An intelligent monitoring system and method for advertising based on multi-source data fusion

Abstract

The present invention discloses an intelligent monitoring system and method for advertising delivery based on multi-source data fusion, which relates to the technical field of intelligent monitoring of advertising delivery. The present invention comprises: S10: generating real-time delivery characteristics of a target advertisement; S20: finding a target associated node and analyzing the optimization index of the target advertisement delivery strategy; S30: determining the strategy optimization object of the target advertisement; and S40: generating the delivery optimization strategy of the target advertisement. The present invention analyzes the real-time optimization factor of the target advertisement delivery strategy through the target associated node, thereby analyzing the user's acceptance of the target advertisement, which is conducive to quickly capturing the real-time delivery effect of the target advertisement. By generating a time management matrix for the target advertisement through key associated nodes, the priority of each effective delivery time period of the target advertisement can be intuitively obtained, which facilitates the rapid adjustment of the delivery time of the target advertisement, thereby reducing the delivery cost of the target advertisement.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring of advertising placement, and in particular to an intelligent monitoring system and method for advertising placement based on multi-source data fusion. Background Art

[0002] With the rapid development of Internet technology, the status of online advertising has become increasingly prominent. Advertisers and advertising agencies are paying more and more attention to the effectiveness of advertising and pursuing accurate and efficient advertising methods.

[0003] However, traditional advertising delivery methods often have problems such as poor delivery effects, high costs, and low delivery efficiency. At the same time, the monitoring of advertising delivery effects only stays on user feedback data. The authenticity and processing efficiency of user feedback data will lead to certain differences between the monitoring results and the actual results. In addition, the analysis of the effective delivery time of advertisements is generally determined by the playback volume in each time period, which cannot achieve rapid adjustment of the effective playback time. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent monitoring system and method for advertising delivery based on multi-source data fusion to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent monitoring method for advertising delivery based on multi-source data fusion, the method comprising:

[0006] S10: When a user clicks a target ad, a window quantization value of the target ad on the user's click interface is collected, carrier features of the user's click interface are obtained, historical delivery data of associated ads generated on each carrier feature is analyzed, and real-time delivery features of the target ad are generated based on the user density of the target ad in a specific environment.

[0007] S20: Finding target related nodes based on the real-time delivery characteristics of the target advertisement, and analyzing the optimization index of the target advertisement delivery strategy;

[0008] S30: Classifying and optimizing the time characteristics of key related nodes, and generating a time management matrix for the target advertisement. Determining the target advertisement's strategy optimization object based on the compatibility between the optimized time of the target advertisement delivery strategy and the time management matrix.

[0009] S40: Generate a delivery optimization strategy for the target advertisement.

[0010] Furthermore, the S10 includes:

[0011] S101: When a user clicks a target ad, a window quantization value T of the target ad on the user's click interface is collected, where T = (S1 / S2) × γ, where S1 represents the window area of ​​the target ad, S2 represents the area of ​​the user's click interface, and γ represents the weight value corresponding to the window position of the target ad.

[0012] S102: Acquire carrier features existing in the user-clicked interface. Carrier features refer to text or images recorded in the area of ​​the user-clicked interface other than the target ad window. Number each acquired carrier feature. The numbering result is: i = 1, 2, ..., n, where n represents the total number of acquired carrier features.

[0013] Collect the average conversion rate gi of the associated ads generated on carrier feature i, calculate the ratio fi between gi-z and 1-exp(u-ri), sum all fi from i=1 to i=n, ​​record the summation result as F, calculate the ratio between F and n, and obtain the conversion index Z of the user clicking on the target ad, where ri represents the average popularity index of carrier feature i on the Internet, z represents the average conversion rate of the target ad when there is no carrier feature, u represents the average popularity index of the target ad when there is no carrier feature, exp() represents the exponent with base e and e=2.73;

[0014] S103: When a user clicks on a target ad, a pop-up window requesting the user's geographic location is displayed on the target ad interface. If the user authorizes the request and does not exit the target ad within a time threshold, the user's geographic location is automatically collected to eliminate user geographic location data generated by forced advertising behavior, thereby improving the prediction accuracy of the target ad's user density. If the collected user geographic location is within the sales and delivery range of the product displayed in the target ad, the corresponding user is marked; otherwise, the user is not marked. The ratio of the marked users to the total number of collected user geographic locations is calculated to obtain the user density Y of the target ad.

[0015] Generate the delivery feature Kt of the target advertisement at time t, Kt={Tt,Zt,Yt}, where Yt represents the user density of the target advertisement at time t, Tt represents the window quantization value of the target advertisement at time t, and Zt represents the conversion index of the target advertisement when the user clicks on the interface at time t.

[0016] Based on the real-time delivery characteristics of the target advertisement, the real-time delivery effect of the target advertisement is intelligently monitored. Compared with the method of achieving the monitoring purpose through user feedback data, it can effectively eliminate some unnecessary monitoring data and further improve the monitoring effect of the advertising delivery effect.

[0017] Furthermore, the S20 includes:

[0018] S201: Generate a user density-time curve L1 and a conversion index-time curve L2 for the target advertisement, record the intersection of the two curves as an associated node, and number each associated node in chronological order. The numbering result is: p = 1, 2, ..., q, where q represents the total number of associated nodes.

[0019] In the time period [pt, pt+h], if the curve L1 shows a downward trend, the associated node p is considered to be the target associated node, where pt represents the appearance time of the associated node p and h represents the monitoring interval. The target associated node is used to analyze the abnormal situation of the target advertisement delivery, and the analysis process is visualized and efficient.

