Cross-application advertisement behavior information application tracking method, device and equipment

Abstract

The invention relates to a cross-application advertisement behavior information application tracking method, device and equipment, and the method comprises the steps: responding to an action of cross-application backup advertisement behavior information, and transmitting application action information for indicating to report the application advertisement behavior information to a second application program; constructing a feature vector according to the application action information and the advertisement behavior information; according to the feature vector, generating an application tracking fingerprint corresponding to the second application program, adding the application tracking fingerprint to the advertisement behavior information, and storing the application tracking fingerprint; backing up the advertisement behavior information added with the application tracking fingerprint to the second application program through a pixel code of the second application program, responding to an application action of the advertisement behavior information, and obtaining application action information corresponding to the application action through the application tracking fingerprint; and generating a matching fingerprint for the application action according to the application action information, and matching the matching fingerprint with the application tracking fingerprint to obtain an application tracking result for the application of the advertisement behavior information.

Description

Technical Field

[0001] The present disclosure relates to the field of advertising technology, and in particular to a method, device and equipment for tracking cross-application advertising behavior information. Background Art

[0002] Tracking technology for advertising behavior information is crucial for advertisers. It can help target audiences more accurately and improve the effectiveness and conversion rate of advertising. However, with the popularity of mobile devices and the frequent multitasking, users may see an ad in one app and then conduct related searches or purchases in another app. Through business cooperation, cross-app advertising behavior tracking can provide more accurate user portraits and achieve more accurate advertising push.

[0003] However, cross-app advertising behavior information tracking has also raised concerns about user privacy protection. Users may want to know how their behavior is used across apps. Summary of the invention

[0004] The purpose of the present invention is to provide a method, device and equipment for tracking advertising behavior information across applications, aiming to enable users to know the technical problem of how the advertising behavior information is specifically used by other applications when the advertising behavior information is backed up across applications.

[0005] In order to achieve the above-mentioned purpose, a first aspect of an embodiment of the present disclosure provides a method for tracking cross-application advertising behavior information, which is applied to a first application, wherein a pixel code for a second application to back up the advertising behavior information from the first application across applications is added to a hypertext transfer protocol of a target advertising playback page of the first application, and the method includes: In response to the action of the second application backing up the advertising behavior information from the target advertisement playback page of the first application across applications through the pixel code, sending application action information to the second application instructing the second application to report the advertising behavior information; receiving the application action information reported by the second application, and constructing a feature vector of the second application according to the application action information and the advertising behavior information reported by the second application; generating the application tracking fingerprint corresponding to the second application according to the feature vector of the second application, adding the application tracking fingerprint to the advertising behavior information backed up by the second application, and saving the application tracking fingerprint corresponding to the second application; The advertising behavior information with the application tracking fingerprint added thereto is backed up to the second application through the pixel code of the second application, and in response to an application action of the second application with respect to the advertising behavior information, application action information corresponding to the application action is obtained through the application tracking fingerprint; A matching fingerprint for the application action of the second application is generated according to the application action information, and the matching fingerprint is matched with the application tracking fingerprint to obtain an application tracking result for the second application for the application of the advertising behavior information.

[0006] In a possible implementation, constructing a feature vector of the second application according to the application action information and the advertising behavior information reported by the second application includes: According to the created advertising behavior vocabulary, each advertising behavior term in the advertising behavior information is converted as a text into a vector, so as to obtain a plurality of advertising behavior term vectors corresponding to the advertising behavior information, wherein the advertising behavior vocabulary stores binary tags in advertising behavior terms without duplication; Arrange the multiple advertising behavior vectors corresponding to the advertising behavior information into a two-dimensional matrix according to the preset advertising behavior terms sequence, wherein the ordinate in the two-dimensional matrix represents different advertising behavior terms, and the abscissa represents the advertising behavior vector of each advertising behavior term; vectorizing the application action information reported by the second application program to obtain an application action vector for each application action entry in the application action information; The advertising behavior vectors of each of the advertising behavior terms and the application action vectors of each of the application action terms in the two-dimensional matrix are normalized and then concatenated to obtain a feature vector of the second application program.

[0007] In a possible implementation, the step of normalizing and concatenating the advertising behavior vectors of each of the advertising behavior terms and the application action vectors of each of the application action terms in the two-dimensional matrix to obtain the feature vector of the second application program includes: Calculating the mean and standard deviation of the advertising behavior vector of each of the advertising behavior terms and the application action vector of each of the application action terms in the two-dimensional matrix; According to the mean and the standard deviation, Z-score normalization is performed on the advertising behavior vector of each advertising behavior term and the application action vector of each application action term; The advertisement behavior vector of each of the advertisement behavior terms and the application action vector of each of the application action terms after Z score standardization are concatenated to obtain a feature vector of the second application program.

[0008] In a possible implementation, in response to the application action of the second application program with respect to the advertising behavior information, obtaining application action information corresponding to the application action through the application tracking fingerprint includes: receiving a report request sent by the second application, wherein the report request is generated by the second application according to the application tracking fingerprint in response to an application action for the advertising behavior information, and sent to the first application according to a first application identifier corresponding to the application tracking fingerprint; Parsing the report request to obtain specific API endpoint information and the application tracking fingerprint carried in the report request for obtaining application action information corresponding to the application action; Comparing the application tracking fingerprint obtained by the analysis with the saved application tracking fingerprint, and sending a packaging instruction to the second application program instructing to package the application action information corresponding to the application action when the comparison result indicates that the two fingerprints match; When feedback information after packaging of the second application program is received, application action information corresponding to the packaged application action is obtained through the specific API endpoint represented by the specific API endpoint information.

[0009] In a possible implementation, generating the application tracking fingerprint corresponding to the second application according to the feature vector of the second application includes: Take the first 32 decimal places of the square roots of multiple prime numbers preset in the mathematical constants to determine the initial hash value; When the number of bits of the feature vector of the second application is padded to reach a preset number of bits, dividing the padded feature vector into a plurality of vector blocks of preset bytes; Decomposing each of the vector blocks corresponding to the feature vector into a plurality of message blocks of a preset length to obtain a basic message word in a hash algorithm; Arranging the information in the application action information of the second application in sequence to generate a specific vector, and determining the extended message words corresponding to the remaining words in the hash algorithm according to the values ​​at the multiple specified positions in the specific vector; Combining the multiple basic message words corresponding to each of the vector blocks with the extended message word respectively to generate hash constants of the multiple message blocks corresponding to each of the vector blocks; Iteratively calculating the hash initial value and the hash constants of the plurality of message blocks corresponding to each of the vector blocks to obtain a hash value of each of the message blocks; The application tracking fingerprint corresponding to the second application is generated according to the hash values ​​of the plurality of message blocks corresponding to each of the vector blocks.

