Multi-dimensional attribution analysis method and system for cross-media advertisement putting effect

Abstract

The invention discloses a multi-dimensional attribution analysis method and system for a cross-media advertisement putting effect, relates to the technical field of data analysis and advertisement, and is used for solving the problems that the advertisement conversion path recognition accuracy is reduced and the putting strategy optimization strategy of an attribution result is low. The method comprises the following steps: collecting contacts, time sequences and advertisement creative information of a user in a multimedia channel through a point burying technology, classifying according to a channel mapping rule, calculating the conversion contribution degree and propagation efficiency of each channel in combination with historical conversion data, obtaining an influence weight, constructing a user cross-channel access path diagram, and counting node switching frequency. And a multi-layer perceptron model is adopted to calculate a key node coupling coefficient, extract a nonlinear interaction relationship, fuse the coupling coefficient, a time sequence and advertisement originality, extract an attribution variable based on principal component analysis, generate a structured attribution result, and realize refined optimization of an advertisement strategy.

Description

Technical Field

[0001] This invention relates to the fields of data analysis and advertising technology, and more specifically, to a multi-dimensional attribution analysis method and system for cross-media advertising effectiveness. Background Technology

[0002] Multi-dimensional attribution analysis methods and systems for cross-media advertising effectiveness can comprehensively evaluate the performance and impact of advertising across multiple media channels. With the rapid development of digital marketing, it has gradually become a research hotspot in the field of advertising technology. However, existing attribution analysis techniques still show certain limitations in dealing with cross-media advertising effectiveness, especially in multi-dimensional data integration, dynamic path analysis, and modeling of complex user behavior patterns. There is room for improvement in their ability to support the needs of precise advertising in modern times.

[0003] The existing technology has the following shortcomings:

[0004] Currently, existing cross-media advertising attribution analysis methods suffer from static problems in user touchpoint path modeling, lack of a quantitative mechanism for the coupling relationship between key nodes in different media channels, and difficulty in integrating multi-source heterogeneous information such as time series and advertising creatives for unified modeling. This leads to reduced accuracy in advertising conversion path identification and insufficient optimization effect of attribution results on advertising placement strategies. Therefore, a multi-dimensional attribution analysis method and system for cross-media advertising placement effect is proposed.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a multi-dimensional attribution analysis method and system for cross-media advertising effectiveness. By constructing a multi-dimensional user touchpoint information model, a media channel influence weight evaluation model, a dynamic switching path modeling mechanism, and a key node coupling analysis method, the method further integrates time series information and advertising creative content to form a multi-source collaborative attribution analysis result to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a multi-dimensional attribution analysis method for cross-media advertising effectiveness, comprising the following steps:

[0008] Step S1: Collect cross-media advertising data to obtain user touchpoints, time series, and ad creative information;

[0009] Step S2: Divide user touchpoints into different media channels, combine historical data to statistically analyze the conversion contribution and dissemination efficiency of each media channel, and calculate the influence weight of each media channel;

[0010] Step S3: Use dynamic path modeling to generate the user's switching path between different media channels, record the key nodes in the switching path and calculate the conversion probability, and analyze the coupling coefficient between each key node;

[0011] Step S4: Combine the coupling coefficient with time series and ad creative information to generate attribution analysis results, and formulate ad placement optimization strategies based on the attribution analysis results.

[0012] In a preferred embodiment, in step S1, the advertising delivery data includes user touchpoints, time series, and advertising creative information;

[0013] User touchpoints are user clicks and interactions on different media.

[0014] The time series represents the time interval between each adjacent user touchpoint;

[0015] Advertising creative information includes advertising type and dissemination method.

[0016] In a preferred embodiment, in step S2, user touchpoints are mapped to media channels according to a preset media channel mapping rule.

[0017] Media channels are categorized into social media, search engines, and video platforms;

[0018] The media channel mapping rules match the domain name field of the touchpoint location to identify the media channel to which the user touchpoint belongs.

[0019] In a preferred embodiment, in step S2, historical data for each type of media channel are collected, including conversion contribution and dissemination efficiency;

[0020] Conversion contribution is the ratio of the number of conversion events achieved by user touchpoints in each media channel to the total number of user touchpoints.