[0020] S202: Based on the user density-time curve L1, collect the user density Y(pt) and Y(pt+h) at the time pt and pt+h respectively, and calculate the difference Y between Y(pt) and Y(pt+h). pt→pt+h Perform calculations;

[0021] In the [pt, pt+h] time period, when the conversion index-time curve L2 shows a downward trend:

[0022] Based on the conversion index-time curve L2, the conversion index Z(pt) and Z(pt+h) are collected at the time pt and pt+h respectively, and the difference Z between Z(pt) and Z(pt+h) is calculated. pt→pt+h Perform calculations;

[0023] If 1-exp(Y pt→pt+h ) and 1-exp(Z pt→pt+h ) is within the error range, then the optimization factor of the target advertising delivery strategy at the time pt+h is W(pt+h)=1-exp(k pt→pt+h );

[0024] In the [pt, pt+h] time period, when the conversion index-time curve L2 shows an upward or horizontal trend:

[0025] The optimization factor of the target advertising strategy is W(pt+h)=1-exp[-(K pt→pt+h -k pt→pt+h )], where K pt→pt+h represents the slope of curve L2 in the time period [pt, pt+h], k pt→pt+h Represents the slope of curve L1 in the time period [pt, pt+h];

[0026] S203: Calculate the product of W(pt+h) and the window quantization value T(pt+h) of the target advertisement at the time pt+h to obtain the optimization index J(pt+h) of the target advertisement delivery strategy at the time pt+h.

[0027] Based on the temporal changes in the various delivery characteristics of the target advertisement, the user's acceptance of the target advertisement is analyzed. Compared with the analysis results obtained through the user's interaction status, residence time, etc., the analysis process of the interactive content is reduced, which is conducive to quickly capturing the delivery effect of the target advertisement.

[0028] Furthermore, the S30 includes:

[0029] S301: When 0.6≤J(pt+h)≤1, it indicates that the target advertisement delivery strategy needs to be optimized at time pt+h; when 0≤J(pt+h)<0.6, it indicates that the target advertisement delivery strategy does not need to be optimized at time pt+h;

[0030] S302: In the time period [pt, pt+h], if both curves L1 and L2 show an upward trend, the associated node p is considered to be a key associated node. The interval between the key associated node and the next adjacent associated node is used as the time feature E of the key associated node, where E = [Ct, Kt], where Ct represents the time when the key associated node appears, and Kt represents the time when the next adjacent associated node appears.

[0031] Collect the time features of all key associated nodes, perform overlapping processing on the time features of each key associated node, and classify the time features with overlapping time into the same category. For time features of the same category, calculate the product of the overlapping time length of the time features and the total number of time features to obtain the priority management coefficient of each type of time feature. The priority management coefficients are numbered according to their numerical values. The numbering results are: v = 1, 2, ..., V, where V represents the total number of priority management coefficients.

[0032] Based on the priority management coefficient, a time management matrix for the target advertisement is generated, and the overlapping time periods corresponding to the priority management coefficient numbered v are stored in the vth quadrant of the time management matrix;

[0033] Based on the priority management coefficients of various time characteristics of key related nodes, a time management matrix for target ads is generated. This facilitates direct viewing of the effective play time periods of target ads and intuitive analysis of the effectiveness index of each play time period. This facilitates direct adjustment of the play time of target ads based on the time management matrix, facilitating the rapid generation of strategy optimization plans without having to analyze the causes of abnormalities in the target ad delivery strategy one by one based on user feedback data.

[0034] S303: When the target advertisement delivery strategy needs to be optimized, the adaptability between the optimized time Ut of the target advertisement delivery strategy and the time management matrix is ​​analyzed. The specific analysis method is:

[0035] Calculate the difference Qv between Ut and the start time of the overlapping time period stored in the vth quadrant, and eliminate the differences with a calculated result less than 0. The quadrant corresponding to the minimum difference minQv is recorded as the target quadrant, and the target quadrant is numbered s, s = 1, 2, ..., V. Calculate the product between 1 / s and exp(-minQv) to obtain the fitness Pt between Ut and the time management matrix at time t, where min represents the minimum value symbol;

[0036] S304: When Pt> the set threshold and the conversion index-time curve L2 shows a downward trend in the [Ut-h, Ut] time period, the target advertisement strategy optimization object is: the carrier characteristics of the target advertisement;

[0037] When Pt> the set threshold and the conversion index-time curve L2 shows an upward or horizontal trend in the [Ut-h, Ut] time period, the target advertising strategy optimization object is: the quality or after-sales service of the product displayed in the target advertisement;

[0038] When Pt is less than the set threshold and the conversion index-time curve L2 shows a downward trend in the [Ut-h, Ut] time period, the target advertising strategy optimization objects are: the carrier characteristics and delivery time of the target advertising;

[0039] When Pt is less than the set threshold and the conversion index-time curve L2 shows an upward or horizontal trend within the time period [Ut-h, Ut], the strategic optimization objects of the target advertising are: the target advertising delivery time and the quality or after-sales service of the product displayed in the target advertising.

