WO2019161718A1 - 归因方法和装置 - Google Patents
归因方法和装置 Download PDFInfo
- Publication number
- WO2019161718A1 WO2019161718A1 PCT/CN2019/072098 CN2019072098W WO2019161718A1 WO 2019161718 A1 WO2019161718 A1 WO 2019161718A1 CN 2019072098 W CN2019072098 W CN 2019072098W WO 2019161718 A1 WO2019161718 A1 WO 2019161718A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- probability
- promotion information
- target
- channel
- matrix
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0277—Online advertisement
Definitions
- the present application relates to the field of Internet data analysis, and in particular to an attribution method and apparatus.
- the promotion information is the advertisement advertisement of an advertisement space
- the payment method is CPM
- the delivery method such as DSP
- the main purpose of the present application is to provide an attribution method and apparatus to solve the problem in the related art that it is difficult to quickly calculate the attribution ratio of each channel.
- an attribution method includes: collecting access data of the target promotion information by using multiple channels in a preset time period, and obtaining an access data set, wherein the multiple channels are channels for the target promotion information, and the access data set includes multiple pieces of data.
- Each piece of data includes at least a channel ID, access time of the target promotion information through each channel; and a source matrix, a jump matrix set, and a transformation matrix are determined according to the data in the access data set, wherein the source matrix is used to represent the access The probability of accessing the target promotion information through the various channels for the first time in the data set; the jump matrix set is used to indicate the probability of jumping between multiple channels when accessing the target promotion information in the access data set; the transformation matrix is used to indicate the access The probability that the target promotion information is directly converted into the preset index through various channels in the data collection; according to the source matrix, the jump matrix set and the transformation matrix, the total conversion probability is obtained, wherein the total probability of conversion is the promotion information of the target through multiple channels.
- the probability of conversion to a preset metric based on the total probability of conversion Calculation of each channel to promote the information into a preset target indicators of the proportion of the contribution.
- the total probability of obtaining the transformation includes:
- X represents the total probability of conversion
- S represents the source matrix
- C represents the transformation matrix
- M j represents the matrix formed from the probability of accessing the target promotion information to the probability that the conversion to the preset indicator is j-jumped between channels
- n represents the highest The number of jumps.
- calculating the contribution ratio of each channel to the target promotion information into the preset indicator includes: setting a channel as the target channel, and calculating the target promotion information after the target channel is removed and converting the target promotion information into the preset indicator Probability; according to the total probability of conversion and the probability that the target promotion information is converted into a preset indicator after removing the target channel, the contribution ratio of the target channel to the target promotion information into the preset indicator is obtained.
- setting a channel as a target channel, and calculating a probability that the target promotion information is converted into a preset indicator after the target channel is removed includes:
- Y represents the total probability of conversion after the target channel is removed
- Sir represents the source matrix after the target channel is removed
- Cir represents the transformation matrix after the target channel is removed
- Mir j represents the removal of the target channel from the access target promotion information to the conversion to the preset A matrix formed by the probability that the indicator passes through j jumps between channels
- n represents the highest number of jumps.
- the contribution ratio of the target channel to the target promotion information into the preset indicator is obtained:
- A is the probability that the target promotion information is converted into a preset indicator after the target channel is removed.
- an attribution device configured to collect access data for accessing the target promotion information through multiple channels in a preset time period, to obtain an access data set, where multiple channels are channels for the target promotion information, and the data set is accessed.
- the method includes multiple pieces of data, each of which includes at least a channel ID, access time of the target promotion information through each channel, and a determining unit configured to determine a source matrix, a jump matrix set, and a transformation matrix according to the data in the access data set.
- the source matrix is used to indicate the probability of accessing the target promotion information through the various channels for the first time in the access data set; the jump matrix set is used to indicate that the target promotion information is accessed between the multiple channels when accessing the target promotion information in the access data set.
- Probability the transformation matrix is used to represent the probability that the target promotion information is directly converted into a preset indicator through each channel in the access data set; the acquisition unit is set to obtain the total conversion probability according to the source matrix, the jump matrix set, and the transformation matrix. , where the total probability of conversion is through multiple channels Promotion of information into a standard indicator of the probability of default; computing unit, set up according to the proportion of the total contribution to the probability of conversion, calculation of each channel for the promotion of information into a preset target indicators.
