Data analysis method, system and device and computer storage medium

Abstract

The invention relates to the technical field of data analysis, and discloses a data analysis method, system and device and a computer storage medium. The method comprises the following steps: acquiring user behavior data; calculating a hierarchical weight according to the user behavior data; performing keyword preference calculation according to the user behavior data to obtain keyword preference scores; calculating information density characteristics according to the keyword preference scores; according to the information density feature, the keyword preference score and the hierarchical weight, calculating a user demand priority; calculating a content and keyword correlation score according to the level weight, the keyword preference score and the user behavior data; and calculating a recommended content global priority according to the content and keyword correlation score and the user demand priority, and recommending the content corresponding to the keyword to the user according to the recommended content global priority. According to the method, the accuracy of data analysis can be improved, and more accurate content recommendation is further realized.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and in particular to a data analysis method, system, device and computer storage medium. Background Art

[0002] Data analysis is the process of processing, analyzing and interpreting collected data through technical means to extract information, reveal patterns and assist in decision-making. The data analysis process usually includes stages such as data acquisition, feature extraction, model calculation and result application. For example, data can come from user behavior records, historical features or other environmental variables. Key indicators are extracted through preprocessing and feature engineering, and then combined with algorithms to calculate specific analysis results. These analysis results can be used to build user portraits, recommend content or optimize resource allocation, and are the core link in modern intelligent systems.

[0003] However, existing data analysis technologies are not without problems, especially in terms of the accuracy of analysis results. For example, analysis based on clustering or classification may be highly sensitive to initial parameters, resulting in unstable results; models such as preference scores may produce errors due to missing data or the presence of noise; in recommendation scenarios, if the weights of the algorithm are not set properly, it will affect the sorting of content priorities and ultimately weaken the user experience.

[0004] In summary, although existing data analysis technologies can improve the depth and breadth of analysis by using different algorithms or models, there are still problems with the accuracy of data analysis. Summary of the invention

[0005] The present invention provides a data analysis method, system, device and computer storage medium, which can improve the accuracy of data analysis and further achieve more accurate content recommendation.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides a data analysis method, comprising: Obtaining user behavior data; wherein the user behavior data includes the number of clicks on each keyword, the similarity between each keyword and the behavior, the historical weight of each keyword, and the total number of keywords; Calculating hierarchical weights based on the user behavior data; Calculate keyword preference based on the user behavior data to obtain a keyword preference score; Calculating information density features according to the keyword preference scores; Calculating user demand priority according to the information density feature, the keyword preference score and the hierarchical weight; Calculating a content-keyword relevance score based on the hierarchical weights, keyword preference scores, and user behavior data; The global priority of the recommended content is calculated according to the content-keyword relevance score and the user demand priority, and the content corresponding to the keyword is recommended to the user according to the global priority of the recommended content.

[0007] Preferably, calculating the hierarchical weight according to the user behavior data includes: The layer weights are calculated using the following formula: In the formula, The historical weight of each keyword; is the total number of keywords; is the level weight; is the historical weight of the ith keyword.

[0008] Preferably, the keyword preference calculation is performed based on the user behavior data to obtain a keyword preference score, including: The keyword preference score is calculated using the following formula: In the formula, The historical weight of each keyword; The number of clicks for each keyword; The similarity between each keyword and behavior; is the total number of keywords; Score keyword preference; is the number of clicks on the i-th keyword; is the historical weight of the ith keyword; is the similarity between the ith keyword and behavior.

[0009] Preferably, calculating the information density feature according to the keyword preference score includes: The information density feature is calculated by the following formula: In the formula, Score keyword preference; The number of clicks for each keyword; is the total number of keywords; is the information density feature; is the number of clicks on the ith keyword.

[0010] Preferably, the calculating of user demand priority according to the information density feature, the keyword preference score and the hierarchical weight includes: The user demand priority is calculated by the following formula: In the formula, Score keyword preference; is the information density feature; is the level weight; and is the weight parameter, satisfying ; Prioritize user needs.

[0011] Preferably, the calculating of the content and keyword relevance score according to the hierarchical weight, keyword preference score and user behavior data includes: The content and keyword relevance score is calculated using the following formula: In the formula, is the level weight; Score keyword preference; The number of clicks for each keyword; is the total number of keywords; and is the correlation weight coefficient, satisfying ; Score the relevance of content to keywords; is the number of clicks on the ith keyword.

[0012] Preferably, the calculating the global priority of the recommended content according to the content-keyword relevance score and the user demand priority includes: The global priority of recommended content is calculated using the following formula: In the formula, Prioritize user needs; Score the relevance of content to keywords; The global priority of the recommended content.

