Advertisement intelligent putting method and system based on automatic annotation information feedback

Abstract

The invention discloses an intelligent advertisement putting method and system based on automatic annotation information feedback, and relates to the technical field of advertisement putting, and the method comprises the steps: obtaining user behavior data, obtaining user behavior keyword information based on automatic annotation according to the user behavior data, and carrying out the intelligent advertisement putting according to the user behavior keyword information based on the user behavior data. Acquiring a behavior keyword influence coefficient corresponding to each user behavior keyword; the interest, behavior habits and consumption preferences of the user are deeply mined through the behavior keyword influence coefficient, the preference condition of the user target is accurately analyzed, different advertisement putting times are allocated through the advertisement putting time node information, the maximum display of the advertisement effect is ensured, and the user experience is improved. Interference of invalid advertisements to users is reduced, the advertisement effect is monitored through the advertisement monitoring information, it is ensured that abnormal advertisement effect is found in time, and the stability of the advertisement effect is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of advertisement delivery, and in particular to an advertisement intelligent delivery method and system based on automatic annotation information feedback. Background Art

[0002] With the rapid development of Internet technology, the advertising industry has rapidly shifted from the traditional offline model to the online model. Online advertising has the advantages of wide coverage, relatively low cost, and high accuracy, and has gradually become the main choice of advertisers. According to statistics, the scale of the global online advertising market has continued to grow at a double-digit annual growth rate in recent years, and is expected to maintain a high growth trend in the next few years. At the same time, the forms of advertising are becoming increasingly diversified, including search engine advertising, social media advertising, video advertising, information flow advertising, etc. Different forms of advertising have their own characteristics in terms of delivery strategy and effect evaluation, which also brings more complexity and challenges to advertising.

[0003] At present, there are still problems in advertising delivery, such as the inability to accurately analyze user behavior and the inability to accurately evaluate user preferences for different content based on user browsing content. Advertisements are often delivered directly based on keywords of user historical content visits. This method has low delivery accuracy and poor advertising effects. However, if advertisements are delivered based on specific analysis of browsing content, it will result in low advertising delivery efficiency, increased data volume, and increased resource utilization. For delivered advertisements, it is impossible to accurately predict the delivery differences between different advertisements, to accurately select appropriate time nodes for delivery, and to adaptively adjust the monitoring cycle based on the delivered advertisements. Summary of the invention

[0004] In order to solve the above technical problems, a method and system for intelligent advertising delivery based on automatic annotation information feedback are provided. This technical solution solves the problems raised in the above background technology that it is impossible to accurately analyze user behavior and accurately evaluate the user's preference for different content based on the user's browsing content. Advertisements are often delivered directly based on keywords of the user's historical access content. This method has low delivery accuracy and poor advertising effect. However, if advertisements are delivered based on specific analysis of browsing content, it will result in low advertising delivery efficiency, increased data volume, and increased resource occupancy. For the delivered advertisements, it is impossible to accurately predict the delivery differences between different advertisements, it is impossible to accurately select the appropriate time node for delivery, and it is impossible to adaptively adjust the monitoring cycle according to the delivered advertisements.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] An intelligent advertising delivery method based on automatic labeling information feedback, comprising:

[0007] Acquire user behavior data, wherein the user behavior data includes user click information and user browsing information;

[0008] According to the user behavior data, based on automatic labeling, user behavior keyword information is obtained, wherein the user behavior keyword information includes a first behavior keyword and a second behavior keyword;

[0009] According to the user behavior keyword information and based on the user behavior data, obtain the behavior keyword influence coefficient corresponding to each user behavior keyword;

[0010] According to the user behavior data, corresponding historical advertising delivery data is obtained, wherein the historical advertising delivery data includes historical advertising delivery time node information, historical advertising click rate information, and historical advertising conversion rate information;

[0011] According to historical advertising data, obtain advertising delivery time node information;

[0012] Acquire information of advertisements to be placed, wherein the information of advertisements to be placed includes keyword information of advertisements to be placed;

[0013] According to the information of the advertisement to be placed and the influence coefficient of the behavioral keywords, the advertisement recommendation index corresponding to each advertisement to be placed is obtained;

[0014] Obtain advertising delivery information based on the advertising recommendation index and advertising delivery time node information;

[0015] Place advertisements according to the advertisement placement information;

[0016] Obtain advertising monitoring information based on historical advertising data and advertising information;

[0017] The delivered advertisements are monitored according to the advertisement monitoring information.

[0018] Preferably, obtaining the behavior keyword influence coefficient corresponding to each user behavior keyword based on the user behavior data according to the user behavior keyword information specifically includes:

[0019] Based on user behavior data, obtain user browsing topic information and user browsing content information;

[0020] Based on automatic annotation, obtain the keywords of user browsing topics and user browsing content;

[0021] The user browsing topic keyword is used as the first behavior keyword, and the user browsing content keyword is used as the second behavior keyword to obtain the user behavior keyword information;

[0022] According to the user behavior keyword information, the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keywords of the behavior is used as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keywords of the behavior is used as the second keyword benchmark influence coefficient;

[0023] According to user behavior data, obtain the user browsing time information corresponding to each user's browsing content;

[0024] Based on the user's browsing content information, predict the user's browsing time at a normal information browsing speed to obtain the browsing prediction time;

[0025] According to the user browsing time information and the browsing prediction time, the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient are adjusted to obtain the behavior keyword influence coefficient.