[0021] Dissemination efficiency is calculated as the proportion of user touchpoints on each media channel to the total number of user touchpoints.

[0022] The weighted average of conversion contribution and dissemination efficiency is used as the influence weight.

[0023] In a preferred embodiment, in step S3, the user's cross-media switching operation is recorded, and the user's switching path graph is constructed. The path nodes are media channels, and the directed edges represent the jumping behavior between media channels.

[0024] The ratio of the number of jumps from one path node to another to the total number of times a path node is used as the starting node for jumps is the transition probability.

[0025] The transition probabilities between any two path nodes in the switching path graph are collected as input features to construct a multilayer perceptron model for nonlinear relationship learning.

[0026] In a preferred embodiment, in step S3, the multilayer perceptron model includes an input layer, a hidden layer, and an output layer;

[0027] The input layer accepts the transition probability as input.

[0028] The hidden layer models nonlinear relationships based on the transition probability using an activation function;

[0029] The output layer generates the coupling coefficients between path nodes.

[0030] In a preferred embodiment, in step S4, the feature matrix required for principal component analysis is constructed based on the coupling coefficient, time series, and advertising creative information.

[0031] Standardize the feature matrix X so that the mean of each feature dimension is 0 and the variance is 1;

[0032] Principal component analysis was applied to reduce the dimensionality of the standardized feature matrix to obtain the covariance matrix.

[0033] By finding the eigenvalues ​​and eigenvectors of the covariance matrix, the main variation directions in the original high-dimensional feature space are extracted, and the principal component matrix is ​​obtained.

[0034] By analyzing the principal component matrix and identifying the main influencing factors in the advertising path through principal component loading analysis, attribution analysis results are obtained.

[0035] In a preferred embodiment, in step S4, an advertising optimization strategy is formulated based on the attribution analysis results, including channel optimization, content optimization, timing optimization, and deployment of combined strategies.

[0036] The multi-dimensional attribution analysis system for cross-media advertising performance includes a data acquisition module, a media classification module, a path modeling module, and an attribution analysis module. The functions of each module are as follows:

[0037] The data acquisition module is used to collect cross-media advertising data, and obtain user touchpoints, time series, and advertising creative information;

[0038] The media classification module divides user touchpoints into different media channels, combines historical data to statistically analyze the conversion contribution and dissemination efficiency of each media channel, and calculates the influence weight of each media channel.

[0039] The path modeling module uses dynamic path modeling to generate user switching paths between different media channels, records key nodes in the switching paths and calculates the switching probability, and analyzes the coupling coefficient between each key node.

[0040] The attribution analysis module integrates the coupling coefficient with time series and ad creative information to generate attribution analysis results, and formulates ad placement optimization strategies based on the attribution analysis results.

[0041] The technical effects and advantages of this invention are as follows:

[0042] This invention utilizes tracking technology to collect user touchpoints, time series data, and advertising creative information across multimedia channels such as social media, search engines, and video platforms. User touchpoints are categorized according to media channel mapping rules. Historical conversion data is combined to calculate the conversion contribution and dissemination efficiency of each media channel. The influence weight of each channel is obtained through a weighted approach. Based on this, a cross-channel media access path graph is constructed, the switching frequency between nodes is statistically analyzed, and the coupling coefficient between key nodes is calculated using a multilayer perceptron model. This captures the nonlinear interaction relationships between different nodes in the path. The coupling coefficient, behavioral time series data, and advertising creative information are used to extract key attribution variables through principal component analysis, generating structured attribution results. Based on principal component loadings, the advertising strategy is refined at the channel, content, and timing levels. This method comprehensively reflects the impact mechanism of different advertising contact factors on the final conversion, improving the accuracy of advertising placement. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating the multi-dimensional attribution analysis method for cross-media advertising effectiveness according to the present invention.