[0010] In a possible implementation, determining, according to the values ​​at the multiple specified positions in the specific vector, the extended message words corresponding to the remaining multiple words in the hash algorithm includes: Calculating square roots of the values ​​at the multiple designated positions in the specific vector to obtain a first value at a preset multiple decimal places corresponding to each of the designated positions; Convert the numerical values ​​at the preset multiple decimal places corresponding to each of the designated positions into binary values ​​represented by binary; According to the sequence of the designated positions, the binary values ​​corresponding to the designated positions are sequentially used as the extended message words corresponding to the remaining multiple words in the hash algorithm.

[0011] In one possible implementation, the application action includes at least one of the following: user portrait construction action, advertisement matching action, product matching action, advertisement behavior expansion action, applicable population matching action, behavior pattern recognition action, and activity statistics action.

[0012] In a possible implementation, the application action information includes at least one of the following: frequency of using the advertising behavior information, application action of using the advertising behavior information, conditions for using the advertising behavior information, and font of the application.

[0013] According to a second aspect of the embodiments of the present disclosure, there is provided a device for tracking cross-application advertising behavior information, which is applied to a first application, wherein a pixel code for a second application to back up the advertising behavior information from the first application across applications is added to a hypertext transfer protocol of a target advertising playback page of the first application, and the device comprises: a sending module configured to send application action information instructing the second application to report the application's advertising behavior information in response to the second application's action of backing up the advertising behavior information across applications from the target advertising playback page of the first application through the pixel code; a receiving and constructing module, configured to receive the application action information reported by the second application, and construct a feature vector of the second application according to the application action information and the advertising behavior information reported by the second application; a generating and saving module, configured to generate the application tracking fingerprint corresponding to the second application according to the feature vector of the second application, add the application tracking fingerprint to the advertising behavior information backed up by the second application, and save the application tracking fingerprint corresponding to the second application; an acquisition module configured to back up the advertising behavior information with the application tracking fingerprint added to the second application through the pixel code of the second application, and in response to an application action of the second application with respect to the advertising behavior information, acquire application action information corresponding to the application action through the application tracking fingerprint; The generation and matching module is configured to generate a matching fingerprint for the application action of the second application based on the application action information, and match the matching fingerprint with the application tracking fingerprint to obtain an application tracking result for the application of the second application for the advertising behavior information.

[0014] According to a third aspect of the present disclosure, an electronic device is provided, including: a memory having a computer program stored thereon; A processor is used to execute the computer program in the memory to implement the steps of any one of the methods in the first aspect.

[0015] The present invention provides a method, device and apparatus for tracking cross-application advertising behavior information. Compared with the prior art, it has the following beneficial effects: By backing up advertising behavior information across applications, users' related behaviors across applications can be accurately tracked. Cross-application backup means that the evaluation of advertising effectiveness is no longer limited to a single application, thereby improving the accuracy and comprehensiveness of the evaluation. Feature vectors are constructed based on the user's behavior in different applications, and corresponding application tracking fingerprints are generated. When the advertising behavior information is used across applications, the application tracking fingerprint can inform the application that is the source of the advertising behavior information of the application status, so that users can know which applications across applications use the behavior information on the first application, and know how the cross-application applications use the advertising behavior information. This ensures the security of the user's personal data while ensuring that the advertising behavior information is backed up across applications, which helps analyze the user's advertising preferences and consumption habits.

[0016] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings: Figure 1 It is a flow chart of a method for tracking cross-application advertising behavior information applications according to an embodiment of the specification.

[0018] Figure 2It is a block diagram of a device for tracking cross-application advertising behavior information applications according to an embodiment of the specification.

[0019] Figure 3 It is a block diagram of another cross-application advertising behavior information application tracking device shown according to an embodiment of the specification. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0022] The present disclosure provides a cross-application advertising behavior information application tracking method, which is applied to a first application, wherein a second application is added to a hypertext transfer protocol of a target advertising playback page of the first application to back up pixel code of the advertising behavior information from the first application across applications. Figure 1 The flowchart of a method for tracking cross-application advertising behavior information according to an embodiment is shown. The method comprises: In step S11, in response to the action of the second application backing up the advertising behavior information from the target advertising playback page of the first application across applications through the pixel code, application action information instructing the second application to report the advertising behavior information is sent to the second application; Among them, pixel code: also known as tracking pixel, web bug or network beacon, is a small piece of code embedded in a web page or application (usually a 1x1 pixel transparent image), which is used to trigger data tracking or send a request when a user visits the page. It can be used to track users' clicks, browsing, ignoring, skipping, and other behaviors on advertisements.

[0023] In the disclosure, the pixel code of the second application is embedded in the HTTP request of the target ad playback page of the first application. When the page of the first application is accessed, a request for the pixel code is triggered, and the request contains relevant information about the ad behavior (such as ad ID, display time, etc.). For example, a user browses an ad in application A (the first application), and the HTTP request of the ad page contains the pixel code of application B (the second application). When the user loads the ad page, application B receives a request from application A containing ad behavior information.

[0024] Furthermore, the second application performs cross-application backup processing on the advertising behavior information received through the pixel code, which may include storage and analysis. For example, after application B receives the advertising behavior information sent by application A, it stores it in a database for subsequent analysis of whether the user has further interaction in application B (such as jumping to the product page in application B after clicking on the advertisement).

[0025] In step S12, the application action information reported by the second application is received, and a feature vector of the second application is constructed according to the application action information and the advertising behavior information reported by the second application; The feature vector may include features corresponding to the user's actions of clicking, browsing, ignoring, skipping, etc. on the advertisement, as well as information on the application action that the second application wants to use the advertisement behavior information. The application action can be understood as a use action or an application action, that is, how the second application uses, applies, or utilizes the advertisement behavior information.

[0026] In the present disclosure, when a user needs to perform an operation related to the advertising behavior information in the first application in the second application (such as building a user profile based on click, browse, ignore, skip in the advertising behavior information), the second application reports the information related to the operation to the first application. For example, a user sees an advertisement in application A and clicks it, and then application B wants to use the clicked advertisement behavior to build a user profile, so it needs to first inform application A of the information on building the user profile.

[0027] Furthermore, based on the application action information and advertising behavior information reported by the second application, a feature vector of the user is constructed. The feature vector is used to subsequently construct an application tracking fingerprint. For example, after application A receives information that application B wants to use the clicked advertising behavior to construct a user portrait, this information is combined with the clicked advertising behavior to generate a feature vector.