[0026] Preferably, the adjusting the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient according to the user browsing time information and the browsing prediction time to obtain the behavior keyword influence coefficient specifically includes:

[0027] The ratio of user browsing time information to browsing prediction time is used as the user preference time performance coefficient;

[0028] Obtain the influence coefficient of the first keyword of the behavior according to the first keyword benchmark influence coefficient and the user preference time performance coefficient;

[0029] Based on the analysis of user browsing time, obtain the user preference time performance coefficient threshold;

[0030] Obtaining user characteristic browsing content according to the user preference time expression coefficient and the user preference time expression coefficient threshold;

[0031] If the user preference time performance coefficient w of the user browsing content is greater than or equal to 0.7, the user browsing content is regarded as the user characteristic browsing content;

[0032] The first keyword of the user's characteristic browsing behavior is used as the calibration keyword;

[0033] Taking the calibration keyword as a benchmark, obtaining the behavioral second keywords that appear simultaneously with the calibration keyword in all the user browsing contents, and constructing a behavioral second keyword set corresponding to the calibration keyword;

[0034] Obtaining a behavior keyword correction coefficient according to the calibration keyword and the behavior second keyword set;

[0035] Obtaining the behavioral second keyword influence coefficient according to the second keyword baseline influence coefficient and the behavioral keyword correction coefficient;

[0036] Obtain the behavior keyword influence coefficient according to the behavior first keyword influence coefficient and the behavior second keyword influence coefficient;

[0037] The influence coefficient of the first keyword of the behavior is specifically:

[0038]

[0039] Where Q(x) is the influence coefficient of the first keyword of the x-th behavior, w(i) represents the user preference time performance coefficient of the i-th user browsing content containing the x-th behavior first keyword, R(x) is the first keyword benchmark influence coefficient of the x-th behavior first keyword, and n is the total number of user browsing content containing the x-th behavior first keyword;

[0040] The influence coefficient of the second keyword of the behavior is specifically:

[0041]

[0042] Where Q(y) is the influence coefficient of the second keyword of the y-th behavior, R(y) is the second keyword benchmark influence coefficient of the second keyword of the y-th behavior, k y (z) is the behavior keyword correction coefficient of the yth behavior second keyword and the zth calibration keyword, w(z) represents the user preference time performance coefficient of the user browsing content corresponding to the zth calibration keyword, d(y, z) represents the number of occurrences of the yth behavior second keyword in the behavior second keyword set corresponding to the zth calibration keyword, d(z) represents the total number of behavior second keywords in the behavior second keyword set corresponding to the zth calibration keyword, and D(z) represents the behavior second keyword set corresponding to the zth calibration keyword.

[0043] Preferably, the step of obtaining advertisement delivery time node information based on historical advertisement delivery data specifically includes:

[0044] Obtain historical advertising keyword information based on historical advertising data;

[0045] Obtain the historical advertising recommendation index based on historical advertising keyword information and behavioral keyword influence coefficient;

[0046] The product of the historical advertisement click-through rate and the historical advertisement conversion rate corresponding to each historical advertisement is used as the advertisement effect index;

[0047] According to the historical advertising delivery data, obtain the historical advertising delivery time node information;

[0048] Obtain the time node impact coefficient based on historical advertising time node information, historical advertising recommendation index and advertising effect index;

[0049] Arrange the historical advertising delivery time nodes in descending order according to the time node influence coefficients to obtain the advertising delivery time node information;

[0050] The calculation formula of the time node influence coefficient is:

[0051]

[0052] Where E is the time node influence coefficient, is the advertising effect index of the g-th advertisement delivered at the advertising delivery time node, and Q(g, h) is the behavioral keyword influence coefficient of the h-th keyword of the g-th advertisement delivered at the advertising delivery time node.

[0053] Preferably, the acquiring of advertisement delivery information according to the advertisement recommendation index and advertisement delivery time node information specifically includes:

[0054] According to the information of the advertisement to be placed and the influence coefficient of the behavioral keywords, the advertisement recommendation index corresponding to each advertisement to be placed is obtained;

[0055] Arrange the advertisements to be placed in descending order of the advertisement recommendation index, and obtain the order information of the advertisements to be placed;

[0056] Match the order information of the advertisements to be placed with the advertisement placement time node information to obtain the advertisement placement information;

[0057] The advertisement recommendation index is specifically:

[0058]

[0059] Where G is the advertising recommendation index, and Q(j) represents the behavioral keyword influence coefficient of the jth advertising keyword.

[0060] Preferably, monitoring the delivered advertisements according to the advertisement monitoring information specifically includes:

[0061] According to the historical advertising delivery data, the historical advertising delivery data and the time node impact coefficient corresponding to each historical advertising delivery time node are obtained;

[0062] The average value of the historical advertising recommendation index of the historically delivered advertisements at each historical advertising delivery time node is used as the standard value of the advertising recommendation index at the advertising delivery time node;

[0063] According to the advertisement delivery information, the advertisement recommendation index of the advertisement corresponding to each advertisement delivery time node is obtained;

[0064] The ratio of the advertising recommendation index to the standard value of the advertising recommendation index is used as the periodic impact coefficient;

[0065] Based on the needs of advertising effect monitoring, obtain the benchmark period for advertising effect monitoring;

[0066] Acquire advertising monitoring information according to the advertising effect monitoring benchmark period and the period influence coefficient, wherein the advertising monitoring information includes advertising effect monitoring period information;

[0067] Monitor the delivered advertisements based on the advertisement monitoring information;

[0068] The advertising effect monitoring cycle is specifically:

[0069]

[0070] Where, T is the advertising effect monitoring period, T 0 is the benchmark period for monitoring advertising effectiveness, G is the advertising recommendation index, and G 0 Recommend index standard values ​​for advertisements.

[0071] Furthermore, an advertisement intelligent delivery system based on automatic annotation information feedback is proposed, which is used to implement the delivery method as described above, including:

[0072] A main control module, the main control module is used to obtain a user preference time performance coefficient according to user browsing time information and browsing prediction time, obtain user characteristic browsing content according to the user preference time performance coefficient and the user preference time performance coefficient threshold, use the behavior first keyword of the user characteristic browsing content as a calibration keyword, and use the calibration keyword as a benchmark to obtain the behavior second keyword that appears simultaneously with the calibration keyword in all user browsing content, construct a behavior second keyword set corresponding to the calibration keyword, obtain the behavior keyword correction coefficient according to the calibration keyword and the behavior second keyword set, arrange the historical advertising delivery time nodes in the order of the time node influence coefficient from large to small, obtain the advertising delivery time node information, obtain the advertising delivery information according to the advertising recommendation index and the advertising delivery time node information, and obtain the advertising monitoring information according to the historical advertising delivery data and the advertising delivery information;

