Personalized recommendation method and device based on user behavior analysis

Abstract

The invention provides a personalized recommendation method and device based on user behavior analysis, and relates to the technical field of user behavior data processing. The method comprises the following steps: analyzing user behavior data acquired in real time to obtain an access frequency evaluation value, and generating a recommendation suggestion according to a user portrait; processing data in the user behavior analysis process are monitored in real time through a data collection tool, the real-time performance of data analysis is evaluated according to the processing data to obtain a data real-time evaluation value, and whether optimization measures are taken or not is judged based on the data real-time evaluation value and a preset real-time threshold value; by the adoption of the method and device, the real-time performance and the overall performance of the system can be improved, the effect of personalized recommendation timeliness is improved, and the problem that in the prior art, data processing real-time performance is not high in the personalized recommendation process is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of user behavior data processing, and in particular to a personalized recommendation method and device based on user behavior analysis. Background Art

[0002] With the development of Internet technology, user behavior data on various websites has become increasingly important. This data can not only help websites understand user needs, but also provide users with more personalized services and recommendations by analyzing user behavior. Currently, many Internet companies are using data collection and analysis technology to improve user experience and business efficiency.

[0003] In the prior art, by analyzing the similarity between contents, other contents similar to the contents that the user has browsed are recommended. If the recommendation method cannot generate recommendation results within a reasonable time, the user may feel that the waiting time is too long, which affects the overall user experience. This problem is particularly prominent in web search recommendation scenarios with high real-time requirements. In addition, if the data cannot be processed in real time, the recommendation results generated based on outdated data will lose their reference value. Therefore, it is necessary to design a personalized recommendation method and device based on user behavior analysis that can process a large amount of data in a short time and generate recommendation results, so as to solve the problem of low real-time data processing in the personalized recommendation process. Summary of the invention

[0004] In order to solve the technical problem of low real-time data processing in the process of personalized recommendation in the prior art, the present invention provides a personalized recommendation method and device based on user behavior analysis. The technical solution is as follows:

[0005] On the one hand, a personalized recommendation method based on user behavior analysis is provided, the method is implemented by a personalized recommendation device based on user behavior analysis, and the method includes:

[0006] S1. Analyze the user behavior data collected in real time to obtain the access frequency evaluation value, and generate recommendation suggestions based on the user portrait. The user behavior data is used to describe the user's operation behavior in different recommended tags, and the user behavior data includes click data, browsing data and search data;

[0007] S2. Real-time monitoring of the processed data in the data analysis process using a data acquisition tool, wherein the processed data includes the number of cache hits, CPU usage, memory usage, and response time;

[0008] S3, evaluating the real-time performance of data analysis according to the processed data to obtain a real-time data evaluation value, wherein the real-time data evaluation value is used to measure the real-time performance of data analysis;

[0009] S4. Determine whether to take optimization measures based on the real-time evaluation value of the data and a preset real-time threshold, wherein the optimization measures include cache strategy adjustment and resource allocation optimization.

[0010] On the other hand, a personalized recommendation device based on user behavior analysis is provided, and the device is applied to a personalized recommendation method based on user behavior analysis, and the device includes:

[0011] A user behavior analysis module is used to analyze the user behavior data collected in real time to obtain the access frequency evaluation value and generate recommendation suggestions based on the user portrait. The user behavior data is used to describe the user's operation behavior in different recommended tags. The user behavior data includes click data, browsing data and search data;

[0012] A processing data acquisition module is used to monitor the processing data in the data analysis process in real time through a data acquisition tool, wherein the processing data includes the number of cache hits, CPU usage, memory usage and response time;

[0013] A data real-time evaluation module, used to evaluate the real-time performance of data analysis according to the processed data to obtain a data real-time evaluation value, wherein the data real-time evaluation value is used to measure the real-time performance of data analysis;

[0014] The data judgment module is used to judge whether to take optimization measures based on the real-time evaluation value of the data and a preset real-time threshold, and the optimization measures include cache strategy adjustment and resource allocation optimization.

[0015] On the other hand, a personalized recommendation device based on user behavior analysis is provided, and the personalized recommendation device based on user behavior analysis includes: a processor; a memory, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned personalized recommendation methods based on user behavior analysis is implemented.

[0016] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned personalized recommendation methods based on user behavior analysis.

[0017] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0018] The embodiment of the present invention performs data analysis on user behavior data collected in real time to obtain an access frequency evaluation value, and generates recommendation suggestions based on user portraits. The user behavior data is used to describe the user's operating behavior in different tags to be recommended, and the user behavior data includes click data, browsing data and search data; the processing data in the data analysis process is monitored in real time through a data collection tool, and the processing data includes the number of cache hits, CPU usage, memory usage and response time; the real-time performance of the data analysis is evaluated based on the processed data to obtain a real-time data evaluation value, and the real-time data evaluation value is used to measure the real-time performance of the data analysis; based on the real-time data evaluation value and a preset real-time threshold, it is determined whether to take optimization measures, and the optimization measures include cache strategy adjustment and resource allocation optimization, thereby improving the real-time performance and overall performance of the system, achieving an improvement in the timeliness of personalized recommendations, and thus improving user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 A flowchart of a personalized recommendation method based on user behavior analysis provided by the present invention;

[0021] Figure 2 A schematic diagram of a change in a visit frequency evaluation value provided by the present invention;

[0022] Figure 3 A structural schematic diagram of a personalized recommendation device based on user behavior analysis provided by the present invention;

[0023] Figure 4 A schematic diagram of the structure of a personalized recommendation device based on user behavior analysis provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0025] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0026] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0027] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0028] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0029] The embodiment of the present invention provides a personalized recommendation method based on user behavior analysis, which can be implemented by a personalized recommendation device based on user behavior analysis, and the personalized recommendation device based on user behavior analysis can be a terminal or a server. Figure 1 As shown, an embodiment of the present invention provides a flowchart of a personalized recommendation method based on user behavior analysis, and the processing flow of the method may include the following steps:

[0030] S1. By analyzing the user behavior data collected in real time, the access frequency evaluation value is obtained, and recommendation suggestions are generated based on the user portrait. The user behavior data is used to describe the user's operation behavior in different recommended tags. The user behavior data includes click data, browsing data and search data.

[0031] Among them, click data includes click events, click locations and click counts; browsing data includes page title, dwell time and content description; search data includes search keywords, search counts and search result titles; and recommendation suggestions indicate content recommendations and interface adjustments based on the user's recommended tags and behavior patterns.