[0044] Figure 2 This is a schematic diagram of the structure of the multi-dimensional data fusion module in the multi-dimensional attribution analysis system for cross-media advertising performance of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Example 1

[0047] Please see Figure 1 The multi-dimensional attribution analysis method for cross-media advertising effectiveness includes the following steps:

[0048] Step S1: Collect cross-media advertising data to obtain user touchpoints, time series, and ad creative information;

[0049] Step S2: Divide user touchpoints into different media channels, combine historical data to statistically analyze the conversion contribution and dissemination efficiency of each media channel, and calculate the influence weight of each media channel;

[0050] Step S3: Use dynamic path modeling to generate the user's switching path between different media channels, record the key nodes in the switching path and calculate the conversion probability, and analyze the coupling coefficient between each key node;

[0051] Step S4: Combine the coupling coefficient with time series and ad creative information to generate attribution analysis results, and formulate ad placement optimization strategies based on the attribution analysis results.

[0052] The specific implementation is as follows:

[0053] In step S1, cross-media advertising data is collected, including user touchpoints, time series, and advertising creative information.

[0054] The system uses tracking technology to fully record user interactions on various media platforms. The tracking system deployed on the media platforms captures user interactions in real time. The tracking system includes a script module embedded in the webpage. The script module automatically collects user interaction records when the user interacts. The user interaction records include user ID, timestamp, behavior type, and touch point location.

[0055] Among them, User ID is an identifier used to identify a unique user, Timestamp is the specific time when the user's behavior occurred, Behavior Type is the type of interaction behavior, including click, browse and like, Touchpoint Location is the ad location that identifies the interaction behavior, and the format is the domain field of the ad.

[0056] Combine the user ID with the corresponding touchpoint location to construct a sequence as the user touchpoint;

[0057] Time series are obtained by calculating the time interval between consecutive user interaction behaviors. Specifically, behavior types are grouped according to user ID, and the grouped data are sorted in ascending order according to timestamp. Then, the time interval between adjacent interaction behaviors is calculated to form a time series.

[0058] By calling web crawlers to capture the structured fields of advertising pages and analyzing the format of advertising pages, the collection of advertising creative information is achieved. The advertising creative information includes advertising type and dissemination form. Among them, advertising types include product, brand and event, and dissemination forms include text embedding, image banner and video display.

[0059] It should be noted that event tracking technology refers to the technique of recording and uploading user behavior data within a system by embedding code snippets at specific user behavior trigger locations in web pages or mobile applications. Web crawlers are automated program modules that access, parse, and extract information from specified web page content based on preset rules. A domain name is a character-based address used on the Internet to identify a specific website or network service resource.

[0060] In step S2, based on user touchpoints, the user touchpoints are mapped to media channels according to preset media channel mapping rules. The media channels are divided into social media, search engines and video platforms according to the media type of advertising.

[0061] The implementation of media channel mapping rules is based on the domain name field of the touchpoint location. By matching the fields, the media channel to which the user touchpoint belongs is identified. User touchpoints containing social media domain names are classified as social media channels, user touchpoints containing search engine domain names are classified as search engine channels, and user touchpoints containing video platform domain names are classified as video platform channels.

[0062] Based on historical conversion data, the conversion contribution and dissemination efficiency of each media channel were statistically calculated.

[0063] To determine conversion contribution, we count the total number of conversion events achieved through user touchpoints on each media channel. A conversion event is defined as an event where a user completes the actual advertising requirement after engaging in interactive behavior. Simultaneously, we count the total number of user touchpoints on each media channel. Dividing the total number of conversion events achieved through user touchpoints on each media channel by the total number of user touchpoints on each media channel yields the conversion contribution of each media channel. The conversion contribution represents the conversion effectiveness of user touchpoints on each media channel and is used to measure the actual contribution of each media channel to the final advertising conversion.

[0064] Based on user touchpoint locations and interaction behaviors, identify users' cross-channel switching paths, count the total number of user touchpoints across all media channels, divide the number of user touchpoints of a single user across all media channels by the total number of user touchpoints of that user across all media channels to obtain the dissemination efficiency. The dissemination efficiency is the dissemination frequency of each media channel, that is, the dissemination capability and dissemination compactness of each media channel among users.

[0065] After calculating the conversion contribution and dissemination efficiency, the influence weight of each media channel is comprehensively calculated based on the normalization results. The influence weight is defined as the weighted average of the conversion contribution and dissemination efficiency. The influence weight of each media channel is calculated as follows:

[0066] W k =α·C k +β·Ek ;

[0067] Among them, W k To influence weighting, it is used to measure the overall influence of each media channel in the advertising delivery path, C k E is the contribution to the conversion. k For propagation efficiency, α and β represent the weighted influence of conversion contribution and propagation efficiency, respectively, satisfying α+β=1, and taking values ​​in the range [0,1].

