in big data based on the obtained portrait features. The specific matching method is: obtain all electronic files and label them as i, and i=1, 2, ..., j, where j represents the number of electronic files. At the same time, obtain the file type corresponding to the electronic file, and use the portrait features as the standard to screen the electronic file i to obtain matching files, which are labeled n, and n=1, 2, ..., m, where m represents the number of matching files.

[0008] As a further solution of the present invention, the specific method of generating the pre-selected archive information in step S2 is: Then, the user's historical data is obtained, and the corresponding browsing history is obtained based on the historical data. At the same time, the electronic files in the browsing history are classified according to the file type to obtain files of the same type, and the files of the same type are labeled as a, and a=1, 2, ..., b, where b represents the type label of the files of the same type. Then, the historical search count Ca and historical download count Xa of the files of the same type a are obtained, and the obtained parameters are summed up to obtain the browsing value Qa of the files of the same type a, and sorted from large to small according to the browsing value, and the matching files are screened from large to small according to the browsing value Qa to obtain pre-selected file information, and the specific screening method is to select matching files that meet the same type of files and record them as pre-selected files.

[0009] As a further solution of the present invention, the specific method of generating the presence analysis signal and the absence analysis signal in step S3 is: Obtain all pre-selected files and label them as o, where o = 1, 2, ..., p, and p represents the number of pre-selected files. Then, obtain the file content of the pre-selected file o and the user's real-time data, obtain the real-time content corresponding to the real-time data, and judge the real-time content. If the real-time content exists in the user's browsing history, an existence analysis signal is generated. Conversely, if the real-time content does not exist in the user's browsing history, an absence analysis signal is generated.

[0010] As a further solution of the present invention, the specific method of processing the presence analysis signal in step S4 is: Obtain all pre-selected files o, and calculate the content relevance between the pre-selected files o and the user's real-time data. The obtained relevance value (cosine similarity) is recorded as Fo. The relevance value here is calculated and analyzed by sequentially analyzing the real-time data and the pre-selected files. For example, if there are 5 pre-selected files o, then the calculated relevance value Fo will also exist 5 times. Then, the relevance value Fo is compared with the threshold Fy, and the specific value of the threshold Fy is set by the operator. If the correlation value Fo is greater than the threshold Fy, the corresponding pre-selected file will be selected and marked as the file to be analyzed. Conversely, if the correlation value Fo is less than the threshold Fy, the corresponding pre-selected file will not be processed, and then the file to be analyzed will be obtained. At the same time, the file type of the file to be analyzed will be obtained, and different recommendations will be made based on the file type to generate recommendation information.

[0011] As a further solution of the present invention, the specific method of generating recommendation information by making different recommendations based on the file type in step S4 is: If there is only one file type, the corresponding recommendation information is generated based on the file to be analyzed. If there are multiple file types, the file type corresponding to the pre-selected file is obtained, and the number of files corresponding to different file types is obtained. Then, the average correlation degree corresponding to files of the same type is calculated, and the recommendation information is generated based on the file of the same type corresponding to the file with the largest average correlation degree.

[0012] As a further solution of the present invention, the specific method of analyzing the absence of the analysis result in step S5 is: Obtain all users within time t as users to be analyzed, and calculate the user similarity between the user to be analyzed and the current user. Then, filter out similar users based on the user similarity and record them as h, where h = 1, 2, ..., g, where g represents the number of similar users. Then, obtain the recommended profiles corresponding to similar users, and compare and analyze the recommended profiles of similar users with the browsing history of the current user. Filter out the browsed recommended profiles and the unbrowsed recommended profiles, and obtain the corresponding number and record them as L1. h and L2 h , and so on to obtain the number of browsed recommended profiles and unbrowsed recommended profiles corresponding to all similar users h; Then according to the number of unviewed recommended files L2 h Sort from large to small, and identify the types of unviewed recommended files corresponding to different similar users h. At the same time, select the files of the same type corresponding to the largest number of unviewed recommended files of the same type as the standard, and generate recommendation information.