- the obtaining unit includes:
- X represents the total probability of conversion
- S represents the source matrix
- C represents the transformation matrix
- M j represents the matrix formed from the probability of accessing the target promotion information to the probability that the conversion to the preset indicator is j-jumped between channels
- n represents the highest The number of jumps.
- the calculating unit includes: a first calculating sub-module, configured to set a channel as a target channel, and calculate a probability that the target promotion information is converted into a preset indicator after the target channel is removed; and the second calculating sub-module is set to be converted according to the conversion The total probability and the probability that the target promotion information is converted into the preset indicator after the target channel is removed, and the contribution ratio of the target channel to the target promotion information into the preset indicator is obtained.
- a storage medium including a stored program, wherein the program executes any one of the above attribution methods.
- a processor for running a program wherein the program is executed to perform any of the above attribution methods.
- the following steps are adopted: collecting access data of the target promotion information through multiple channels in a preset time period, and obtaining an access data set, wherein multiple channels are channels for the target promotion information to be accessed, and accessing the data set
- the method includes multiple pieces of data, each of which includes at least a channel ID, access time of the target promotion information through each channel, and a source matrix, a jump matrix set, and a transformation matrix according to the data in the access data set, wherein the source matrix is used
- the probability of converting the target promotion information into a preset indicator According to the total transformation probability is
- FIG. 1 is a flow chart of an attribution method provided in accordance with an embodiment of the present application.
- FIG. 2 is a schematic diagram of an attribution device provided in accordance with an embodiment of the present application.
- FIG. 1 is a flow diagram of an attribution method in accordance with an embodiment of the present invention. As shown in FIG. 1, the method includes the following steps:
- step S101 the access data of the target promotion information is accessed through multiple channels in the preset time period, and the access data set is obtained, wherein the multiple channels are channels for the target promotion information, and the access data set includes multiple pieces of data.
- Each piece of data includes at least a channel ID, and access time of the target promotion information through each channel;
- the preset time period is 3 months
- the target promotion information is the A brand mobile phone advertisement
- the merchants put the A brand mobile phone advertisement in the b browser and the g browser;
- each user accesses the access data of the A brand mobile phone advertisement through the b browser and the g browser, and obtains the access data set, wherein the access data set includes a plurality of data, and at least the b browser is included in each data. And the ID of the g browser and the access time of the mobile phone advertisement of the A brand through the b browser and the g browser.
- Step S102 Determine a source matrix, a jump matrix set, and a transformation matrix according to data in the access data set, where the source matrix is used to indicate a probability of accessing the target promotion information through the respective channels for the first time in the access data set; The probability of jumping between multiple channels when accessing the target promotion information in the access data set; the transformation matrix is used to indicate the probability that the target promotion information is directly converted into the preset indicator through each channel in the access data set;
- the source matrix is used to indicate the probability of accessing the A-brand mobile phone advertisement through the b browser and the g browser for the first time in the access data set.
- the source matrix is recorded as (sb sg), which is a matrix of 1*2. Only one row represents the starting point s, and the two columns represent two channels.
- the corresponding value sb in the source matrix indicates the probability of the starting point s to the b browser, that is, the first time to access the A brand mobile phone through the b browser.
- the probability of advertising, sg represents the probability of starting point s to g browser, that is, the probability of accessing the A brand mobile phone advertisement through the g browser for the first time;
- the conversion matrix is used to indicate the probability of directly converting a brand-name mobile phone advertisement into a preset indicator through the b browser and the g browser in the access data set, and the conversion matrix is recorded as For a 2*1 matrix, 2 rows represent two channels, only one column represents the conversion point c, and the corresponding value bc in the transformation matrix represents the probability of the b browser to the conversion point c, that is, browsing through b
- the probability of directly converting the A-brand mobile phone advertisement into a preset indicator for example, the preset indicator can be used to view the purchase of the brand-name mobile phone after the A-brand mobile phone advertisement is generated, and the gc represents the g browser to the conversion.