[0040] Furthermore, the method of generating the target advertisement delivery optimization strategy in S40 is as follows:

[0041] When the target advertisement's strategy optimization object is the target advertisement's carrier feature, the carrier feature with a low heat index existing in the user click interface is replaced with the carrier feature with a high heat index;

[0042] When the strategic optimization target of the target advertisement is the quality or after-sales service of the product displayed in the target advertisement, add quality or after-sales improvement comparison content in the target advertisement;

[0043] When the target advertisement strategy optimization object is the target advertisement delivery time, the target advertisement delivery time is controlled within the overlapping time periods stored in the first and second quadrants according to the time management matrix;

[0044] According to the strategy optimization objects of the target advertisement determined in S304 and the strategy optimization contents corresponding to each strategy optimization object, a delivery optimization strategy for the target advertisement is generated.

[0045] An intelligent monitoring system for advertising delivery based on multi-source data fusion, the system includes an advertising delivery feature generation module, an optimization index analysis and prediction module, a strategy optimization object determination module, and an advertising delivery optimization strategy generation module;

[0046] The advertisement delivery feature generation module is used to collect the window quantization value of the target advertisement on the user click interface when the user clicks the target advertisement, obtain the carrier features existing in the user click interface, analyze the historical delivery data of the associated advertisements generated on each carrier feature, combine the user density of the target advertisement in a specific environment, and generate the real-time delivery features of the target advertisement;

[0047] The optimization index analysis and prediction module is used to find target related nodes and analyze the optimization index of the target advertising delivery strategy;

[0048] The strategy optimization object determination module is used to classify and optimize the management analysis of the time characteristics of key related nodes, and generate a time management matrix for the target advertisement. According to the adaptation between the optimized time of the target advertisement delivery strategy and the time management matrix, the strategy optimization object of the target advertisement is determined;

[0049] The delivery optimization strategy generation module is used to generate a delivery optimization strategy for the target advertisement;

[0050] Furthermore, the advertisement delivery feature generation module includes a window quantization value calculation unit, a conversion index calculation unit, a user density calculation unit and an advertisement delivery feature generation unit;

[0051] When a user clicks on a target advertisement, the window quantization value calculation unit calculates the window quantization value of the target advertisement based on the proportion of the target advertisement window in the user click interface compared to the user click interface and the weight value corresponding to the window position of the target advertisement;

[0052] The conversion index calculation unit analyzes the average conversion rate of the associated advertisements generated by the carrier features present on the user click interface, the popularity of each carrier feature on the Internet, and the historical delivery data of the target advertisement without the carrier feature, to obtain the conversion index of the user click interface to the target advertisement;

[0053] The user density calculation unit automatically collects the user's geographic location when the user authorizes and the user does not exit the target advertisement within a time threshold, and calculates the user density of the target advertisement based on the result of determining whether the user's geographic location is within the sales and distribution range of the product displayed in the target advertisement;

[0054] The advertisement delivery feature generation unit generates real-time delivery features of the target advertisement according to the user density, the window quantization value and the conversion index.

[0055] Furthermore, the optimization index analysis and prediction module includes a target associated node search unit, an optimization factor analysis unit and an optimization index prediction unit;

[0056] The target associated node search unit uses the intersection of the user density-time curve L1 and the conversion index-time curve L2 of the target advertisement as the associated node, and searches for the target associated node according to the change trend of the curve L1 within a monitoring period determined according to the associated node;

[0057] The optimization factor analysis unit analyzes the optimization factor of the target advertising delivery strategy at the end of the monitoring period based on the change trend of the curve L2;

[0058] The optimization index prediction unit predicts the optimization index of the target advertisement delivery strategy according to the optimization factor of the target advertisement delivery strategy at the end of the monitoring period and the window quantization value of the target advertisement at the end of the monitoring period.

[0059] Furthermore, the strategy optimization object determination module includes an optimization judgment unit, a time management matrix generation unit, a fitness analysis unit and a strategy optimization object determination unit;

[0060] The optimization judgment unit judges whether the target advertisement delivery strategy needs to be optimized according to the optimization index of the target advertisement delivery strategy;

[0061] The time management matrix generating unit classifies the time features of the key associated nodes according to the overlap of the time features of the key associated nodes, calculates the priority management coefficient of each time feature according to the overlapping time length of each time feature and the total number of time features, and generates a time management matrix for the target advertisement based on the calculation results;

[0062] When the target advertisement delivery strategy needs to be optimized, the adaptability analysis unit analyzes the adaptability between the optimized time of the target advertisement delivery strategy and the time management matrix;

[0063] The strategy optimization object determination unit determines the strategy optimization object of the target advertisement according to the fitness analysis result and the change trend of the curve L2.

[0064] Furthermore, the delivery optimization strategy generation module includes an optimization strategy analysis unit and an optimization strategy generation unit;

[0065] The optimization strategy analysis unit is used to analyze the strategy optimization content of each strategy optimization object of the target advertisement;

[0066] The optimization strategy generating unit generates a delivery optimization strategy for the target advertisement according to the strategy optimization content analysis result of the target advertisement and the strategy optimization object determined by the strategy optimization object determining unit.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] 1. The present invention intelligently monitors the real-time delivery effect of the target advertisement based on the window quantization value, conversion index and user density of the target advertisement. This process ensures the authenticity and validity of the collection results by limiting the collection conditions of the user's geographic location. In addition, this process monitors the delivery effect of the target advertisement through the heat index of the carrier characteristics existing in the user click interface and the average conversion of each carrier characteristic to the associated advertisement, thereby reflecting the impact of the advertisement content on the delivery effect from the root. There is no need to analyze the quality of the advertisement content based on the user's viewing of the target advertisement, thereby improving the system's monitoring accuracy and efficiency of the advertisement delivery effect.