[0013] An electronic archive recommendation system based on big data, comprising a user information acquisition unit, an archive pre-selection and analysis unit, a recommendation information generation unit, and a recommendation information output unit; A user information acquisition unit, which is used to collect basic information of the user and transmit the acquired basic information to the profile pre-selection and analysis unit; The profile pre-selection analysis unit is used to create a user profile based on the acquired user basic information, and to match the user profile with the electronic profile in the big data to obtain a matching profile. Furthermore, a secondary match is performed with the user's historical data to generate pre-selected profile information, and the pre-selected profile information is transmitted to the recommendation information generation unit; A recommendation information generating unit, which is used to analyze the acquired pre-selected profiles and simultaneously analyze the user's real-time data, and to generate a presence analysis signal and a non-presence analysis signal by analyzing the profile content of the pre-selected profiles; Processing the obtained presence analysis signal, calculating the content relevance between the pre-selected profile and the user's real-time data, and generating recommendation information based on the relevance analysis, while transmitting the recommendation information to the recommendation information output unit; Analyze the obtained non-existence analysis results, identify similar users by calculating the similarity between users, generate recommendation information based on the recommendation profiles corresponding to the similar users, and transmit the recommendation information to the recommendation information output unit; The recommendation information output unit is used to display the acquired recommendation information to the corresponding operator.

[0014] The present invention provides a method and system for recommending electronic archives based on big data. Compared with the existing technology, it has the following advantages: The present invention collects basic user information from multiple dimensions to construct a user profile, which can more accurately capture the user's interests and potential needs, provide more accurate guidance for subsequent profile matching, and greatly improve the accuracy and reliability of the initial matching. When processing pre-selected profiles, by calculating the correlation between the profile content and the user's real-time data, and making intelligent recommendations based on the profile type and correlation threshold, the recommended content can be dynamically adjusted according to the user's current browsing context, adapting to different real-time demand scenarios, thereby enhancing the flexibility and adaptability of the recommendation system. When the content browsed by a user in real time is not within the historical range, similar users are identified by calculating user similarity, and recommendation information is generated based on the recommendation profiles of similar users. This fully taps the commonalities and differences between user groups, makes up for the shortcomings of individual data with the help of group wisdom, further expands the source and scope of recommendations, improves the comprehensiveness and diversity of recommendations, and provides users with more electronic archive resources that may be of interest but have not been discovered, thereby improving the performance and service quality of the entire electronic archive recommendation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a diagram of the steps and methods of the present invention; Figure 2 This is a block diagram of the system principle of the present invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0017] For example 1, please refer to Figure 1 , this application provides an electronic archive recommendation method based on big data, which specifically includes the following steps: Step S1, collect the basic information of the user and build a corresponding user portrait based on the basic information. The basic information here includes age, gender, occupation, interests and hobbies, etc. At the same time, the construction of user portraits is an existing technology and will not be elaborated here. For details, please refer to the patent with publication number CN105608171B.

[0018] Step S2: Match the user profile with the electronic files in the big data to obtain a matching file, and perform a secondary match based on the user's historical data to generate pre-selected file information.