[0028] In step S13, based on the feature vector of the second application, the application tracking fingerprint corresponding to the second application is generated, and the application tracking fingerprint is added to the advertising behavior information backed up by the second application, and the application tracking fingerprint corresponding to the second application is saved; In the present disclosure, a unique application tracking fingerprint is generated based on the feature vector of the second application using one or more hash algorithms (such as SHA-256, MD5, etc.). This fingerprint is calculated based on the feature vector of the second application, so it can uniquely identify the application and its application action on the advertising behavior information. Subsequently, this application tracking fingerprint is associated with the advertising behavior information backed up by the second application, that is, this fingerprint is added to the advertising behavior information, and the advertising behavior information with the fingerprint added is sent to the second application, and the corresponding application tracking fingerprint is saved in the database corresponding to the first application.

[0029] In step S14, the advertising behavior information with the application tracking fingerprint added is backed up to the second application through the pixel code of the second application, and in response to the application action of the second application with respect to the advertising behavior information, application action information corresponding to the application action is obtained through the application tracking fingerprint; In the present disclosure, the advertising behavior information with application tracking fingerprints added is backed up to the second application through the pixel code of the second application. In this way, the second application can use the advertising behavior information generated by the first application to perform operations such as user portrait construction, and after performing these operations, the application action information corresponding to the application action can be reported to the first application through the application tracking fingerprint.

[0030] In step S15, a matching fingerprint of the application action for the second application is generated based on the application action information, and the matching fingerprint is matched with the application tracking fingerprint to obtain an application tracking result for the second application for the application of the advertising behavior information.

[0031] In the present disclosure, a matching fingerprint is generated based on the application action information of the second application using the same hash algorithm (the same as when generating the application tracking fingerprint). This matching fingerprint is calculated based on a specific part of the application action information (such as the name of the action of building a user profile, etc.). Then, this matching fingerprint is matched with the application tracking fingerprint to verify whether the application action is the same as the previously reported application action information. If the match is successful, the application action is considered to be a valid application for the previously reported application action information, and a corresponding application tracking result is generated.

[0032] For example, based on the application action information corresponding to the user's operation of building a user profile using advertising behavior information in application B and the clicked advertising behavior, a matching fingerprint "ABC789" is generated using the SHA-256 algorithm. This matching fingerprint is then matched with the previously generated application tracking fingerprint. Since both fingerprints are calculated based on the clicked advertising behavior and the application action information corresponding to building the user profile, they can be successfully matched. Based on this, it is judged that this application action is a valid application of the previously reported application action information, and the corresponding application tracking result is generated.

[0033] In the embodiment of the present disclosure, when the application tracking result indicates that the application action of using the advertising behavior information across applications has not been obtained, a reminder action can be performed. A corresponding application log can also be generated according to the application tracking result, so that the user can view the log and know how the cross-application application uses the advertising behavior information when the cross-application advertising behavior information is backed up across applications.

[0034] The above technical solution can accurately track the user's related behaviors across applications by backing up advertising behavior information across applications. Cross-application backup makes the evaluation of advertising effectiveness no longer limited to a single application, thereby improving the accuracy and comprehensiveness of the evaluation. A feature vector is constructed based on the user's behavior in different applications, and a corresponding application tracking fingerprint is generated. When the advertising behavior information is used across applications, the application tracking fingerprint can inform the application that is the source of the advertising behavior information of the application status, so that the user can know which applications across applications use the behavior information on the first application, and know how the cross-application applications use the advertising behavior information. On the basis of ensuring that the advertising behavior information is backed up across applications, which helps to analyze the user's advertising preferences and consumption habits, the security of the user's personal data is ensured.

[0035] In a possible implementation, in step S12, constructing a feature vector of the second application according to the application action information and the advertising behavior information reported by the second application includes: In step S121, according to the created advertising behavior vocabulary, each advertising behavior term in the advertising behavior information is converted into a vector as a text, so as to obtain a plurality of advertising behavior term vectors corresponding to the advertising behavior information, wherein the advertising behavior vocabulary stores binary tags in advertising behavior terms without duplication; In the present disclosure, a pre-created Ad Behavior Vocabulary is used to convert each ad behavior term (such as "click on an ad", "browse product details", etc.) in the ad behavior information into a vector. This vocabulary is a set of all possible ad behavior terms, and each term has a unique binary tag in the vocabulary (such as represented by an index). The conversion process usually uses word embedding technology, such as TF-IDF, Word2Vec, etc., but here, since the vocabulary is predefined and relatively small in size, each term can be directly mapped to a fixed-dimensional vector (such as one-hot encoding or a simple index vector).

[0036] For example, suppose the advertising behavior vocabulary contains the following terms: {"click on the ad": 0, "browse product details": 1, "add to cart": 2, "purchase product": 3}. For a piece of advertising behavior information "the user clicked on the ad and browsed product details", the advertising behavior terms "click on the ad" and "browse product details" are first identified, and then converted into vectors [1, 0, 0, 0] and [0, 1, 0, 0] respectively according to the vocabulary.

[0037] In step S122, multiple advertising behavior vectors corresponding to the advertising behavior information are arranged into a two-dimensional matrix according to a preset advertising behavior term sequence, wherein the ordinate in the two-dimensional matrix represents different advertising behavior terms, and the abscissa represents the advertising behavior vector of each advertising behavior term; In the present disclosure, the vectors corresponding to all advertising behavior terms are arranged into a two-dimensional matrix according to the preset advertising behavior term order (i.e., the order in the vocabulary). The ordinate of this matrix represents different advertising behavior terms, and the abscissa corresponds to the advertising behavior vector of each term. In this way, the system can integrate a series of discrete advertising behavior information into a structured data representation, which is convenient for subsequent feature matching and tracking analysis.

[0038] Continuing with the above example, suppose there is a vocabulary containing four advertising behavior terms, and a corresponding vector has been generated for each term in a piece of advertising behavior information. Then, these vectors can be arranged into a 4xN two-dimensional matrix in the order of the vocabulary (i.e. "click on the ad", "browse product details", "add to shopping cart", "purchase product"), where N is the dimension of the vector.

[0039] In step S123, the application action information reported by the second application is vectorized to obtain an application action vector of each application action entry in the application action information; In the present disclosure, the application action information reported by the second application is vectorized. The purpose of this step is to convert unstructured text or event information into a structured numerical vector. Vectorization can be based on a predefined application action vocabulary, which contains all possible application action terms and assigns a unique identifier or index to each term.