[0069] 2. The present invention identifies target association nodes and key association nodes through the intersection and change trends of the user density-time curve and the conversion index-time curve, and analyzes the real-time optimization factors of the target advertising delivery strategy through the target association nodes to realize the analysis of users' acceptance of the target advertisements. Compared with traditional analysis methods, the present invention reduces the analysis process of user interaction, residence time, etc., which is conducive to quickly capturing the real-time delivery effect of the target advertisement. The time management matrix of the target advertisement is generated through the key association nodes, which can intuitively obtain the priority of each effective delivery time period of the target advertisement, facilitates the rapid adjustment of the delivery time of the target advertisement, and thus reduces the delivery cost of the target advertisement.

[0070] 3. The present invention locks the strategy optimization object of the target advertisement according to the adaptability between the optimized time of the target advertisement delivery strategy and the time management matrix, as well as the changing trend of the curve L2. Based on the locking result, a new delivery optimization strategy is generated to achieve accurate delivery of the target advertisement. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a schematic diagram of the workflow of an intelligent monitoring method for advertising placement based on multi-source data fusion according to the present invention;

[0072] Figure 2 Schematic diagram of the time management matrix of the present invention. DETAILED DESCRIPTION

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0074] Example: Figure 1-Figure 2 As shown, the present invention provides a system and method for intelligently monitoring advertising delivery based on multi-source data fusion, and a method for intelligently monitoring advertising delivery based on multi-source data fusion, the method comprising:

[0075] S10: When a user clicks a target ad, a window quantization value of the target ad on the user's click interface is collected, carrier features of the user's click interface are obtained, historical delivery data of associated ads generated on each carrier feature is analyzed, and real-time delivery features of the target ad are generated based on the user density of the target ad in a specific environment.

[0076] The S10 includes:

[0077] S101: When a user clicks a target ad, collect the window quantization value T of the target ad on the user's click interface, where T = (S1 / S2) × γ, where S1 represents the window area of ​​the target ad, S2 represents the area of ​​the user's click interface, and γ represents the weight value corresponding to the window position of the target ad, where γ = 1 / (1 + the shortest distance between the target ad's window position and the user's key operation position). The key operation position refers to the area where the user performs core functions or high-frequency operations.

[0078] S102: Acquire carrier features existing in the user-clicked interface. Carrier features refer to text or images recorded in the area of ​​the user-clicked interface other than the target ad window. Number each acquired carrier feature. The numbering result is: i = 1, 2, ..., n, where n represents the total number of acquired carrier features.

[0079] Collect the average conversion rate gi of the associated ads generated on carrier feature i. If the product promoted in ad A is of the same type as the product promoted in the target ad, then ad A is called the associated ad of the target ad. Conversion rate = (number of conversions / number of ad exposures or clicks) × 100%. The average conversion rate refers to first summing the conversions of each associated ad, then dividing the sum by the number of associated ads, and calculating the ratio fi between gi-z and 1-exp(u-ri). Sum all fi from i=1 to i=n, ​​and record the sum as The result is F. The ratio between F and n is calculated to obtain the conversion index Z of the user clicking on the interface to the target advertisement, where ri represents the average popularity index of the carrier feature i on the Internet, z represents the average conversion rate of the target advertisement without carrier features, u represents the average popularity index of the target advertisement without carrier features, exp() represents the index with e as the base and e=2.73, popularity index = (search volume × 50% + discussion volume × 30% + dissemination volume × 20%) × interaction rate. The calculation principle of the average popularity index is the same as that of the average conversion rate.

[0080] S103: When a user clicks on a target ad, a pop-up window requesting the user's geographic location is displayed on the target ad interface. If the user authorizes the request and the user does not exit the target ad within a time threshold, the user's geographic location is automatically collected. The time threshold is calculated as the target ad playback duration / 5. If the collected user's geographic location is within the sales and delivery range of the product displayed in the target ad, the corresponding user is marked. Otherwise, the user is not marked. The ratio of the marked users to the total number of collected user geographic locations is calculated to obtain the user density Y of the target ad.

[0081] Generate the delivery feature Kt of the target ad at time t, Kt={Tt,Zt,Yt}, where Yt represents the user density of the target ad at time t, Tt represents the window quantization value of the target ad at time t, and Zt represents the conversion index of the user clicking on the interface to the target ad at time t;

[0082] S20: Finding target related nodes based on the real-time delivery characteristics of the target advertisement, and analyzing the optimization index of the target advertisement delivery strategy;

[0083] The S20 includes:

[0084] S201: Generate a user density-time curve L1 and a conversion index-time curve L2 for the target advertisement, record the intersection of the two curves as an associated node, and number each associated node in chronological order. The numbering result is: p = 1, 2, ..., q, where q represents the total number of associated nodes.