[0019] Obtain user portraits and extract the corresponding portrait features of the user portraits. The portrait features here are established according to the user's corresponding interests and hobbies. For text data, use the topic model in natural language processing, such as the LDA (Latent Dirichlet Allocation) algorithm, to obtain the user's portrait features by analyzing these data, and match them with the electronic files in the big data based on the obtained portrait features. The specific matching method is: obtain all electronic files and label them as i, and i=1, 2, ..., j, where j represents the number of electronic files, and obtain the file type corresponding to the electronic file, and use the portrait features as The standard screens the electronic archive i to obtain matching archives, which are denoted as n, and n=1, 2, ..., m, where m represents the number of matching archives. For example, taking the portrait characteristics of scientific researchers as an example, when matching, the system will screen out archives with academic research types and content themes highly related to quantum physics from all electronic archives as matching archives. These matching archives are numbered n in sequence (n ranges from 1 to m, and m is the number of archives that are finally successfully matched). For example, a series of electronic archives on the explanation of new quantum physics theories, the release of the latest results of quantum physics experiments, etc. may be screened out. These archives will be recommended to scientific researchers with the portrait characteristics of "strong interest in quantum physics research". Then, the user's historical data is obtained, and the corresponding browsing history is obtained based on the historical data. The browsing history here refers to the browsing records within a time period t. The specific value of the specific time t is set by the operator. At the same time, the electronic archives in the browsing history are classified according to the archive type to obtain archives of the same type, and the archives of the same type are labeled as a, and a=1, 2, ..., b, where b represents the type label of the archive of the same type. For example, a=1 represents an electronic archive of an academic research type, and a=2 represents an electronic archive of a historical and cultural type. The archive type corresponding to the specific label is set by the operator. Then, the historical search count Ca and historical download count Xa of the archive a of the same type are obtained. At the same time, the obtained parameters are summed and calculated to obtain the browsing value Qa of the archive a of the same type, and the archives are sorted from large to small according to the browsing value; The matching files are screened from large to small according to the browsing value Qa to obtain the pre-selected file information, and here it means that the matching files are matched and screened with the files of the same type corresponding to the browsing value Qa, and the matching files that meet the same type of files are selected and recorded as pre-selected files. Specifically, the matching is selected and matched in sequence according to the browsing value.

[0020] Step S3: obtaining a pre-selected file and analyzing it in combination with the user's real-time data. By analyzing the file content of the pre-selected file, an existence analysis signal and an absence analysis signal are generated.

[0021] Obtain all pre-selected files and label them as o, where o = 1, 2, ..., p, and p represents the number of pre-selected files. Then, obtain the file content of the pre-selected file o and the user's real-time data, where the user's real-time data is the file currently being browsed by the user, obtain the real-time content corresponding to the real-time data, and judge the real-time content. If the real-time content exists in the user's browsing history, a presence analysis signal is generated; otherwise, if the real-time content does not exist in the user's browsing history, a non-existence analysis signal is generated, and the generated presence analysis signal and non-existence analysis signal are processed separately; Step S4: Processing the obtained presence analysis signal, calculating the content relevance between the pre-selected archive and the user's real-time data, and generating recommendation information based on the analysis of the relevance.

[0022] Obtain all pre-selected files o and calculate the content relevance between the pre-selected files o and the user's real-time data. The specific calculation method is to analyze them through the Bag-of-Words model (Bag-of-Words) and TF-IDF (Term Frequency-Inverse Document Frequency). For example, for an electronic file A, the content is "Application of Artificial Intelligence in Medical Image Diagnosis", and the content of another electronic file B is "New Advances in Artificial Intelligence Technology in the Medical Field". After converting them into bags of words, A may contain words such as "artificial intelligence, medical image diagnosis, application", and B contains words such as "artificial intelligence, medical field, new progress". Then, calculate the TF-IDF value, where F refers to the frequency of a word appearing in a document, and IDF refers to the general importance of a word (through Calculate the inverse of the number of documents containing the word). For two documents, calculate the TF-IDF vectors of all their words. Measure the relevance by calculating the cosine similarity of the two vectors. The range of cosine similarity is [-1, 1]. The closer the value is to 1, the higher the relevance. For example, after calculating the cosine similarity of the above-mentioned documents A and B, it may be around 0.7, indicating that they have a high relevance. The obtained relevance value (cosine similarity) is recorded as Fo. The relevance value here is calculated and analyzed by comparing the real-time data with the pre-selected files in sequence. For example, if there are 5 pre-selected files o, then the calculated relevance value Fo will also exist for 5. Then the relevance value Fo is compared with the threshold Fy, and the specific value of the threshold Fy is set by the operator. If the correlation value Fo is greater than the threshold Fy, the corresponding pre-selected file is selected and marked as the file to be analyzed. On the contrary, if the correlation value Fo is less than the threshold Fy, the corresponding pre-selected file is not processed. Then the file to be analyzed is obtained, and the file type of the file to be analyzed is obtained. Different recommendations are made based on the file type to generate recommendation information; If there is only one file type, the corresponding recommendation information is generated based on the file to be analyzed. If there are multiple file types, the file type corresponding to the pre-selected file is obtained, and the number of files corresponding to different file types is obtained. The number of files here is the number of pre-selected files. Then, the mean correlation value corresponding to files of the same type is calculated. The specific calculation method is: obtain all pre-selected files of the same type and the corresponding correlation values, and calculate the sum of all correlation values. At the same time, the average value is calculated based on the sum value, and the recommendation information is generated based on the file of the same type corresponding to the largest mean correlation value.