- the probability of point c that is, the probability of directly placing an order to purchase the brand mobile phone after viewing the A brand mobile phone advertisement through the g browser;
- the jump matrix set is used to indicate the probability of jumping between multiple channels when accessing the target advertisement in the access data set, from the starting point s directly to the conversion point c, or from the starting point s to the conversion point c only by one Channel, in these two cases, there is no jump between channels, no jump matrix is generated; that is, the jump matrix contains the jump matrix from the starting point s to the conversion point c, and the probability of a channel jump is Among them, bb represents the probability of jumping from b browser to b browser itself when viewing mobile advertising of brand A.
- the calculation rule is to find out the number of paths divided by the total number of jump paths, such as a.
- b->b->c->b->g->b->b->c extract the number of occurrences of b->b as the numerator, the total number of path jumps of all user paths as the denominator, bb Indicates the probability of jumping from the b browser to the b browser itself when viewing the mobile advertisement of the A brand.
- the number of users accessing the A brand mobile advertisement through the b browser is m, after leaving the b browser, at this m Among the users, m 1 users then visited the A brand mobile phone advertisement through the b browser again.
- the number of users accessing the A brand mobile phone advertisement by the browser is m. After leaving the b browser, among the m users, m 2 users then access the A brand mobile phone advertisement through the g browser.
- gb indicates the probability of jumping from the g browser to the b browser when viewing the mobile advertising of the A brand, specifically, browsing through g a number of users accessing the mobile advertising for brand n, g after leaving the browser, users in the n, n 1 users then have access to the a brand of mobile advertising through the b browser, That is the probability of jumping from the g browser to the b browser when viewing the mobile advertising of the A brand;
- gg means the probability of jumping from the g browser to the g browser itself when viewing the mobile advertising of the A brand, specifically, by g
- the number of users accessing the A brand mobile phone advertisement by the browser is n.
- n 2 users in the n users then access the A brand mobile phone advertisement through the g browser again. That is, the probability of jumping from the g browser to the g browser itself when viewing the mobile phone advertisement of the A brand; compared with the graph attribution method, the attribution method of the present application allows jumping from the b browser to the b browser or from the g browser
- the browser jumps to the self-loop of the g browser, and the computational complexity is not increased;
- the jump matrix consisting of the probability of going through the two channel jumps from the starting point s to the conversion point c is The jump matrix consisting of the probability of jumping from the starting point s to the conversion point c and passing through three channels on the way, And so on, until the transition matrix from the starting point s to the conversion point c, the probability of n channel jumps on the way, is
- the above matrix constitutes a set of jump matrices.
- Step S103 obtaining a total conversion probability according to the source matrix, the jump matrix set, and the transformation matrix.
- the total conversion probability is to convert the A brand mobile phone promotion information into a preset index by using a b browser and a g browser.
- the preset indicator that the promotion information is converted into is a behavior in which the user purchases the mobile phone advertisement of the A brand and then orders the mobile phone to purchase the brand mobile phone, according to the source matrix (sb sg), the jump matrix set and the transformation matrix.
- sb sg source matrix
- the jump matrix set transformation matrix.
- Step S104 calculating, according to the total conversion probability, a proportion of contribution of each channel to the target promotion information into a preset indicator
- the probability that the mobile phone advertisement of the A brand is placed in the b browser and the g browser to purchase the brand mobile phone after logging in to the official website after the user accesses the advertisement is calculated.
- the total probability of obtaining the transformation according to the source matrix, the jump matrix set, and the transformation matrix includes:
- X represents the total probability of conversion
- S represents the source matrix
- C represents the transformation matrix
- M j represents the matrix formed from the probability of accessing the target promotion information to the probability that the conversion to the preset indicator is j-jumped between channels
- n represents the highest The number of jumps.
- the total probability of conversion is a result obtained by multiplying a jump matrix composed of a probability of a jump from a starting point s to a transition point c through different channel jumps, and multiplying the source matrix and the transformation matrix: P0 is not used.
- P1 is the probability of going through a channel jump from the starting point s to the conversion point c, and P2 is the probability of going through two channel jumps from the starting point s to the conversion point c
- Pn is the probability that the n-channel jumps from the starting point s to the conversion point c;
- Pn s ⁇ b
- P0 to Pn are summed and summed to obtain the total conversion probability X, which is the probability that the mobile phone advertisement of the A brand is accessed through the b browser and the g browser, and the order is purchased on the official website to purchase the brand mobile phone.