[0085] In the time period [pt, pt+h], if the curve L1 shows a downward trend, and the curve direction includes a downward trend, an upward trend, and a horizontal trend, then the associated node p is considered to be the target associated node, where pt represents the appearance time of the associated node p and h represents the monitoring interval;

[0086] S202: Based on the user density-time curve L1, collect the user density Y(pt) and Y(pt+h) at the time pt and pt+h respectively, and calculate the difference Y between Y(pt) and Y(pt+h). pt→pt+h Perform calculations;

[0087] In the [pt, pt+h] time period, when the conversion index-time curve L2 shows a downward trend:

[0088] Based on the conversion index-time curve L2, the conversion index Z(pt) and Z(pt+h) are collected at the time pt and pt+h respectively, and the difference Z between Z(pt) and Z(pt+h) is calculated. pt→pt+h Perform calculations;

[0089] If 1-exp(Y pt→pt+h ) and 1-exp(Z pt→pt+h ) is within the error range, then the optimization factor of the target advertising delivery strategy at the time pt+h is W(pt+h)=1-exp(k pt→pt+h );

[0090] In the [pt, pt+h] time period, when the conversion index-time curve L2 shows an upward or horizontal trend:

[0091] The optimization factor of the target advertising strategy is W(pt+h)=1-exp[-(K pt→pt+h -k pt→pt+h )], where K pt→pt+h represents the slope of curve L2 in the time period [pt, pt+h], k pt→pt+h Represents the slope of curve L1 in the time period [pt, pt+h];

[0092] S203: Calculate the product of W(pt+h) and the window quantization value T(pt+h) of the target advertisement at time pt+h to obtain the optimization index J(pt+h) of the target advertisement delivery strategy at time pt+h;

[0093] S30: Classifying and optimizing the time characteristics of key related nodes, and generating a time management matrix for the target advertisement. Determining the target advertisement's strategy optimization object based on the compatibility between the optimized time of the target advertisement delivery strategy and the time management matrix.

[0094] The S30 includes:

[0095] S301: When 0.6≤J(pt+h)≤1, it indicates that the target advertisement delivery strategy needs to be optimized at time pt+h; when 0≤J(pt+h)<0.6, it indicates that the target advertisement delivery strategy does not need to be optimized at time pt+h;

[0096] S302: In the time period [pt, pt+h], if both curves L1 and L2 show an upward trend, the associated node p is considered to be a key associated node. The interval between the key associated node and the next adjacent associated node is used as the time feature E of the key associated node. The two endpoints of the interval occur within the same natural day (24 hours). E = [Ct, Kt], where Ct represents the time when the key associated node appears, and Kt represents the time when the next adjacent associated node appears.

[0097] Collect the time features of all key associated nodes, perform overlapping processing on the time features of each key associated node, and classify the time features with overlapping time into the same category. For time features of the same category, calculate the product of the overlapping time length of the time features and the total number of time features to obtain the priority management coefficient of each type of time feature. The priority management coefficients are numbered according to their numerical values. The numbering results are: v = 1, 2, ..., V, where V represents the total number of priority management coefficients.

[0098] Based on the priority management coefficient, a time management matrix for the target advertisement is generated, and the overlapping time periods corresponding to the priority management coefficient numbered v are stored in the vth quadrant of the time management matrix;

[0099] S303: When the target advertisement delivery strategy needs to be optimized, the adaptability between the optimized time Ut of the target advertisement delivery strategy and the time management matrix is ​​analyzed. The specific analysis method is:

[0100] Calculate the difference Qv between Ut and the start time of the overlapping time period stored in the vth quadrant, and eliminate the differences with a calculated result less than 0. The quadrant corresponding to the minimum difference minQv is recorded as the target quadrant, and the target quadrant is numbered s, s = 1, 2, ..., V. Calculate the product between 1 / s and exp(-minQv) to obtain the fitness Pt between Ut and the time management matrix at time t, where min represents the minimum value symbol;

[0101] S304: When Pt> the set threshold and the conversion index-time curve L2 shows a downward trend in the [Ut-h, Ut] time period, the target advertisement strategy optimization object is: the carrier characteristics of the target advertisement;

[0102] When Pt> the set threshold and the conversion index-time curve L2 shows an upward or horizontal trend in the [Ut-h, Ut] time period, the target advertising strategy optimization object is: the quality or after-sales service of the product displayed in the target advertisement;

[0103] When Pt is less than the set threshold and the conversion index-time curve L2 shows a downward trend in the [Ut-h, Ut] time period, the target advertising strategy optimization objects are: the carrier characteristics and delivery time of the target advertising;

[0104] When Pt is less than the set threshold and the conversion index-time curve L2 shows an upward or horizontal trend in the [Ut-h, Ut] time period, the target advertising strategy optimization objects are: the target advertising delivery time and the quality or after-sales service of the product displayed in the target advertisement; the set threshold is set manually;

[0105] S40: generating a delivery optimization strategy for the target advertisement;

[0106] The method for S40 to generate target advertising delivery optimization strategy is as follows:

[0107] When the target advertisement's strategy optimization object is the target advertisement's carrier feature, the carrier feature with a low heat index existing in the user click interface is replaced with the carrier feature with a high heat index;

[0108] When the strategic optimization target of the target advertisement is the quality or after-sales service of the product displayed in the target advertisement, add quality or after-sales improvement comparison content in the target advertisement;

[0109] When the target advertisement strategy optimization object is the target advertisement delivery time, the target advertisement delivery time is controlled within the overlapping time periods stored in the first and second quadrants according to the time management matrix;

[0110] According to the strategy optimization objects of the target advertisement determined in S304 and the strategy optimization contents corresponding to each strategy optimization object, a delivery optimization strategy for the target advertisement is generated.