[0023] For example, among the pre-selected archives on a comprehensive academic information platform, 30 belong to the natural sciences, 20 to the social sciences, and 15 to engineering and technology. The average correlation value for archives of the same type is then calculated. Specifically, for each archive type, all pre-selected archives of that type and their corresponding correlation values ​​are collected, these correlation values ​​are summed, and then the sum is divided by the number of pre-selected archives of that type to obtain the average value. For example, if the sum of the correlation values ​​of the 30 pre-selected archives in the natural sciences is 150, then the average correlation value for that type is 150 ÷ ​​30 = 5. For example, for engineering and technology, if the sum of the correlation values ​​is 60, then the average is 60 ÷ 15 = 4. Based on the calculated average correlation value, the pre-selected archives of the natural sciences type are selected to generate recommended information.

[0024] Step S5: Analyze the obtained non-existence analysis results, identify similar users by calculating the similarity between users, and generate recommendation information based on the recommendation profiles corresponding to the similar users.

[0025] All users within the acquisition time t are recorded as users to be analyzed, and the users here do not include the current user. At the same time, the user similarity between the user to be analyzed and the current user is calculated. For example, the Pearson correlation coefficient is used to measure the similarity of users' ratings of electronic archives. If user A and user B have very similar rating patterns for a series of electronic archives, then they are considered to be similar users. Similar users are screened based on user similarity and recorded as h, and h=1, 2, ..., g, where g represents the number of similar users. Then, the recommended archives corresponding to the similar users are obtained. Here, one group of similar users is used as the analysis object for analysis, and the recommended archives of similar users are compared with the browsing history of the current user. Browsed recommended archives and unbrowsed recommended archives are screened, and the browsed recommended archives indicate that the corresponding browsing records exist in the browsing history of the current user, while the unbrowsed recommended archives indicate that the corresponding browsing records do not exist in the browsing history of the current user, and the corresponding numbers are obtained and recorded as L1. h and L2 h, and so on to obtain the number of browsed recommended profiles and unbrowsed recommended profiles corresponding to all similar users h; Then according to the number of unviewed recommended files L2 h Sort from large to small, and identify the types of unviewed recommended files corresponding to different similar users h. At the same time, select the files of the same type corresponding to the largest number of unviewed recommended files of the same type as the standard, and generate recommendation information.

[0026] For example, on a music electronic archive platform, similar user A has 20 unviewed recommended profiles, similar user B has 15, and similar user C has 30. Then, after sorting, they are similar user C, similar user A, and similar user B. For example, in the above example of the music platform, similar user C's 30 unviewed recommended profiles may include 10 rock music profiles, 8 classical music profiles, 12 pop music profiles, and so on. After identifying all the types of unviewed recommended profiles, the total number of unviewed recommended profiles of each type for different similar users is counted. Assume that the total number of rock music types in the unviewed recommended profiles of similar users A, B, and C is 15, the total number of classical music types is 12, and the total number of pop music types is 20. Then, select the type with the largest number of unviewed recommended profiles of the same type—in this case, pop music. Finally, generate recommendation information based on the unviewed recommended profiles of pop music.

[0027] For example 2, please refer to Figure 2 This application provides an electronic archive recommendation system based on big data, including: a user information acquisition unit, an archive pre-selection and analysis unit, a recommendation information generation unit and a recommendation information output unit, and Figure 2 It can be known that the functional units are electrically connected in a unidirectional manner.