[0032] Among them, the tag to be recommended represents the topic that the user is interested in, such as medical topics; the click event represents the content clicked by the user; the click position represents the specific location clicked by the user, for example, the title in the page; the number of clicks represents the frequency of users clicking on the tag to be recommended; the page title represents the name of the page; the dwell time represents the time the user stays on the page; the content description represents a detailed description of the content displayed on the page; the search keyword represents the search term entered by the user; the number of searches represents the frequency of users searching; the search result title represents the title displayed in the search results; the preset real-time threshold is used to determine whether the real-time performance of data analysis meets the requirements; recommendations are generated based on user portraits, but the recommendations are not output; through the above steps, the timeliness of personalized recommendations is improved, thereby improving user satisfaction and platform usage efficiency.

[0033] Optionally, S1 performs data analysis on the user behavior data collected in real time to obtain a visit frequency evaluation value, including:

[0034] S11, real-time collection of user behavior data of users in different tags to be recommended;

[0035] Among them, the user behavior data of users in different recommended tags are collected in real time by embedding scripts in the website;

[0036] S12, classifying and storing the user behavior data according to preset standards, and performing pre-processing;

[0037] Among them, classified storage means storing user behavior data in different areas of the database according to event types, and preprocessing is the same as ensuring the accuracy of user behavior data;

[0038] Optionally, the specific implementation process of the preprocessing in S12 includes:

[0039] S121, determining whether there is noise in the database, if there is noise, performing noise smoothing on the user behavior data, otherwise executing S122;

[0040] Among them, noise smoothing means smoothing the data using filtering techniques;

[0041] S122, determining whether there are missing values ​​in the database, if there are missing values, filling the missing values ​​in the user behavior data, otherwise executing S123;

[0042] Among them, missing value filling means filling the missing fields in the database with the mean of similar data;

[0043] S123. Standardize the units of the data in the database. Unit standardization means unifying the units of the user behavior data.

[0044] S13. Obtaining a visit frequency evaluation value of the user in different tags to be recommended based on the user behavior data collected in real time. The visit frequency evaluation value is used to measure the visit frequency of the user in the tags to be recommended.

[0045] Optionally, the specific restricted expression of the access frequency evaluation value is the following formula (1):

[0046]

[0047] Where i is the user number, i = 1, 2, ..., N, N is the total number of users, ij is the number of the recommended tags visited by the ith user, ij = i.1, i.2, ..., iJ, iJ is the total number of recommended tags visited by the ith user, FX i.j represents the access frequency evaluation value of the ijth recommended tag visited by the ith user, TL i.j represents the residence time of the ijth recommended tag corresponding to the ith user, DC i.j represents the number of clicks on the ijth recommended tag corresponding to the ith user, SC i.j Indicates the number of searches for the ijth tag to be recommended corresponding to the ith user.

[0048] Among them, the definition Among them, T i.j is the residence time coefficient of the ijth recommended tag corresponding to the ith user, and is defined as Among them, D i.j is the click coefficient of the ijth recommended tag corresponding to the ith user, and is defined as Among them, S i.j is the search frequency coefficient of the ijth tag to be recommended corresponding to the ith user. The algorithm of the access frequency evaluation value is combined with the "stay time coefficient T of the ijth tag to be recommended corresponding to the ith user" i.j ”, “The click count coefficient D of the ijth recommended tag corresponding to the ith user i.j ” and “the search frequency coefficient S of the ijth recommended tag corresponding to the ith user i.j ”, comprehensive analysis obtains the “visit frequency evaluation value FX of the ijth recommended tag visited by the ith user” i.j ", and each parameter is related to each other, not independent, it is a whole, and it is not possible to get the "visit frequency evaluation value FX of the ijth recommended tag visited by the ith user" by simply combining and adding them. i.j ”, for example, in this formula, “the residence time coefficient T of the ijth recommended tag corresponding to the ith useri.j " may lead to an increase in the "visit frequency evaluation value FX of the ijth recommended tag visited by the ith user" i.j " does not necessarily mean that the access frequency evaluation value FX of the ijth recommended tag visited by the ith user increases. i.j " is large, and at this time, we also need to consider the "click coefficient D of the ijth recommended tag corresponding to the ith user i.j ” and “the search frequency coefficient S of the ijth recommended tag corresponding to the ith user i.j " can also determine the frequency of users' visits to the recommended tags at the next moment. It takes into account the "residence time coefficient T of the ijth recommended tag corresponding to the ith user" in a quantitative way. i.j ”, “The click count coefficient D of the ijth recommended tag corresponding to the ith user i.j ”, “The search frequency coefficient S of the ijth recommended tag corresponding to the ith user i.j ” and “the access frequency evaluation value FX of the ijth recommended tag visited by the ith user i.j "The correlation and mutual influence between them.

[0049] Among them, the recommended tags with more clicks tend to attract users to stay longer, which in turn prompts users to perform more search operations. The higher the number of clicks, the higher the user's visit frequency, so the visit frequency evaluation value is larger; the longer the stay time, the greater the possibility of clicking, and the greater the possibility of performing the search operation, which in turn prompts users to visit the tag more frequently, so the visit frequency evaluation value is larger; search operations are often accompanied by click behaviors, because users need to click to view search results or visit related pages. Frequent search operations may mean that users are looking for the tag on the website, which may prolong the user's stay time. The more searches may prompt users to visit more frequently, so the larger the visit frequency evaluation value is. The algorithm itself is obtained through comprehensive analysis, which can accurately and quantitatively analyze the user's visit frequency in the recommended tags, so accurate analysis is performed by establishing mathematical forms; effective evaluation of the user's visit frequency in the recommended tags is achieved, thereby avoiding performance differences caused by low accuracy of the visit frequency evaluation value.

[0050] In a feasible implementation manner, JavaScript is a scripting language widely used in web page development, which can enhance the interactivity and dynamic functions of web pages; JavaScript scripts are used to be embedded in websites so as to collect user behavior data on the website in real time; user behavior data represents data generated by users browsing the website and interacting with the website, including clicks and stays; preset standards include click data, browsing data and search data, that is, user behavior data is classified and stored in different areas of the database according to preset standards; user behavior data is input into the Apriori algorithm to obtain labels; user behavior data is input into hierarchical clustering to obtain user behavior patterns, hierarchical clustering includes bottom-up (agglomerated) hierarchical clustering and top-down (divisive) hierarchical clustering, and top-down hierarchical clustering is selected in this embodiment; through the above steps, an effective evaluation of the frequency of user visits is achieved.