- calculating the contribution ratio of each channel to the target promotion information into the preset indicator includes: setting a channel as the target channel, and calculating the removal target. The probability that the target promotion information is converted into the preset indicator after the channel; according to the total probability of conversion and the probability that the target promotion information is converted into the preset indicator after the target channel is removed, the contribution ratio of the target channel to the target promotion information into the preset indicator is obtained.
- the b browser is set as the target channel, and the probability that the user accesses the brand-name mobile phone advertisement after removing the b-type browser and then logs in to the official website to purchase the brand mobile phone is calculated; according to the total probability of conversion and the user purchase after removing the b-browser The probability of the brand mobile phone, the b browser to the user to access the A brand of mobile advertising, after logging into the official website to place an order to purchase the proportion of the brand's mobile phone.
- a channel is set as a target channel, and a probability that the target promotion information is converted into a preset indicator after the target channel is removed is included;
- Y represents the total probability of conversion after the target channel is removed
- Sir represents the source matrix after the target channel is removed
- Cir represents the transformation matrix after the target channel is removed
- Mir j represents the removal of the target channel from the access target promotion information to the conversion to the preset A matrix formed by the probability that the indicator passes through j jumps between channels, and n represents the highest number of jumps;
- b Jump matrix Mir 1 (gg) after the probability of a channel jump after the browser, calculate the total conversion probability Y after removing the b browser according to Sir, Cir and Mir j .
- the target channel is converted into the preset indicator by the target channel.
- the contribution ratios include:
- A is the probability that the target promotion information is converted into a preset indicator after the target channel is removed.
- the mobile phone advertisement of the A brand is launched, and the proportion of the contribution of the brand mobile phone to the user who has accessed the brand mobile phone advertisement after logging in to the official website is calculated.
- the browser and the g browser log in to the official website to place an order to purchase the brand mobile phone.
- the user accesses the brand-name mobile phone advertisement through the g browser and then logs in to the official website to purchase the brand. The difference in the probability of the phone.
- the attribution method of the present application removes the dependency of the graph database, and all the data participates in the computation through the mathematical matrix in the memory, and the attribution result is quickly issued, and the attribution performance is greatly improved.
- the solution provided by any of the above embodiments is calculated by using a mathematical matrix when calculating the contribution ratio of each channel by using a computer, which simplifies the calculation process and can be executed in the computer memory without using an external graph database.
- Multi-layer complex recursive computing can effectively improve the processing speed of the computer and the efficiency of memory resource utilization, that is, greatly improve the processing performance of the computer, and then quickly obtain the attribution result.
- the attribution method provided by the embodiment of the present application obtains an access data set by accessing access data of the target promotion information through multiple channels in a preset time period, wherein multiple channels are channels for the target promotion information to be accessed.
- the data set includes a plurality of pieces of data, each of which includes at least a channel ID, access time of the target promotion information through each channel, and a source matrix, a jump matrix set, and a transformation matrix according to the data in the access data set, wherein
- the source matrix is used to indicate the probability of accessing the target promotion information through the various channels for the first time in the access data set;
- the jump matrix set is used to indicate the probability of jumping between multiple channels when accessing the target promotion information in the access data set;
- the transformation matrix is used to represent the probability that the target promotion information is directly converted into the preset indicator through each channel in the access data set; according to the source matrix, the jump matrix set and the transformation matrix, the total conversion probability is obtained, wherein the total conversion probability is passed Converting target promotion information to prese
- the embodiment of the present application further provides an attribution device. It should be noted that the attribution device of the embodiment of the present application can be used to perform the attribution method provided by the embodiment of the present application. The attribution device provided by the embodiment of the present application is introduced below.
- the apparatus includes: an acquisition unit 10, a determination unit 20, an acquisition unit 30, and a calculation unit 40.
- the collecting unit 10 is configured to collect access data of the target promotion information by using multiple channels in a preset time period to obtain an access data set, where multiple channels are channels for the target promotion information, and the data set is accessed.