[0013] In a second aspect, the present invention provides a data analysis device, comprising: A data acquisition module is used to acquire user behavior data; wherein the user behavior data includes the number of clicks on each keyword, the similarity between each keyword and the behavior, the historical weight of each keyword and the total number of keywords; A weight calculation module, used to calculate the hierarchical weight according to the user behavior data; A preference calculation module, used to perform keyword preference calculation based on the user behavior data to obtain a keyword preference score; A density calculation module, used to calculate information density features according to the keyword preference scores; A demand calculation module, used to calculate the user demand priority according to the information density feature, the keyword preference score and the level weight; A content relevance calculation module, used to calculate the content and keyword relevance scores based on the hierarchical weights, keyword preference scores and user behavior data; The recommendation calculation module is used to calculate the global priority of the recommended content according to the content and keyword relevance score and the user demand priority, and recommend the content corresponding to the keyword to the user according to the recommended content global priority.

[0014] In a third aspect, the present invention further provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements any one of the above-mentioned data analysis methods when executing the computer program.

[0015] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the data analysis methods described above.

[0016] Compared with the prior art, the present invention has the following beneficial effects: the embodiments of the present invention provide a data analysis method, system, device and computer storage medium. The method includes: obtaining user behavior data; wherein the user behavior data includes the number of clicks on each keyword, the similarity between each keyword and behavior, the historical weight of each keyword and the total number of keywords; calculating the hierarchical weight according to the user behavior data; performing keyword preference calculation according to the user behavior data to obtain a keyword preference score; calculating the information density feature according to the keyword preference score; calculating the user demand priority according to the information density feature, the keyword preference score and the hierarchical weight; calculating the content and keyword relevance score according to the hierarchical weight, the keyword preference score and the user behavior data; calculating the global priority of the recommended content according to the content and keyword relevance score and the user demand priority, and recommending the keyword corresponding content to the user according to the global priority of the recommended content.

[0017] In the present invention, the method is based on user behavior data, comprehensively considers the number of keyword clicks, behavior similarity, historical weight multi-dimensional information, and constructs a hierarchical weight and preference score calculation model. By gradually calculating the keyword preference score, information density characteristics and user demand priority, and combining the content relevance score, the global priority of the recommended content is finally determined. The method uses a layered and progressive analysis method to explore the potential correlation between data, and can also dynamically adjust the analysis weight to adapt to changes in data characteristics, thereby improving the accuracy of the analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of a data analysis method provided by the first embodiment of the present invention; Figure 2 It is a structural diagram of a data analysis device provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] Reference Figure 1 The first embodiment of the present invention provides a data analysis method, comprising the following steps: S11, obtaining user behavior data.

[0021] S12, calculating the hierarchical weight according to the user behavior data.

[0022] S13, performing keyword preference calculation based on the user behavior data to obtain a keyword preference score.

[0023] S14, calculating information density features according to the keyword preference scores.

[0024] S15, calculating the user demand priority according to the information density feature, the keyword preference score and the hierarchical weight.

[0025] S16, calculating the content and keyword relevance score according to the hierarchical weight, keyword preference score and user behavior data.

[0026] S17, calculating the global priority of the recommended content according to the content-keyword relevance score and the user demand priority, and recommending the content corresponding to the keyword to the user according to the global priority of the recommended content.

[0027] It is worth noting that data analysis is the process of processing, analyzing and interpreting collected data through technical means to extract information, reveal patterns and assist in decision-making. The data analysis process usually includes stages such as data acquisition, feature extraction, model calculation and result application. For example, data can come from user behavior records, historical features or other environmental variables. Key indicators are extracted through preprocessing and feature engineering, and then combined with algorithms to calculate specific analysis results. These analysis results can be used to build user portraits, recommend content or optimize resource allocation, and are the core link in modern intelligent systems.

[0028] In order to facilitate the understanding of the present invention, some preferred embodiments of the present invention are further described below.

[0029] In step S11, user behavior data is obtained.

[0030] Preferably, the user behavior data includes the number of clicks on each keyword, the similarity between each keyword and the behavior, the historical weight of each keyword, and the total number of keywords.

[0031] Specifically, the number of clicks on each keyword can be obtained by analyzing the user behavior log to extract the keywords and click records, and then counting the click frequency of each keyword. For example, in an e-commerce scenario, the log record may contain the user ID, keyword, and timestamp, and the number of clicks on each keyword can be counted through SQL query. If the time dimension is required in the analysis, the number of clicks can also be aggregated by day, week, month, etc.