[0051] In a feasible implementation, taking NJ as 1.4 as an example, where N is 1, indicating that the total number of users is 1, and J is 4, indicating that the total number of recommended tags corresponding to each user is 4, the change statistics of the access frequency evaluation value are shown in Table 1:

[0052] Table 1 Statistics of changes in access frequency evaluation values

[0053]

[0054] Among them, from the table, it can be seen that when the proportion of stay time and the proportion of search times remain unchanged, the larger the click coefficient, the larger the visit frequency evaluation value, such as the third group of data and the fourth group of data; from the first group of data and the second group of data, it can be seen that when the proportion of stay time and the proportion of clicks remain unchanged, the larger the search coefficient, the larger the visit frequency evaluation value; such as the second and third groups, when the proportion of clicks and the proportion of searches remain unchanged, the larger the stay time coefficient, the larger the visit frequency evaluation value, therefore, the visit frequency evaluation value is positively correlated with the stay time coefficient, the click coefficient and the search coefficient.

[0055] Optionally, after the step of classifying and storing the user behavior data according to preset standards, the method further includes:

[0056] The user behavior data is transferred to the database and the transmission process of the user behavior data is encrypted using the transport layer security protocol;

[0057] Divide the user behavior data into behavior fragment data according to preset standards, and encrypt the behavior fragment data using a key generated by a symmetric encryption algorithm;

[0058] Among them, the preset standard is used to classify user behavior data according to the user's behavior type including click, browse and search; Transport Layer Security (TLS) is a security protocol used to provide confidentiality and data integrity between two communication applications; TLS is often used to ensure that data transmission between a client and a server is encrypted, thereby protecting the data from eavesdropping or tampering; a symmetric encryption algorithm is an algorithm that uses the same key for encryption and decryption, such as Advanced Encryption Standard (AES) and Data Encryption Standard (DES); in this embodiment, the AES algorithm is selected; the behavior fragment data represents the small segments of data into which the user behavior data is divided according to the classification results, for example, the user's click data is regarded as a behavior fragment, the search data is regarded as a behavior fragment, and the browsing data is regarded as a behavior fragment; the encrypted behavior fragment data is classified and stored in the corresponding area of ​​the database according to the user's behavior type including click data, browsing data and search data; through the above steps, the security of the user behavior data during transmission and storage is improved, thereby preventing data leakage or illegal access.

[0059] Among them, behavioral fragment data includes click data, browsing data and search data;

[0060] The encrypted behavior fragment data is classified and stored in a preset area in the database.

[0061] Among them, classified storage means that the data of each category is stored in the database accordingly.

[0062] Optionally, S1 generates recommendations based on the user profile, including:

[0063] According to the access frequency evaluation value, the corresponding users are divided into user groups, and user profiles are constructed based on user behavior data. User groups include high-frequency user groups and low-frequency user groups. User profiles include tags to be recommended and behavior patterns. Behavior patterns include access frequency evaluation values ​​and access time.

[0064] Predict user behavior based on user behavior data to obtain behavior prediction results;

[0065] Similarity is calculated based on the user portrait and behavior prediction results to obtain content similarity, which indicates the degree of similarity between the user portrait and the recommended content.

[0066] It is determined whether the content similarity is not less than a preset similarity threshold. If the content similarity is not less than the preset similarity threshold, the corresponding content is recorded as a recommendation suggestion, and the recommended content is adjusted according to the recommendation suggestion. Otherwise, the corresponding content is recorded as a non-recommended suggestion.

[0067] In a feasible implementation, the access frequency evaluation value is input into the K-means clustering algorithm to obtain the user group; the preset access frequency is obtained from the preset database, and the preset access frequency is represented by the average value of the historical access frequency; the high-frequency user group represents users with a greater than preset access frequency; the low-frequency user group represents users with a less than preset access frequency; the recommended tags represent content related to the user's interests; the behavior pattern is used to describe the pattern of user behavior, including access time and access frequency; a machine learning algorithm is applied to perform user profiling and behavior prediction; the user behavior data is input into the decision tree to obtain the user portrait, and the statistics of the user portrait are displayed using charts or graphs. The method comprises the following steps: calculating the similarity between the user portrait and the recommended content, determining whether the recommended content meets the user's interests and needs, and obtaining the preset similarity threshold. The preset similarity threshold is obtained from the preset database, and the preset similarity threshold is set to the mean of the content similarity plus two standard deviations. For example, if the similarity of the content labeled as medical is 0.5 and the preset similarity threshold is 0.4, the content labeled as medical will be recorded as a recommendation. Through the above steps, the accuracy of personalized recommendations is achieved, thereby improving user satisfaction.

[0068] In a feasible implementation, a vocabulary is first created based on all the words in the content to be recommended and the browsed content, and then the content to be recommended and the browsed content are converted into vectors through a bag-of-words model. The content similarity is obtained based on the feature vector of the content to be recommended and the feature vector of the browsed content. For example, the content to be recommended is "machine learning is a subfield of artificial intelligence" and the browsed content is "artificial intelligence includes machine learning and deep learning". The vocabulary is {"machine", "learning", "artificial", "intelligence", "sub", "field", "includes", "depth"}. Therefore, the feature vector of the content to be recommended is [1, 1, 1, 1, 1, 1, 0, 0], and the feature vector of the browsed content is [1, 2, 1, 1, 0, 0, 1, 1]. The vector represents the number of times the corresponding word in the vocabulary appears in the content. The content similarity is calculated by the following formula (2):

[0069]

[0070] In the formula, NR i.j Indicates the content similarity of the ijth recommended tag corresponding to the ith user, TN i.jThe feature vector of the recommended content of the ijth recommended tag corresponding to the ith user, TZ i.j The feature vector representing the browsed content of the ijth tag to be recommended corresponding to the ith user, TN i.j *TZ i.j The dot product between the feature vector of the content to be recommended and the feature vector of the content that has been viewed, ||TN i ||The modulus of the feature vector representing the recommended content, ||TZ i.j || represents the modulus of the feature vector of the content description; the feature vector of the content to be recommended represents the number of times the corresponding word in the vocabulary appears in the recommended content; the feature vector of the browsed content represents the number of times the corresponding word in the vocabulary appears in the browsed content; TN i.j *TZ i.j It is used to measure the similarity between the feature vector of the content to be recommended and the feature vector of the browsed content. i.j *TZ i.j The larger the value, the closer the feature vector of the content to be recommended is to the feature vector of the browsed content in the feature space, indicating that the similarity between the feature vector of the content to be recommended and the feature vector of the browsed content is higher; ||TN i ||*||TZ i.j || is used to normalize the size of the vector to eliminate the influence of the vector length, thereby ensuring that the calculation of similarity is not affected by the vector size.