[0040] For each application action term, a variety of methods can be used for vectorization, such as one-hot encoding, TF-IDF, word embedding, etc. However, since the application action vocabulary is usually much smaller than the vocabulary in natural language processing and the action types are relatively fixed, a simple one-hot encoding or index vector may be sufficient.

[0041] For example, suppose the application action vocabulary contains the following entries: {"name of user portrait construction": 0, "frequency of user portrait construction": 1, "font of user portrait construction": 2}. For the application action "construct user portrait", the application action entries "name of user portrait construction", "frequency of user portrait construction", and "font of user portrait construction" are first identified, and then converted into vectors [1, 0, 0, 0], [0, 1, 0, 0], and [0,0, 1, 0] respectively according to the vocabulary.

[0042] In step S124, the advertising behavior vectors of each of the advertising behavior terms and the application action vectors of each of the application action terms in the two-dimensional matrix are normalized and then concatenated to obtain a feature vector of the second application program.

[0043] In the present disclosure, the advertising behavior term vector obtained in step S122 and the application action term vector obtained in step S123 are standardized. Standardization can help eliminate the dimensional differences between different features, so that each feature is at the same numerical level, thereby improving the accuracy and efficiency of subsequent analysis.

[0044] There are many standardization methods, the most common ones are Min-Max Scaling, Z-score standardization (also known as standard deviation standardization), etc. After standardization, the advertising behavior term vector and the application action term vector are concatenated in a certain order to form a longer vector, which is the feature vector of the second application. This feature vector combines the application's advertising behavior information and the user's application action information, and can fully reflect the behavior characteristics of the application.

[0045] Continuing with the above example, there is a two-dimensional matrix for advertising behavior terms (for simplicity, only one user's behavior is considered here, so the matrix degenerates into a vector) and a vector for application action terms. The advertising behavior vector may be [1, 1, 0, 0] (corresponding to the behaviors of "clicking on an ad" and "browsing an ad"), and the application action vector is [1, 1, 1, 0] (corresponding to "the name of the user portrait", "the frequency of building a user portrait", and "the font of the user portrait"). In the standardization stage, since these vectors are already one-hot encoded or similar simple forms, additional numerical transformations may not be required. Finally, these two vectors are concatenated to obtain the feature vector of the second application [1, 1, 0, 0, 1, 1, 1, 0].

[0046] In a possible implementation, in step S124, the advertising behavior vectors of each of the advertising behavior terms and the application action vectors of each of the application action terms in the two-dimensional matrix are normalized and then concatenated to obtain a feature vector of the second application, including: In step S1241, the means and standard deviations of the advertising behavior vectors of the advertising behavior terms and the application action vectors of the application action terms in the two-dimensional matrix are calculated; In the present disclosure, the mean and standard deviation of the advertising behavior vectors of all advertising behavior terms and the application action vectors of all application action terms in the two-dimensional matrix are calculated respectively. The mean reflects the central tendency of the data, while the standard deviation measures the degree of dispersion of the data. For example: suppose there is a two-dimensional matrix containing advertising behavior information of multiple users, and a series of application action vectors. For the advertising behavior vectors, the average behavior intensity of all users on each advertising behavior term can be calculated (if the vector is one-hot encoded, it may need to be converted to other forms of representation, such as frequency or duration), as well as the standard deviation of these behavior intensities. Similarly, for the application action vectors, the average number of behaviors or duration of all users on each application action term, as well as the corresponding standard deviation, can also be calculated.

[0047] In step S1242, Z-score normalization is performed on the advertising behavior vector of each advertising behavior term and the application action vector of each application action term according to the mean and the standard deviation; In the present disclosure, after obtaining the mean and standard deviation, the Z-score standardization (also called standard deviation standardization) method is used to standardize the advertising behavior vector of each advertising behavior term and the application action vector of the application action term. After Z-score standardization, the data will conform to the standard normal distribution, that is, the mean is 0 and the standard deviation is 1. In this way, the dimensional differences between different features are eliminated.

[0048] For example: Continuing with the above example, suppose there is an advertising behavior vector [2,3,0,1] (indicating the user's behavior intensity on four advertising behavior terms), with a mean μ=1.5 and a standard deviation σ=1.29. Then, after Z-score normalization, the vector will become [(2−1.5) / 1.29, (3−1.5) / 1.29, (0−1.5) / 1.29, (1−1.5) / 1.29], that is, [0.39, 1.16, −1.16, −0.39]. Similarly, the application action vector will also be normalized in this way.

[0049] In step S1243, the advertising behavior vector of each of the advertising behavior terms and the application action vector of each of the application action terms after Z-score normalization are concatenated to obtain a feature vector of the second application program.

[0050] In the present disclosure, after completing the Z score normalization, the advertising behavior vectors of all advertising behavior terms and the application action vectors of the application action terms are spliced ​​into a longer vector in a certain order, that is, the feature vector of the second application. This feature vector combines the advertising behavior information of the application and the application action information of the user, and can fully reflect the behavior characteristics of the application and the user.

[0051] For example: Assume that after normalization, we get two vectors: the ad behavior vector [0.39, 1.16, −1.16, −0.39] and the application action vector [0.5, −0.8, 1.2, 0] (these values ​​are just examples). Then, concatenate these two vectors to get the feature vector of the second application [0.39, 1.16, −1.16, −0.39, 0.5, −0.8, 1.2, 0].

[0052] In a possible implementation, in step S14, in response to the application action of the second application program with respect to the advertising behavior information, obtaining application action information corresponding to the application action through the application tracking fingerprint includes: In step S141, receiving a report request sent by the second application, wherein the report request is generated by the second application in response to an application action for the advertising behavior information according to the application tracking fingerprint, and sent to the first application according to a first application identifier corresponding to the application tracking fingerprint; In the present disclosure, when an application action associated with advertising behavior information is executed in the second application, the second application generates a report request according to the preset logic. This request contains key information for identifying user behavior, such as application tracking fingerprint (ATF), and then based on the source of the application tracking fingerprint (ATF), that is, the first application identifier (that is, the source of advertising behavior information and the first application). Then, the second application sends this report request to the first application through a communication protocol (such as HTTP request).

[0053] In step S142, the report request is parsed to obtain specific API endpoint information and the application tracking fingerprint carried in the report request for obtaining application action information corresponding to the application action; In the present disclosure, after receiving the report request, the first application will parse it to extract the specific API endpoint information (API location for subsequent acquisition of application action information) and application tracking fingerprint (ATF) contained therein. For example: the first application parses the report request and obtains the API endpoint information (such as / api / track / action) and ATF (a string of unique identifiers).