- the method includes multiple pieces of data, and each piece of data includes at least a channel ID, and access time of the target promotion information through each channel;
- the determining unit 20 is configured to determine a source matrix, a jump matrix set, and a transformation matrix according to the data in the access data set, wherein the source matrix is used to indicate the probability of accessing the target promotion information through the respective channels for the first time in the access data set;
- the matrix set is used to indicate the probability of jumping between multiple channels when accessing the target promotion information in the access data set;
- the transformation matrix is used to indicate that the target promotion information is directly converted into the preset indicator through each channel in the access data set. The probability;
- the obtaining unit 30 is configured to obtain a total conversion probability according to the source matrix, the jump matrix set, and the transformation matrix, wherein the total conversion probability is a probability that the target promotion information is converted into the preset indicator by using multiple channels;
- the calculating unit 40 is configured to calculate a contribution ratio of each channel to the target promotion information into a preset indicator according to the total conversion probability.
- the obtaining unit 30 includes:
- X represents the total probability of conversion
- S represents the source matrix
- C represents the transformation matrix
- M j represents the matrix formed from the probability of accessing the target promotion information to the probability that the conversion to the preset indicator is j-jumped between channels
- n represents the highest The number of jumps.
- the calculation unit 40 includes: a first calculation sub-module, configured to set a channel as a target channel, and calculate a target promotion information to be converted into a preset after the target channel is removed.
- the probability of the indicator the second calculation sub-module is configured to obtain the contribution of the target channel to the target promotion information to be converted into the preset indicator according to the total probability of conversion and the probability that the target promotion information is converted into the preset indicator after removing the target channel. proportion.
- the first calculation submodule includes:
- Y represents the total probability of conversion after the target channel is removed
- Sir represents the source matrix after the target channel is removed
- Cir represents the transformation matrix after the target channel is removed
- Mir j represents the removal of the target channel from the access target promotion information to the conversion to the preset A matrix formed by the probability that the indicator passes through j jumps between channels
- n represents the highest number of jumps.
- the second calculation submodule includes:
- A is the probability that the target promotion information is converted into a preset indicator after the target channel is removed.
- the attribution device collects access data of the target promotion information through multiple channels in a preset time period, and obtains an access data set, where multiple channels are used for the target promotion information.
- the access data set includes a plurality of data, each of the data includes at least a channel ID, and access time of the target promotion information is accessed through each channel;
- the determining unit 20 determines the source matrix and the jump according to the data in the access data set. a matrix set and a transformation matrix, wherein the source matrix is used to represent the probability of accessing the target promotion information through the various channels for the first time in the access data set; the jump matrix set is used to indicate that the target promotion information is accessed in multiple channels when accessing the data collection.
- the transformation matrix is used to indicate the probability of directly converting the target promotion information into a preset indicator through each channel in the access data set; the obtaining unit 30, according to the source matrix, the jump matrix set, and the transformation matrix, Get the total probability of conversion, where the total probability of conversion is through multiple The probability of converting the target promotion information into the preset indicator; the calculating unit 40 calculates the contribution ratio of each channel to the target promotion information into the preset indicator according to the total conversion probability, and solves the difficulty in quickly calculating each channel in the related technology.
- the problem of attribution ratio By matrixing the first-time access probability, the jump probability and the conversion probability, through the operation between the matrices, the contribution ratio of the target promotion information to the preset index is obtained, and the attribution of each channel is quickly calculated. The effect of the ratio.
- the attribution device includes a processor and a memory, and the above-mentioned acquisition unit 10, determination unit 20, acquisition unit 30, calculation unit 40, and the like are all stored as a program unit in a memory, and the processor executes the above-mentioned program unit stored in the memory. Implement the corresponding functions.
- the processor contains a kernel, and the kernel removes the corresponding program unit from the memory.
- the kernel can be set to one or more, and the kernel parameters can be adjusted to quickly calculate the attribution ratio for each channel.
- the memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory (flash RAM), the memory including at least one Memory chip.
- RAM random access memory
- ROM read only memory
- flash RAM flash memory
- Embodiments of the present invention provide a storage medium on which a program is stored, which is implemented by a processor to implement the attribution method.
- the embodiment of the invention provides a processor for running a program, wherein the attribution method is executed when the program runs.
- An embodiment of the present invention provides a device, including a processor, a memory, and a program stored on the memory and executable on the processor.