[0071] S2. Use data collection tools to monitor the processing data in the data analysis process in real time. The processing data includes the number of cache hits, CPU usage, memory usage, and response time.

[0072] Among them, data collection tools include New Relic and perf tools; processing data includes cache hit counts, CPU usage, memory usage and response time; cache hit counts indicate the number of attempts to retrieve data from the cache within a specific time period and the number of successful data retrievals; the higher the cache hit count, the better the cache effectiveness and the higher the system performance will generally be; the higher the cache evaluation value, the higher the real-time nature of the cache content; the higher the processing evaluation value, the faster the data processing speed and the more effective it is in meeting user needs; this ensures real-time nature in the data analysis process, thereby improving system performance and user experience.

[0073] S3. Evaluate the real-time performance of data analysis based on the processed data to obtain a real-time data evaluation value, which is used to measure the real-time performance of data analysis.

[0074] Optionally, S3 evaluates the real-time performance of data analysis according to the processed data to obtain a real-time data evaluation value, including:

[0075] The data collected during the user behavior analysis is monitored in real time by a data collection tool. The real-time performance of the data analysis is evaluated based on the processed data to obtain a real-time data evaluation value, which is used to measure the real-time performance of the data analysis.

[0076] The data real-time evaluation value includes a cache evaluation value and a processing evaluation value. The cache evaluation value is used to measure the real-time performance of the cache content, and the processing evaluation value is used to measure the timeliness of data processing.

[0077] Optionally, the step of obtaining the cache evaluation value includes:

[0078] The cache evaluation value is calculated using the following formula (3) based on the cache hit rate, access frequency evaluation value, and the number of searches of user behavior data during the user behavior analysis process:

[0079]

[0080] Where i is the user number, i = 1, 2, ..., N, N is the total number of users, ij is the number of the recommended tags visited by the ith user, ij = i.1, i.2, ..., iJ, iJ is the total number of recommended tags visited by the ith user, HC i.j HM represents the cache evaluation value of the ijth tag to be recommended visited by the ith user, i.j represents the cache hit count of the ijth recommended tag corresponding to the ith user, FX i.j represents the access frequency evaluation value of the ijth recommended tag corresponding to the ith user, SC i.j It represents the number of searches for the ijth tag to be recommended corresponding to the ith user, where the number of searches is not zero. The processing evaluation value is obtained by denormalizing the memory usage, CPU usage, response time, and cache factor during the user analysis process. The cache factor represents the impact of the cache on data processing performance.

[0081] The step of obtaining the evaluation value includes: obtaining the evaluation value through the memory usage rate, CPU usage rate, response time and cache factor after de-unitization during the user analysis process, wherein the cache factor indicates the influence of the cache on the data processing performance;

[0082] The processing evaluation value is calculated by the following formula (4):

[0083]

[0084] Where t is the number of searches, t = 1, 2, ..., T, T is the total number of searches, CL i.trepresents the processing evaluation value of the t-th search corresponding to the i-th user, CP i.t represents the CPU usage of the tth search corresponding to the i-th user, NC i.t represents the memory usage of the t-th search corresponding to the i-th user, XY i.t represents the response time of the t-th search corresponding to the i-th user, and θ represents the cache factor.

[0085] S4. Determine whether to take optimization measures based on the real-time evaluation value of the data and the preset real-time threshold. The optimization measures include cache strategy adjustment and resource allocation optimization.

[0086] Optionally, judging whether to take optimization measures based on the real-time evaluation value of the data and a preset real-time threshold includes:

[0087] S41, determining whether the cache evaluation value is not less than a preset cache threshold. If the cache evaluation value is not less than the preset cache threshold, executing S42, otherwise adjusting the cache strategy until the cache evaluation value is not less than the preset cache threshold. The cache strategy adjustment includes shortening the cache expiration time and increasing the cache capacity.

[0088] S42, judging whether the processing evaluation value is not less than a preset processing threshold value, if the processing evaluation value is not less than the preset processing threshold value, outputting the recommendation, otherwise executing S43;

[0089] S43. Extract the CPU usage and memory usage of each component corresponding to the processing evaluation value that is less than the preset processing threshold, and arrange them in ascending order to obtain the CPU usage extreme value and the memory usage extreme value. Assign the task corresponding to the maximum CPU usage to the component with the minimum CPU usage for execution, and assign the task corresponding to the maximum memory usage to the component with the minimum memory usage for execution, until the processing evaluation value is not less than the preset processing threshold. The CPU usage extreme value includes the maximum CPU usage and the minimum CPU usage, and the memory usage extreme value includes the maximum memory usage and the minimum memory usage.

[0090] Among them, only when the real-time evaluation value of the data is not less than the preset real-time threshold, the generated recommendation suggestion is output; the preset cache threshold is used to compare with the cache evaluation value to determine whether the cache system needs to be optimized; if the cache evaluation value is lower than the preset cache threshold, it means that the cache system may not be efficient enough and needs to be adjusted; shortening the cache expiration time is used to allow the cache data to be updated or replaced faster, and increasing the cache capacity is used to store more data to reduce the access to the original data source; the preset processing threshold is used to compare with the processing evaluation value to determine whether the system processing capacity needs to be optimized; the preset cache threshold and the preset processing threshold are obtained from the preset database, the preset cache threshold is set to 70% of the maximum value of the historical cache evaluation value under normal operating conditions, and the preset processing threshold is set to 70% of the maximum value of the processing evaluation value under normal operating conditions; the CPU usage rate indicates the percentage of the CPU occupied in a period of time to process tasks; the higher the CPU usage rate may indicate that the system is processing a large number of tasks, but it may also indicate that there is a performance bottleneck; the memory usage rate refers to the proportion of the system memory occupied; the higher the memory usage rate may indicate that the system is using a large amount of memory resources, but it may also indicate that the memory is insufficient and needs to be optimized; the data processing speed is improved and the system delay is reduced, thereby achieving an improved user experience.