[0054] In step S143, the application tracking fingerprint obtained by parsing is compared with the saved application tracking fingerprint, and when the comparison result indicates that the two fingerprints match, a packaging instruction for packaging the application action information corresponding to the application action is sent to the second application program; In the present disclosure, the first application compares the parsed application tracking fingerprint (ATF) with the application tracking fingerprint stored in a local or remote database. If the comparison result shows that the two fingerprints match, then the report request can be considered valid. For example: the server compares the received ATF with the ATF stored in the database, and after confirming that they match, it considers that the report request is legal and prepares to send an encapsulation instruction to the second application.

[0055] In step S144, when feedback information after the packaging of the second application is received, the application action information corresponding to the packaged application action is obtained through the specific API endpoint represented by the specific API endpoint information.

[0056] After confirming that the report request is valid, the first application sends an encapsulation instruction (such as an HTTP response or message) to the second application, requesting the second application to encapsulate the application action information into a specific format and send it back through the specific API endpoint information parsed previously.

[0057] In a possible implementation, in step S13, generating the application tracking fingerprint corresponding to the second application according to the feature vector of the second application includes: In step S131, the first 32 decimal places of the square roots of the preset prime numbers in the mathematical constants are taken to determine the initial hash value; In the present disclosure, multiple prime numbers preset in mathematical constants (such as Fermat numbers, Mersenne primes, etc.) are selected, their square roots are calculated, and the first 32 decimal places are taken as the initial value of the hash algorithm. This selection method utilizes the unique mathematical properties of prime numbers and helps to increase the complexity and unpredictability of hash values.

[0058] In step S132, when the number of bits of the feature vector of the second application is padded to reach a preset number of bits, the padded feature vector is divided into a plurality of vector blocks of preset bytes; In the present disclosure, it is checked whether the feature vector of the second application reaches a preset number of digits. If not, a specific padding method (such as zero padding, random padding, etc.) is used to expand the feature vector to the required number of digits. Then, the padded feature vector is divided into multiple vector blocks according to a preset byte size for subsequent processing.

[0059] For example: Assume that the original feature vector is 128 bits long, but at least 256 bits are required. Use zero padding to expand the feature vector to 256 bits, and then split the 256 bits into four 64-bit vector blocks.

[0060] In step S133, each of the vector blocks corresponding to the feature vector is decomposed into a plurality of message blocks of a preset length to obtain a basic message word in a hash algorithm; In the present disclosure, each vector block is further decomposed into multiple message blocks of preset length (also called basic message words), which will be used as inputs to the hash algorithm. The decomposition method can be simple byte-by-byte or bit-by-bit splitting, depending on the requirements of the hash algorithm. For example: Continuing with the above example, each 64-bit vector block is decomposed into 8 8-bit message blocks, each of which is used as a basic input to the hash algorithm.

[0061] In step S134, each information in the application action information of the second application is arranged in sequence to generate a specific vector, and the extended message words corresponding to the remaining multiple words in the hash algorithm are determined according to the values ​​at multiple specified positions in the specific vector; In the present disclosure, a specific vector is generated according to the application action information of the second application program, and then, multiple values ​​at specified positions are selected from the specific vector as extended message words of the remaining words (i.e., additional inputs other than the basic message words) in the hash algorithm.

[0062] In step S135, the multiple basic message words corresponding to each of the vector blocks are respectively combined with the extended message word to generate hash constants of the multiple message blocks corresponding to each of the vector blocks; In the present disclosure, for each vector block, the system combines the basic message word with the extended message word to generate hash constants of multiple message blocks corresponding to the vector block. The combination can be a simple concatenation or a more complex function mapping. For example: each 8-bit basic message word is concatenated with the corresponding extended message word (which may be one or more numerical values) to form a longer message block as the input of the hash algorithm.

[0063] In step S136, the hash initial value is iterated with the hash constants of the plurality of message blocks corresponding to each of the vector blocks to obtain a hash value of each of the message blocks; In the present disclosure, a hash algorithm (such as MD5, SHA-256, etc.) is used to iteratively calculate the hash initial value and the hash constant of the message block corresponding to each vector block. The specific steps and number of iterative operations depend on the implementation details of the hash algorithm. In the end, each message block will get a corresponding hash value. For example: using the SHA-256 algorithm, the hash initial value is iteratively calculated with the message block of the first vector block to obtain the first hash value. Then, use the first hash value as the new initial value and iteratively calculate with the message block of the second vector block, and so on, until all vector blocks are processed.

[0064] In step S137, the application tracking fingerprint corresponding to the second application is generated according to the hash values ​​of the plurality of message blocks corresponding to each of the vector blocks.

[0065] In the present disclosure, the hash values ​​corresponding to all the vector blocks are merged (such as concatenation, XOR, etc.) to generate a final string or number as the application tracking fingerprint (ATF) corresponding to the second application. This ATF will be used to uniquely identify the application and its behavior. For example, concatenating the hash values ​​corresponding to all the vector blocks into a long string is the application tracking fingerprint corresponding to the second application.

[0066] In a possible implementation, in step S134, determining the extended message words corresponding to the remaining words in the hash algorithm according to the values ​​at the multiple specified positions in the specific vector includes: In step S1341, the square roots of the values ​​at the multiple designated positions in the specific vector are calculated to obtain the first value at the preset multiple decimal places corresponding to each of the designated positions; In the present disclosure, for each value at a specified position in a specific vector, its square root is first calculated. Then, the values ​​at a preset number of decimal places are extracted from the results of these square roots. The square roots and their decimal parts are selected because they can provide another mathematical representation of the original values, which helps to increase the complexity and unpredictability of the data.

[0067] For example: Assume that the square root of a value at a specified position in a specific vector is 3.14159... Then, extract the first N digits of this decimal part (such as N=4) to obtain 0141 as the first value.

[0068] In step S1342, the numerical values ​​at the preset multiple decimal places corresponding to each of the designated positions are converted into binary values ​​represented by binary; In the present disclosure, after obtaining the first numerical value at each specified position, these numerical values ​​are converted into binary values ​​represented in binary. This is because the computer stores and processes data in binary form, and converting the numerical value into a binary value can facilitate the subsequent use in the hash algorithm.

[0069] In step S1343, according to the order of the designated positions, the binary values ​​corresponding to the designated positions are sequentially used as the extended message words corresponding to the remaining multiple words in the hash algorithm.

