Advertising audience and delivery path analysis system and method based on big data

Abstract

The present invention discloses an advertising audience and delivery path analysis system and method based on big data, and relates to the field of advertising delivery technology. The system includes a user data collection module, a user interest analysis module and an advertising delivery module that are connected to each other; the user data collection module is used to collect user behavior data through the device used by the user; the user interest analysis module is used to analyze the user's access behavior based on the user behavior data and identify the topics that the user is interested in. The present invention can identify the collected user data, separate the user sessions belonging to the device owner, reduce the impact of the data generated when others use it on the accuracy of the interest analysis of the device owner, and improve the effect of advertising delivery; through two clustering, it is possible to conveniently express the operations performed by the user on a certain page, and improve the accuracy of analyzing the user's special interest in the functional module corresponding to the touch screen gesture in the loading page.

Description

Technical Field

[0001] The present invention relates to the technical field of advertising delivery, and in particular to an advertising audience and delivery path analysis system and method based on big data. Background Art

[0002] In the field of digital marketing, the advertising backend delivery system is an important bridge between advertising companies, brands and advertising platforms. This system not only carries the release and management of advertising content, but also involves multiple links such as data analysis and strategy formulation.

[0003] In the early days, there were few source channels and data volumes for user data, and user portraits were mainly studied based on statistical data. As the industry generated more and more data, leading Internet companies had their own databases. Each company focused on how to apply the data in the database to mine the various browsing behaviors and interest preferences of existing users on the Internet to obtain more refined information-tagged user portraits. Information-tagged user portraits focus on data mining and the construction of a user tag system to achieve precision marketing.

[0004] In the prior art, a device is used by the device owner most of the time. In special circumstances, it may be used by people other than the device owner. If the data generated by others when using the device is not separated when analyzing user interests, the accuracy of the interest analysis of the device owner will be reduced, resulting in poor advertising effectiveness. Summary of the invention

[0005] The purpose of the present invention is to provide an advertising audience and delivery path analysis system and method based on big data to solve the problems raised in the prior art.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solution: an advertising audience and delivery path analysis method based on big data, the method comprising:

[0007] Step S1, collecting user behavior data through the device used by the user;

[0008] Step S2: Analyze the user's access behavior based on the user behavior data to identify topics that the user is interested in;

[0009] Step S3: Determine the advertising audience based on the topics that users are interested in and develop a delivery path.

[0010] The step S1 of collecting user behavior data through the device used by the user includes:

[0011] Step S11: In a user session, a user performs a click action on an application, and the user session D is recorded in a data set containing n objects and stored in a log;

[0012] User session D=<(P 1 , D 1 , T 1 , X 1 , t 1 ),…,(P n , D n , T n , X n , t n ), (P end , D end , T end , X n+1 , t n+1 )>

[0013] Among them, P n Indicates the name of the loaded page, D n Indicates a specific action on the page, including searching, browsing products, adding to shopping cart, submitting orders, completing payments, and canceling orders. n Indicates the loading time of the corresponding page name, X n Indicates the touch screen gesture data under the corresponding page name, t n Indicates the residence time on the page; P end The page indicating the end of the session, D end Indicates the final action taken by the user on the termination page, T end Indicates the time when the session is terminated;

[0014] Step 12: Pre-process the user behavior data. If the dwell time is less than t1, it means that the user clicked by mistake or is not interested in the page, and the behavior data is removed. If the dwell time is greater than t2, it means that the page failed to load or the user has left, and the behavior data is removed.

[0015] Step 13: Reversely select the user session belonging to the device owner based on the touch screen gesture data, and hide the user sessions other than the device owner.

[0016] The step 13 comprises:

[0017] Step 131: judging the state of the user using the device according to the stability of the touch screen gesture data in a user session; if the stability is low, the user is walking to operate the device; if the stability is medium, the user is standing to operate the device; if the stability is high, the user is standing to operate the device;

[0018] Step 132: When the corresponding user is using the device, the touch screen gesture data is modeled: the center point of the display screen of the device is taken as the origin O, and the y-axis and z-axis are drawn through the origin O in the horizontal and vertical directions perpendicular to each other, and the x-axis is drawn through the origin O in the vertical direction perpendicular to the display screen. The values ​​in the y-axis and z-axis directions represent the touch point positions, and the line connecting all the touch point positions (y, z) in a user session is the finger sliding trajectory line. The value in the x-axis direction represents the touch pressure value at the touch point (y, z); the touch point data = (x, y, z), and the line connecting all the touch point data in a user session is the pressure change trajectory.

