Big data-based online course intelligent recommendation method and system

Abstract

The present application relates to big data intelligent recommendation technical field, especially online course intelligent recommendation method and system based on big data, the present application obtains the behavior tendency characteristic of user end before the purchase behavior and the change of each sub-feature of the behavior tendency characteristic, can be based on the change analysis of each sub-feature of user end behavior tendency characteristic to each user end sensitive sub-feature, improve the efficiency of online course intelligent recommendation, through the analysis course sensitive tendency distinguishes sensitive tendency category, customizes each user end online course push frequency. Through the analysis whether adjustment push mode, and determine the online course page that the user end has visited can determine the visited online course category, through each visited online course category can effectively adjust the course push frequency for each online course category, so as to further improve the efficiency of push frequency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data intelligent recommendation, in particular to an online course intelligent recommendation method and system based on big data. BACKGROUND

[0002] With the rapid development of Internet technology, the online education industry has rapidly risen. The flexibility and convenience of online courses have been favored by the majority of teachers and students. Online education platforms have become an important way for many learners to acquire knowledge. Due to the sharp increase in the number of courses on online education platforms, various fields and disciplines are covered, providing learners with a wide range of choices. However, this has led to students facing a vast amount of course information, making it difficult for them to choose the most suitable course for themselves. This has resulted in an increase in user personalized needs. Different learners have different learning styles, including visual, auditory, and problem-solving types. In addition, learners' knowledge levels, learning abilities, and learning progress vary greatly. The traditional "one-size-fits-all" teaching mode and course recommendation method has become difficult to meet the diverse learning needs of students. Personalized recommendation systems have become increasingly important. The problem of how to tailor learning resources for each learner has become increasingly urgent.

[0003] Chinese Patent Publication No. CN115423569B discloses a recommendation method based on big data and a recommendation system based on big data. For the obtained to-be-recommended e-commerce business data; by determining not less than two kinds of e-commerce business data knowledge vectors of the to-be-recommended e-commerce business data, it is convenient to explore the target e-commerce business data several times in the pre-configured e-commerce business data set through not less than two kinds of e-commerce business data knowledge vectors; thus, based on not less than two kinds of e-commerce business data knowledge vectors, several times of selection in the pre-configured e-commerce business data set can select target e-commerce business data more associated with the to-be-recommended e-commerce business data; therefore, through not less than two kinds of knowledge vectors, the specified e-commerce business data set can be effectively selected, which can improve the efficiency and accuracy of selecting target e-commerce business data; thereby, data pushing can be more accurately performed, and the accuracy and reliability of data pushing can be improved.

[0004] However, the existing technology still has the following problems:

[0005] The existing technology does not consider the difference in user behavior tendency characteristics, which leads to different user sensitive tendency categories. Due to the balanced pushing of big data to each user's online course, the frequency of online course placement is not suitable, which leads to the problem of low efficiency of online course intelligent recommendation. SUMMARY

[0006] To this end, the application provides an online course intelligent recommendation method and system based on big data, to overcome the problem of low online course intelligent recommendation efficiency caused by the fact that the user end behavior tendency category is different due to the fact that the user end behavior tendency feature difference is not considered in the prior art, and the fact that the user end behavior tendency course delivery frequency is not suitable due to the fact that big data balances the online course delivery of each user end.

[0007] To achieve the above-mentioned purpose, the application provides an online course intelligent recommendation method based on big data, comprising:

[0008] Obtaining the behavior tendency feature of the user end before the online course purchase behavior and the change of the sub-feature in the behavior tendency feature;

[0009] Analyzing the sensitive sub-feature of each user end based on the change of the sub-feature in the user end behavior tendency feature, constructing a sensitive sub-feature set, and analyzing the sensitive tendency representation value based on the corresponding sensitive sub-feature set of the user end;

[0010] Determining the sensitive tendency category of the user end based on the sensitive tendency representation value, analyzing the real-time data, and pushing the course, comprising,