[0032] Specifically, the feature vectors of keywords and behaviors are constructed using user click behavior data, and their relevance is calculated through a similarity algorithm. That is, the similarity between each keyword and behavior needs to be calculated through the keyword vector and the behavior vector to calculate the correlation between them. Exemplarily, the method that can be used to calculate the correlation can be: using cosine similarity to calculate the correlation, cosine similarity measures the cosine value of the angle between two vectors, with a numerical range of 0-1, which can be used to measure the similarity between keywords and user behaviors. First, construct a vector representation of keywords and user behaviors, such as the user's click frequency or weight matrix on keywords; second, use the cosine similarity calculation formula to calculate the cosine similarity; finally, normalize the result to the range of 0-1. The method can also use the Jaccard similarity coefficient, which can calculate the ratio of the intersection and union of two sets, which is suitable for the case where the keyword and behavior data are in the form of sets. First, convert the keyword and user behavior data into a set form, such as a list of clicked keywords; second, calculate the size of the intersection and union; finally, compare the intersection and union to obtain the similarity. For example, the Pearson correlation coefficient can also be used to calculate similarity, which can measure the linear correlation of two variables and is applicable to the case where the keyword weight and behavior characteristics have statistical distribution. Specifically, the mean of the keyword and behavior data is first calculated, and then the covariance and standard deviation are calculated, and then the covariance and standard deviation are compared to obtain the Pearson correlation coefficient.

[0033] Specifically, to obtain the historical weight of each keyword, the following methods and formulas can be used: 1) Time decay method: assign a weight that decreases over time to each click. The earlier the click occurs, the smaller the weight. The total historical weight is calculated by multiplying the number of each click by a decay factor related to the time difference. For example, suppose that among the three clicks of "mobile phone", the most recent click has a greater weight, and the three clicks are 10, 5 and 15 respectively, which gradually decrease in chronological order. Its historical weight is the sum of the products of these click values ​​and the time decay factor. 2) Weighted moving average method: assign different weights to the number of clicks in different time periods, such as a higher weight for the number of clicks in the last month and a lower weight for three months ago. Multiply the number of clicks in each time period by the weight of the time period, add them up, and then divide them by the sum of all weights to obtain the historical weight. For example, the clicks in the last month are assigned 3 times the weight, and the clicks three months ago are assigned 1 times the weight. The total value is calculated by accumulating these segmented weights. 3) Linear decay method: assign decreasing weights to the number of clicks in chronological order from high to low, and the more recent the clicks, the greater the contribution. For example, the weight of the most recent click can be set to 1, and then decrease to 0.5, 0.2, and so on. The number of clicks is multiplied by the corresponding weight and then added up to obtain the total weight.

[0034] Specifically, to obtain the total number of keywords, the following methods can be used: 1) Direct deduplication statistics method: Count all keywords that have appeared, and remove duplicates to obtain the total number of keywords after deduplication. For example, if the record contains "mobile phone", "computer", "mobile phone", and "headphones", after deduplication, it is "mobile phone", "computer", and "headphones", a total of 3 keywords. 2) Hierarchical statistics method: When the keywords have a hierarchical structure, the number of keywords at a certain level can be counted. For example, if the keyword is "electronic products> mobile phone> iPhone", the number of top-level categories "electronic products" can be counted, or the number of sub-categories "mobile phones" can be counted. The corresponding level can be selected for statistics as needed. 3) Time window statistics method: Limit the time range of statistics and only count keywords that appear within a certain period of time. For example, if the total number of keywords in the past month is counted, only the keywords that appeared in that time period are considered and deduplicated to obtain the number of keywords that meet the conditions.

[0035] In step S12, the hierarchical weights are calculated based on the user behavior data.

[0036] Preferably, calculating the hierarchical weight according to the user behavior data includes: The layer weights are calculated using the following formula: In the formula, The historical weight of each keyword; is the total number of keywords; is the level weight; is the historical weight of the ith keyword.

[0037] It is worth noting that hierarchical weight is an indicator used to measure the relative importance of an element (such as a keyword) in the overall system or set, which is usually obtained through normalization. It reflects the influence or priority of an element in a specific level or structure, and is often used in hierarchical data processing, analysis and decision-making systems. Hierarchical weight is a relative proportion calculated based on a certain indicator (such as the number of clicks, historical weight, feature importance), which is used to indicate the contribution ratio of a keyword or element in the overall set. Its characteristic is that hierarchical weight is usually calculated by normalization method so that the sum of all weights is 1 (i.e. relative value). It is also a relative indicator, which does not indicate absolute importance, but reflects the importance of the element compared with other elements. Exemplarily, hierarchical weight can be calculated by the following method: divide the specific attribute value of each element (such as historical weight) by the sum of all element attribute values ​​to obtain the weight of each element. If the time factor is considered, the historical data can be given a decay factor, and the more recent the data, the higher the weight. When elements are distributed in different levels (such as categories or themes), weights can be allocated according to the contribution ratio of each level.