[0111] An intelligent monitoring system for advertising delivery based on multi-source data fusion, the system includes an advertising delivery feature generation module, an optimization index analysis and prediction module, a strategy optimization object determination module, and an advertising delivery optimization strategy generation module;

[0112] The advertisement delivery feature generation module is used to collect the window quantization value of the target advertisement on the user click interface when the user clicks the target advertisement, obtain the carrier features existing in the user click interface, analyze the historical delivery data of the associated advertisements generated on each carrier feature, combine the user density of the target advertisement in a specific environment, and generate the real-time delivery features of the target advertisement;

[0113] The advertisement delivery feature generation module includes a window quantization value calculation unit, a conversion index calculation unit, a user density calculation unit and an advertisement delivery feature generation unit;

[0114] When a user clicks on a target advertisement, the window quantization value calculation unit calculates the window quantization value of the target advertisement based on the proportion of the target advertisement window in the user click interface compared to the user click interface and the weight value corresponding to the window position of the target advertisement;

[0115] The conversion index calculation unit analyzes the average conversion rate of the associated advertisements generated by the carrier features present on the user click interface, the popularity of each carrier feature on the Internet, and the historical delivery data of the target advertisement without the carrier feature, to obtain the conversion index of the user click interface to the target advertisement;

[0116] The user density calculation unit automatically collects the user's geographic location when the user authorizes and the user does not exit the target advertisement within a time threshold, and calculates the user density of the target advertisement based on the result of determining whether the user's geographic location is within the sales and distribution range of the product displayed in the target advertisement;

[0117] The advertisement delivery feature generation unit generates real-time delivery features of the target advertisement according to the user density, the window quantization value and the conversion index.

[0118] The optimization index analysis and prediction module is used to find target related nodes and analyze the optimization index of the target advertising delivery strategy;

[0119] The optimization index analysis and prediction module includes a target associated node search unit, an optimization factor analysis unit, and an optimization index prediction unit;

[0120] The target associated node search unit uses the intersection of the user density-time curve L1 and the conversion index-time curve L2 of the target advertisement as the associated node, and searches for the target associated node according to the change trend of the curve L1 within the monitoring period determined according to the associated node;

[0121] The optimization factor analysis unit analyzes the optimization factor of the target advertising delivery strategy at the end of the monitoring period based on the change trend of the curve L2;

[0122] The optimization index prediction unit predicts the optimization index of the target advertisement delivery strategy according to the optimization factor of the target advertisement delivery strategy at the end of the monitoring period and the window quantization value of the target advertisement at the end of the monitoring period.

[0123] The strategy optimization object determination module is used to classify and optimize the management analysis of the time characteristics of key related nodes, and generate a time management matrix for the target advertisement. Based on the adaptation between the optimized time of the target advertisement delivery strategy and the time management matrix, the strategy optimization object of the target advertisement is determined;

[0124] The strategy optimization object determination module includes an optimization judgment unit, a time management matrix generation unit, a fitness analysis unit and a strategy optimization object determination unit;

[0125] The optimization judgment unit judges whether the target advertisement delivery strategy needs to be optimized according to the optimization index of the target advertisement delivery strategy;

[0126] The time management matrix generation unit classifies the time features of the key associated nodes according to the overlap of the time features of the key associated nodes, calculates the priority management coefficient of each time feature according to the overlapping time length and the total number of time features of each type of time feature, and generates a time management matrix for the target advertisement based on the calculation results;

[0127] When the target advertisement delivery strategy needs to be optimized, the adaptability analysis unit analyzes the adaptability between the optimized time of the target advertisement delivery strategy and the time management matrix;

[0128] The strategy optimization object determination unit determines the strategy optimization object of the target advertisement based on the fitness analysis result and the change trend of the curve L2;

[0129] The delivery optimization strategy generation module is used to generate a delivery optimization strategy for target advertisements;

[0130] The delivery optimization strategy generation module includes an optimization strategy analysis unit and an optimization strategy generation unit;

[0131] The optimization strategy analysis unit is used to analyze the strategy optimization content of each strategy optimization object of the target advertisement;

[0132] The optimization strategy generating unit generates a delivery optimization strategy for the target advertisement according to the strategy optimization content analysis result of the target advertisement and the strategy optimization object determined by the strategy optimization object determining unit.

[0133] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. An intelligent advertising monitoring method based on multi-source data fusion, characterized by: The method comprises: S101: When a user clicks a target ad, a window quantization value T of the target ad on the user's click interface is collected, where T = (S1 / S2) × γ, where S1 represents the window area of ​​the target ad, S2 represents the area of ​​the user's click interface, and γ represents the weight value corresponding to the window position of the target ad. S102: Acquire carrier features existing in the user-clicked interface. Carrier features refer to text or images recorded in the area of ​​the user-clicked interface other than the target ad window. Number each acquired carrier feature. The numbering result is: i = 1, 2, ..., n, where n represents the total number of acquired carrier features. Collect the average conversion rate gi of the associated ads generated on carrier feature i, calculate the ratio fi between gi-z and 1-exp(u-ri), sum all fi from i=1 to i=n, ​​record the summation result as F, calculate the ratio between F and n, and obtain the conversion index Z of the user clicking on the target ad, where ri represents the average popularity index of carrier feature i on the Internet, z represents the average conversion rate of the target ad when there is no carrier feature, u represents the average popularity index of the target ad when there is no carrier feature, exp() represents the exponent with base e and e=2.73; S103: When a user clicks on a target ad, a pop-up window requesting the user's geographic location is displayed on the target ad interface. If the user authorizes the request and the user does not exit the target ad within a time threshold, the user's geographic location is automatically collected. If the collected user's geographic location is within the sales and delivery range of the product displayed in the target ad, the corresponding user is marked; otherwise, the user is not marked. The ratio of the marked users to the total number of collected user geographic locations is calculated to obtain the user density Y of the target ad. Generate the delivery feature Kt of the target ad at time t, Kt={Tt,Zt,Yt}, where Yt represents the user density of the target ad at time t, Tt represents the window quantization value of the target ad at time t, and Zt represents the conversion index of the user clicking on the interface to the target ad at time t; S20: Finding target related nodes based on the real-time delivery characteristics of the target advertisement, and analyzing the optimization index of the target advertisement delivery strategy. S20 includes: S201: Generate a user density-time curve L1 and a conversion index-time curve L2 for the target advertisement, record the intersection of the two curves as an associated node, and number each associated node in chronological order. The numbering result is: p = 1, 2, ..., q, where q represents the total number of associated nodes. In the time period [pt, pt+h], if the curve L1 shows a downward trend, the associated node p is considered to be the target associated node, where pt represents the appearance time of the associated node p and h represents the monitoring interval; S30: In the time period [pt, pt+h], if both curves L1 and L2 show an upward trend, the associated node p is considered to be a key associated node. The time characteristics of the key associated nodes are classified and optimized and managed. A time management matrix for the target advertisement is generated. Based on the compatibility between the optimized time of the target advertisement delivery strategy and the time management matrix, the target advertisement strategy optimization object is determined. S40: Generate a delivery optimization strategy for the target advertisement.