[0091] In a feasible implementation, by calculating the Z score, that is, the difference between the data point and the mean divided by the standard deviation, the threshold of the Z score (such as ±3) is used to determine which data points may be noise; missing values ​​are identified by built-in functions in the database, for example, in Python's Pandas, the isnull() or isna() functions can be used to check missing values; noise data may be generated due to a variety of reasons, such as interference during data transmission, etc., and noise data will interfere with the results of data analysis and reduce data accuracy; the purpose of noise smoothing is to reduce or eliminate noise in the data, making the data smoother and thus more accurately reflecting the actual situation; by applying filtering techniques including: moving average, median filtering and high Gaussian filtering, etc., can smooth out abnormal values ​​or fluctuations in the data and make the data trend more obvious. In this embodiment, Gaussian filtering technology is selected. Missing values ​​may be caused by various reasons including: data not collected, data lost, and data damaged. The purpose of missing value filling is to fill the missing values ​​in the data set through certain methods, including: the mean, median, mode and interpolation of similar data, so as to make the data set complete. In this embodiment, the mean of similar data is selected for filling. Unit standardization helps to eliminate the problems caused by inconsistent units and improve the accuracy and comparability of data analysis. For example, the unit of residence time is unified as minutes. Through the above steps, the accuracy and consistency of user behavior data are improved.

[0092] like Figure 2As shown, it is a schematic diagram of the change of the cache evaluation value provided by an embodiment of the present invention. From the figure, it can be seen that when the number of searches and the access frequency evaluation value remain unchanged, the more cache hits there are, the larger the cache evaluation value is; when the number of cache hits and the access frequency evaluation value remain unchanged, the more searches there are, the smaller the cache evaluation value is; when the number of searches and the number of cache hits remain unchanged, the larger the access frequency evaluation value is, the larger the cache evaluation value is. Therefore, the cache evaluation value is positively correlated with the number of cache hits and the access frequency evaluation value, and negatively correlated with the number of searches.

[0093] In a feasible implementation, the cache evaluation algorithm is combined with the "cache hit count HM of the ijth recommended tag corresponding to the ith user" i.j ”, “The number of searches for the ijth recommended tag corresponding to the ith user SC i.j ” and “the access frequency evaluation value FX of the ijth recommended tag corresponding to the ith user i.j ”, comprehensive analysis obtains “the cache evaluation value HC of the ijth recommended tag visited by the ith user i.j ", and the various parameters are related to each other, not independent, it is a whole, and it is not possible to get the "cache evaluation value HC of the ijth recommended tag visited by the ith user" by simply combining and adding them. i.j ”, for example, in this formula, “the number of cache hits of the ijth recommended tag corresponding to the ith user HM i.j " may cause the cache evaluation value HC of the ijth recommended tag visited by the ith user to increase i.j " does not necessarily mean that the cache evaluation value HC of the ijth recommended tag visited by the ith user i.j " is large, and at this time, we also need to consider "the number of searches for the ijth recommended tag corresponding to the ith user SC i.j ” and “the access frequency evaluation value FX of the ijth recommended tag corresponding to the ith user i.j " can also determine the real-time nature of the cached content at the next moment. It takes into account the "number of cache hits HM of the ijth recommended tag corresponding to the ith user" in a quantitative way. i.j ”, “The number of searches for the ijth recommended tag corresponding to the ith user SC i.j ”, “FX, the access frequency evaluation value of the ijth recommended tag corresponding to the ith user i.j ” and “the cache evaluation value HC of the ijth recommended tag visited by the ith user i.j" Generally speaking, when the number of searches increases, if the cache strategy is effective, the number of cache hits will increase, resulting in an increase in the cache evaluation value, but it may also increase the number of cache misses, resulting in a decrease in the cache evaluation value; if the access frequency evaluation value increases, it means that the user's access frequency to the ijth recommended tag is high, and these data will be more likely to remain in the cache, thereby improving the cache hit rate and increasing the cache evaluation value. The algorithm itself is obtained through comprehensive analysis, which can accurately and quantitatively analyze the real-time nature of the cache content, indicating the effect of accurately evaluating the cache evaluation value, so accurate analysis is performed by establishing a mathematical form.

[0094] The algorithm for processing the evaluation value combines the CPU usage CP of the t-th search corresponding to the i-th user. i.t ”, “The memory usage of the t-th search corresponding to the i-th user NC i.t " and "the response time XY of the t-th search corresponding to the i-th user i.t ", comprehensive analysis to obtain "the processing evaluation value CL of the t-th search corresponding to the i-th user i.t ", and the various parameters are related to each other, not independent, it is a whole, and it is not possible to get the "processing evaluation value CL of the t-th search corresponding to the i-th user" by simply combining and adding them. i.t ”, for example, in this formula, “the CPU usage CP of the t-th search corresponding to the i-th user i.t " may lead to an increase in the processing evaluation value CL of the t-th search corresponding to the i-th user i.t " does not necessarily mean that the processing evaluation value CL of the t-th search corresponding to the i-th user i.t " is large, and at this time we also need to consider "the memory usage rate NC of the t-th search corresponding to the i-th user i.t " and "the response time XY of the t-th search corresponding to the i-th user i.t " can also determine the real-time nature of data processing at the next moment. It takes into account the "CPU usage CP of the t-th search corresponding to the i-th user" in a quantitative way. i.t ”, “The memory usage of the t-th search corresponding to the i-th user NC i.t ”, “The response time XY for the tth search corresponding to the i-th user i.t ” and “the processing evaluation value CL of the t-th search corresponding to the i-th user i.t"The correlation and mutual influence between them. Generally speaking, the higher the memory usage rate, the more likely it is to increase the burden on the CPU, resulting in an increase in CPU usage rate. It may also cause the system to enter swap mode (swap), which increases the response time, thereby affecting the real-time performance of data processing and resulting in a decrease in the processing evaluation value. If the CPU usage rate is higher, the memory usage may also increase, which will also lead to an increase in response time, because the system's ability to process requests is limited, thereby affecting the real-time performance of data processing and resulting in a decrease in the processing evaluation value. The response time directly reflects the real-time performance of the system in processing data. A lower response time usually means a higher real-time performance of data processing, resulting in an increase in the processing evaluation value. The algorithm obtained through comprehensive analysis can accurately and quantitatively analyze the real-time performance of data processing, indicating the effect of accurately evaluating the processing evaluation value. Therefore, accurate analysis is performed by establishing a mathematical form. Effective evaluation of the real-time performance of cache content and data processing is achieved, thereby avoiding performance deviations caused by inaccurate cache evaluation values ​​and processing evaluation values.