[0038] Specifically, the historical weight of each keyword is used to measure the degree to which the importance of click behavior changes over time or other characteristics (such as user behavior preferences). For example, in a recommendation system, user interests are usually time-sensitive, and recent clicks are more representative of current needs than in the past, so it is necessary to give higher weights to clicks that are more recent. Hierarchical weights are used to measure the relative importance of a keyword among all keywords. Because the historical weight of each keyword may vary due to factors such as the number of clicks and time decay, direct use will make it difficult to compare data of different dimensions. Therefore, it is necessary to convert the historical weight of each keyword into a relative proportion through normalization, so that the sum of the weights of all keywords is 1, which is convenient for analysis and comparison. The advantage of normalization is that it can eliminate the influence of the absolute value of the weight on the calculation result. For example, when the weight value is very large, it can be unified to a limited range (such as 0 to 1) through normalization.

[0039] Specifically, the calculated hierarchical weight represents the relative importance of the keyword among all keywords. For example, if the historical weight of the keyword "mobile phone" is 20, and the total weight of other keywords is 100, the hierarchical weight of "mobile phone" is 0.2, indicating that it occupies 20% of the relative importance. This calculation can be used to guide scenarios such as recommendation and sorting, such as giving priority to keywords with high hierarchical weights. Its formula is essentially a normalization operation, which reflects the proportion of individual (single keyword) characteristics in the whole. In step S13, a keyword preference calculation is performed based on the user behavior data to obtain a keyword preference score.

[0040] Preferably, the keyword preference calculation is performed based on the user behavior data to obtain a keyword preference score, including: The keyword preference score is calculated using the following formula: In the formula, The historical weight of each keyword; The number of clicks for each keyword; The similarity between each keyword and behavior; is the total number of keywords; Score keyword preference; is the number of clicks on the i-th keyword; is the historical weight of the ith keyword; is the similarity between the ith keyword and behavior.

[0041] Specifically, the formula It is the historical weight of each keyword, which is used to measure the importance of the keyword in past data; The number of clicks on each keyword indicates the frequency of users using the keyword; is the similarity between each keyword and behavior, which is used to reflect the matching degree between the keyword and the user behavior. The denominator is the sum of the comprehensive scores of all keywords. This step realizes the normalization operation, so that the preference score of each keyword is between 0 and 1, which is convenient for comparing the priorities of different keywords. By dividing the numerator and the denominator, the relative preference score of the kth keyword is calculated. This score represents the relative importance or priority of the keyword among all keywords.

[0042] In step S14, the information density feature is calculated according to the keyword preference score.

[0043] Preferably, calculating the information density feature according to the keyword preference score includes: The information density feature is calculated by the following formula: In the formula, Score keyword preference; The number of clicks for each keyword; is the total number of keywords; is the information density feature; is the number of clicks on the ith keyword.

[0044] Specifically, It is the information density feature, which reflects the weighted result of the comprehensive preference of keywords and the distribution of click behavior. It is the keyword preference score, indicating the importance of the keyword. The number of clicks on each keyword reflects the frequency of users using the keyword. is the sum of all keyword clicks, which is used for normalization. The calculation process first calculates the proportion of each keyword's clicks to the total number of clicks, that is, , which represents the weight of the click behavior of the kth keyword in the whole. Secondly, multiply the preference score of each keyword by its normalized click count to get the weighted contribution value of the keyword in the whole. Finally, sum the weighted contribution values ​​of all keywords to get the information density feature.

[0045] In step S15, the user demand priority is calculated according to the information density feature, the keyword preference score and the hierarchical weight.

[0046] Preferably, the calculating of user demand priority according to the information density feature, the keyword preference score and the hierarchical weight includes: The user demand priority is calculated by the following formula: In the formula, Score keyword preference; is the information density feature; is the level weight; and is the weight parameter, satisfying ; Prioritize user needs.