2. The method for intelligently monitoring advertising placement based on multi-source data fusion according to claim 1, characterized in that: The S20 further includes: S202: Based on the user density-time curve L1, collect the user density Y(pt) and Y(pt+h) at the time pt and pt+h respectively, and calculate the difference Y between Y(pt) and Y(pt+h). pt→pt+h Perform calculations; In the [pt, pt+h] time period, when the conversion index-time curve L2 shows a downward trend: Based on the conversion index-time curve L2, the conversion index Z(pt) and Z(pt+h) are collected at the time pt and pt+h respectively, and the difference Z between Z(pt) and Z(pt+h) is calculated. pt→pt+h Perform calculations; If 1-exp(Y pt→pt+h ) and 1-exp(Z pt→pt+h ) is within the error range, then the optimization factor of the target advertising delivery strategy at the time pt+h is W(pt+h)=1-exp(k pt→pt+h ); In the [pt, pt+h] time period, when the conversion index-time curve L2 shows an upward or horizontal trend: The optimization factor of the target advertising strategy is W(pt+h)=1-exp[-(K pt→pt+h -k pt→pt+h )], where K pt→pt+h represents the slope of curve L2 in the time period [pt, pt+h], k pt→pt+h Represents the slope of curve L1 in the time period [pt, pt+h]; S203: Calculate the product of W(pt+h) and the window quantization value T(pt+h) of the target advertisement at the time pt+h to obtain the optimization index J(pt+h) of the target advertisement delivery strategy at the time pt+h.

3. The method for intelligently monitoring advertising placement based on multi-source data fusion according to claim 2, characterized in that: The S30 includes: S301: When 0.6≤J(pt+h)≤1, it indicates that the target advertisement delivery strategy needs to be optimized at time pt+h; when 0≤J(pt+h)<0.6, it indicates that the target advertisement delivery strategy does not need to be optimized at time pt+h; S302: The interval time between the key associated node and the next adjacent associated node is used as the time feature E of the key associated node, where E=[Ct, Kt], where Ct represents the time when the key associated node appears, and Kt represents the time when the next adjacent associated node appears. Collect the time features of all key associated nodes, perform overlapping processing on the time features of each key associated node, and classify the time features with overlapping time into the same category. For time features of the same category, calculate the product of the overlapping time length of the time features and the total number of time features to obtain the priority management coefficient of each type of time feature. The priority management coefficients are numbered according to their numerical values. The numbering results are: v = 1, 2, ..., V, where V represents the total number of priority management coefficients. Based on the priority management coefficient, a time management matrix for the target advertisement is generated, and the overlapping time periods corresponding to the priority management coefficient numbered v are stored in the vth quadrant of the time management matrix; S303: When the target advertisement delivery strategy needs to be optimized, the adaptability between the optimized time Ut of the target advertisement delivery strategy and the time management matrix is ​​analyzed. The specific analysis method is: Calculate the difference Qv between Ut and the start time of the overlapping time period stored in the vth quadrant, and eliminate the differences with a calculated result less than 0. The quadrant corresponding to the minimum difference minQv is recorded as the target quadrant, and the target quadrant is numbered s, s = 1, 2, ..., V. Calculate the product between 1 / s and exp(-minQv) to obtain the fitness Pt between Ut and the time management matrix at time t, where min represents the minimum value symbol; S304: When Pt> the set threshold and the conversion index-time curve L2 shows a downward trend in the [Ut-h, Ut] time period, the target advertisement strategy optimization object is: the carrier characteristics of the target advertisement; When Pt> the set threshold and the conversion index-time curve L2 shows an upward or horizontal trend in the [Ut-h, Ut] time period, the target advertising strategy optimization object is: the quality or after-sales service of the product displayed in the target advertisement; When Pt is less than the set threshold and the conversion index-time curve L2 shows a downward trend in the [Ut-h, Ut] time period, the target advertising strategy optimization objects are: the carrier characteristics and delivery time of the target advertising; When Pt is less than the set threshold and the conversion index-time curve L2 shows an upward or horizontal trend within the time period [Ut-h, Ut], the strategic optimization objects of the target advertising are: the target advertising delivery time and the quality or after-sales service of the product displayed in the target advertising.