[0019] The start and end of the finger sliding trajectory are point A and point A1 respectively, and the start and end of the pressure change trajectory are point B and point B1 respectively. Point A and point B, point A1 and point B1 are connected to form a closed graph to represent the geometric user session, and the new geometric user session is compared with the geometric user sessions of users pre-stored in the database for similarity;

[0020] Step 132: The absolute value of the value obtained by subtracting the straight line length L1 at both ends of the finger sliding trajectory line from the straight line length L2 at both ends of the preset standard finger sliding trajectory line is the trajectory difference; the absolute value of the value obtained by subtracting the area S1 of the geometric user session from the area S2 of the preset standard geometric user session is the pressure difference; different weights are assigned to the pressure difference and the trajectory difference and then added to obtain a detection value M; the detection value M1 is compared with the preset detection value threshold {α, β}; user sessions corresponding to the detection value M1 exceeding {α, β} are hidden; and user sessions corresponding to the detection value M1 not exceeding {α, β} are confirmed as user sessions belonging to the device owner.

[0021] The step S2 analyzes the access behaviors of multiple users based on the user behavior data to identify topics of interest in the user cluster, including:

[0022] Step S21: performing first-level session clustering based on the calculated similarity of user sessions;

[0023] Step S22: Based on the first-level session clustering, the user is clustered into a second-level session cluster according to the distribution of all sessions generated in different session clusters of each user.

[0024] The step S21 comprises:

[0025] Step S211: Mark all objects in the user session D of the device owner as unvisited, and preset a cluster S; 1 Select an object as the access object P and mark it as visited;

[0026] Step S212: If the neighborhood E of the access object P contains at least a objects, where a is the neighborhood density threshold, create a new cluster class C i1 , and add the access object P and the objects contained in its neighborhood E to the cluster class C i1 ; Select an object from the user session D 2 as the access object P, obtaining a new cluster class C i2 , and so on;

[0027] If the number of common elements in two cluster classes is not less than k times the number of elements in the two cluster classes, where 0 < k < 1, merge the two cluster classes into a new cluster C, and at the same time add the cluster C to the cluster set S;

[0028] If the number of common elements in two cluster classes is less than k times the number of elements in the two cluster classes, merge the two cluster classes into two independent clusters, and at the same time add the cluster C to the cluster set S.

[0029] The said step S22 includes:

[0030] Step S221: There are w clusters in the cluster set S, select one cluster as the sample Q;

[0031] Step S222: Initialize the membership degree matrix U of the sample Q to the cluster center, and calculate the eigenvector C of the cluster center point i (i = 1, 2, 3...);

[0032] Step S223: Calculate the objective function based on the fuzzy algorithm of the objective function, update the membership degree matrix through the eigenvector C of the center point, and determine whether the objective function value satisfies the iteration termination condition; if it satisfies, end; if it does not satisfy, continue to execute step S222; i Step S224: Determine the cluster index to which each sample belongs according to the maximum membership degree in the dataset. The users in the cluster index have special interests in the products in the page name.

[0033] An advertising audience and delivery path analysis system based on big data, the system includes a connected user data collection module, a user interest analysis module, and an advertising delivery module;

[0034] The said user data collection module is used to collect user behavior data through the devices used by users;

[0035] The said user interest analysis module is used to analyze the access behavior of users based on user behavior data and identify the topics that users are interested in;

[0036] The said advertising delivery module is used to determine the advertising audience based on the topics that users are interested in and formulate the delivery path.

[0037] The said advertising delivery module is used to determine the advertising audience based on the topics that users are interested in and formulate the delivery path.

[0038] The user data collection module comprises:

[0039] The data storage module is used to store the click actions performed by the user on the application in the user session, and records the user session D in the form of a data set containing n objects and stores it in the log;

[0040] User session D=<(P 1 , D 1 , T 1 , X 1 , t 1 ),…,(P n , D n , T n , X n , t n ), (P end , D end , T end , X n+1 , t n+1 )>

[0041] Among them, P n Indicates the name of the loaded page, D n Indicates a specific action on the page, including searching, browsing products, adding to shopping cart, submitting orders, completing payments, and canceling orders. n Indicates the loading time of the corresponding page name, X n Indicates the touch screen gesture data under the corresponding page name, t n Indicates the residence time on the page; P end The page indicating the end of the session, D end Indicates the final action taken by the user on the termination page, T end Indicates the time when the session is terminated;

[0042] The data preprocessing module is used to preprocess the user behavior data. If the dwell time is less than t1, it means that the user clicked by mistake or is not interested in the page, and the behavior data is removed; if the dwell time is greater than t2, it means that the page failed to load or the user has left, and the behavior data is removed;

[0043] The data identification module is used to reversely select the user sessions belonging to the device owner based on the touch screen gesture data, and hide the user sessions other than the device owner.