[0047] It is worth mentioning that and is the weight parameter, satisfying The determination method can be: data-driven method, based on historical data, determine the weight value through statistical analysis or machine learning method, where statistical analysis refers to analyzing the correlation between user behavior characteristics and hierarchical weights in historical data, calculating their contribution ratio to the recommendation results, and using it as the basis for determination; machine learning refers to building a model (such as linear regression, decision tree, etc.) to optimize the recommendation effect. and This method is more accurate, but it requires certain historical data support. Another method of determination is to use optimization algorithms (such as genetic algorithms, particle swarm optimization, etc.) to automatically find the optimal weight parameters: construct an objective function and use the indicator of the recommendation system (such as click-through rate) as the optimization target. Under the constraint of α+β=1, automatically adjust the weight value to find the optimal solution.

[0048] Specifically, the formula is used to calculate the priority of user needs. The core of the formula is to combine the keyword's preference score, information density characteristics and hierarchical weight to ultimately determine the importance of the keyword to the user's needs. It is the keyword preference score, reflecting the individual importance of the keyword. is the information density feature, which represents the density characteristics of the keyword set in the world, combining the click distribution and preference weight of the keyword. and After multiplication, it represents the contribution value of the kth keyword in the global information density. Then multiply it by the weight parameter , adjust its impact on the priority. is the hierarchical weight, which represents the relative importance of keywords in a specific hierarchy. Multiplied by the weight parameter β, it is used to balance the impact of hierarchical weight on user demand priority.

[0049] In step S16, the content and keyword relevance scores are calculated based on the hierarchical weights, keyword preference scores and user behavior data.

[0050] Preferably, the calculating of the content and keyword relevance score according to the hierarchical weight, keyword preference score and user behavior data includes: The content and keyword relevance score is calculated using the following formula: In the formula, is the level weight; Score keyword preference; The number of clicks for each keyword; is the total number of keywords; and is the correlation weight coefficient, satisfying ; Score the relevance of content to keywords; is the number of clicks on the ith keyword.

[0051] It is worth mentioning that and is the correlation weight coefficient, satisfying , the determination method can be: based on the contribution analysis of historical data, calculate the actual contribution of hierarchical weights, keyword preference scores and clicks to the recommendation effect (such as click-through rate, conversion rate, etc.) from historical data: respectively count the contribution ratio of the three indicators to the target effect; take the contribution ratio of each indicator as the initial value of its corresponding weight; finally ensure that the sum of the three weight coefficients is 1 through normalization. Or use machine learning models (such as random forests, gradient boosting trees, etc.) to evaluate the feature importance of hierarchical weights, keyword preference scores and clicks in the recommendation effect: respectively calculate the feature importance of the three indicators in the model; convert these importance values ​​into weights; adjust and normalize the proportions to obtain and You can also determine the weights through experiments and optimization, use different weight allocation strategies, and compare the recommendation effects through experiments to determine the best weights: set up multiple weight allocation schemes and adjust the importance of each indicator respectively; evaluate the recommendation effect of each scheme (such as user click-through rate or dwell time) through online A / B testing or offline simulation testing; select the weight allocation scheme with the best effect and ensure that its sum is 1.

[0052] Specifically, the first part of the formula is relevance weight 1 × hierarchical weight, where hierarchical weight reflects the relative importance of keywords in the hierarchical structure. Relevance weight 1 adjusts the strength of the influence of hierarchical weight on the relevance score. The second part is relevance weight 2 × keyword preference score, where keyword preference score indicates the degree of match between keywords and user needs. Relevance weight 2 controls the contribution of keyword preference score to the relevance calculation. The third part is relevance weight 3 × (keyword clicks ÷ sum of all keyword clicks), where keyword clicks indicate the frequency of use of the keyword in user behavior, and after normalization (i.e., the number of clicks on a keyword divided by the total number of clicks on all keywords), its share in the total clicks is obtained. Relevance weight 3 adjusts the impact of the share of clicks on the relevance score. It is worth noting that the sum of relevance weight 1, relevance weight 2 and relevance weight 3 is 1, which is used to balance the contribution of these three parts in the relevance score. The weighted values ​​of hierarchical weight, keyword preference score and share of clicks are calculated in turn, and then added together to obtain the relevance score of the content and keywords. This score combines the hierarchical structure characteristics of keywords, user preference and the distribution of behavioral data.

[0053] In step S17, the global priority of the recommended content is calculated according to the content-keyword relevance score and the user demand priority, and the content corresponding to the keyword is recommended to the user according to the global priority of the recommended content.

[0054] Preferably, the calculating the global priority of the recommended content according to the content-keyword relevance score and the user demand priority includes: The global priority of recommended content is calculated using the following formula: In the formula, Prioritize user needs; Score the relevance of content to keywords; The global priority of the recommended content.