[0028] A user information acquisition unit, which is used to collect basic information of the user and transmit the acquired basic information to the profile pre-selection and analysis unit; A profile pre-selection analysis unit is used to create a user profile based on the acquired basic user information, and to match the user profile with the electronic profile in the big data to obtain a matching profile. Furthermore, a secondary match is performed in combination with the user's historical data to generate pre-selected profile information, and the pre-selected profile information is transmitted to the recommendation information generation unit. The processing method here is similar to the processing process of step S2 in the embodiment; A recommendation information generating unit is used to analyze the acquired pre-selected profiles and analyze them in combination with the user's real-time data. By analyzing the profile contents of the pre-selected profiles, a presence analysis signal and an absence analysis signal are generated. The processing method here is similar to the processing process of step S3 in the first embodiment. The obtained presence analysis signal is processed, and the content relevance between the pre-selected profile and the user's real-time data is calculated, and recommendation information is generated based on the relevance analysis. The recommendation information is transmitted to the recommendation information output unit. The processing here is similar to the processing method of step S4 in embodiment 1. The obtained non-existence analysis results are analyzed, similar users are identified by calculating the similarity between users, and recommendation information is generated according to the recommendation profiles corresponding to the similar users, and the recommendation information is transmitted to the recommendation information output unit. The processing here is similar to the processing method of step S5 in embodiment 1; The recommendation information output unit is used to display the acquired recommendation information to the corresponding operator.

[0029] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0030] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for recommending electronic archives based on big data, characterized in that: The method specifically comprises the following steps: Step S1, collecting basic information of the user and building a corresponding user portrait based on the basic information; Step S2, matching the user portrait with the electronic files in the big data to obtain a matching file, and performing secondary matching with the user's historical data to generate pre-selected file information; Step S3, obtaining a pre-selected file, and analyzing it in combination with the user's real-time data, and generating an existence analysis signal and an absence analysis signal by analyzing the file content of the pre-selected file; Step S4, processing the obtained existence analysis signal, calculating the content relevance between the pre-selected archive and the user's real-time data, and generating recommendation information based on the analysis of the relevance; Step S5, analyzing the obtained non-existence analysis results, identifying similar users by calculating the similarity between users, and generating recommendation information according to the recommendation profiles corresponding to the similar users.

2. The electronic archive recommendation method based on big data according to claim 1, characterized in that: The specific method of obtaining the matching file in step S2 is: The user portrait is obtained, and the portrait features corresponding to the user portrait are extracted, and the obtained portrait features are matched with the electronic files in the big data. The specific matching method is: all electronic files are obtained and labeled as i, and i=1, 2, ..., j, where j represents the number of electronic files, and the file type corresponding to the electronic file is obtained, and the electronic file i is screened according to the portrait features to obtain the matching file, which is recorded as n, and n=1, 2, ..., m, where m represents the number of matching files.

3. The electronic archive recommendation method based on big data according to claim 1, characterized in that: The specific method of generating the pre-selected archive information in step S2 is: Then, the user's historical data is obtained, and the corresponding browsing history is obtained based on the historical data. At the same time, the electronic archives in the browsing history are classified according to the archive type to obtain archives of the same type, and the archives of the same type are labeled as a, and a=1, 2, ..., b, where b represents the type label of the same type of archive. Then, the historical search times Ca and historical download times Xa of the same type of archive a are obtained, and the obtained parameters are summed and calculated to obtain the browsing value Qa of the same type of archive a, and the browsing values ​​are sorted from large to small, and the matching archives are screened from large to small according to the browsing value Qa to obtain pre-selected archive information, and the specific screening method is to select matching archives that meet the same type of archives and record them as pre-selected archives.

4. The electronic archive recommendation method based on big data according to claim 1, characterized in that: The specific method of generating the presence analysis signal and the absence analysis signal in step S3 is: Obtain all pre-selected files and label them as o, where o=1, 2, ..., p, and p represents the number of pre-selected files. Then, obtain the file content of the pre-selected file o and the real-time data of the user, obtain the real-time content corresponding to the real-time data, and judge the real-time content. If the real-time content exists in the user's browsing history, an existence analysis signal is generated. Conversely, if the real-time content does not exist in the user's browsing history, a non-existence analysis signal is generated.