[0011] Performing delivery verification and pushing, obtaining the response of each sub-feature after the delivery verification and pushing, analyzing whether to adjust the pushing mode, and pushing the course in the corresponding pushing mode;

[0012] Or, collecting the page closing rate of the user end, analyzing whether the user end has an aversion behavior based on the page closing rate, and reducing the pushing frequency and then verifying again;

[0013] The behavior tendency feature includes the course trial viewing time of the user end, the trial viewing completion rate and the click rate; the delivery verification and pushing includes pushing the course at a predetermined pushing frequency.

[0014] Further, the process of analyzing the sensitive sub-feature of each user end based on the change of the sub-feature in the user end behavior tendency feature comprises,

[0015] Calculating the difference ratio mean of the data of the sub-feature in the user end behavior tendency feature and the preset threshold of the corresponding sub-feature, and determining the difference ratio mean as a feature factor;

[0016] Comparing the feature factors of each sub-feature;

[0017] Determining the two sub-features with larger values in the compared feature factors as the sensitive sub-features.

[0018] Further, the process of analyzing the sensitive tendency representation value based on the sensitive sub-feature set comprises,

[0019] Calculate the ratio of the first sensitive sub-feature of the user end to the corresponding first sensitive sub-feature threshold value to obtain the first recommendation influencing factor;

[0020] Calculate the ratio of the second sensitive sub-feature of the user end to the corresponding second sensitive sub-feature threshold value to obtain the second recommendation influencing factor;

[0021] Determine the sum of the first recommendation influencing factor and the second recommendation influencing factor as the sensitive tendency representation value.

[0022] Further, determine the sensitive tendency category of the user end based on the sensitive tendency representation value, comprising,

[0023] If the sensitive tendency representation value is greater than the behavior tendency feature threshold value, determine the first sensitive tendency category;

[0024] If the sensitive tendency representation value is less than or equal to the behavior tendency feature threshold value, determine the second sensitive tendency category.

[0025] Further, analyze the real-time data to push the course, comprising,

[0026] If the division result is the first sensitive tendency category, perform a delivery verification push, obtain the response of each sub-feature after the delivery verification push, analyze whether to adjust the push mode, and push the course in the corresponding push mode;

[0027] If the division result is the second sensitive tendency category, collect the page closing rate of the user end, analyze whether the user end has an aversion behavior based on the page closing rate, and reduce the push frequency and then re-verify.

[0028] Further, the process of obtaining the response of each sub-feature after the delivery verification push and analyzing whether to adjust the push mode comprises,

[0029] Extract the reduction amount of each sensitive sub-feature in the time period;

[0030] If the reduction amount of any sensitive sub-feature is greater than or equal to a predetermined change threshold value, determine to adjust the push mode to push the course in the corresponding push mode.

[0031] Further, adjusting the push mode to push the course in the corresponding push mode comprises,

[0032] Determine the keywords in each of the online course pages;

[0033] Determine the online course category to which each of the keywords belongs;

[0034] Wherein, the attribution relationship between each keyword and the online course category is pre-set;

[0035] determining remaining unvisited online course categories based on the visited online course categories;

[0036] determining to adjust the online course push frequency for each of the unvisited online course categories;

[0037] determining to increase the online course push frequency.

[0038] Further, based on the behavior characteristic analysis, whether the user end appears the aversion behavior is determined, including,

[0039] extracting the page closing rate of the user end before the online course purchase behavior;

[0040] if the page closing rate is greater than or equal to a predetermined page closing threshold, it is determined that the user end appears the aversion behavior, and the push frequency is reduced and then verified again.

[0041] Further, the process of reducing the push frequency includes,

[0042] solving the mean value of the reduction amount of each sensitive sub-feature;

[0043] determining that the reduction amount of the online course push frequency is positively correlated with the mean value of the reduction amount.