[0055] Specifically, the global priority of recommended content is a measurement value calculated by combining the user demand priority and the content-keyword relevance score, which is used to determine the order of recommended content. The user demand priority indicates the importance of a keyword to user demand, and is calculated based on the keyword's preference score, information density characteristics, and hierarchical weight. It reflects the global matching degree between keywords and user needs. The content-keyword relevance score indicates the matching degree between content and keywords, and is calculated by combining the keyword's hierarchical weight, preference score, and user behavior (click-through ratio), reflecting the relevance of content under specific keywords. The user demand priority is multiplied by the content-keyword relevance score to obtain the global priority of recommended content. This step combines the user's demand intensity for keywords and the relevance of content to keywords. The calculation of the global priority of recommended content is essentially a trade-off between user demand and content matching. If the user demand priority is high but the relevance is low, or the relevance is high but the user demand priority is low, the global priority will be reduced. Only when both are high will the content get a higher priority.

[0056] It is worth noting that the method of recommending keyword-corresponding content to users according to the global priority of recommended content can be achieved from the following aspects: In one implementation, a ranking recommendation method can be used. According to the calculated global priority of the recommended content, the content corresponding to all keywords is ranked by priority, and the content with the highest priority is recommended to the user first. This method ensures that the content seen by the user is the most suitable for his or her needs and behaviors. First, all content is sorted in descending order according to the global priority. Then, the content with the highest priority is selected for recommendation according to the set recommendation quantity limit (such as Top-N). The user behavior data is updated in real time during the process, and the ranking results are adjusted dynamically.

[0057] In another implementation, a hierarchical recommendation method can be used. According to the priority of the recommended content, the hierarchy is divided into high priority (Top 10%), medium priority (Top 30%) and low priority (the remaining content), and the content is gradually displayed to the user according to the hierarchy. First, the hierarchy is divided according to the global priority. Then the high priority content is displayed first. If the user does not select it, the medium priority and low priority content are gradually recommended to the user. Finally, the recommendation strategy is adjusted according to user feedback.

[0058] In another implementation, a scene matching recommendation method can be used. Combined with the scene or current behavior of the user, the keywords and content related to the scene are filtered, and matching content is recommended based on priority. For example, when a user browses a certain category of goods, content related to the category of goods is recommended first. First, the user's current scene (such as browsing page, time, and location) is extracted. Secondly, the content matching the scene in the keywords is filtered. Finally, the matching content is sorted according to the global priority and recommended.

[0059] The following describes the working process of the present invention by taking a relatively common scenario as an example. The working process is as follows: The system extracts keyword-related data from the user's behavior records, including the number of clicks, similarity with behavior, historical weight, and total number of keywords. These data serve as the basis for subsequent analysis.

[0060] According to the historical weight of each keyword and the total number of keywords, the hierarchical weight is normalized to reflect the relative importance of the keyword in the whole.

[0061] Combining each keyword's historical weight, number of clicks, and similarity with behavior, a keyword preference score is calculated to reflect the user's preference for the keyword.

[0062] The preference scores and click distributions of all keywords are combined to calculate the information density feature, which reflects the global density of the keyword set and the distribution of user attention.

[0063] Based on the keyword preference score, information density characteristics and hierarchical weight, the user demand priority is calculated to determine the degree of match between the keyword and the user demand.

[0064] The correlation between content and keywords is evaluated through hierarchical weights, keyword preference scores and click distribution, reflecting the adaptability of content to keywords.

[0065] Combining the user demand priority and the relevance score between content and keywords, the global priority of the content is comprehensively determined for ranking and recommendation.

[0066] In summary, the present invention provides a data analysis method, including: obtaining user behavior data; wherein the user behavior data includes the number of clicks on each keyword, the similarity between each keyword and the behavior, the historical weight of each keyword and the total number of keywords; calculating the hierarchical weight according to the user behavior data; performing keyword preference calculation according to the user behavior data to obtain a keyword preference score; calculating the information density feature according to the keyword preference score; calculating the user demand priority according to the information density feature, the keyword preference score and the hierarchical weight; calculating the content and keyword relevance score according to the hierarchical weight, the keyword preference score and the user behavior data; calculating the global priority of recommended content according to the content and keyword relevance score and the user demand priority, and recommending the keyword corresponding content to the user according to the global priority of recommended content.

[0067] The working process of the present invention starts with obtaining user behavior data. The system analyzes the user behavior log to extract the number of keyword clicks, the similarity between keywords and behaviors, the historical weight of keywords, and the total number of keywords. These data provide the basis for subsequent analysis and calculation. After the data is acquired, the system performs a normalization operation based on the historical weight of the keyword and the total number of keywords to calculate the hierarchical weight of each keyword. The hierarchical weight is used to reflect the relative importance of the keyword in the entire keyword set to ensure that the sum of the weights of the keywords is one.