[0068] In step S3, a media access sequence is constructed by grouping users by ID and sorting them by time using timestamps. In the media access sequence, each change of media channel constitutes a cross-media switching operation.

[0069] The media channels in the media access sequence are deduplicated, that is, adjacent identical channels are merged into one access, and each switching behavior is recorded to construct a switching path graph of the user between multiple channels. The path nodes are media channels, and the directed edges represent the jumping behavior between media channels.

[0070] The frequency of occurrence of path nodes and the number of directed edges between path nodes are statistically analyzed, i.e., the hopping behavior between path nodes. For any two path nodes, the transition probability is defined as the ratio of the number of hops from one path node to the total number of times a path node is used as the starting node for a hop. The specific calculation formula is as follows:

[0071]

[0072] Among them, P(M x →M y N(M) represents the transition probability, used to measure the likelihood that a user will switch from one media channel to another during their behavioral path. x →M y ) represents the user's path from path node M in the user path graph. x Jump to M y The number of times, N(M) x ) represents path node M x The total number of jump start nodes;

[0073] After completing the user path graph construction and obtaining the transition probabilities between each path node, the coupling relationship between the path nodes is further modeled and analyzed. The transition probabilities between any two path nodes in the switching path graph are collected as input features to construct a multilayer perceptron model for nonlinear relationship learning and output the coupling coefficient between the path nodes.

[0074] The transition probabilities are used as feature vectors input into a multilayer perceptron model for modeling and analysis. The multilayer perceptron model includes an input layer, at least one hidden layer, and an output layer. The input layer accepts a single real-valued input representing the transition probability; the hidden layer models nonlinear relationships using activation functions; and the output layer generates coupling coefficients between path nodes.

[0075] The network structure is defined as follows:

[0076] The input layer has a dimension of 1, corresponding to a single input feature;

[0077] The number of hidden layers is not fixed, each layer contains neurons, and the ReLU function is used as the activation function to enhance nonlinear modeling capabilities;

[0078] The output layer outputs real values, i.e., path node M. i With path node M j The coupling coefficient between them is specifically calculated using the following expression:

[0079] C i,j =σ(W n ·h n +b n );

[0080] Among them, C i,j For path node M i With path node M j The coupling coefficient between them, h n W represents the output of the last hidden layer. n b is the weight parameter. n σ is the bias term, and σ is the Sigmoid function, which ensures that the output coupling coefficient is in the interval [0,1].

[0081] It should be noted that a multilayer perceptron is a feedforward neural network structure, consisting of an input layer, one or more hidden layers, and an output layer connected sequentially. Each layer contains several neurons, and information is transmitted between neurons in different layers through fully connected connections. The activation function is a non-linear mapping function in the neural network, which acts on the linearly weighted input values ​​received by the neurons to enhance the network's ability to non-linearly express input features, thereby improving the model's fitting effect on complex mapping relationships. The ReLU function is a commonly used activation function.

[0082] In step S4, based on coupling coefficients, time series, and advertising creative information, principal component analysis is performed on key factors in the advertising conversion path to extract representative attribution variables, generate attribution analysis results, and formulate advertising optimization strategies accordingly.

[0083] First, in the completed user switching path graph across multiple channels, the feature information contained in each switching path is uniformly structured to form a feature matrix required for attribution analysis. Each row in the feature matrix corresponds to a complete cross-media advertising behavior path of a user and includes the following feature items:

[0084] The elements of the coupling coefficient matrix represent the coupling strength between each node in the path, which are the coupling coefficients calculated by the multilayer perceptron model in step S3.

[0085] Time series represents the time interval sequence between adjacent interactive behaviors in a user behavior path;

[0086] Advertising creative information includes advertising type and dissemination format. This type of information is processed by one-hot encoding to form a quantitative feature vector.

[0087] The above features are concatenated column-wise to construct a feature matrix. Where m represents the total number of user path switching, and d represents the total number of feature dimensions. The feature matrix X is standardized so that the mean of each feature dimension is 0 and the variance is 1.