[0044] Further, the present application provides an online course intelligent recommendation system based on big data, including,

[0045] a data collector configured to acquire the behavior tendency characteristic of the user end before the online course purchase behavior and the change amount of each sub-feature of the behavior tendency characteristic;

[0046] a feature analyzer connected with the data collector, configured to analyze the sensitive sub-feature of each user end based on the change of the sub-feature in the user end behavior tendency characteristic, and analyze the sensitive tendency representation value based on the sensitive sub-feature set;

[0047] a course recommendation component connected with the data collector and the feature analyzer respectively, including a clustering unit and a control unit, the clustering unit is configured to distinguish the sensitive tendency category according to the sensitive tendency representation value, including,

[0048] the control unit is configured to determine the sensitive tendency category of the user end based on the sensitive tendency representation value, analyze the real-time data, and push the course, including,

[0049] performing the verification push, acquiring the response of each sub-feature after the verification push, analyzing whether to adjust the push mode, and pushing the course according to the push mode;

[0050] Or, collect the page closing rate of the user end, analyze whether the user end appears aversion behavior based on the behavior characteristic, re-verify after reducing the push frequency;

[0051] The behavior tendency characteristic includes a course preview time length of the user end, a preview completion rate and a click rate, and the push verification push includes pushing the course at a predetermined push frequency.

[0052] Compared with the prior art, the application can analyze sensitive sub-features for each user end based on the change of each sub-feature of the user end behavior tendency characteristic, improve the efficiency of online course intelligent recommendation, distinguish sensitive tendency categories by analyzing course sensitive tendency, customize online course push frequency for each user end, reduce the occurrence of user end aversion to online push by analyzing the page closing rate, determine the online course pages visited by the user end, determine the online course categories visited based on each online course page, and effectively adjust the course push frequency for each online course category based on each online course category visited, thereby further improving the efficiency of online course push frequency.

[0053] Especially, the application provides an online course intelligent recommendation system based on big data, which includes a data collector, a feature analyzer and a course recommendation component. The application can obtain the behavior tendency characteristic before the user end purchase behavior and the change amount of each sub-feature of the user end behavior tendency characteristic through the data collector, improve the efficiency of course push frequency through the behavior tendency characteristic and the sensitive tendency representation value of the user end, distinguish the first sensitive tendency category and the second sensitive tendency category through the course recommendation component, accurately analyze whether to adjust the push mode through the change amplitude of each sensitive sub-feature of the user end, and therefore the application considers the online course pages visited by the user end, considers the online course categories with potential tendency that the user end has not browsed, mines the potential tendency of the user end, determines the online course pages visited by the user end, determines the online course categories visited based on each online course page, effectively adjusts the course push frequency for each online course category based on each online course category visited, further improves the efficiency of course push frequency, and improves the accuracy of online course recommendation.

[0054] Especially, for the second sensitive tendency category of course sensitive tendency, the page closing rate of the user end is collected, the user end can be analyzed whether to appear aversion behavior based on the page closing rate, re-verification is performed after reducing the push frequency, the occurrence of user end aversion anomaly is reduced, and the efficiency of online course push is improved. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 A step flow chart of the online course intelligent recommendation method based on big data of the embodiment of the present application;

[0056] Figure 2 A step flow chart of the sensitive sub-feature of each user terminal based on the change analysis of each sub-feature of the user terminal behavior tendency feature of the embodiment of the present application;

[0057] Figure 3 A step flow chart of the sensitive tendency representation value based on the sensitive sub-feature set analysis of the embodiment of the present application;

[0058] Figure 4 A structure block diagram of the online course intelligent recommendation system based on big data of the embodiment of the present application. DETAILED DESCRIPTION

[0059] In order to make the objects and advantages of the present application clearer, the present application will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0060] The preferred embodiments of the present application will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application and are not intended to limit the protection scope of the present application.