[0068] Next, the system calculates a preference score for each keyword based on the historical weight of the keyword, the number of clicks, and the similarity between the keyword and user behavior. This preference score is normalized so that the importance of each keyword can be quantified and compared. Then, based on the distribution of preference scores and keyword clicks, the system calculates the information density feature. The information density feature measures the overall characteristics of the keyword set, including the importance of each keyword and the comprehensive distribution of click behavior.

[0069] After calculating the information density characteristics, the system further combines the keyword preference score, information density characteristics and hierarchical weights to calculate the user demand priority. The user demand priority is to adjust the weight ratio of each factor to comprehensively determine the degree of influence of each keyword on user demand. On this basis, the system combines the adaptability of content and keywords, and calculates the relevance score of content and keywords through hierarchical weights, preference scores and click behavior distribution. This score reflects the matching degree of a certain content under a specific keyword.

[0070] Finally, the system comprehensively calculates the global priority of the recommended content based on the user demand priority and the content-keyword relevance score. The global priority reflects both the user's demand intensity for a certain keyword and the adaptability of the content to the keyword. The system sorts the content according to the global priority, and gives priority to recommending the content that best meets user needs and has the highest keyword match. In actual applications, the recommended content will be dynamically updated according to the user's real-time behavior to ensure the accuracy and real-time nature of the recommendation. The present invention achieves accurate characterization of user needs and accurate recommendation of content through the deep fusion of multi-dimensional data and a progressive calculation process, thereby improving the accuracy of the analysis results.

[0071] In the present invention, the method is based on user behavior data, comprehensively considers the number of keyword clicks, behavior similarity, historical weight multi-dimensional information, and constructs a hierarchical weight and preference score calculation model. By gradually calculating the keyword preference score, information density characteristics and user demand priority, and combining the content relevance score, the global priority of the recommended content is finally determined. The method uses a layered and progressive analysis method to explore the potential correlation between data, and can also dynamically adjust the analysis weight to adapt to changes in data characteristics, thereby improving the accuracy of the analysis results.

[0072] Reference Figure 2 A second embodiment of the present invention provides a data analysis device, comprising: A data acquisition module is used to acquire user behavior data; wherein the user behavior data includes the number of clicks on each keyword, the similarity between each keyword and the behavior, the historical weight of each keyword and the total number of keywords; A weight calculation module, used to calculate the hierarchical weight according to the user behavior data; A preference calculation module, used to perform keyword preference calculation based on the user behavior data to obtain a keyword preference score; A density calculation module, used to calculate information density features according to the keyword preference scores; A demand calculation module, used to calculate the user demand priority according to the information density feature, the keyword preference score and the level weight; A content relevance calculation module, used to calculate the content and keyword relevance scores based on the hierarchical weights, keyword preference scores and user behavior data; The recommendation calculation module is used to calculate the global priority of the recommended content according to the content and keyword relevance score and the user demand priority, and recommend the content corresponding to the keyword to the user according to the recommended content global priority.

[0073] In an optional implementation, the weight calculation module is specifically used to: The layer weights are calculated using the following formula: In the formula, The historical weight of each keyword; is the total number of keywords; is the level weight; is the historical weight of the ith keyword.

[0074] In an optional implementation, the preference calculation module is specifically used to: The keyword preference score is calculated using the following formula: In the formula, The historical weight of each keyword; The number of clicks for each keyword; The similarity between each keyword and behavior; is the total number of keywords; Score keyword preference; is the number of clicks on the i-th keyword; is the historical weight of the ith keyword; is the similarity between the ith keyword and behavior.

[0075] In an optional implementation, the density calculation module is specifically used for: The information density feature is calculated by the following formula: In the formula, Score keyword preference; The number of clicks for each keyword; is the total number of keywords; is the information density feature; is the number of clicks on the ith keyword.

[0076] In an optional implementation, the demand calculation module is specifically used to: The user demand priority is calculated by the following formula: In the formula, Score keyword preference; is the information density feature; is the level weight; and is the weight parameter, satisfying ; Prioritize user needs.

[0077] In an optional implementation manner, the content relevance calculation module is specifically used to: The content and keyword relevance score is calculated using the following formula: In the formula, is the level weight; Score keyword preference; The number of clicks for each keyword; is the total number of keywords; and is the correlation weight coefficient, satisfying ; Score the relevance of content to keywords; is the number of clicks on the ith keyword.