4. The method for intelligently monitoring advertising placement based on multi-source data fusion according to claim 3, characterized in that: The method of generating the target advertisement delivery optimization strategy in S40 is as follows: When the target advertisement's strategy optimization object is the target advertisement's carrier feature, the carrier feature with a low heat index existing in the user click interface is replaced with the carrier feature with a high heat index; When the strategic optimization target of the target advertisement is the quality or after-sales service of the product displayed in the target advertisement, add quality or after-sales improvement comparison content in the target advertisement; When the target advertisement strategy optimization object is the target advertisement delivery time, the target advertisement delivery time is controlled within the overlapping time periods stored in the first and second quadrants according to the time management matrix; According to the strategy optimization objects of the target advertisement determined in S304 and the strategy optimization contents corresponding to each strategy optimization object, a delivery optimization strategy for the target advertisement is generated.

5. An intelligent monitoring system for advertising placement based on multi-source data fusion, applied to the intelligent monitoring method for advertising placement based on multi-source data fusion according to any one of claims 1 to 4, characterized in that: The system includes an advertisement delivery feature generation module, an optimization index analysis and prediction module, a strategy optimization object determination module, and a delivery optimization strategy generation module; The advertisement delivery feature generation module is used to collect the window quantization value of the target advertisement on the user click interface when the user clicks the target advertisement, obtain the carrier features existing in the user click interface, analyze the historical delivery data of the associated advertisements generated on each carrier feature, combine the user density of the target advertisement in a specific environment, and generate the real-time delivery features of the target advertisement; The optimization index analysis and prediction module is used to find target related nodes and analyze the optimization index of the target advertising delivery strategy; The strategy optimization object determination module is used to classify and optimize the management analysis of the time characteristics of key related nodes, and generate a time management matrix for the target advertisement. According to the adaptation between the optimized time of the target advertisement delivery strategy and the time management matrix, the strategy optimization object of the target advertisement is determined; The delivery optimization strategy generation module is used to generate a delivery optimization strategy for a target advertisement.

6. The intelligent monitoring system for advertising placement based on multi-source data fusion according to claim 5, characterized in that: The advertisement delivery feature generation module includes a window quantization value calculation unit, a conversion index calculation unit, a user density calculation unit and an advertisement delivery feature generation unit; When a user clicks on a target advertisement, the window quantization value calculation unit calculates the window quantization value of the target advertisement based on the proportion of the target advertisement window in the user click interface compared to the user click interface and the weight value corresponding to the window position of the target advertisement; The conversion index calculation unit analyzes the average conversion rate of the associated advertisements generated by the carrier features present on the user click interface, the popularity of each carrier feature on the Internet, and the historical delivery data of the target advertisement without the carrier feature, to obtain the conversion index of the user click interface to the target advertisement; The user density calculation unit automatically collects the user's geographic location when the user authorizes and the user does not exit the target advertisement within a time threshold, and calculates the user density of the target advertisement based on the result of determining whether the user's geographic location is within the sales and distribution range of the product displayed in the target advertisement; The advertisement delivery feature generation unit generates real-time delivery features of the target advertisement according to the user density, the window quantization value and the conversion index.

7. The intelligent monitoring system for advertising placement based on multi-source data fusion according to claim 6, characterized in that: The optimization index analysis and prediction module includes a target associated node search unit, an optimization factor analysis unit and an optimization index prediction unit; The target associated node search unit uses the intersection of the user density-time curve L1 and the conversion index-time curve L2 of the target advertisement as the associated node, and searches for the target associated node according to the change trend of the curve L1 within a monitoring period determined according to the associated node; The optimization factor analysis unit analyzes the optimization factor of the target advertising delivery strategy at the end of the monitoring period based on the change trend of the curve L2; The optimization index prediction unit predicts the optimization index of the target advertisement delivery strategy according to the optimization factor of the target advertisement delivery strategy at the end of the monitoring period and the window quantization value of the target advertisement at the end of the monitoring period.

8. The intelligent monitoring system for advertising placement based on multi-source data fusion according to claim 7, characterized in that: The strategy optimization object determination module includes an optimization judgment unit, a time management matrix generation unit, a fitness analysis unit and a strategy optimization object determination unit; The optimization judgment unit judges whether the target advertisement delivery strategy needs to be optimized according to the optimization index of the target advertisement delivery strategy; The time management matrix generating unit classifies the time features of the key associated nodes according to the overlap of the time features of the key associated nodes, calculates the priority management coefficient of each time feature according to the overlapping time length of each time feature and the total number of time features, and generates a time management matrix for the target advertisement based on the calculation results; When the target advertisement delivery strategy needs to be optimized, the adaptability analysis unit analyzes the adaptability between the optimized time of the target advertisement delivery strategy and the time management matrix; The strategy optimization object determination unit determines the strategy optimization object of the target advertisement according to the fitness analysis result and the change trend of the curve L2.

9. The intelligent monitoring system for advertising placement based on multi-source data fusion according to claim 8, characterized in that: The delivery optimization strategy generation module includes an optimization strategy analysis unit and an optimization strategy generation unit; The optimization strategy analysis unit is used to analyze the strategy optimization content of each strategy optimization object of the target advertisement; The optimization strategy generating unit generates a delivery optimization strategy for the target advertisement according to the strategy optimization content analysis result of the target advertisement and the strategy optimization object determined by the strategy optimization object determining unit.

Images