[0061] It should be noted that in the description of the present application, unless otherwise explicitly specified and limited, the term "connection" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, or the internal communication of two elements. Those skilled in the art can understand the specific meaning of the above-mentioned term in the present application according to the specific circumstances.

[0062] Please refer to Figure 1 As shown in the figure, it is a step flow chart of the online course intelligent recommendation method based on big data of the embodiment of the present application, the present application provides an online course intelligent recommendation method based on big data, comprising:

[0063] Step S1, obtaining the behavior tendency features of the user terminal before the online course purchase behavior and the change of the sub-features in the behavior tendency features;

[0064] Step S2, analyzing the sensitive sub-features of each user terminal based on the change of the sub-features in the user terminal behavior tendency features, constructing a sensitive sub-feature set, and analyzing the sensitive tendency representation value based on the sensitive sub-feature set corresponding to the user terminal;

[0065] Step S3, determining the sensitive tendency category of the user terminal based on the sensitive tendency characteristic value, analyzing the real-time data, and pushing the course, including,

[0066] performing a delivery verification push, obtaining the response of each sub-feature after the delivery verification push, and analyzing whether to adjust the push mode to push the course according to the push mode;

[0067] Or, collect the page closing rate of the user terminal, analyze whether the user terminal has an aversion behavior based on the page closing rate, and reduce the push frequency and then verify again;

[0068] The behavior tendency feature includes the course trial viewing time, trial viewing completion rate and click rate of the user terminal; the delivery verification push includes pushing the course at a predetermined push frequency.

[0069] Specifically, the online course is a pre-recorded high school entrance examination tutoring video course, which can also be other forms of pre-recorded video courses, which will not be repeated here.

[0070] Specifically, the division of the online course category is not limited, and in the implementation, the online course category is mathematics, physics, chemistry, biology, Chinese, English, and other forms can also be used, which will not be repeated here.

[0071] Specifically, the sub-features of the user terminal behavior tendency feature have three, which are all data obtained within a predetermined time before the purchase behavior, and the predetermined time can be selected within [5min, 25min].

[0072] Specifically, each sub-feature is a single course trial viewing time, a single trial viewing completion rate, and a single course corresponding click rate.

[0073] Specifically, the way to obtain the user terminal behavior tendency feature is not limited, and related tracking software can be used to obtain, for example, one of ClickMagick, Trackier, of course, other forms can also be used as long as the behavior tendency feature can be obtained.

[0074] Specifically, the application can improve the efficiency of online course intelligent recommendation by obtaining the behavior tendency characteristics of the user end before the purchase of online courses and the change amount of each sub-feature of the behavior tendency characteristics, and analyzing the sensitive sub-features of each user end based on the change of each sub-feature of the user end behavior tendency characteristics. The sensitive tendency categories are distinguished by analyzing the sensitive tendency of the course, and the online course push frequency of each user end is customized. By analyzing whether to adjust the push mode, determining the accessed online course categories based on each online course page, and effectively adjusting the course push frequency for each online course category based on each accessed online course category, the push frequency is reduced by the page closing rate of the user end, reducing the occurrence of user end aversion to online push, thereby improving the efficiency of online course push frequency.

[0075] Please refer to Figure 2 As shown in the figure, it is a step flow chart for analyzing the sensitive sub-features of each user end based on the change of each sub-feature of the user end behavior tendency characteristics. The process of analyzing the sensitive sub-features of each user end based on the change of each sub-feature of the user end behavior tendency characteristics includes,

[0076] Step S11, calculate the difference ratio mean of the difference between the data of each sub-feature of the user end behavior tendency characteristics and the corresponding preset threshold value of the sub-feature, and determine the difference ratio mean as a feature factor;

[0077] Step S12, compare the feature factors of each sub-feature;

[0078] Step S13, determine the two sub-features with larger values in the compared feature factors as sensitive sub-features.