[0070] In the present disclosure, according to the order of the designated positions, the binary value corresponding to each designated position is sequentially used as the extended message word corresponding to the remaining multiple words in the hash algorithm. These extended message words will participate in the calculation process of the hash value together with the basic message word.

[0071] For example, suppose there are three specified positions in a specific vector, and the corresponding binary values ​​are 1111, 0101, and 1010. According to the order of these positions, these binary values ​​are used as extended message words in sequence. During the execution of the hash algorithm, these extended message words will be processed together with the basic message words (i.e., the message blocks obtained by vector block decomposition) to generate the final hash value.

[0072] In one possible implementation, the application action includes at least one of the following: user portrait construction action, advertisement matching action, product matching action, advertisement behavior expansion action, applicable population matching action, behavior pattern recognition action, and activity statistics action.

[0073] Among them, the user portrait construction action refers to creating one or more detailed descriptions or models of user characteristics by collecting, organizing and analyzing multi-dimensional information of users (such as basic attributes, behavioral data, preferences, consumption habits, etc.).

[0074] In the disclosed embodiment, the data used to construct the user portrait may include data information obtained by the application itself and data information backed up from non-applications across applications, wherein the data information may include, for example, user registration information, browsing history, purchase history, social media activities, etc. The construction of the user portrait may include data cleaning (removing duplicate, erroneous or irrelevant data) and data analysis (using statistical methods, machine learning algorithms, etc.) to extract key features of the user. The constructed user portrait can better understand user needs, thereby providing more personalized services or product recommendations.

[0075] Ad matching refers to the process of matching the most appropriate ads to target users based on the ad behavior information backed up from non-apps across apps. It usually involves building an ad library based on the ad behavior information backed up across apps, classifying and labeling ad content, and matching ads with ad behavior information through algorithms (such as collaborative filtering, content recommendation, etc.). The matching results are based on multiple factors such as user interests, needs, and historical behaviors, and are intended to improve ad click-through rates, conversion rates, and user satisfaction.

[0076] Product matching is similar to ad matching, which is to match the most suitable products to target users based on the ad behavior information backed up from non-apps across apps, but focuses on recommending the most suitable products or services to users. By analyzing the ad behavior information backed up from non-apps across apps, combined with product features, prices, reviews and other information, recommendation algorithms (such as content-based recommendations, collaborative filtering-based recommendations, etc.) are used to generate personalized product recommendation lists for users. This process aims to improve user experience, increase purchase conversion rates, and promote the formation of user loyalty.

[0077] Ad behavior expansion actions predict and expand the types of ads or content that users may be interested in based on ad behavior information backed up from non-apps across apps. By analyzing ad clicks, browsing time, conversion behavior and other data in ad behavior information backed up from non-apps across apps, machine learning models are used to predict users' future ad preferences. Based on these predictions, more relevant or innovative ad content is pushed to users to increase ad exposure and user engagement.

[0078] Applicable population matching is to accurately match specific products, services or information to the target population that is most likely to be interested or benefit. It relies on a deep understanding of the characteristics of the target population, including demographic characteristics, psychological characteristics, behavioral patterns, etc. By building a detailed target population portrait and combining it with product characteristics for matching analysis, companies can more effectively carry out market positioning and promotion, and improve the efficiency and effectiveness of marketing activities.

[0079] The behavior pattern recognition action is to identify the user's unique behavior patterns or habits by analyzing the advertising behavior information backed up from non-apps across applications in the long or short term, or combining the advertising behavior information obtained from the application in the long or short term and the advertising behavior information backed up from non-apps across applications. Behavioral patterns may include, for example, click time, click frequency, and click type. In-depth mining and pattern recognition of user behavior data aims to discover the regularity, periodicity, or abnormality of user behavior. The identified behavior patterns can be used to optimize user experience (such as personalized recommendations, interface optimization), risk assessment (such as fraud detection), and market strategy adjustments (such as pricing strategy, promotion strategy), etc.

[0080] The activity statistics action is an action to collect statistics based on the frequency, time point or duration of the advertisement played on the target advertisement playback page of the first application program, which is indicated in the advertisement behavior information by the user, such as viewing, clicking or ignoring the advertisement.

[0081] In a possible implementation, the application action information includes at least one of the following: frequency of using the advertising behavior information, application action of using the advertising behavior information, conditions for using the advertising behavior information, and font of the application.

[0082] The frequency of using the advertising behavior information is the frequency of the second application using the advertising behavior information backed up across applications. The application actions using the advertising behavior information are the specific actions for which the second application uses the advertising behavior information (e.g., user profile building, advertising matching, product matching, advertising behavior expansion, applicable population matching, behavior pattern recognition, activity statistics). These actions may also include, but are not limited to, filling out forms, etc.

[0083] The condition for using the advertising behavior information refers to the circumstances under which different application actions will use the advertising behavior information backed up from the first application program across applications.

[0084] The embodiment of the present disclosure also provides a device for tracking cross-application advertising behavior information, which is applied to a first application, wherein a pixel code for a second application to back up the advertising behavior information from the first application across applications is added to the hypertext transfer protocol of the target advertising playback page of the first application, see Figure 2 As shown, the device comprises: The sending module 210 is configured to send application action information instructing the second application to report the application's advertising behavior information in response to the second application's action of backing up the advertising behavior information across applications from the target advertising playback page of the first application through the pixel code to the second application; The receiving and constructing module 220 is configured to receive the application action information reported by the second application, and construct a feature vector of the second application according to the application action information and the advertising behavior information reported by the second application; The generating and saving module 230 is configured to generate the application tracking fingerprint corresponding to the second application according to the feature vector of the second application, add the application tracking fingerprint to the advertising behavior information backed up by the second application, and save the application tracking fingerprint corresponding to the second application; The acquisition module 240 is configured to back up the advertising behavior information with the application tracking fingerprint added to the second application through the pixel code of the second application, and in response to the application action of the second application with respect to the advertising behavior information, acquire application action information corresponding to the application action through the application tracking fingerprint; The generation and matching module 250 is configured to generate a matching fingerprint for the application action of the second application based on the application action information, and match the matching fingerprint with the application tracking fingerprint to obtain an application tracking result for the application of the second application for the advertising behavior information.