[0073] An information acquisition module, the information acquisition module is used to acquire user behavior data, user click information and user browsing information, and acquire user behavior keyword information, behavior first keyword and behavior second keyword based on user behavior data and automatic annotation, and acquire corresponding historical advertising delivery data, historical advertising delivery time node information, historical advertising click rate information and historical advertising conversion rate information based on user behavior data, and acquire advertising delivery time node information, and acquire information on advertisements to be delivered and keyword information on advertisements to be delivered based on historical advertising delivery data;

[0074] An evaluation module, wherein the evaluation module is used to use the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keyword of the behavior as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keyword of the behavior as the second keyword benchmark influence coefficient according to the user behavior keyword information, adjust the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient according to the user browsing time information and the browsing prediction time, and obtain the behavior keyword influence coefficient, obtain the historical advertising recommendation index according to the historical advertising keyword information and the behavior keyword influence coefficient, use the product of the historical advertising click-through rate and the historical advertising conversion rate corresponding to each historical advertisement as the advertising effect index, obtain the historical advertising delivery time node information according to the historical advertising delivery data, and obtain the time node influence coefficient according to the historical advertising delivery time node information, the historical advertising recommendation index and the advertising effect index;

[0075] The display module interacts with the main control module and is used to output the display behavior keyword influence coefficient, advertisement delivery time node information, advertisement delivery information and advertisement monitoring information.

[0076] Optionally, the main control module specifically includes:

[0077] A control unit, the control unit is used to use a first keyword of a user's characteristic browsing content as a calibration keyword, and based on the calibration keyword, obtain a second keyword of a behavior that appears simultaneously with the calibration keyword in all the user's browsing content, construct a set of the second keyword of a behavior that corresponds to the calibration keyword, obtain a correction coefficient of the behavior keyword based on the calibration keyword and the set of the second keyword of the behavior, arrange the historical advertising delivery time nodes in descending order of the time node influence coefficient, obtain advertising delivery time node information, obtain advertising delivery information based on the advertising recommendation index and the advertising delivery time node information, and obtain advertising monitoring information based on the historical advertising delivery data and the advertising delivery information;

[0078] An information receiving unit, which interacts with the information acquisition module and the evaluation module to receive data and transmit it to the judgment unit;

[0079] A judgment unit is used to obtain a user preference time expression coefficient according to user browsing time information and browsing prediction time, and obtain user characteristic browsing content according to the user preference time expression coefficient and the user preference time expression coefficient threshold.

[0080] Optionally, the information acquisition module specifically includes:

[0081] A first acquisition unit, the first acquisition unit is used to acquire user behavior data, user click information and user browsing information, and acquire user behavior keyword information, a first behavior keyword and a second behavior keyword based on the user behavior data and automatic annotation;

[0082] The second acquisition unit is used to obtain corresponding historical advertising delivery data, historical advertising delivery time node information, historical advertising click-through rate information and historical advertising conversion rate information based on user behavior data, obtain advertising delivery time node information based on historical advertising delivery data, and obtain information about advertisements to be delivered and keyword information about advertisements to be delivered.

[0083] Optionally, the evaluation module specifically includes:

[0084] A keyword evaluation unit, the keyword evaluation unit is used to use the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keywords of the behavior as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keywords of the behavior as the second keyword benchmark influence coefficient according to the user behavior keyword information, and adjust the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient according to the user browsing time information and the browsing prediction time to obtain the behavior keyword influence coefficient;

[0085] A time node evaluation unit, wherein the time node evaluation unit is used to obtain a historical advertising recommendation index based on historical advertising keyword information and a behavioral keyword influence coefficient, and to use the product of the historical advertising click-through rate and the historical advertising conversion rate corresponding to each historical advertisement as the advertising effect index, and to obtain historical advertising delivery time node information based on historical advertising delivery data, and to obtain the time node influence coefficient based on the historical advertising delivery time node information, the historical advertising recommendation index and the advertising effect index.

[0086] Compared with the prior art, the present invention has the following beneficial effects:

[0087] The present invention proposes an intelligent advertising delivery method and system based on automatic annotation information feedback, which deeply mines the interests, behavioral habits, and consumption preferences of users through behavioral keyword influence coefficients, accurately analyzes the preference status of user targets, and allocates different advertising delivery times through advertising delivery time node information, thereby ensuring the maximum display of advertising effects, reducing the interference of invalid advertisements on users, making users browse information more smoothly, and improving user experience. Advertising effects are monitored through advertising monitoring information to ensure timely discovery of advertising effect anomalies and ensure the stability of advertising effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1This is a flow chart of an intelligent advertising delivery method based on automatic annotation information feedback proposed by the present invention;

[0089] Figure 2 A flowchart for obtaining the influence coefficient of the behavior keyword in the present invention;

[0090] Figure 3 This is a flowchart for obtaining advertisement delivery time node information in the present invention;

[0091] Figure 4 This is a flowchart of obtaining advertisement monitoring information in the present invention;

[0092] Figure 5 This is a structural block diagram of an advertising intelligent delivery system based on automatic annotation information feedback proposed by the present invention. DETAILED DESCRIPTION

[0093] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0094] Reference Figure 1 - Figure 4 As shown, an advertisement intelligent delivery method based on automatic annotation information feedback in an embodiment of the present invention includes:

[0095] Acquire user behavior data, wherein the user behavior data includes user click information and user browsing information;

[0096] According to the user behavior data, based on automatic labeling, user behavior keyword information is obtained, wherein the user behavior keyword information includes a first behavior keyword and a second behavior keyword;

[0097] According to the user behavior keyword information and based on the user behavior data, obtain the behavior keyword influence coefficient corresponding to each user behavior keyword;

[0098] Specifically, according to the user behavior keyword information and based on the user behavior data, the behavior keyword influence coefficient corresponding to each user behavior keyword is obtained, including:

[0099] Based on user behavior data, obtain user browsing topic information and user browsing content information;

[0100] Based on automatic annotation, obtain the keywords of user browsing topics and user browsing content;

[0101] The user browsing topic keyword is used as the first behavior keyword, and the user browsing content keyword is used as the second behavior keyword to obtain the user behavior keyword information;

[0102] According to the user behavior keyword information, the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keywords of the behavior is used as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keywords of the behavior is used as the second keyword benchmark influence coefficient;

[0103] According to user behavior data, obtain the user browsing time information corresponding to each user's browsing content;

[0104] Based on the user's browsing content information, predict the user's browsing time at a normal information browsing speed to obtain the browsing prediction time;

[0105] According to the user browsing time information and the browsing prediction time, the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient are adjusted to obtain the behavior keyword influence coefficient.