[0044] The user interest analysis module includes:

[0045] The device status detection module is used to determine the state of the user using the device according to the stability of the touch screen gesture data in a user session; if the stability is low, the user is walking to operate the device; if the stability is medium, the user is standing to operate the device; if the stability is high, the user is standing to operate the device;

[0046] The data modeling module is used to model the touch screen gesture data: the center point of the device's display screen is taken as the origin O, the y-axis and z-axis are drawn through the origin O in the horizontal and vertical directions perpendicular to each other, and the x-axis is drawn through the origin O in the vertical direction perpendicular to the display screen. The values ​​in the y-axis and z-axis directions represent the touch point positions. The line connecting all the touch point positions (y, z) in a user session is the finger sliding trajectory line. The value in the x-axis direction represents the touch pressure value at the touch point (y, z); the touch point data = (x, y, z), and the line connecting all the touch point data in a user session is the pressure change trajectory.

[0047] The start and end of the finger sliding trajectory are point A and point A1 respectively, the start and end of the pressure change trajectory are point B and point B1 respectively, and point A and point B, point A1 and point B1 are connected to form a closed graph representing a geometric user session.

[0048] The user interest analysis module also includes:

[0049] The device owner judgment module is used to compare the similarity of the new geometric user session with the geometric user sessions of the users pre-stored in the database, hide the user session corresponding to the detection value M1 that exceeds the detection value threshold {α, β}, and confirm the user session corresponding to the detection value M1 that does not exceed the detection value threshold {α, β} as the user session belonging to the device owner.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] The present invention can identify the collected user data, separate the user sessions belonging to the device owner, reduce the influence of the data generated by others when using the device on the accuracy of the interest analysis of the device owner, and improve the effect of advertising delivery; pre-process the data to remove data with low analysis value, and reduce the time spent on subsequent analysis; through two clustering, it is possible to conveniently express the operations performed by the user on a certain page, and improve the accuracy of analyzing the user's special interest in the functional module corresponding to the touch screen gesture in the loading page. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a flow chart of the advertising audience and delivery path analysis method based on big data of the present invention;

[0053] Figure 2 This is a composition diagram of the advertising audience and delivery path analysis system based on big data of the present invention. DETAILED DESCRIPTION

[0054] 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.

[0055] Example: Figure 1 As shown, the present invention provides an advertising audience and delivery path analysis method based on big data, the method comprising:

[0056] Step S1: Collecting user behavior data through the device used by the user; for example, monitoring the user device through motion sensors such as accelerometers and gyroscopes, and position sensors such as magnetometers;

[0057] Step S2: Analyze the user's access behavior based on the user behavior data to identify topics that the user is interested in;

[0058] Step S3: Determine the advertising audience based on the topics that users are interested in and develop a delivery path.

[0059] For example, advertisers use topics that users are interested in to place products from merchants that correspond to the topics on the user's loading page, thereby recommending appropriate advertising content to users and performing periodic optimization based on data updates.

[0060] The step S1 of collecting user behavior data through the device used by the user includes:

[0061] Step S11: In a user session, a user performs a click action on an application, and the user session D is recorded in a data set containing n objects and stored in a log;

[0062] User session D=<(P 1 , D 1 , T 1 , X 1 , t 1 ),…,(P n , D n , T n , X n , t n ), (P end , D end , T end , X n+1 , t n+1 )>

[0063] Among them, P nIndicates the name of the loaded page, D n Indicates a specific action on the page, including searching, browsing products, adding to shopping cart, submitting orders, completing payments, and canceling orders. n Indicates the loading time of the corresponding page name, X n Indicates the touch screen gesture data under the corresponding page name, t n Indicates the residence time on the page; P end The page indicating the end of the session, D end Indicates the final action taken by the user on the termination page, T end Indicates the time when the session is terminated;

[0064] Step 12: Pre-process the user behavior data. If the dwell time is less than t1, it means that the user clicked by mistake or is not interested in the page, and the behavior data is removed. If the dwell time is greater than t2, it means that the page failed to load or the user has left, and the behavior data is removed. Pre-process the data to remove data with low analysis value and reduce the time spent on subsequent analysis.