5. The electronic archive recommendation method based on big data according to claim 1, characterized in that: The specific method of processing the existence analysis signal in step S4 is: Obtain all pre-selected files o, and calculate the content relevance between the pre-selected files o and the user's real-time data, and record the obtained relevance value as Fo, and the relevance value here is calculated and analyzed by sequentially analyzing the real-time data and the pre-selected files. If there are 5 pre-selected files o, then there are 5 calculated relevance values ​​Fo, and then compare the relevance value Fo with the threshold value Fy, and the specific value of the threshold value Fy is set by the operator; If the correlation value Fo is greater than the threshold value Fy, the corresponding pre-selected file is selected and marked as the file to be analyzed. Otherwise, if the correlation value Fo is less than the threshold value Fy, the corresponding pre-selected file is not processed, and then the file to be analyzed is obtained. At the same time, the file type of the file to be analyzed is obtained, and different recommendations are made based on the file type to generate recommendation information.

6. The electronic archive recommendation method based on big data according to claim 5, characterized in that: The specific method of generating recommendation information by making different recommendations based on the file type in step S4 is: If there is only one file type, the corresponding recommendation information is generated based on the file to be analyzed. If there are multiple file types, the file type corresponding to the pre-selected file is obtained, and the number of files corresponding to different file types is obtained. Then the mean correlation degree corresponding to files of the same type is calculated, and the recommendation information is generated based on the file of the same type with the largest mean correlation degree.

7. The electronic archive recommendation method based on big data according to claim 1, characterized in that: The specific method of analyzing the non-existence of the analysis result in step S5 is: All users within time t are obtained as the users to be analyzed, and the user similarity between the users to be analyzed and the current user is calculated. Similar users are screened according to the user similarity and recorded as h, where h=1, 2, ..., g, where g represents the number of similar users. Then, the recommended files corresponding to the similar users are obtained, and the recommended files of the similar users are compared and analyzed with the browsing history of the current user. The browsed recommended files and the unbrowsed recommended files are screened, and the corresponding numbers are obtained and recorded as L1. h and L2 h , and so on to obtain the number of browsed recommended profiles and unbrowsed recommended profiles corresponding to all similar users h; Then according to the number of unviewed recommended files L2 h Sort from large to small, and identify the types of unviewed recommended archives corresponding to different similar users h. At the same time, select the archives of the same type corresponding to the largest number of unviewed recommended archives of the same type as the standard, and generate recommendation information.

8. An electronic archive recommendation system based on big data, used to execute the electronic archive recommendation method based on big data according to any one of claims 1 to 7, characterized in that: It includes a user information acquisition unit, a profile pre-selection and analysis unit, a recommendation information generation unit and a recommendation information output unit; A user information acquisition unit, which is used to collect basic information of the user and transmit the acquired basic information to the profile pre-selection analysis unit; A profile pre-selection analysis unit is used to establish a user profile based on the acquired user basic information, and to match the user profile with the electronic profile in the big data to obtain a matching profile. A secondary match is performed in combination with the user's historical data to generate pre-selected profile information, and the pre-selected profile information is transmitted to the recommendation information generation unit; A recommendation information generating unit, which is used to analyze the acquired pre-selected profiles and analyze them in combination with the real-time data of the user, and to generate an existence analysis signal and an non-existence analysis signal by analyzing the profile contents of the pre-selected profiles; Processing the obtained existence analysis signal, calculating the content relevance between the pre-selected archive and the user's real-time data, and generating recommendation information based on the relevance analysis, and transmitting the recommendation information to the recommendation information output unit; Analyze the obtained non-existence analysis results, identify similar users by calculating the similarity between users, generate recommendation information according to the recommendation profiles corresponding to the similar users, and transmit the recommendation information to the recommendation information output unit; The recommendation information output unit is used to display the acquired recommendation information to the corresponding operator.

Images