[0106] It can be understood that the normal information browsing speed refers to the browsing speed of users under normal conditions, for example, the average reading speed of adults is about 200-300 words per minute (wpm). When users view pictures, they usually stay on each picture for about 1-3 seconds. In this embodiment, the text reading speed is 250 words per minute, the picture reading speed is 2 seconds per picture, and the predicted browsing time for video content is the video time.

[0107] Specifically, according to the user browsing time information and the browsing prediction time, the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient are adjusted to obtain the behavior keyword influence coefficient, which specifically includes:

[0108] The ratio of user browsing time information to browsing prediction time is used as the user preference time performance coefficient;

[0109] Obtain the influence coefficient of the first keyword of the behavior according to the first keyword benchmark influence coefficient and the user preference time performance coefficient;

[0110] Based on the analysis of user browsing time, obtain the user preference time performance coefficient threshold;

[0111] Obtaining user characteristic browsing content according to the user preference time expression coefficient and the user preference time expression coefficient threshold;

[0112] If the user preference time performance coefficient w of the user browsing content is greater than or equal to 0.7, the user browsing content is regarded as the user characteristic browsing content;

[0113] The first keyword of the user's characteristic browsing behavior is used as the calibration keyword;

[0114] Taking the calibration keyword as a benchmark, obtaining the behavioral second keywords that appear simultaneously with the calibration keyword in all the user browsing contents, and constructing a behavioral second keyword set corresponding to the calibration keyword;

[0115] Obtaining a behavior keyword correction coefficient according to the calibration keyword and the behavior second keyword set;

[0116] Obtaining the behavioral second keyword influence coefficient according to the second keyword baseline influence coefficient and the behavioral keyword correction coefficient;

[0117] Obtain the behavior keyword influence coefficient according to the behavior first keyword influence coefficient and the behavior second keyword influence coefficient;

[0118] The influence coefficient of the first keyword of the behavior is specifically:

[0119]

[0120] Where Q(x) is the influence coefficient of the first keyword of the x-th behavior, w(i) represents the user preference time performance coefficient of the i-th user browsing content containing the x-th behavior first keyword, R(x) is the first keyword benchmark influence coefficient of the x-th behavior first keyword, and n is the total number of user browsing content containing the x-th behavior first keyword;

[0121] The influence coefficient of the second keyword of the behavior is specifically:

[0122]

[0123] Where Q(y) is the influence coefficient of the second keyword of the y-th behavior, R(y) is the second keyword benchmark influence coefficient of the second keyword of the y-th behavior, k y (z) is the behavior keyword correction coefficient of the yth behavior second keyword and the zth calibration keyword, w(z) represents the user preference time performance coefficient of the user browsing content corresponding to the zth calibration keyword, d(y, z) represents the number of occurrences of the yth behavior second keyword in the behavior second keyword set corresponding to the zth calibration keyword, d(z) represents the total number of behavior second keywords in the behavior second keyword set corresponding to the zth calibration keyword, and D(z) represents the behavior second keyword set corresponding to the zth calibration keyword.

[0124] In this scheme, by taking the user's browsing topic keyword as the first keyword of the behavior, taking the user's browsing content keyword as the second keyword of the behavior, taking the ratio of the number of occurrences of each first keyword of the behavior to the total number of the first keywords of the behavior as the first keyword benchmark influence coefficient, taking the ratio of the number of occurrences of each second keyword of the behavior to the total number of the second keywords of the behavior as the second keyword benchmark influence coefficient, the frequency of keywords often directly reflects the user's preference status, therefore, the frequency of different keywords is used to indicate the user's preference for keywords;

[0125] It is understandable that although keywords directly represent the user's preference, the user's specific preference for different content is still inaccurate. Compared with the keywords used by users to browse content, the keywords in the content topics can often better represent the user's preference. When users browse content because of the keywords in the topics, the frequency of the keywords in the content will also increase, affecting the user's preference analysis. At the same time, the user's preference is closely related to the user's browsing time. If the user's click-through rate on the content is high but the browsing time is short, then if you place an advertisement, it will not work.

[0126] In this embodiment, It means that the second keyword of the yth behavior does not belong to the set of second keywords of the behavior corresponding to the zth marked keyword.

[0127] According to the user behavior data, corresponding historical advertising delivery data is obtained, wherein the historical advertising delivery data includes historical advertising delivery time node information, historical advertising click rate information, and historical advertising conversion rate information;

[0128] According to historical advertising data, obtain advertising delivery time node information;

[0129] Specifically, based on historical advertising data, information about advertising delivery time nodes is obtained, including:

[0130] Obtain historical advertising keyword information based on historical advertising data;

[0131] Obtain the historical advertising recommendation index based on historical advertising keyword information and behavioral keyword influence coefficient;

[0132] The product of the historical advertisement click-through rate and the historical advertisement conversion rate corresponding to each historical advertisement is used as the advertisement effect index;

[0133] According to the historical advertising delivery data, obtain the historical advertising delivery time node information;

[0134] Obtain the time node impact coefficient based on historical advertising time node information, historical advertising recommendation index and advertising effect index;

[0135] Arrange the historical advertising delivery time nodes in descending order according to the time node influence coefficients to obtain the advertising delivery time node information;

[0136] The calculation formula of the time node influence coefficient is:

[0137]

[0138] Where E is the time node influence coefficient, is the advertising effect index of the g-th advertisement delivered at the advertising delivery time node, and Q(g, h) is the behavioral keyword influence coefficient of the h-th keyword of the g-th advertisement delivered at the advertising delivery time node.