[0065] Step 13: Reversely select the user session belonging to the device owner based on the touch screen gesture data, and hide the user sessions other than the device owner.

[0066] The step 13 comprises:

[0067] Step 131: judging the state of the user using the device according to the stability of the touch screen gesture data in a user session; if the stability is low, the user is walking to operate the device; if the stability is medium, the user is standing to operate the device; if the stability is high, the user is standing to operate the device;

[0068] Step 132: When the corresponding user is using the device, the touch screen gesture data is modeled: the center point of the display screen of the device is taken as the origin O, and the y-axis and z-axis are drawn through the origin O in the horizontal and vertical directions perpendicular to each other, and the x-axis is drawn through the origin O in the vertical direction perpendicular to the display screen. The values ​​in the y-axis and z-axis directions represent the touch point positions, and the line connecting all the touch point positions (y, z) in a user session is the finger sliding trajectory line. The value in the x-axis direction represents the touch pressure value at the touch point (y, z); the touch point data = (x, y, z), and the line connecting all the touch point data in a user session is the pressure change trajectory.

[0069] The start and end of the finger sliding trajectory are point A and point A1 respectively, and the start and end of the pressure change trajectory are point B and point B1 respectively. Point A and point B, point A1 and point B1 are connected to form a closed graph to represent the geometric user session, and the new geometric user session is compared with the geometric user sessions of users pre-stored in the database for similarity;

[0070] In step 132, the absolute value of the difference obtained by subtracting the straight-line length L1 at both ends of the finger sliding trajectory line from the straight-line length L2 at both ends of the preset standard finger sliding trajectory line is the trajectory difference, and the absolute value of the difference obtained by subtracting the area S1 of the geometric user session from the area S2 of the preset standard geometric user session is the pressure difference. Different weights are assigned to the pressure difference and the trajectory difference and then added together to obtain the detection value M. The detection value M1 is compared with the preset detection value thresholds {α, β}, and the user session corresponding to the detection value M1 that exceeds {α, β} is hidden, and the user session corresponding to the detection value M1 that does not exceed {α, β} is confirmed as the user session belonging to the device owner. The collected user data can be identified to separate the user sessions belonging to the device owner, reducing the impact of the data generated by others on the accuracy of the interest analysis of the device owner and improving the advertising delivery effect.

[0071] In step S2, the analysis of the access behaviors of multiple users based on the user behavior data to identify the interesting topics in the user cluster includes:

[0072] Step S21: Perform the first-layer session clustering based on the calculated similarity of user sessions;

[0073] Step S22: Perform the second-layer session clustering on users based on the distribution of all sessions generated in different session clusters of each user according to the first-layer session clustering.

[0074] The step S21 includes:

[0075] Step S211: Mark all objects in the user session D of the device owner as unvisited, and preset the cluster set S; select an object from the user session D 1 as the accessed object P and mark it as accessed;

[0076] Step S212: If the neighborhood E of the accessed object P contains at least a objects, where a is the neighborhood density threshold, then create a new cluster class C i1 , add the accessed object P and the objects contained in its neighborhood E to the cluster class C i1 ; select an object from the user session D 2 as the accessed object P to obtain a new cluster class C i2 , and so on;

[0077] If the number of common elements in two cluster classes is not less than k times the number of elements in the two cluster classes, where 0 < k < 1, then merge the two cluster classes into a new cluster C, and at the same time add the cluster C to the cluster set S;

[0078] If the number of common elements in two clusters is less than k times the number of elements in the two clusters, the two clusters are merged into two independent clusters and added to the cluster set S at the same time.

[0079] For example: If the cluster type C i1 and cluster class C i2 The common elements in the cluster are not less than C. i and cluster class C j If the number of elements in is 0.6 times, then the cluster class C is merged i1 and cluster class C i2 is a new cluster C, and cluster C is added to cluster set S;

[0080] If the cluster type C i1 and cluster class C i2 There are fewer common elements in the cluster than in the cluster class C i1 and cluster class C i2 0.6 times the number of elements in the cluster, then cluster class C i and cluster class C j Become two independent clusters and join the cluster set S at the same time.