[0085] In a possible implementation, the receiving and constructing module 220 is configured to: According to the created advertising behavior vocabulary, each advertising behavior term in the advertising behavior information is converted as a text into a vector, so as to obtain a plurality of advertising behavior term vectors corresponding to the advertising behavior information, wherein the advertising behavior vocabulary stores binary tags in advertising behavior terms without duplication; Arrange the multiple advertising behavior vectors corresponding to the advertising behavior information into a two-dimensional matrix according to the preset advertising behavior terms sequence, wherein the ordinate in the two-dimensional matrix represents different advertising behavior terms, and the abscissa represents the advertising behavior vector of each advertising behavior term; vectorizing the application action information reported by the second application program to obtain an application action vector for each application action entry in the application action information; The advertising behavior vectors of each of the advertising behavior terms and the application action vectors of each of the application action terms in the two-dimensional matrix are normalized and then concatenated to obtain a feature vector of the second application program.

[0086] In a possible implementation, the receiving and constructing module 220 is configured to: Calculating the mean and standard deviation of the advertising behavior vector of each of the advertising behavior terms and the application action vector of each of the application action terms in the two-dimensional matrix; According to the mean and the standard deviation, Z-score normalization is performed on the advertising behavior vector of each advertising behavior term and the application action vector of each application action term; The advertisement behavior vector of each of the advertisement behavior terms and the application action vector of each of the application action terms after Z score standardization are concatenated to obtain a feature vector of the second application program.

[0087] In a possible implementation, the acquisition module 240 is configured to: receiving a report request sent by the second application, wherein the report request is generated by the second application according to the application tracking fingerprint in response to an application action for the advertising behavior information, and sent to the first application according to a first application identifier corresponding to the application tracking fingerprint; Parsing the report request to obtain specific API endpoint information and the application tracking fingerprint carried in the report request for obtaining application action information corresponding to the application action; Comparing the application tracking fingerprint obtained by the analysis with the saved application tracking fingerprint, and sending a packaging instruction to the second application program instructing to package the application action information corresponding to the application action when the comparison result indicates that the two fingerprints match; When feedback information after packaging of the second application program is received, application action information corresponding to the packaged application action is obtained through the specific API endpoint represented by the specific API endpoint information.

[0088] In a possible implementation, the generating and saving module 230 is configured to: Take the first 32 decimal places of the square roots of multiple prime numbers preset in the mathematical constants to determine the initial hash value; When the number of bits of the feature vector of the second application is padded to reach a preset number of bits, dividing the padded feature vector into a plurality of vector blocks of preset bytes; Decomposing each of the vector blocks corresponding to the feature vector into a plurality of message blocks of a preset length to obtain a basic message word in a hash algorithm; Arranging the information in the application action information of the second application in sequence to generate a specific vector, and determining the extended message words corresponding to the remaining words in the hash algorithm according to the values ​​at the multiple specified positions in the specific vector; Combining the multiple basic message words corresponding to each of the vector blocks with the extended message word respectively to generate hash constants of the multiple message blocks corresponding to each of the vector blocks; Iteratively calculating the hash initial value and the hash constants of the plurality of message blocks corresponding to each of the vector blocks to obtain a hash value of each of the message blocks; The application tracking fingerprint corresponding to the second application is generated according to the hash values ​​of the plurality of message blocks corresponding to each of the vector blocks.

[0089] In a possible implementation, the generating and saving module 230 is configured to: Calculating square roots of the values ​​at the multiple designated positions in the specific vector to obtain a first value at a preset multiple decimal places corresponding to each of the designated positions; Convert the numerical values ​​at the preset multiple decimal places corresponding to each of the designated positions into binary values ​​represented by binary; According to the sequence of the designated positions, the binary values ​​corresponding to the designated positions are sequentially used as the extended message words corresponding to the remaining multiple words in the hash algorithm.

[0090] In one possible implementation, the application action includes at least one of the following: user portrait construction action, advertisement matching action, product matching action, advertisement behavior expansion action, applicable population matching action, behavior pattern recognition action, and activity statistics action.

[0091] In a possible implementation, the application action information includes at least one of the following: frequency of using the advertising behavior information, application action using the advertising behavior information, conditions for using the advertising behavior information, and font of the application.

[0092] The present disclosure also provides an electronic device, including: a memory having a computer program stored thereon; A processor is used to execute the computer program in the memory to implement the steps of the method in any one of the aforementioned embodiments.

[0093] Figure 3 The cross-application advertising behavior information application tracking device 100 shown includes: a processor 1001 and a memory 1003. Among them, the processor 1001 and the memory 1003 are connected, such as through a bus 1002. Optionally, the cross-application advertising behavior information application tracking device 100 may also include a communication component, which can be used for data interaction between the device 100 and other devices, such as data sending and / or data receiving. It should be noted that in actual scheduling, the communication component is not limited to one, and the structure of the cross-application advertising behavior information application tracking device 100 does not constitute a limitation on the embodiments of the present application.

[0094] Processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 1001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0095] The bus 1002 may include a path to transmit information between the above components. The bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 1002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0096] The memory 1003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), magnetic disk storage medium, other magnetic storage devices, or any other medium that can be used to carry or store program code and can be read by a computer, without limitation herein.

[0097] The memory 1003 is used to store program codes for executing the embodiments of the present disclosure, and the execution is controlled by the processor 1001. The processor 1001 is used to execute the program codes stored in the memory 1003 to implement the steps shown in the embodiment of the above-mentioned cross-application advertising behavior information application tracking method.

[0098] The disclosed embodiments also provide a computer-readable storage medium having program code stored thereon. When the program code is executed by a processor, the steps and corresponding contents of the aforementioned cross-application advertising behavior information application tracking method embodiment can be implemented.

[0099] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments; within the technical concept of the present disclosure, various changes, modifications, substitutions and variations may be made to these embodiments, and these changes, modifications, substitutions and variations all fall within the protection scope of the present disclosure.

[0100] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any appropriate manner without contradiction, and they should also be regarded as the contents disclosed in this disclosure. In order to avoid unnecessary repetition, this disclosure will not further describe various possible combinations. The technical scope of this application is not limited to the contents in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A method for tracking cross-application advertising behavior information, characterized in that: Applied to a first application, wherein a pixel code used by a second application to back up the advertising behavior information from the first application across applications is added to the hypertext transfer protocol of the target advertisement playing page of the first application, and the method comprises: In response to the action of the second application backing up the advertising behavior information from the target advertisement playback page of the first application across applications through the pixel code, sending application action information to the second application instructing the second application to report the advertising behavior information; receiving the application action information reported by the second application, and constructing a feature vector of the second application according to the application action information and the advertising behavior information reported by the second application; generating the application tracking fingerprint corresponding to the second application according to the feature vector of the second application, adding the application tracking fingerprint to the advertising behavior information backed up by the second application, and saving the application tracking fingerprint corresponding to the second application; The advertising behavior information with the application tracking fingerprint added thereto is backed up to the second application through the pixel code of the second application, and in response to an application action of the second application with respect to the advertising behavior information, application action information corresponding to the application action is obtained through the application tracking fingerprint; A matching fingerprint for the application action of the second application is generated according to the application action information, and the matching fingerprint is matched with the application tracking fingerprint to obtain an application tracking result for the second application for the application of the advertising behavior information.