[0139] In this solution, the product of the historical advertisement click-through rate and the historical advertisement conversion rate corresponding to each historical advertisement is used as the advertisement effect index, and the historical advertisement delivery time node information is obtained according to the historical advertisement delivery data. The time node influence coefficient is obtained according to the historical advertisement delivery time node information, the historical advertisement recommendation index and the advertisement effect index. The historical advertisement delivery time nodes are arranged in descending order according to the time node influence coefficient to obtain the advertisement delivery time node information. The complex relationship between the advertisement delivery time and the advertisement effect is fully considered. Different from the existing method that may ignore the time factor or simply divide the time interval, the influence of different time nodes on the advertisement effect is accurately quantified.

[0140] It is understandable that different delivery time nodes have a great impact on the effectiveness of advertising, but when analyzing through historical advertising data, the advertising effect of the advertisement itself will also have an impact on the final delivery effect of the advertisement. Therefore, the time node impact coefficient can be used to accurately analyze the impact of advertising effects at different delivery time nodes.

[0141] It should be noted that, in this embodiment, the historical advertisement recommendation index represents the sum of the behavioral keyword influence coefficients of all keywords in the historical advertisements, and the calculation method is the same as that of the advertisement recommendation index, so it is not further described in the solution.

[0142] Acquire information of advertisements to be placed, wherein the information of advertisements to be placed includes keyword information of advertisements to be placed;

[0143] According to the information of the advertisement to be placed and the influence coefficient of the behavioral keywords, the advertisement recommendation index corresponding to each advertisement to be placed is obtained;

[0144] Obtain advertising delivery information based on the advertising recommendation index and advertising delivery time node information;

[0145] Specifically, according to the advertisement recommendation index and advertisement delivery time node information, advertisement delivery information is obtained, including:

[0146] According to the information of the advertisement to be placed and the influence coefficient of the behavioral keywords, the advertisement recommendation index corresponding to each advertisement to be placed is obtained;

[0147] Arrange the advertisements to be placed in descending order of the advertisement recommendation index, and obtain the order information of the advertisements to be placed;

[0148] Match the order information of the advertisements to be placed with the advertisement placement time node information to obtain the advertisement placement information;

[0149] The advertisement recommendation index is specifically:

[0150]

[0151] Where G is the advertising recommendation index, and Q(j) represents the behavioral keyword influence coefficient of the jth advertising keyword.

[0152] In this solution, the advertisement recommendation index corresponding to each advertisement to be placed is obtained according to the advertisement information to be placed and the influence coefficient of the behavior keyword, and the advertisements to be placed are arranged in descending order of the advertisement recommendation index to obtain the order information of the advertisements to be placed, and the order information of the advertisements to be placed is matched with the advertisement placement time node information to obtain the advertisement placement information;

[0153] It is understandable that different time nodes have completely different effects on advertising effects. By matching the order information of the ads to be delivered with the time node information of the ads to be delivered, that is, delivering the first ad to be delivered at the first time node of the ads to be delivered, the advertising impact is maximized and the overall advertising effect is improved. It should be noted that the time nodes of ads delivery are different time nodes when users browse content. For example, the reading time of short text is generally about 1-2 minutes. Ads can be delivered when the user has read 70% to 80% of the progress, that is, 0.7-1.6 minutes. The reading time of medium-length text is about 3-8 minutes. It is recommended to deliver ads when the user has read 40% to 60% of the text, that is, between 1.2 and 4.8 minutes. If the reading time exceeds 8 minutes, ads can be delivered once when the user has read 30% to 50%, that is, around 2.4 to 4 minutes, and then delivered again when the user has read 70% to 85%, that is, around 5.6 to 6.8 minutes, etc.

[0154] Place advertisements according to the advertisement placement information;

[0155] Obtain advertising monitoring information based on historical advertising data and advertising information;

[0156] The delivered advertisements are monitored according to the advertisement monitoring information.

[0157] Specifically, the advertisements placed are monitored based on the advertisement monitoring information, including:

[0158] According to the historical advertising delivery data, the historical advertising delivery data and the time node impact coefficient corresponding to each historical advertising delivery time node are obtained;

[0159] The average value of the historical advertising recommendation index of the historically delivered advertisements at each historical advertising delivery time node is used as the standard value of the advertising recommendation index at the advertising delivery time node;

[0160] According to the advertisement delivery information, the advertisement recommendation index of the advertisement corresponding to each advertisement delivery time node is obtained;

[0161] The ratio of the advertising recommendation index to the standard value of the advertising recommendation index is used as the periodic impact coefficient;

[0162] Based on the needs of advertising effect monitoring, obtain the benchmark period for advertising effect monitoring;

[0163] Acquire advertising monitoring information according to the advertising effect monitoring benchmark period and the period influence coefficient, wherein the advertising monitoring information includes advertising effect monitoring period information;

[0164] Monitor the delivered advertisements based on the advertisement monitoring information;

[0165] The advertising effect monitoring cycle is specifically:

[0166]

[0167] Where, T is the advertising effect monitoring period, T 0 is the benchmark period for monitoring advertising effectiveness, G is the advertising recommendation index, and G 0 Recommend index standard values ​​for advertisements.

[0168] In this scheme, the average value of the historical advertising recommendation index of the historically delivered advertisements at each historical advertising delivery time node is used as the standard value of the advertising recommendation index at the advertising delivery time node, and the ratio of the advertising recommendation index to the standard value of the advertising recommendation index is used as the cycle impact coefficient. Based on the advertising effect monitoring needs, the advertising effect monitoring benchmark cycle is obtained. According to the advertising effect monitoring benchmark cycle and the cycle impact coefficient, the advertising effect monitoring cycle is obtained, and problems in the advertising delivery process are discovered in time, thereby improving the accuracy and effect of advertising delivery, avoiding waste of resources, and improving the efficiency and return on investment of advertising delivery.

[0169] It should be noted that in this embodiment, the advertising effect monitoring benchmark period is 7 days, the advertising standard click rate is 2.5%, and the advertising standard conversion rate is 3%. The standard click-through rate of an ad, or the conversion rate of an ad served If the advertising standard conversion rate is not met, the advertising delivery will be adjusted.