[0081] The step S22 comprises:

[0082] Step S221, there are w clusters in the cluster set S, and one cluster is selected as sample Q;

[0083] Step S222: Initialize the membership relationship matrix U of the sample Q to the cluster center, and calculate the eigenvector C of the cluster center point. i (i=1,2,3…);

[0084] Step S223: Calculate the target function based on the fuzzy algorithm of the target function, and use the feature vector C of the center point i Update the membership degree matrix and determine whether the objective function value satisfies the iteration termination condition; if so, terminate; if not, continue to execute step S222;

[0085] Step S224, determine the cluster index to which each sample belongs according to the maximum member membership in the data set, and the user in the cluster index has a special interest in the product in the page name. Through two clusterings, it is convenient to express the operation performed by the user on a certain page, and improve the accuracy of analyzing the user's special interest in the functional module corresponding to the touch screen gesture in the loading page.

[0086] Figure 2 , an advertising audience and delivery path analysis system based on big data, the system includes a user data collection module, a user interest analysis module and an advertising delivery module connected to each other;

[0087] The user data collection module is used to collect user behavior data through the device used by the user;

[0088] The user interest analysis module is used to analyze the user's access behavior based on the user behavior data and identify the topics that the user is interested in;

[0089] The advertisement delivery module is used to determine the advertisement audience and formulate a delivery path based on the topics that the users are interested in.

[0090] The user data collection module comprises:

[0091] The data storage module is used to store the click actions performed by the user on the application in the user session, and records the user session D in the form of a data set containing n objects and stores it in the log;

[0092] User session D=<(P 1 , D 1 , T 1 , X 1 , t 1 ),…,(P n , D n , T n , X n , t n ), (P end , D end , T end , X n+1 , t n+1 )>

[0093] Among them, P n Indicates the name of the loaded page, D n Indicates a specific action on the page, including searching, browsing products, adding to shopping cart, submitting orders, completing payments, and canceling orders. n Indicates the loading time of the corresponding page name, X n Indicates the touch screen gesture data under the corresponding page name, t n Indicates the residence time on the page; P end The page indicating the end of the session, D end Indicates the final action taken by the user on the termination page, T end Indicates the time when the session is terminated;

[0094] The data preprocessing module is used to preprocess the user behavior data. If the dwell time is less than t1, it means that the user clicked by mistake or is not interested in the page, and the behavior data is removed; if the dwell time is greater than t2, it means that the page failed to load or the user has left, and the behavior data is removed;

[0095] The data identification module is used to reversely select the user sessions belonging to the device owner based on the touch screen gesture data, and hide the user sessions other than the device owner.

[0096] The user interest analysis module includes:

[0097] The device status detection module is used to determine the state of the user using the device according to the stability of the touch screen gesture data in a user session; if the stability is low, the user is walking to operate the device; if the stability is medium, the user is standing to operate the device; if the stability is high, the user is standing to operate the device;

[0098] The data modeling module is used to model the touch screen gesture data: the center point of the device's display screen is taken as the origin O, the y-axis and z-axis are drawn through the origin O in the horizontal and vertical directions perpendicular to each other, and the x-axis is drawn through the origin O in the vertical direction perpendicular to the display screen. The values ​​in the y-axis and z-axis directions represent the touch point positions. The line connecting all the touch point positions (y, z) in a user session is the finger sliding trajectory line. The value in the x-axis direction represents the touch pressure value at the touch point (y, z); the touch point data = (x, y, z), and the line connecting all the touch point data in a user session is the pressure change trajectory.

[0099] The start and end of the finger sliding trajectory are point A and point A1 respectively, the start and end of the pressure change trajectory are point B and point B1 respectively, and point A and point B, point A1 and point B1 are connected to form a closed graph representing a geometric user session.

[0100] The user interest analysis module also includes:

[0101] The device owner judgment module is used to compare the similarity of the new geometric user session with the geometric user sessions of the users pre-stored in the database, hide the user session corresponding to the detection value M1 that exceeds the detection value threshold {α, β}, and confirm the user session corresponding to the detection value M1 that does not exceed the detection value threshold {α, β} as the user session belonging to the device owner.