[0095] The embodiment of the present invention performs data analysis on user behavior data collected in real time to obtain an access frequency evaluation value, and generates recommendation suggestions based on user portraits. The user behavior data is used to describe the user's operating behavior in different tags to be recommended, and the user behavior data includes click data, browsing data and search data; the processing data in the data analysis process is monitored in real time through a data collection tool, and the processing data includes the number of cache hits, CPU usage, memory usage and response time; the real-time performance of the data analysis is evaluated based on the processed data to obtain a real-time data evaluation value, and the real-time data evaluation value is used to measure the real-time performance of the data analysis; based on the real-time data evaluation value and a preset real-time threshold, it is determined whether to take optimization measures, and the optimization measures include cache strategy adjustment and resource allocation optimization, thereby improving the real-time performance and overall performance of the system, achieving an improvement in the timeliness of personalized recommendations, and thus improving user satisfaction.

[0096] like Figure 3 FIG. 1 is a schematic diagram of a personalized recommendation device based on user behavior analysis provided by an embodiment of the present invention, wherein the device is used in a personalized recommendation method based on user behavior analysis. Figure 3 The device includes: a user behavior analysis module 310, a processing data acquisition module 320, a real-time data evaluation module 330 and a data judgment module 340. Among them:

[0097] User behavior analysis module 310, used to obtain access frequency evaluation values ​​by analyzing user behavior data collected in real time, and generate recommendation suggestions based on user portraits. User behavior data is used to describe user operation behaviors in different tags to be recommended. User behavior data includes click data, browsing data and search data. Click data includes click events, click positions and click times. Browsing data includes page titles, dwell time and content descriptions. Search data includes search keywords, search times and search result titles. Recommendation suggestions indicate content recommendations and interface adjustments based on the user's tags to be recommended and behavior patterns.

[0098] Processing data acquisition module 320, used to monitor the processing data in the data analysis process in real time through data acquisition tools, the processing data includes the number of cache hits, CPU usage, memory usage and response time;

[0099] A data real-time evaluation module 330 is used to evaluate the real-time performance of data analysis according to the processed data to obtain a data real-time evaluation value, where the data real-time evaluation value is used to measure the real-time performance of data analysis;

[0100] The data judgment module 340 is used to judge whether to take optimization measures based on the real-time evaluation value of the data and a preset real-time threshold, and the optimization measures include cache strategy adjustment and resource allocation optimization.

[0101] Optionally, the step of analyzing the user behavior data collected in real time to obtain the access frequency evaluation value includes:

[0102] Collect user behavior data in different recommended tags in real time;

[0103] Classify and store user behavior data according to preset standards and perform pre-processing;

[0104] The access frequency evaluation value of the user in different tags to be recommended is obtained according to the user behavior data collected in real time, and the access frequency evaluation value is used to measure the access frequency of the user in the tags to be recommended.

[0105] Optionally, the specific restricted expression of the access frequency evaluation value is the following formula (1):

[0106]

[0107] Where i is the user number, i = 1, 2, ..., N, N is the total number of users, ij is the number of the recommended tags visited by the ith user, ij = i.1, i.2, ..., iJ, iJ is the total number of recommended tags visited by the ith user, FX i.jrepresents the access frequency evaluation value of the ijth recommended tag visited by the ith user, TL i.j represents the residence time of the ijth recommended tag corresponding to the ith user, DC i.j represents the number of clicks on the ijth recommended tag corresponding to the ith user, SC i.j Indicates the number of searches for the ijth tag to be recommended corresponding to the ith user.

[0108] Optionally, after the step of classifying and storing the user behavior data according to preset standards, the method further includes:

[0109] The user behavior data is transferred to the database and the transmission process of the user behavior data is encrypted using the transport layer security protocol;

[0110] Divide the user behavior data into behavior fragment data according to preset standards, and encrypt the behavior fragment data using a key generated by a symmetric encryption algorithm;

[0111] The encrypted behavior fragment data is classified and stored in a preset area in the database.

[0112] Optionally, the specific implementation process of the preprocessing includes:

[0113] S121, determining whether there is noise in the database, if there is noise, performing noise smoothing on the user behavior data, otherwise executing S122;

[0114] S122, determining whether there are missing values ​​in the database, if there are missing values, filling the missing values ​​in the user behavior data, otherwise executing S123;

[0115] S123. Standardize the units of the data in the database. The unit standardization means unifying the units of the user behavior data.

[0116] Optionally, generating a recommendation based on the user portrait includes:

[0117] The corresponding users are divided into user groups according to the access frequency evaluation value, and a user profile is constructed in combination with the user behavior data. The user profile includes the tags to be recommended and the behavior pattern. The behavior pattern includes the access frequency evaluation value and the access time.

[0118] Predict user behavior based on user behavior data to obtain behavior prediction results;

[0119] Calculate similarity based on user portraits and behavior prediction results to obtain content similarity;

[0120] It is determined whether the content similarity is not less than a preset similarity threshold. If the content similarity is not less than the preset similarity threshold, the corresponding content is recorded as a recommendation suggestion, and the recommended content is adjusted according to the recommendation suggestion. Otherwise, the corresponding content is recorded as a non-recommended suggestion.

[0121] Optionally, the real-time data evaluation module 330 is used to:

[0122] The processed data in the user behavior analysis process is monitored in real time by a data collection tool, and the real-time performance of the data analysis is evaluated based on the processed data to obtain a real-time data evaluation value, wherein the real-time data evaluation value is used to measure the real-time performance of the data analysis;

[0123] The data real-time evaluation value includes a cache evaluation value and a processing evaluation value. The cache evaluation value is used to measure the real-time performance of cache content, and the processing evaluation value is used to measure the timeliness of data processing.

[0124] Optionally, the step of obtaining the cache evaluation value includes:

[0125] The cache hit rate, access frequency evaluation value and the number of searches of user behavior data in the user behavior analysis process are used to calculate the cache evaluation value. The cache evaluation value is specifically calculated using the following formula (2):

[0126]

[0127] Where i is the user number, i = 1, 2, ..., N, N is the total number of users, ij is the number of the recommended tags visited by the ith user, ij = i.1, i.2, ..., iJ, iJ is the total number of recommended tags visited by the ith user, HC i.j HM represents the cache evaluation value of the ijth tag to be recommended visited by the ith user, i.j represents the cache hit count of the ijth recommended tag corresponding to the ith user, FX i.j represents the access frequency evaluation value of the ijth recommended tag corresponding to the ith user, SC i.j Indicates the number of searches for the ijth tag to be recommended corresponding to the ith user;

[0128] The step of obtaining the processing evaluation value includes: obtaining the processing evaluation value by de-unitizing the memory usage, CPU usage, response time and cache factor during the user analysis process, wherein the cache factor indicates the influence of the cache on the data processing performance; wherein the processing evaluation value is calculated by the following formula (3):

[0129]

[0130] Where t is the number of searches, t = 1, 2, ..., T, T is the total number of searches, CL i.t represents the processing evaluation value of the t-th search corresponding to the i-th user, CP i.t represents the CPU usage of the tth search corresponding to the i-th user, NC i.t represents the memory usage of the t-th search corresponding to the i-th user, XY i.t represents the response time of the t-th search corresponding to the i-th user, and θ represents the cache factor.