[0170] Reference Figure 5 As shown, further, in combination with the above-mentioned advertising intelligent delivery method based on automatic annotation information feedback, an advertising intelligent delivery system based on automatic annotation information feedback is proposed, including:

[0171] A main control module, the main control module is used to obtain a user preference time performance coefficient according to user browsing time information and browsing prediction time, obtain user characteristic browsing content according to the user preference time performance coefficient and the user preference time performance coefficient threshold, use the behavior first keyword of the user characteristic browsing content as a calibration keyword, and use the calibration keyword as a benchmark to obtain the behavior second keyword that appears simultaneously with the calibration keyword in all user browsing content, construct a behavior second keyword set corresponding to the calibration keyword, obtain the behavior keyword correction coefficient according to the calibration keyword and the behavior second keyword set, arrange the historical advertising delivery time nodes in the order of the time node influence coefficient from large to small, obtain the advertising delivery time node information, obtain the advertising delivery information according to the advertising recommendation index and the advertising delivery time node information, and obtain the advertising monitoring information according to the historical advertising delivery data and the advertising delivery information;

[0172] An information acquisition module, the information acquisition module is used to acquire user behavior data, user click information and user browsing information, and acquire user behavior keyword information, behavior first keyword and behavior second keyword based on user behavior data and automatic annotation, and acquire corresponding historical advertising delivery data, historical advertising delivery time node information, historical advertising click rate information and historical advertising conversion rate information based on user behavior data, and acquire advertising delivery time node information, and acquire information on advertisements to be delivered and keyword information on advertisements to be delivered based on historical advertising delivery data;

[0173] An evaluation module, wherein the evaluation module is used to use the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keyword of the behavior as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keyword of the behavior as the second keyword benchmark influence coefficient according to the user behavior keyword information, adjust the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient according to the user browsing time information and the browsing prediction time, and obtain the behavior keyword influence coefficient, obtain the historical advertising recommendation index according to the historical advertising keyword information and the behavior keyword influence coefficient, use the product of the historical advertising click-through rate and the historical advertising conversion rate corresponding to each historical advertisement as the advertising effect index, obtain the historical advertising delivery time node information according to the historical advertising delivery data, and obtain the time node influence coefficient according to the historical advertising delivery time node information, the historical advertising recommendation index and the advertising effect index;

[0174] The display module interacts with the main control module and is used to output the display behavior keyword influence coefficient, advertisement delivery time node information, advertisement delivery information and advertisement monitoring information.

[0175] Main control module, specifically including:

[0176] A control unit, the control unit is used to use a first keyword of a user's characteristic browsing content as a calibration keyword, and based on the calibration keyword, obtain a second keyword of a behavior that appears simultaneously with the calibration keyword in all the user's browsing content, construct a set of the second keyword of a behavior that corresponds to the calibration keyword, obtain a correction coefficient of the behavior keyword based on the calibration keyword and the set of the second keyword of the behavior, arrange the historical advertising delivery time nodes in descending order of the time node influence coefficient, obtain advertising delivery time node information, obtain advertising delivery information based on the advertising recommendation index and the advertising delivery time node information, and obtain advertising monitoring information based on the historical advertising delivery data and the advertising delivery information;

[0177] An information receiving unit, which interacts with the information acquisition module and the evaluation module to receive data and transmit it to the judgment unit;

[0178] A judgment unit is used to obtain a user preference time expression coefficient according to user browsing time information and browsing prediction time, and obtain user characteristic browsing content according to the user preference time expression coefficient and the user preference time expression coefficient threshold.

[0179] Information acquisition module, specifically including:

[0180] A first acquisition unit, the first acquisition unit is used to acquire user behavior data, user click information and user browsing information, and acquire user behavior keyword information, a first behavior keyword and a second behavior keyword based on the user behavior data and automatic annotation;

[0181] The second acquisition unit is used to obtain corresponding historical advertising delivery data, historical advertising delivery time node information, historical advertising click-through rate information and historical advertising conversion rate information based on user behavior data, obtain advertising delivery time node information based on historical advertising delivery data, and obtain information about advertisements to be delivered and keyword information about advertisements to be delivered.

[0182] Assessment modules include:

[0183] A keyword evaluation unit, the keyword evaluation unit is used to use the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keywords of the behavior as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keywords of the behavior as the second keyword benchmark influence coefficient according to the user behavior keyword information, and adjust the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient according to the user browsing time information and the browsing prediction time to obtain the behavior keyword influence coefficient;

[0184] A time node evaluation unit, wherein the time node evaluation unit is used to obtain a historical advertising recommendation index based on historical advertising keyword information and a behavioral keyword influence coefficient, and to use the product of the historical advertising click-through rate and the historical advertising conversion rate corresponding to each historical advertisement as the advertising effect index, and to obtain historical advertising delivery time node information based on historical advertising delivery data, and to obtain the time node influence coefficient based on the historical advertising delivery time node information, the historical advertising recommendation index and the advertising effect index.

[0185] In summary, the advantages of the present invention are: through user behavior keyword information, based on user behavior data, the behavior keyword influence coefficient corresponding to each user behavior keyword is obtained, the user's interests, behavior habits, and consumption preferences are deeply mined through the behavior keyword influence coefficient, and the preference status of the user target is accurately analyzed, and the advertising delivery time node information is obtained through historical advertising delivery data, and different advertising delivery times are allocated to ensure the maximum display of advertising effects, reduce the interference of invalid advertisements on users, make users browse information more smoothly, and improve user experience, obtain advertising monitoring information through historical advertising delivery data and advertising delivery information, monitor advertising effects through advertising monitoring information, ensure timely detection of advertising effect abnormalities, and ensure the stability of advertising effects.

[0186] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. An intelligent advertising delivery method based on automatic labeling information feedback, characterized in that: include: Acquire user behavior data, wherein the user behavior data includes user click information and user browsing information; According to the user behavior data, based on automatic labeling, user behavior keyword information is obtained, wherein the user behavior keyword information includes a first behavior keyword and a second behavior keyword; According to the user behavior keyword information and based on the user behavior data, obtain the behavior keyword influence coefficient corresponding to each user behavior keyword; According to the user behavior data, corresponding historical advertising delivery data is obtained, wherein the historical advertising delivery data includes historical advertising delivery time node information, historical advertising click rate information, and historical advertising conversion rate information; According to historical advertising data, obtain advertising delivery time node information; Acquire information of advertisements to be placed, wherein the information of advertisements to be placed includes keyword information of advertisements to be placed; According to the information of the advertisement to be placed and the influence coefficient of the behavioral keywords, the advertisement recommendation index corresponding to each advertisement to be placed is obtained; Obtain advertising delivery information based on the advertising recommendation index and advertising delivery time node information; Place advertisements according to the advertisement placement information; Obtain advertising monitoring information based on historical advertising data and advertising information; The delivered advertisements are monitored according to the advertisement monitoring information.