[0102] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and range of equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. The advertising audience and delivery path analysis method based on big data is characterized by: The method includes: Step S1, collecting user behavior data through the device used by the user; Step S2: Analyze the user's access behavior based on the user behavior data to identify topics that the user is interested in; Step S3: Determine the advertising audience and formulate a delivery path based on the topics that users are interested in; The step S1 of collecting user behavior data through the device used by the user includes: Step S11: In a user session, a user performs a click action on an application, and the user session D is recorded in a data set containing n objects and stored in a log; User session D=<(P1,D1,T1,X1,t1,…,(P n , D n , T n , X n , t n ), (P end , D end , T end , X n+1 , t n+1 )>; Among them, P n Indicates the name of the loaded page, D n Indicates a specific action on the page, including searching, browsing products, adding to shopping cart, submitting orders, completing payments, and canceling orders. n Indicates the loading time of the corresponding page name, X n Indicates the touch screen gesture data under the corresponding page name, t n Indicates the residence time on the page; P end The page indicating the end of the session, D end Indicates the final action taken by the user on the termination page, T end Indicates the time when the session is terminated; Step 12: Pre-process the user behavior data. If the dwell time is less than t1, it means that the user clicked by mistake or is not interested in the page, and the behavior data is removed. If the dwell time is greater than t2, it means that the page failed to load or the user has left, and the behavior data is removed. Step 13: Reversely select the user session belonging to the device owner based on the touch screen gesture data, and hide the user sessions other than the device owner; The step 13 comprises: Step 131: judging the state of the user using the device according to the stability of the touch screen gesture data in a user session; if the stability is low, the user is walking to operate the device; if the stability is medium, the user is standing to operate the device; if the stability is high, the user is standing to operate the device; Step 132: When the corresponding user is using the device, the touch screen gesture data is modeled: the center point of the display screen of the device is taken as the origin O, the y-axis and z-axis are drawn through the origin O and are perpendicular to each other in the horizontal and vertical directions, and the x-axis is drawn through the origin O and is perpendicular to the display screen. The values ​​in the y-axis and z-axis directions represent the touch point positions, the line connecting all the touch point positions (y, z) in a user session is the finger sliding trajectory line, and the value in the x-axis direction represents the touch pressure value at the touch point (y, z); touch point data = (x, y, z), and the line connecting all the touch point data in a user session is the pressure change trajectory. The start and end of the finger sliding trajectory are point A and point A1 respectively, and the start and end of the pressure change trajectory are point B and point B1 respectively. Point A and point B, point A1 and point B1 are connected to form a closed graph to represent the geometric user session, and the new geometric user session is compared with the geometric user sessions of users pre-stored in the database for similarity; Step 132: The absolute value of the value obtained by subtracting the straight line length L1 at both ends of the finger sliding trajectory line from the straight line length L2 at both ends of the preset standard finger sliding trajectory line is the trajectory difference; the absolute value of the value obtained by subtracting the area S1 of the geometric user session from the area S2 of the preset standard geometric user session is the pressure difference; different weights are assigned to the pressure difference and the trajectory difference and then added to obtain a detection value M; the detection value M1 is compared with the preset detection value threshold {α, β}; user sessions corresponding to the detection value M1 exceeding {α, β} are hidden; and user sessions corresponding to the detection value M1 not exceeding {α, β} are confirmed as user sessions belonging to the device owner.

2. The advertising audience and delivery path analysis method based on big data according to claim 1 is characterized by: In step S2, the access behaviors of multiple users are analyzed based on the user behavior data to identify topics of interest in the user cluster, including: Step S21: Perform the first - layer session clustering based on the calculated similarity of user sessions; Step S22: Based on the first - layer session clustering, perform the second - layer session clustering on users according to the distribution of all sessions generated in different session clusters of each user.

3. The method for analyzing advertising audience and delivery path based on big data according to claim 2, characterized in that: The step S21 includes: Step S211: Mark all objects in the user session D of the device owner as unvisited, and preset the cluster set S; Select an object from the user session D1 as the visited object P and mark it as visited; Step S212: If the neighborhood E of the accessed object P contains at least a objects, where a is the neighborhood density threshold, a new cluster C is created. i1 , add the visited object P and the objects contained in its neighborhood E to the cluster class C i1 In the example above, an object is selected from user session D2 as the access object P, and a new cluster class C is obtained. i2 , and so on; If the number of common elements in two cluster classes is not less than k times the number of elements in the two cluster classes, where 0 < k < 1, then merge the two cluster classes into a new cluster C, and at the same time add the cluster C to the cluster set S; If the number of common elements in two cluster classes is less than k times the number of elements in the two cluster classes, then merge the two cluster classes into two independent clusters, and at the same time add them to the cluster set S.