[0079] In implementation, the user end has several corresponding behavior tendency characteristics, which are obtained based on several purchase behavior records of the user end, which will not be repeated here.

[0080] In implementation, in the behavior tendency characteristics, the preset threshold value corresponding to the try-to-complete rate sub-feature is determined in advance. The click frequency mean of several user ends is calculated in a historical period, the preset threshold value is set as the product of the try-to-complete rate mean and the precision coefficient, the precision coefficient is selected in the interval [1.1, 1.2], and the historical period is the same as the predetermined time length.

[0081] Similarly, the preset threshold value of each sub-feature is derived from the product of the historical period of big data corresponding sub-feature behavior tendency characteristics and the precision coefficient, which will not be repeated here.

[0082] The difference ratio is the ratio of the difference between two values to the corresponding mean of the two values, which will not be repeated here.

[0083] In implementation, before the user end occurs a purchase behavior, the feature factors corresponding to the sensitive sub-features of each sub-feature are respectively:

[0084] The feature factor corresponding to the single course trial viewing time length sub-feature is 0.62;

[0085] The feature factor corresponding to the single trial viewing completion rate sub-feature is 0.58;

[0086] The feature factor corresponding to the single course corresponding click rate sub-feature is 0.60;

[0087] Since 0.62>0.60>0.58, the user end sensitive sub-feature of the embodiment is the course trial viewing time length sub-feature and the course corresponding click rate sub-feature, which indicates that the user end has strong data representation in the course trial viewing time length and the course corresponding click rate sub-feature before the purchase behavior, and the numerical change is relatively obvious, so the online course intelligent recommendation system will adaptively analyze the sensitive sub-feature held by the user end, and then the first sensitive sub-feature is the course trial viewing time length, and the second sensitive sub-feature is the click rate. The customized recommendation of the application improves the efficiency of online course recommendation.

[0088] Please refer to Figure 3 As shown in the figure, it is a step flow chart for analyzing sensitive tendency representation value based on a sensitive sub-feature set, and the process of analyzing the sensitive tendency representation value based on the sensitive sub-feature set comprises,

[0089] Step S21, calculating the ratio of the first sensitive sub-feature of the user end to the corresponding first sensitive sub-feature threshold value to obtain a first recommendation influence factor;

[0090] Step S22, calculating the ratio of the second sensitive sub-feature of the user end to the corresponding second sensitive sub-feature threshold value to obtain a second recommendation influence factor;

[0091] Step S23, determining the sum of the first recommendation influence factor and the second recommendation influence factor as the sensitive tendency representation value.

[0092] Specifically, the first sensitive sub-feature threshold value is 1.15 times the average value of the first sensitive sub-feature value of the user end in the system in the historical period; similarly, the second sensitive sub-feature threshold value is 1.25 times the average value of the second sensitive sub-feature of the user in the three months before the system recommendation behavior in the historical period

[0093] Specifically, the sensitive sub-feature threshold value corresponding to different sub-features is determined in advance,

[0094] For the single course trial viewing time length sub-feature, the average value of the single course trial viewing time length of the user end in each historical period is determined;

[0095] For the single trial viewing completion rate sub-feature, the average value of the single trial viewing completion rate of the user end in each historical period is determined;

[0096] For the click rate sub-feature corresponding to a single course, the average value of the click rate of the user end corresponding to the single course in each historical period is determined.

[0097] Specifically, determining the sensitive tendency category of the user end based on the sensitive tendency representation value includes,

[0098] If the sensitive tendency representation value is greater than the behavior tendency feature threshold value, it is determined as the first sensitive tendency category.

[0099] If the sensitive tendency representation value is less than or equal to the behavior tendency feature threshold value, it is determined as the second sensitive tendency category.

[0100] Specifically, since the data reference value of the sensitive sub-feature is high, the actual behavior tendency feature is determined by the form of ratio of the corresponding sensitive sub-feature threshold value, and the change of the actual behavior tendency feature relative to the normal situation is determined, and then the user end with obvious behavior sensitive tendency is identified.