2. The method for intelligent advertising delivery based on automatic annotation information feedback according to claim 1, characterized in that: The step of obtaining the behavior keyword influence coefficient corresponding to each user behavior keyword based on the user behavior data according to the user behavior keyword information specifically includes: Based on user behavior data, obtain user browsing topic information and user browsing content information; Based on automatic annotation, obtain the keywords of user browsing topics and user browsing content; The user browsing topic keyword is used as the first behavior keyword, and the user browsing content keyword is used as the second behavior keyword to obtain the user behavior keyword information; According to the user behavior keyword information, the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keywords of the behavior is used as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keywords of the behavior is used as the second keyword benchmark influence coefficient; According to user behavior data, obtain the user browsing time information corresponding to each user's browsing content; Based on the user's browsing content information, predict the user's browsing time at a normal information browsing speed to obtain the browsing prediction time; According to the user browsing time information and the browsing prediction time, the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient are adjusted to obtain the behavior keyword influence coefficient.

3. The method for intelligent advertising delivery based on automatic annotation information feedback according to claim 2, characterized in that: The step of adjusting the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient according to the user browsing time information and the browsing prediction time to obtain the behavior keyword influence coefficient specifically includes: The ratio of user browsing time information to browsing prediction time is used as the user preference time performance coefficient; Obtain the influence coefficient of the first keyword of the behavior according to the first keyword benchmark influence coefficient and the user preference time performance coefficient; Based on the analysis of user browsing time, obtain the user preference time performance coefficient threshold; Obtaining user characteristic browsing content according to the user preference time expression coefficient and the user preference time expression coefficient threshold; If the user preference time performance coefficient w of the user browsing content is greater than or equal to 0.7, the user browsing content is regarded as the user characteristic browsing content; The first keyword of the user's characteristic browsing behavior is used as the calibration keyword; Taking the calibration keyword as a benchmark, obtaining the behavioral second keywords that appear simultaneously with the calibration keyword in all the user browsing contents, and constructing a behavioral second keyword set corresponding to the calibration keyword; Obtaining a behavior keyword correction coefficient according to the calibration keyword and the behavior second keyword set; Obtaining the behavioral second keyword influence coefficient according to the second keyword baseline influence coefficient and the behavioral keyword correction coefficient; Obtain the behavior keyword influence coefficient according to the behavior first keyword influence coefficient and the behavior second keyword influence coefficient; The influence coefficient of the first keyword of the behavior is specifically: Where Q(x) is the influence coefficient of the first keyword of the x-th behavior, w(i) represents the user preference time performance coefficient of the i-th user browsing content containing the x-th behavior, R(x) is the first keyword benchmark influence coefficient of the x-th behavior, and n is the total number of user browsing content containing the x-th behavior; The influence coefficient of the second keyword of the behavior is specifically: Where Q(y) is the influence coefficient of the second keyword of the y-th behavior, R(y) is the second keyword benchmark influence coefficient of the second keyword of the y-th behavior, k y (z) is the behavior keyword correction coefficient of the yth behavior second keyword and the zth calibration keyword, w(z) represents the user preference time performance coefficient of the user browsing content corresponding to the zth calibration keyword, d(y, z) represents the number of occurrences of the yth behavior second keyword in the behavior second keyword set corresponding to the zth calibration keyword, d(z) represents the total number of behavior second keywords in the behavior second keyword set corresponding to the zth calibration keyword, and D(z) represents the behavior second keyword set corresponding to the zth calibration keyword.

4. The method for intelligent advertising delivery based on automatic annotation information feedback according to claim 1, characterized in that: The step of obtaining advertisement delivery time node information based on historical advertisement delivery data specifically includes: Obtain historical advertising keyword information based on historical advertising data; Obtain the historical advertising recommendation index based on historical advertising keyword information and behavioral keyword influence coefficient; The product of the historical advertisement click-through rate and the historical advertisement conversion rate corresponding to each historical advertisement is used as the advertisement effect index; According to the historical advertising delivery data, obtain the historical advertising delivery time node information; Obtain the time node impact coefficient based on historical advertising time node information, historical advertising recommendation index and advertising effect index; Arrange the historical advertising delivery time nodes in descending order according to the time node influence coefficients to obtain the advertising delivery time node information; The calculation formula of the time node influence coefficient is: Where E is the time node influence coefficient, is the advertising effect index of the g-th advertisement delivered at the advertising delivery time node, and Q(g, h) is the behavioral keyword influence coefficient of the h-th keyword of the g-th advertisement delivered at the advertising delivery time node.

5. The method for intelligent advertising delivery based on automatic annotation information feedback according to claim 1, characterized in that: The step of obtaining advertisement delivery information according to the advertisement recommendation index and advertisement delivery time node information specifically includes: According to the information of the advertisement to be placed and the influence coefficient of the behavioral keywords, the advertisement recommendation index corresponding to each advertisement to be placed is obtained; Arrange the advertisements to be placed in descending order of the advertisement recommendation index, and obtain the order information of the advertisements to be placed; Match the order information of the advertisements to be placed with the advertisement placement time node information to obtain the advertisement placement information; The advertisement recommendation index is specifically: Where G is the advertising recommendation index, and Q(j) represents the behavioral keyword influence coefficient of the jth advertising keyword.