- the processor executes the program, the following steps are implemented: collecting multiple channels through a preset time period Accessing the access data of the target promotion information, and obtaining the access data set, wherein the plurality of channels are channels for the target promotion information, and the access data set includes multiple pieces of data, and each piece of data includes at least a channel ID, and each channel includes each channel Accessing the access time of the target promotion information; determining a source matrix, a jump matrix set, and a transformation matrix according to the data in the access data set, wherein the source matrix is used to indicate the probability of accessing the target promotion information through the respective channels for the first time in the access data set;
- the jump matrix set is used to indicate the probability of jumping between multiple channels when accessing the target promotion information in the access data set;
- the transformation matrix is used to indicate that the target promotion information is directly converted into the pre-process through the respective channels in the access
- the probability of the indicator is the probability of converting the target promotion information into the preset indicator through multiple channels; and calculating the contribution of each channel to the target promotion information into the preset indicator according to the total conversion probability proportion.
- the total probability of obtaining the transformation includes:
- X represents the total probability of conversion
- S represents the source matrix
- C represents the transformation matrix
- M j represents the matrix formed from the probability of accessing the target promotion information to the probability that the conversion to the preset indicator is j-jumped between channels
- n represents the highest The number of jumps.
- calculating the contribution ratio of each channel to the target promotion information into the preset indicator includes: setting a channel as the target channel, and calculating the target promotion information after the target channel is removed and converting the target promotion information into the preset indicator Probability; according to the total probability of conversion and the probability that the target promotion information is converted into a preset indicator after removing the target channel, the contribution ratio of the target channel to the target promotion information into the preset indicator is obtained.
- setting a channel as a target channel, and calculating a probability that the target promotion information is converted into a preset indicator after the target channel is removed includes:
- Y represents the total probability of conversion after the target channel is removed
- Sir represents the source matrix after the target channel is removed
- Cir represents the transformation matrix after the target channel is removed
- Mir j represents the removal of the target channel from the access target promotion information to the conversion to the preset A matrix formed by the probability that the indicator passes through j jumps between channels
- n represents the highest number of jumps.
- the contribution ratio of the target channel to the target promotion information into the preset indicator is obtained:
- A is the probability that the target promotion information is converted into a preset indicator after the target channel is removed.
- the devices in this document can be servers, PCs, PADs, mobile phones, and the like.
- the present application further provides a computer program product, when executed on a data processing device, is adapted to perform a process of initializing the following method steps: collecting access data of the target promotion information through multiple channels within a preset time period, and obtaining Accessing a data set, wherein the plurality of channels are channels for the target promotion information, and the access data set includes a plurality of data, each of the data includes at least a channel ID, and access time of the target promotion information through each channel;
- the data in the access data set determines a source matrix, a jump matrix set, and a transformation matrix, wherein the source matrix is used to indicate the probability of accessing the target promotion information through the various channels for the first time in the access data set; the jump matrix set is used to indicate the access The probability of jumping between multiple channels when accessing the target promotion information in the data set; the transformation matrix is used to indicate the probability of directly converting the target promotion information into the preset indicator through each channel in the access data set; according to the source matrix, Jump matrix set and transformation matrix to get the total conversion
- the total probability of obtaining the transformation includes:
- X represents the total probability of conversion
- S represents the source matrix
- C represents the transformation matrix
- M j represents the matrix formed from the probability of accessing the target promotion information to the probability that the conversion to the preset indicator is j-jumped between channels
- n represents the highest The number of jumps.
- calculating the contribution ratio of each channel to the target promotion information into the preset indicator includes: setting a channel as the target channel, and calculating the target promotion information after the target channel is removed and converting the target promotion information into the preset indicator Probability; according to the total probability of conversion and the probability that the target promotion information is converted into a preset indicator after removing the target channel, the contribution ratio of the target channel to the target promotion information into the preset indicator is obtained.
- setting a channel as a target channel, and calculating a probability that the target promotion information is converted into a preset indicator after the target channel is removed includes:
- Y represents the total probability of conversion after the target channel is removed
- Sir represents the source matrix after the target channel is removed
- Cir represents the transformation matrix after the target channel is removed
- Mir j represents the removal of the target channel from the access target promotion information to the conversion to the preset
- the matrix formed by the probability that the indicator has passed j jumps between channels, and n represents the highest number of jumps.