[0131] Optionally, the data determination module 340 is used to:

[0132] S41, determining whether the cache evaluation value is not less than a preset cache threshold, if the cache evaluation value is not less than the preset cache threshold, executing S42, otherwise adjusting the cache strategy until the cache evaluation value is not less than the preset cache threshold, the cache strategy adjustment includes shortening the cache expiration time and increasing the cache capacity;

[0133] S42, judging whether the processing evaluation value is not less than a preset processing threshold value, if the processing evaluation value is not less than the preset processing threshold value, outputting the recommendation, otherwise executing S43;

[0134] S43. Extract the CPU usage and memory usage of each component corresponding to the processing evaluation value that is less than the preset processing threshold, and arrange them in ascending order to obtain the CPU usage extreme value and the memory usage extreme value. Assign the task corresponding to the maximum CPU usage to the component with the minimum CPU usage for execution, and assign the task corresponding to the maximum memory usage to the component with the minimum memory usage for execution, until the processing evaluation value is not less than the preset processing threshold, the CPU usage extreme value includes the maximum CPU usage and the minimum CPU usage, and the memory usage extreme value includes the maximum memory usage and the minimum memory usage.

[0135] The embodiment of the present invention performs data analysis on user behavior data collected in real time to obtain an access frequency evaluation value, and generates recommendation suggestions based on user portraits. The user behavior data is used to describe the user's operating behavior in different tags to be recommended, and the user behavior data includes click data, browsing data and search data; the processing data in the data analysis process is monitored in real time through a data collection tool, and the processing data includes the number of cache hits, CPU usage, memory usage and response time; the real-time performance of the data analysis is evaluated based on the processed data to obtain a real-time data evaluation value, and the real-time data evaluation value is used to measure the real-time performance of the data analysis; based on the real-time data evaluation value and a preset real-time threshold, it is determined whether to take optimization measures, and the optimization measures include cache strategy adjustment and resource allocation optimization, thereby improving the real-time performance and overall performance of the system, achieving an improvement in the timeliness of personalized recommendations, and thus improving user satisfaction.

[0136] Figure 4 is a structural diagram of a personalized recommendation device based on user behavior analysis provided by an embodiment of the present invention, such as Figure 4 As shown, the personalized recommendation device based on user behavior analysis may include the above Figure 3 The personalized recommendation device based on user behavior analysis shown. Optionally, the personalized recommendation device based on user behavior analysis 410 may include a first processor 2001 .

[0137] Optionally, the personalized recommendation device 410 based on user behavior analysis may further include a memory 2002 and a transceiver 2003 .

[0138] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0139] Combine the following Figure 4 The components of the personalized recommendation device 410 based on user behavior analysis are specifically introduced as follows:

[0140] The first processor 2001 is the control center of the personalized recommendation measurement device 410 based on user behavior analysis, which can be a processor or a general term for multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (FPGAs).

[0141] Optionally, the first processor 2001 may execute various functions of the personalized recommendation device 410 based on user behavior analysis by running or executing a software program stored in the memory 2002 and calling data stored in the memory 2002 .

[0142] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 4 CPU0 and CPU1 are shown in FIG.

[0143] In a specific implementation, as an embodiment, the personalized recommendation device 410 based on user behavior analysis may also include multiple processors, such as Figure 4The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0144] The memory 2002 is used to store the software program for executing the solution of the present invention, and the first processor 2001 controls the execution. The specific implementation method can refer to the above method embodiment and will not be elaborated here.

[0145] Optionally, the memory 2002 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 2002 may be integrated with the first processor 2001, or may exist independently, and may be connected to the first processor 2001 through the interface circuit ( Figure 4 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0146] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0147] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 4 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0148] Optionally, the transceiver 2003 may be integrated with the first processor 2001, or may exist independently, and communicate with the first processor 2001 through the interface circuit ( Figure 4 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0149] It should be noted that Figure 4 The structure of the personalized recommendation device 410 based on user behavior analysis shown in the figure does not constitute a limitation on the router, and the actual knowledge structure recognition device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0150] In addition, the technical effects of the personalized recommendation device 410 based on user behavior analysis can refer to the technical effects of the personalized recommendation method based on user behavior analysis described in the above method embodiment, which will not be elaborated here.

[0151] It should be understood that the first processor 2001 in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0152] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0153] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0154] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0155] In the present invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0156] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0157] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0158] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be elaborated here.

[0159] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0160] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0161] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0162] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0163] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A personalized recommendation method based on user behavior analysis, characterized in that: The method comprises: S1. Analyze the user behavior data collected in real time to obtain the access frequency evaluation value, and generate recommendation suggestions based on the user portrait. The user behavior data is used to describe the user's operation behavior in different recommended tags, and the user behavior data includes click data, browsing data and search data; S2. Real-time monitoring of the processed data in the data analysis process using a data acquisition tool, wherein the processed data includes the number of cache hits, CPU usage, memory usage, and response time; S3, evaluating the real-time performance of data analysis according to the processed data to obtain a real-time data evaluation value, wherein the real-time data evaluation value is used to measure the real-time performance of data analysis; S4. Determine whether to take optimization measures based on the real-time evaluation value of the data and a preset real-time threshold, wherein the optimization measures include cache strategy adjustment and resource allocation optimization.

2. The personalized recommendation method based on user behavior analysis according to claim 1, characterized in that: S1 obtains a visit frequency evaluation value by analyzing the user behavior data collected in real time, including: S11, real-time collection of user behavior data of users in different tags to be recommended; S12, classifying and storing the user behavior data according to preset standards, and performing pre-processing; S13. Obtaining a user's access frequency evaluation value in different tags to be recommended based on the user behavior data collected in real time, wherein the access frequency evaluation value is used to measure the user's access frequency in the tags to be recommended.

3. The personalized recommendation method based on user behavior analysis according to claim 2, characterized in that: The specific restricted expression of the access frequency evaluation value is the following formula (1): Where i is the user number, i = 1, 2, ..., N, N is the total number of users, ij is the number of the recommended tags visited by the ith user, ij = i.1, i.2, ..., iJ, iJ is the total number of recommended tags visited by the ith user, FX i.j represents the access frequency evaluation value of the ijth recommended tag visited by the ith user, TL i.j represents the residence time of the ijth recommended tag corresponding to the ith user, DC i.j represents the number of clicks on the ijth recommended tag corresponding to the ith user, SC i.j Indicates the number of searches for the ijth tag to be recommended corresponding to the ith user.