4. The method for analyzing advertising audience and delivery path based on big data according to claim 3 is characterized by: The step S22 includes: Step S221: There are w clusters in the cluster set S, select one cluster as the sample Q; Step S222: Initialize the membership relationship matrix U of the sample Q to the cluster center, and calculate the eigenvector C of the cluster center point. i (i=1,2,3…); Step S223: Calculate the target function based on the fuzzy algorithm of the target function, and use the feature vector C of the center point i Update the membership degree matrix and determine whether the objective function value satisfies the iteration termination condition; if so, terminate; if not, continue to execute step S222; Step S224: Determine the cluster index to which each sample belongs according to the maximum membership degree in the dataset. The users in the cluster index have a special interest in the products in the page name.

5. The advertising audience and delivery path analysis system based on big data is applied to the advertising audience and delivery path analysis method based on big data as described in any one of claims 1 to 4 above, characterized in that: The system includes a connected user data collection module, a user interest analysis module, and an advertisement placement module; The user data collection module is used to collect user behavior data through the device used by the user; The user interest analysis module is used to analyze the user's access behavior based on the user behavior data and identify the topics that the user is interested in; The advertisement placement module is used to determine the advertisement audience based on the topics that the user is interested in and formulate the placement path.

6. The advertising audience and delivery path analysis system based on big data according to claim 5 is characterized by: The user data collection module includes: The data storage module is used to store the click actions of the user on the application in the user session, record the user session D in the form of a dataset containing n objects, and store it in the log; User session D=<(P1,D1,T1,X1,t1,…,(P n , D n , T n , X n , t n ), (P end , D end , T end , X n+1 , t n+1 )>; Among them, P n Indicates the name of the loaded page, D n Indicates a specific action on the page, including searching, browsing products, adding to shopping cart, submitting orders, completing payments, and canceling orders. n Indicates the loading time of the corresponding page name, X n Indicates the touch screen gesture data under the corresponding page name, t n Indicates the residence time on the page; P end The page indicating the end of the session, D end Indicates the final action taken by the user on the termination page, T end Indicates the time when the session is terminated; The data pre - processing module is used to pre - process the user behavior data. If the residence time is less than t1, it means that the user clicks by mistake or is not interested in the page, and this behavior data is removed; If the residence time is greater than t2, it means that the page fails to load or the user has left, and this behavior data is removed; The data identification module is used to inversely select the user session belonging to the device owner based on the touch - screen gesture data and hide the user sessions other than the device owner.

7. The advertising audience and delivery path analysis system based on big data according to claim 6 is characterized by: The user interest analysis module includes: The device state detection module is used to judge the state of the user using the device according to the stability degree of the touch - screen gesture data in a user session; If the stability degree is low, it means that the user is operating the device while walking; If the stability degree is medium, it means that the user is standing still to operate the device; If the stability degree is high, it means that the user is operating the device while the device is stationary; A data modeling module is used to model touch screen gesture data: the center point of the device's display screen is taken as the origin O, and the y-axis and z-axis are drawn through the origin O in the horizontal and vertical directions perpendicular to each other. The x-axis is drawn through the origin O in the vertical direction perpendicular to the display screen. The values ​​in the y-axis and z-axis directions represent the touch point positions. The line connecting all the touch point positions (y, z) in a user session is the finger sliding trajectory line. The value in the x-axis direction represents the touch pressure value at the touch point (y, z); touch point data = (x, y, z), and the line connecting all the touch point data in a user session is the pressure change trajectory. The start and end of the finger sliding trajectory are point A and point A1 respectively, the start and end of the pressure change trajectory are point B and point B1 respectively, and point A and point B, point A1 and point B1 are connected to form a closed graph representing a geometric user session.

8. The advertising audience and delivery path analysis system based on big data according to claim 7, characterized in that: The user interest analysis module also includes: The device owner judgment module is used to compare the similarity of the new geometric user session with the geometric user sessions of the users pre-stored in the database, hide the user session corresponding to the detection value M1 that exceeds the detection value threshold {α, β}, and confirm the user session corresponding to the detection value M1 that does not exceed the detection value threshold {α, β} as the user session belonging to the device owner.

Images