[0088] Principal component analysis (PCA) is applied to reduce the dimensionality of the standardized feature matrix. PCA extracts the main directions of variation in the original high-dimensional feature space by finding the eigenvalues ​​and eigenvectors of the covariance matrix, forming linearly independent principal components. The formula for calculating the principal components is as follows:

[0089] Z = X·W;

[0090] Where X is the standardized feature matrix, The eigenvector matrix composed of the principal component directions Let k be the principal component matrix after dimensionality reduction, and k be the number of principal components.

[0091] The obtained principal component matrix Z is analyzed. Each principal component represents the comprehensive factors affecting the advertising conversion effect. By analyzing the principal component loadings, that is, the weights of the original variables in the principal components, the main influencing factors in the advertising path are identified, which is the attribution analysis result.

[0092] For example, if the coupling coefficient class variable has a high loading in a principal component, it indicates that path coupling contributes significantly to the transformation; if the time interval class variable has a high loading, it indicates that the timeliness of the behavior is a key factor.

[0093] Based on the attribution analysis results, an advertising optimization strategy is formulated, which specifically includes:

[0094] Channel optimization: Media channels with high coupling coefficients and frequent occurrences of corresponding paths in the principal component loads are marked as high-value channels and prioritized for advertising budget allocation;

[0095] Content optimization: For types and formats with high ad creative variable loadings in the principal components, prioritize creating new ad content that is consistent with that type of creative;

[0096] Rhythm optimization: Based on the time interval characteristics of high loads in the principal components of the time series, adjust the frequency of ad pushes and the rhythm of user contact to match the high conversion efficiency period;

[0097] Deployment of combined strategies: Utilize the aforementioned optimization results to construct a multi-dimensional optimization model, forming a combined optimization strategy for advertising channels, creative content, and delivery time.

[0098] By using the above method, principal component analysis is used to model multidimensional attribution variables, extract potential factors that are representative of advertising conversion, and finally generate structured and interpretable attribution analysis results. Based on this, refined decision-making and optimized deployment of the advertising process can be achieved.

[0099] It should be noted that Principal Component Analysis (PCA) is a linear transformation method used for dimensionality reduction of high-dimensional feature space data. It aims to extract several orthogonal and independent principal component variables through linear combination while preserving the main information of the original data. One-hot coding is a coding method that converts categorical variables into numerical feature vectors, used to introduce non-numerical discrete variables into mathematical models. The direction of variation refers to the direction in which the data exhibits the maximum variance in the feature space during PCA, that is, the sample points are most dispersed along this direction. This will not be elaborated upon here.

[0100] Example 2: A multi-dimensional attribution analysis system for cross-media advertising effectiveness, such as... Figure 2 As shown, a multi-dimensional attribution analysis method for achieving cross-media advertising effectiveness includes a data acquisition module, a media classification module, a path modeling module, and an attribution analysis module. The modules are interconnected by signals, and their functions are as follows:

[0101] The data acquisition module is used to collect cross-media advertising data, and obtain user touchpoints, time series, and advertising creative information;

[0102] The media classification module divides user touchpoints into different media channels, combines historical data to statistically analyze the conversion contribution and dissemination efficiency of each media channel, and calculates the influence weight of each media channel.

[0103] The path modeling module uses dynamic path modeling to generate user switching paths between different media channels, records key nodes in the switching paths and calculates the switching probability, and analyzes the coupling coefficient between each key node.

[0104] The attribution analysis module integrates the coupling coefficient with time series and ad creative information to generate attribution analysis results, and formulates ad placement optimization strategies based on the attribution analysis results.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0106] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a…" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0108] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0109] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-dimensional attribution analysis method for cross-media advertising effectiveness, characterized by: Includes the following steps: Step S1: Collect cross-media advertising data to obtain user touchpoints, time series, and advertising creative information; Step S2: Divide user touchpoints into different media channels, combine historical data to statistically analyze the conversion contribution and dissemination efficiency of each media channel, and calculate the influence weight of each media channel; Step S3: Use dynamic path modeling to generate the user's switching path between different media channels, record the key nodes in the switching path and calculate the conversion probability, and analyze the coupling coefficient between each key node; Step S4: Combine the coupling coefficient with time series and ad creative information to generate attribution analysis results, and formulate ad placement optimization strategies based on the attribution analysis results.