6. The method for intelligent advertising delivery based on automatic annotation information feedback according to claim 1, characterized in that: The monitoring of the delivered advertisements according to the advertisement monitoring information specifically includes: According to the historical advertising delivery data, the historical advertising delivery data and the time node impact coefficient corresponding to each historical advertising delivery time node are obtained; The average value of the historical advertising recommendation index of the historically delivered advertisements at each historical advertising delivery time node is used as the standard value of the advertising recommendation index at the advertising delivery time node; According to the advertisement delivery information, the advertisement recommendation index of the advertisement corresponding to each advertisement delivery time node is obtained; The ratio of the advertising recommendation index to the standard value of the advertising recommendation index is used as the periodic impact coefficient; Based on the needs of advertising effect monitoring, obtain the benchmark period for advertising effect monitoring; Acquire advertising monitoring information according to the advertising effect monitoring benchmark period and the period influence coefficient, wherein the advertising monitoring information includes advertising effect monitoring period information; Monitor the delivered advertisements based on the advertisement monitoring information; The advertising effect monitoring cycle is specifically: In the formula, T is the advertising effect monitoring period, T0 is the advertising effect monitoring benchmark period, G is the advertising recommendation index, and G0 is the standard value of the advertising recommendation index.

7. An intelligent advertising delivery system based on automatic annotation information feedback, used to implement the delivery method according to any one of claims 1 to 6, characterized in that: include: A main control module, the main control module is used to obtain a user preference time performance coefficient according to user browsing time information and browsing prediction time, obtain user characteristic browsing content according to the user preference time performance coefficient and the user preference time performance coefficient threshold, use the behavior first keyword of the user characteristic browsing content as a calibration keyword, and use the calibration keyword as a benchmark to obtain the behavior second keyword that appears simultaneously with the calibration keyword in all user browsing content, construct a behavior second keyword set corresponding to the calibration keyword, obtain the behavior keyword correction coefficient according to the calibration keyword and the behavior second keyword set, arrange the historical advertising delivery time nodes in the order of the time node influence coefficient from large to small, obtain the advertising delivery time node information, obtain the advertising delivery information according to the advertising recommendation index and the advertising delivery time node information, and obtain the advertising monitoring information according to the historical advertising delivery data and the advertising delivery information; An information acquisition module, the information acquisition module is used to acquire user behavior data, user click information and user browsing information, and acquire user behavior keyword information, behavior first keyword and behavior second keyword based on user behavior data and automatic annotation, and acquire corresponding historical advertising delivery data, historical advertising delivery time node information, historical advertising click rate information and historical advertising conversion rate information based on user behavior data, and acquire advertising delivery time node information, and acquire information on advertisements to be delivered and keyword information on advertisements to be delivered based on historical advertising delivery data; An evaluation module, wherein the evaluation module is used to use the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keyword of the behavior as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keyword of the behavior as the second keyword benchmark influence coefficient according to the user behavior keyword information, adjust the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient according to the user browsing time information and the browsing prediction time, and obtain the behavior keyword influence coefficient, obtain the historical advertising recommendation index according to the historical advertising keyword information and the behavior keyword influence coefficient, use the product of the historical advertising click-through rate and the historical advertising conversion rate corresponding to each historical advertisement as the advertising effect index, obtain the historical advertising delivery time node information according to the historical advertising delivery data, and obtain the time node influence coefficient according to the historical advertising delivery time node information, the historical advertising recommendation index and the advertising effect index; The display module interacts with the main control module and is used to output the display behavior keyword influence coefficient, advertisement delivery time node information, advertisement delivery information and advertisement monitoring information.

8. The intelligent advertising delivery system based on automatic annotation information feedback according to claim 7, characterized in that: The main control module specifically includes: A control unit, the control unit is used to use a first keyword of a user's characteristic browsing content as a calibration keyword, and based on the calibration keyword, obtain a second keyword of a behavior that appears simultaneously with the calibration keyword in all the user's browsing content, construct a set of the second keyword of a behavior that corresponds to the calibration keyword, obtain a correction coefficient of the behavior keyword based on the calibration keyword and the set of the second keyword of the behavior, arrange the historical advertising delivery time nodes in descending order of the time node influence coefficient, obtain advertising delivery time node information, obtain advertising delivery information based on the advertising recommendation index and the advertising delivery time node information, and obtain advertising monitoring information based on the historical advertising delivery data and the advertising delivery information; An information receiving unit, which interacts with the information acquisition module and the evaluation module to receive data and transmit it to the judgment unit; A judgment unit is used to obtain a user preference time expression coefficient according to user browsing time information and browsing prediction time, and obtain user characteristic browsing content according to the user preference time expression coefficient and the user preference time expression coefficient threshold.

9. The intelligent advertising delivery system based on automatic annotation information feedback according to claim 7, characterized in that: The information acquisition module specifically includes: A first acquisition unit, the first acquisition unit is used to acquire user behavior data, user click information and user browsing information, and acquire user behavior keyword information, a first behavior keyword and a second behavior keyword based on the user behavior data and automatic annotation; The second acquisition unit is used to obtain corresponding historical advertising delivery data, historical advertising delivery time node information, historical advertising click-through rate information and historical advertising conversion rate information based on user behavior data, obtain advertising delivery time node information based on historical advertising delivery data, and obtain information about advertisements to be delivered and keyword information about advertisements to be delivered.

10. The intelligent advertising delivery system based on automatic annotation information feedback according to claim 7, characterized in that: The evaluation module specifically includes: A keyword evaluation unit, the keyword evaluation unit is used to use the ratio of the number of occurrences of the first keyword of each behavior to the total number of the first keywords of the behavior as the first keyword benchmark influence coefficient, and the ratio of the number of occurrences of the second keyword of each behavior to the total number of the second keywords of the behavior as the second keyword benchmark influence coefficient according to the user behavior keyword information, and adjust the first keyword benchmark influence coefficient and the second keyword benchmark influence coefficient according to the user browsing time information and the browsing prediction time to obtain the behavior keyword influence coefficient; A time node evaluation unit, wherein the time node evaluation unit is used to obtain a historical advertising recommendation index based on historical advertising keyword information and a behavioral keyword influence coefficient, and to use the product of the historical advertising click-through rate and the historical advertising conversion rate corresponding to each historical advertisement as the advertising effect index, and to obtain historical advertising delivery time node information based on historical advertising delivery data, and to obtain the time node influence coefficient based on the historical advertising delivery time node information, the historical advertising recommendation index and the advertising effect index.

Images