[0078] In an optional implementation, the recommendation calculation module is specifically used to: The global priority of recommended content is calculated using the following formula: In the formula, Prioritize user needs; Score the relevance of content to keywords; The global priority of the recommended content.

[0079] It should be noted that a data analysis device provided in an embodiment of the present invention is used to execute all process steps of a data analysis method in the above embodiment, and the working principles and beneficial effects of the two correspond one to one, so they will not be described in detail.

[0080] The embodiment of the present invention further provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a data analysis method program. When the processor executes the computer program, the steps in the above-mentioned data analysis method embodiments are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of the modules / units in the above-mentioned device embodiments are realized, such as the recommendation calculation module.

[0081] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0082] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. The electronic device may include more or fewer components than the above components, or may combine certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0083] The processor may be a central processing unit (CPU), 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. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device.

[0084] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the electronic device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0085] Wherein, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0086] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0087] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A data analysis method, characterized in that: include: Obtaining user behavior data; wherein the user behavior data includes the number of clicks on each keyword, the similarity between each keyword and the behavior, the historical weight of each keyword, and the total number of keywords; Calculating hierarchical weights based on the user behavior data; Calculate keyword preference based on the user behavior data to obtain a keyword preference score; Calculating information density features according to the keyword preference scores; Calculating user demand priority according to the information density feature, the keyword preference score and the hierarchical weight; Calculating a content-keyword relevance score based on the hierarchical weights, keyword preference scores, and user behavior data; The global priority of the recommended content is calculated according to the content-keyword relevance score and the user demand priority, and the content corresponding to the keyword is recommended to the user according to the global priority of the recommended content.

2. The data analysis method according to claim 1, characterized in that: The step of calculating the hierarchical weight according to the user behavior data includes: The layer weights are calculated using the following formula: In the formula, The historical weight of each keyword; is the total number of keywords; is the level weight; is the historical weight of the ith keyword.

3. The data analysis method according to claim 1, characterized in that: The step of calculating keyword preference based on the user behavior data to obtain a keyword preference score includes: The keyword preference score is calculated using the following formula: In the formula, The historical weight of each keyword; The number of clicks for each keyword; The similarity between each keyword and behavior; is the total number of keywords; Score keyword preference; is the number of clicks on the i-th keyword; is the historical weight of the ith keyword; is the similarity between the ith keyword and behavior.

4. The data analysis method according to claim 1, characterized in that: The calculating of information density features according to the keyword preference scores includes: The information density feature is calculated by the following formula: In the formula, Score keyword preference; The number of clicks for each keyword; is the total number of keywords; is the information density feature; is the number of clicks on the ith keyword.

5. The data analysis method according to claim 1, characterized in that: The calculating the user demand priority according to the information density feature, the keyword preference score and the hierarchical weight includes: The user demand priority is calculated by the following formula: In the formula, Score keyword preference; is the information density feature; is the level weight; and is the weight parameter, satisfying ; Prioritize user needs.

6. The data analysis method according to claim 1, characterized in that: The step of calculating the content and keyword relevance scores according to the hierarchical weights, keyword preference scores and user behavior data includes: The content and keyword relevance score is calculated using the following formula: In the formula, is the level weight; Score keyword preference; The number of clicks for each keyword; is the total number of keywords; and is the correlation weight coefficient, satisfying ; Score the relevance of content to keywords; is the number of clicks on the ith keyword.

7. The data analysis method according to claim 1, characterized in that: The calculating the global priority of the recommended content according to the content-keyword relevance score and the user demand priority includes: The global priority of recommended content is calculated using the following formula: In the formula, Prioritize user needs; Score the relevance of content to keywords; The global priority of the recommended content.

8. A data analysis device, characterized in that: include: A data acquisition module is used to acquire user behavior data; wherein the user behavior data includes the number of clicks on each keyword, the similarity between each keyword and the behavior, the historical weight of each keyword and the total number of keywords; A weight calculation module, used to calculate the hierarchical weight according to the user behavior data; A preference calculation module, used to perform keyword preference calculation based on the user behavior data to obtain a keyword preference score; A density calculation module, used to calculate information density features according to the keyword preference scores; A demand calculation module, used to calculate the user demand priority according to the information density feature, the keyword preference score and the level weight; A content relevance calculation module, used to calculate the content and keyword relevance scores based on the hierarchical weights, keyword preference scores and user behavior data; The recommendation calculation module is used to calculate the global priority of the recommended content according to the content and keyword relevance score and the user demand priority, and recommend the content corresponding to the keyword to the user according to the global priority of the recommended content.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the data analysis method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the data analysis method according to any one of claims 1 to 7.

Images