[0101] Specifically, the real-time data is analyzed, and the course is pushed, including,

[0102] If the division result is the first sensitive tendency category, the verification push is performed, the response of each sub-feature after the verification push is obtained, and whether the push mode is adjusted is analyzed to push the course corresponding to the push mode;

[0103] If the division result is the second sensitive tendency category, the page closing rate of the user end is collected, and whether the user end appears aversion behavior is analyzed based on the page closing rate, so as to reduce the push frequency and then re-verify.

[0104] In implementation, in response to the division result of the sensitive tendency category, the online course category required by the user end for online course recommendation is determined, the online course page corresponding to the behavior tendency feature of the user end is determined, the online course category to which each keyword belongs is determined by the keyword of the online course page, the course corresponding to the online course category is recommended, and then whether the push mode is adjusted is analyzed according to the change range of each sensitive sub-feature of the user end after the online course recommendation, so as to reduce the online course push frequency.

[0105] Specifically, the process of obtaining the response of each sub-feature after the verification push and analyzing whether the push mode is adjusted includes,

[0106] The reduction amount of each sensitive sub-feature in the time period is extracted;

[0107] If the reduction amount of any sensitive sub-feature is greater than or equal to the predetermined change threshold value, it is determined that the push mode is adjusted to push the course corresponding to the push mode.

[0108] In implementation, the change threshold of the sensitive sub-feature is determined based on the sensitive sub-feature threshold of the corresponding sub-feature, and is set to between 1.1 times and 1.2 times of the sensitive sub-feature threshold.

[0109] Specifically, the pushing manner is adjusted to push the courses in the corresponding pushing manner, including,

[0110] determining the keywords in each of the online course pages;

[0111] determining the online course categories to which the keywords belong;

[0112] wherein the attribution relationship between each keyword and the online course category is pre-set;

[0113] determining the remaining unvisited online course categories based on the visited online course categories;

[0114] determining the online course pushing frequency for each of the unvisited online course categories;

[0115] determining to increase the online course pushing frequency.

[0116] Specifically, the online course pushing frequency is increased to between 1.5 times and 1.8 times of the initial online course pushing frequency.

[0117] Specifically, for the second sensitive tendency category of the course sensitive tendency, the browsing behavior data is not representative, and the behavior sensitive tendency or tendency is not obvious, so the present application considers the online course pages visited by the user end, considers the online course categories with potential tendency that the user end has not browsed, mines the potential tendency of the user end, and improves the online course recommendation accuracy.

[0118] Specifically, the behavior feature is analyzed to determine whether the user end has aversion behavior, including,

[0119] extracting the page closing rate of the user end before the online course purchase behavior;

[0120] if the page closing rate is greater than or equal to a predetermined page closing threshold, it is determined that the user end has aversion behavior, and the pushing frequency is reduced and then verified again.

[0121] Specifically, the process of reducing the pushing frequency includes,

[0122] solving the mean value of the reduction amount of each sensitive sub-feature;

[0123] determining that the reduction amount of the online course pushing frequency is positively correlated with the mean value of the reduction amount.

[0124] In implementation, the ratio of the reduction amount mean value and the Euclidean change threshold value is solved, and the obtained ratio value is multiplied by the initial online course pushing frequency to obtain the online course pushing frequency that needs to be reduced.

[0125] Referring to Figure 4 As shown in the structure block diagram of the online course intelligent recommendation system based on big data according to the embodiment of the present application, the present application further provides an online course intelligent recommendation system based on big data, which comprises,

[0126] The data collector is used to acquire the behavior tendency characteristics before the user end purchases the online course and the change amount of each sub-characteristic of the behavior tendency characteristics.