2. The cross-application advertising behavior information application tracking method according to claim 1, characterized in that: The constructing a feature vector of the second application according to the application action information and the advertising behavior information reported by the second application includes: According to the created advertising behavior vocabulary, each advertising behavior term in the advertising behavior information is converted as a text into a vector, so as to obtain a plurality of advertising behavior term vectors corresponding to the advertising behavior information, wherein the advertising behavior vocabulary stores binary tags in advertising behavior terms without duplication; Arrange the multiple advertising behavior vectors corresponding to the advertising behavior information into a two-dimensional matrix according to the preset advertising behavior terms sequence, wherein the ordinate in the two-dimensional matrix represents different advertising behavior terms, and the abscissa represents the advertising behavior vector of each advertising behavior term; vectorizing the application action information reported by the second application program to obtain an application action vector for each application action entry in the application action information; The advertising behavior vectors of each of the advertising behavior terms and the application action vectors of each of the application action terms in the two-dimensional matrix are normalized and then concatenated to obtain a feature vector of the second application program.

3. The cross-application advertising behavior information application tracking method according to claim 2, characterized in that: The step of normalizing and concatenating the advertising behavior vectors of each of the advertising behavior terms and the application action vectors of each of the application action terms in the two-dimensional matrix to obtain the feature vector of the second application program includes: Calculating the mean and standard deviation of the advertising behavior vector of each of the advertising behavior terms and the application action vector of each of the application action terms in the two-dimensional matrix; According to the mean and the standard deviation, Z-score normalization is performed on the advertising behavior vector of each advertising behavior term and the application action vector of each application action term; The advertisement behavior vector of each of the advertisement behavior terms and the application action vector of each of the application action terms after Z score standardization are concatenated to obtain a feature vector of the second application program.

4. The cross-application advertising behavior information application tracking method according to claim 1, characterized in that: The step of obtaining application action information corresponding to the application action of the second application program with respect to the advertising behavior information by using the application tracking fingerprint includes: receiving a report request sent by the second application, wherein the report request is generated by the second application according to the application tracking fingerprint in response to an application action for the advertising behavior information, and sent to the first application according to a first application identifier corresponding to the application tracking fingerprint; Parsing the report request to obtain specific API endpoint information and the application tracking fingerprint carried in the report request for obtaining application action information corresponding to the application action; Comparing the application tracking fingerprint obtained by the analysis with the saved application tracking fingerprint, and sending a packaging instruction to the second application program instructing to package the application action information corresponding to the application action when the comparison result indicates that the two fingerprints match; When feedback information after packaging of the second application program is received, application action information corresponding to the packaged application action is obtained through the specific API endpoint represented by the specific API endpoint information.

5. The cross-application advertising behavior information application tracking method according to claim 1, characterized in that: The step of generating the application tracking fingerprint corresponding to the second application according to the feature vector of the second application includes: Take the first 32 decimal places of the square roots of multiple prime numbers preset in the mathematical constants to determine the initial hash value; When the number of bits of the feature vector of the second application is padded to reach a preset number of bits, dividing the padded feature vector into a plurality of vector blocks of preset bytes; Decomposing each of the vector blocks corresponding to the feature vector into a plurality of message blocks of a preset length to obtain a basic message word in a hash algorithm; Arranging the information in the application action information of the second application in sequence to generate a specific vector, and determining the extended message words corresponding to the remaining words in the hash algorithm according to the values ​​at the multiple specified positions in the specific vector; Combining the multiple basic message words corresponding to each of the vector blocks with the extended message word respectively to generate hash constants of the multiple message blocks corresponding to each of the vector blocks; Iteratively calculating the hash initial value and the hash constants of the plurality of message blocks corresponding to each of the vector blocks to obtain a hash value of each of the message blocks; The application tracking fingerprint corresponding to the second application is generated according to the hash values ​​of the plurality of message blocks corresponding to each of the vector blocks.

6. The method according to claim 5, characterized in that The step of determining, according to the values ​​at the multiple specified positions in the specific vector, the extended message words corresponding to the remaining multiple words in the hash algorithm comprises: Calculating square roots of the values ​​at the multiple designated positions in the specific vector to obtain a first value at a preset multiple decimal places corresponding to each of the designated positions; Convert the numerical values ​​at the preset multiple decimal places corresponding to each of the designated positions into binary values ​​represented by binary; According to the sequence of the designated positions, the binary values ​​corresponding to the designated positions are sequentially used as the extended message words corresponding to the remaining multiple words in the hash algorithm.

7. The method according to any one of claims 1 to 6, characterized in that The application action includes at least one of the following: user portrait construction action, advertisement matching action, product matching action, advertisement behavior expansion action, applicable population matching action, behavior pattern recognition action, and activity statistics action.

8. The method according to any one of claims 1 to 6, characterized in that The application action information includes at least one of the following: frequency of using the advertising behavior information, application action using the advertising behavior information, conditions for using the advertising behavior information, and a font of the application.

9. A device for tracking cross-application advertising behavior information, characterized in that: Applied to a first application, wherein a pixel code for a second application to back up the advertising behavior information from the first application across applications is added to the hypertext transfer protocol of the target advertisement playing page of the first application, and the device comprises: a sending module configured to send application action information instructing the second application to report the advertising behavior information to the second application in response to the action of the second application backing up the advertising behavior information across applications from the target advertising playback page of the first application through the pixel code; a receiving and constructing module, configured to receive the application action information reported by the second application, and construct a feature vector of the second application according to the application action information and the advertising behavior information reported by the second application; a generating and saving module, configured to generate the application tracking fingerprint corresponding to the second application according to the feature vector of the second application, add the application tracking fingerprint to the advertising behavior information backed up by the second application, and save the application tracking fingerprint corresponding to the second application; an acquisition module configured to back up the advertising behavior information with the application tracking fingerprint added to the second application through the pixel code of the second application, and in response to an application action of the second application with respect to the advertising behavior information, acquire application action information corresponding to the application action through the application tracking fingerprint; The generation and matching module is configured to generate a matching fingerprint for the application action of the second application based on the application action information, and match the matching fingerprint with the application tracking fingerprint to obtain an application tracking result for the application of the second application for the advertising behavior information.

10. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 8.

Images