4. The personalized recommendation method based on user behavior analysis according to claim 2, characterized in that: After the step of classifying and storing the user behavior data according to the preset standard, the method further includes: The user behavior data is transferred to the database and the transmission process of the user behavior data is encrypted using the transport layer security protocol; Divide the user behavior data into behavior fragment data according to preset standards, and encrypt the behavior fragment data using a key generated by a symmetric encryption algorithm; The encrypted behavior fragment data is classified and stored in a preset area in the database.

5. The personalized recommendation method based on user behavior analysis according to claim 2, characterized in that: The specific implementation process of the preprocessing of S12 includes: S121, determining whether there is noise in the database, if there is noise, performing noise smoothing on the user behavior data, otherwise executing S122; S122, determining whether there are missing values ​​in the database, if there are missing values, filling the missing values ​​in the user behavior data, otherwise executing S123; S123. Standardize the units of the data in the database. The unit standardization means unifying the units of the user behavior data.

6. The personalized recommendation method based on user behavior analysis according to claim 1, characterized in that: The step S1 generates recommendations based on the user profile, including: The corresponding users are divided into user groups according to the access frequency evaluation value, and a user profile is constructed in combination with the user behavior data. The user profile includes the tags to be recommended and the behavior pattern. The behavior pattern includes the access frequency evaluation value and the access time. Predict user behavior based on user behavior data to obtain behavior prediction results; Calculate similarity based on user portraits and behavior prediction results to obtain content similarity; It is determined whether the content similarity is not less than a preset similarity threshold. If the content similarity is not less than the preset similarity threshold, the corresponding content is recorded as a recommendation suggestion, and the recommended content is adjusted according to the recommendation suggestion. Otherwise, the corresponding content is recorded as a non-recommended suggestion.

7. The personalized recommendation method based on user behavior analysis according to claim 1, characterized in that: The step S3 evaluates the real-time performance of data analysis according to the processed data to obtain a real-time data evaluation value, including: Monitor the processed data in the user behavior analysis process in real time through a data collection tool, evaluate the real-time performance of the data analysis based on the processed data, and obtain a real-time data evaluation value, which is used to measure the real-time performance of the data analysis; The data real-time evaluation value includes a cache evaluation value and a processing evaluation value. The cache evaluation value is used to measure the real-time performance of cache content, and the processing evaluation value is used to measure the timeliness of data processing.

8. The personalized recommendation method based on user behavior analysis according to claim 7, characterized in that: The step of obtaining the cache evaluation value comprises: The cache hit rate, access frequency evaluation value and the number of searches of user behavior data in the user behavior analysis process are used to calculate the cache evaluation value using the following formula (2): Where i is the user number, i = 1, 2, ..., N, N is the total number of users, ij is the number of the recommended tags visited by the ith user, ij = i.1, i.2, ..., iJ, iJ is the total number of recommended tags visited by the ith user, HC i.j HM represents the cache evaluation value of the ijth tag to be recommended visited by the ith user, i.j represents the cache hit count of the ijth recommended tag corresponding to the ith user, FX i.j represents the access frequency evaluation value of the ijth recommended tag corresponding to the ith user, SC i.j Indicates the number of searches for the ijth tag to be recommended corresponding to the ith user; The step of obtaining the processing evaluation value includes: obtaining the evaluation value through the memory usage rate, CPU usage rate, response time and cache factor after de-unitization during the user analysis process, wherein the cache factor indicates the influence of the cache on the data processing performance; The processing evaluation value is calculated by the following formula (3): Where t is the number of searches, t = 1, 2, ..., T, T is the total number of searches, CL i.t represents the processing evaluation value of the t-th search corresponding to the i-th user, CP i.t represents the CPU usage of the tth search corresponding to the i-th user, NC i.t represents the memory usage of the t-th search corresponding to the i-th user, XY i.t represents the response time of the t-th search corresponding to the ith user, and θ represents the cache factor.

9. The personalized recommendation method based on user behavior analysis according to claim 1, characterized in that: The step S4 of judging whether to take optimization measures based on the real-time evaluation value of the data and a preset real-time threshold value includes: S41, determining whether the cache evaluation value is not less than a preset cache threshold, if the cache evaluation value is not less than the preset cache threshold, executing S42, otherwise adjusting the cache strategy until the cache evaluation value is not less than the preset cache threshold, the cache strategy adjustment includes shortening the cache expiration time and increasing the cache capacity; S42, judging whether the processing evaluation value is not less than a preset processing threshold value, if the processing evaluation value is not less than the preset processing threshold value, outputting the recommendation, otherwise executing S43; S43. Extract the CPU usage and memory usage of each component corresponding to the processing evaluation value that is less than the preset processing threshold, and arrange them in ascending order to obtain the CPU usage extreme value and the memory usage extreme value. Assign the task corresponding to the maximum CPU usage to the component with the minimum CPU usage for execution, and assign the task corresponding to the maximum memory usage to the component with the minimum memory usage for execution, until the processing evaluation value is not less than the preset processing threshold, the CPU usage extreme value includes the maximum CPU usage and the minimum CPU usage, and the memory usage extreme value includes the maximum memory usage and the minimum memory usage.

10. A personalized recommendation device based on user behavior analysis, the personalized recommendation device based on user behavior analysis is used to implement the personalized recommendation method based on user behavior analysis as claimed in any one of claims 1 to 9, characterized in that: The device comprises: A user behavior analysis module is used to obtain access frequency evaluation values ​​by analyzing user behavior data collected in real time, and to generate recommendation suggestions based on user portraits. The user behavior data is used to describe the user's operation behavior in different tags to be recommended, and the user behavior data includes click data, browsing data, and search data; A processing data acquisition module is used to monitor the processing data in the data analysis process in real time through a data acquisition tool, wherein the processing data includes the number of cache hits, CPU usage, memory usage and response time; A data real-time evaluation module, used for evaluating the real-time performance of data analysis according to the processed data to obtain a data real-time evaluation value, wherein the data real-time evaluation value is used to measure the real-time performance of data analysis; The data judgment module is used to judge whether to take optimization measures based on the real-time evaluation value of the data and a preset real-time threshold, and the optimization measures include cache strategy adjustment and resource allocation optimization.

Images