[0127] The feature analyzer is connected with the data collector and is used to analyze the sensitive sub-characteristics of each user end based on the change of the sub-characteristics in the user end behavior tendency characteristics, and is used to analyze the sensitive tendency representation value based on the sensitive sub-characteristics set.

[0128] The course recommendation component is connected with the data collector and the feature analyzer respectively and comprises a clustering unit and a control unit.

[0129] The control unit determines the sensitive tendency category of the user end based on the sensitive tendency representation value, analyzes the real-time data, and pushes the course, which comprises,

[0130] The verification pushing is performed, the response of each sub-characteristic after the verification pushing is acquired, and whether the pushing mode is adjusted is analyzed to push the course according to the pushing mode.

[0131] Or, the page closing rate of the user end is collected, whether the user end appears the aversion behavior is analyzed based on the behavior characteristics, and the pushing frequency is reduced to re-verify.

[0132] The behavior tendency characteristics comprise the course trial watching time length, the trial watching completion rate and the click rate of the user end, and the verification pushing comprises pushing the course at a predetermined pushing frequency.

[0133] Specifically, the specific structure of the data collector, the feature analyzer and the course recommendation component is not limited, and each unit thereof can be composed of a logic component, which comprises a field programmable component, a computer or a microprocessor in the computer.

[0134] The application provides an online course intelligent recommendation system based on big data, which comprises a data collector, a feature analyzer and a course recommendation component. The application can obtain the behavior tendency characteristics of the user end before the purchase behavior, the change amount of each sub-feature of the user end behavior tendency characteristics and the course subscription characteristics through the data collector. The efficiency of the course pushing frequency is improved through the behavior tendency characteristics and the user end sensitive tendency characteristic value. The first sensitive tendency category and the second sensitive tendency category can be distinguished through the course recommendation component. Whether the pushing mode is adjusted can be accurately analyzed through the change amplitude of each sensitive sub-feature of the user end. The online course categories that have been accessed are determined based on each online course page. The course pushing frequency for each online course category can be effectively adjusted based on each accessed online course category. The efficiency of the course pushing frequency is improved. Meanwhile, the user end aversion abnormality is reduced through the page closing rate analysis of the user end.

[0135] Up to now, the technical scheme of the present application has been described in combination with the preferred embodiments shown in the drawings, but it is easy for those skilled in the art to understand that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to the related technical features without departing from the principles of the present application, and the technical schemes after these changes or replacements will all fall within the protection scope of the present application.

Claims

1. A method for intelligent recommendation of online courses based on big data, characterized in that: include: Acquire user behavior characteristics before purchasing online courses, as well as changes in sub-characteristics among several behavioral characteristics; Based on the changes in sub-features of user behavior tendency characteristics, we analyze the sensitive sub-features of each user terminal, construct a set of sensitive sub-features, and analyze the sensitive tendency representation value based on the set of sensitive sub-features corresponding to the user terminal. Based on the sensitivity tendency characterization values, the user's sensitivity tendency category is determined, real-time data is analyzed, and courses are pushed accordingly, including... If the sensitivity tendency characteristic value is greater than the behavioral tendency characteristic threshold, it is determined to be the first sensitivity tendency category; If the sensitivity tendency characterization value is less than or equal to the behavioral tendency characteristic threshold, it is determined to be the second sensitivity tendency category; If the classification result is the first sensitive tendency category, then the delivery verification push will be carried out, the response of each sub-feature after the delivery verification push will be obtained, and the push method will be analyzed to determine whether to adjust the push method and push the course in the corresponding push method. If the classification result is the second sensitive tendency category, the page closure rate of the user terminal is collected, and the user terminal is analyzed based on the page closure rate to determine whether there is aversion behavior. The push frequency is reduced and the verification is performed again. The behavioral characteristics include the user's course trial viewing duration, trial viewing completion rate, and click-through rate; the delivery verification push includes pushing courses at a predetermined push frequency.