2. The multi-dimensional attribution analysis method for cross-media advertising effectiveness according to claim 1, characterized in that: In step S1, the advertising delivery data includes user touchpoints, time series, and advertising creative information; User touchpoints are user clicks and interactions on different media. The time series represents the time interval between each adjacent user touchpoint; Advertising creative information includes advertising type and dissemination method.

3. The multi-dimensional attribution analysis method for cross-media advertising effectiveness according to claim 1, characterized in that: In step S2, user touchpoints are mapped to media channels according to preset media channel mapping rules; Media channels are categorized into social media, search engines, and video platforms; The media channel mapping rules match the domain name field of the touchpoint location to identify the media channel to which the user touchpoint belongs.

4. The multi-dimensional attribution analysis method for cross-media advertising effectiveness according to claim 1, characterized in that: In step S2, historical data for each type of media channel is collected, including conversion contribution and dissemination efficiency; Conversion contribution is the ratio of the number of conversion events achieved by user touchpoints in each media channel to the total number of user touchpoints. Dissemination efficiency is calculated as the proportion of user touchpoints on each media channel to the total number of user touchpoints. The weighted average of conversion contribution and dissemination efficiency is used as the influence weight.

5. The multi-dimensional attribution analysis method for cross-media advertising effectiveness according to claim 2, characterized in that: In step S3, the user's cross-media switching operation is recorded, and the user's switching path graph is constructed. The path nodes are media channels, and the directed edges represent the jumping behavior between media channels. The ratio of the number of jumps from one path node to another to the total number of times a path node is used as the starting node for jumps is the transition probability. The transition probabilities between any two path nodes in the switching path graph are collected as input features to construct a multilayer perceptron model for nonlinear relationship learning.

6. The multi-dimensional attribution analysis method for cross-media advertising effectiveness according to claim 5, characterized in that: In step S3, the multilayer perceptron model includes an input layer, a hidden layer, and an output layer; The input layer accepts the transition probability as input. The hidden layer models nonlinear relationships based on the transition probability using an activation function; The output layer generates the coupling coefficients between path nodes.

7. The multi-dimensional attribution analysis method for cross-media advertising effectiveness according to claim 6, characterized in that: In step S4, the feature matrix required for principal component analysis is constructed based on the coupling coefficient, time series, and advertising creative information. Standardize the feature matrix X so that the mean of each feature dimension is 0 and the variance is 1; Principal component analysis was applied to reduce the dimensionality of the standardized feature matrix to obtain the covariance matrix. By finding the eigenvalues ​​and eigenvectors of the covariance matrix, the main variation directions in the original high-dimensional feature space are extracted, and the principal component matrix is ​​obtained. By analyzing the principal component matrix and identifying the main influencing factors in the advertising path through principal component loading analysis, attribution analysis results are obtained.

8. The multi-dimensional attribution analysis method for cross-media advertising effectiveness according to claim 7, characterized in that: In step S4, based on the attribution analysis results, an advertising optimization strategy is formulated, including channel optimization, content optimization, timing optimization, and deployment of combined strategies.

9. A multi-dimensional attribution analysis system for cross-media advertising performance, used to implement the multi-dimensional attribution analysis method for cross-media advertising performance as described in any one of claims 1-8, characterized in that: It includes a data acquisition module, a media classification module, a path modeling module, and an attribution analysis module. The functions of each module are as follows: The data acquisition module is used to collect cross-media advertising data, and obtain user touchpoints, time series, and advertising creative information; The media classification module divides user touchpoints into different media channels, combines historical data to statistically analyze the conversion contribution and dissemination efficiency of each media channel, and calculates the influence weight of each media channel. The path modeling module uses dynamic path modeling to generate user switching paths between different media channels, records key nodes in the switching paths and calculates the switching probability, and analyzes the coupling coefficient between each key node. The attribution analysis module integrates the coupling coefficient with time series and ad creative information to generate attribution analysis results, and formulates ad placement optimization strategies based on the attribution analysis results.

Images