2. The intelligent online course recommendation method based on big data according to claim 1, characterized in that, The process of analyzing the changes in sub-features of user-end behavioral tendencies includes the following: Calculate the average difference ratio between the data of the sub-features in the user behavior tendency features and the preset threshold of the corresponding sub-features, and determine the average difference ratio as the feature factor; Compare the feature factors of each sub-feature; The two sub-features with the larger values ​​among the compared feature factors are identified as sensitive sub-features.

3. The intelligent online course recommendation method based on big data according to claim 1, characterized in that, The process of analyzing sensitivity tendency characterization values ​​based on the aforementioned set of sensitive sub-features includes: Calculate the ratio of the first sensitive sub-feature on the user side to the corresponding threshold of the first sensitive sub-feature to obtain the first recommendation influencing factor; Calculate the ratio of the second sensitive sub-feature on the user side to the corresponding threshold of the second sensitive sub-feature to obtain the second recommendation influencing factor; The sum of the first and second recommended influencing factors is determined as the sensitivity tendency characterization value.

4. The intelligent online course recommendation method based on big data according to claim 1, characterized in that, The process of obtaining the response status of each sub-feature after the delivery verification push, and analyzing whether to adjust the push method, includes: The amount of reduction in each sensitive sub-feature within the extraction time period; If the reduction in any sensitive sub-feature is greater than or equal to a predetermined change threshold, then the push method is adjusted, and the course is pushed in the corresponding push method.

5. The intelligent online course recommendation method based on big data according to claim 1, characterized in that, Adjust the push notification method to deliver courses accordingly. include, Identify the keywords in each of the online course pages; Determine the online course category to which each of the aforementioned keywords belongs; The system pre-sets the relationship between each keyword and the online course category; Based on the categories of online courses that have been visited, determine the remaining categories of online courses that have not been visited; It has been determined that the frequency of online course push notifications for each of the aforementioned unaccessed online course categories needs to be adjusted; It was decided to increase the frequency of the online course push notifications.

6. The intelligent online course recommendation method based on big data according to claim 1, characterized in that, Based on behavioral characteristics analysis, we can determine whether users exhibit aversion to certain behaviors, including: Extract the page close rate before users make an online course purchase; If the page closure rate is greater than or equal to the predetermined page closure threshold, it is determined that the user has engaged in abusive behavior, and the push frequency is reduced before re-verification is performed.

7. The intelligent online course recommendation method based on big data according to claim 1, characterized in that, The process of reducing push frequency includes, Solve for the mean decrease in the feature of each sensitive sub-sub ... The decrease in the frequency of online course pushes is positively correlated with the mean of the decrease.

8. A system using the big data-based intelligent recommendation method for online courses according to any one of claims 1 to 7, characterized in that, include, A data collector is used to acquire behavioral tendencies of users prior to purchasing online courses, as well as the changes in each sub-feature of those behavioral tendencies. A feature analyzer, connected to the data collector, is used to analyze the sensitive sub-features of each user based on the changes in the sub-features in the user behavior tendency features, and to analyze the sensitive tendency representation value based on the set of sensitive sub-features. A course recommendation component, which is connected to the data collector and the feature analyzer respectively, includes a clustering unit and a control unit. The clustering unit is used to distinguish sensitive tendency categories based on the sensitive tendency characterization value. The control unit determines the user's sensitivity category based on the sensitivity tendency characterization value, analyzes real-time data, and pushes courses accordingly. include, Conduct delivery verification push, obtain the response status of each sub-feature after delivery verification push, analyze whether to adjust the push method, and push the course in the corresponding push method; Alternatively, collect the page closure rate on the user's end, analyze whether the user's end exhibits aversion behavior based on behavioral characteristics, and then re-verify after reducing the push frequency; The behavioral characteristics include the user's course trial viewing duration, trial viewing completion rate, and click-through rate; the delivery verification push includes pushing courses at a predetermined push frequency.

Images