- the contribution ratio of the target channel to the target promotion information into the preset indicator is obtained:
- A is the probability that the target promotion information is converted into a preset indicator after the target channel is removed.
- embodiments of the present application can be provided as a method, system, or computer program product.
- the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment in combination of software and hardware.
- the application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
- the solution provided by any of the above embodiments is calculated by using a mathematical matrix when calculating the contribution ratio of each channel by using a computer, which simplifies the calculation process and can be executed in the computer memory without using an external graph database.
- Multi-layer complex recursive computing can effectively improve the processing speed of the computer and the efficiency of memory resource utilization, that is, greatly improve the processing performance of the computer, and then quickly obtain the attribution result.
- the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
- the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
- These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
- the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.
- a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
- processors CPUs
- input/output interfaces network interfaces
- memory volatile and non-volatile memory
- the memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory.
- RAM random access memory
- ROM read only memory
- Memory is an example of a computer readable medium.
- Computer readable media includes both permanent and non-persistent, removable and non-removable media.
- Information storage can be implemented by any method or technology.
- the information can be computer readable instructions, data structures, modules of programs, or other data.
- Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory. (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disk read only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, Magnetic tape cartridges, magnetic tape storage or other magnetic storage devices or any other non-transportable media can be used to store information that can be accessed by a computing device.
- computer readable media does not include temporary storage of computer readable media, such as modulated data signals and carrier waves.
- embodiments of the present application can be provided as a method, system, or computer program product.
- the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment in combination of software and hardware.
- the application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
Abstract
一种归因方法和装置。该方法包括：采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合（S101）；根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵（S102）；根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率（S103）；根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例（S104）。通过该方法，解决了相关技术中难以快速计算出每个渠道的归因比例的问题。
Description
本申请涉及互联网数据分析领域，具体而言，涉及一种归因方法和装置。
目前，在互联网时代的推广信息投放中，例如在线广告、推广软文等的投放，一个客户往往会通过多个媒体渠道进行投放，并且在各个媒体渠道的投放形式会有不同，例如在有的媒体投放的是搜索广告，付费方式是CPC，在有的媒体投放的推广信息是某广告位的展现广告，付费方式是CPM，还有DSP等投放方式，甚至并不能确定具体投放到了哪个具体的网站的具体广告位。面对这么多的推广信息投放媒体渠道，有些媒体投放的推广信息转化效果非常好，有些媒体投放的推广信息却几乎没有转化，但是，没有对推广信息进行直接转化的媒体并不一定对转化没有任何作用，例如，用户可能通过A媒体的广告，对广告涉及的产品有了初步了解，然后在看到了B媒体的该产品的广告时，点击并进行了深入的了解，最后在媒体C进行了产品搜索，根据搜索结果进入了官网，并产生了购买该产品的行为。这种情况下如何评价各个媒体渠道在这次推广信息转化中起到的作用，如果停止某个媒体的推广信息投放，会对最终的转化产生多大的影响，怎么明确各个渠道关联作用、合理利用历史数据有效评价每个媒体的推广信息投放对整体转化产生了多大的贡献，出现了图归因的相关方法，即将所有的数据打通，连成一张图，存储在neo4j这样的图数据库里，再分别针对每个节点，计算其对推广信息整体转化的影响值。但是，图数据库的递归计算耗时高，对于唯一值比较多的时候，比如超过100个或是1000个，便会出现性能问题，而且由于该性能问题，难以实时地根据数据源来计算其归因比例，计算前需要一连串复杂的调度：数据清洗，数据入库，图关系设置，图归因计算等。
针对相关技术中难以快速计算出每个渠道的归因比例的问题，目前尚未提出有效的解决方案。
发明内容
本申请的主要目的在于提供一种归因方法和装置，以解决相关技术中难以快速计算出每个渠道的归因比例的问题。
为了实现上述目的，根据本申请的一个方面，提供了一种归因方法。该方法包括：采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合， 其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目标推广信息转化为预设指标的概率；根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，其中，转化总概率为通过多个渠道将目标推广信息转化为预设指标的概率；根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例。
进一步地，根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率包括：
其中，X表示转化总概率，S表示源头矩阵，C表示转化矩阵，M
j表示从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
进一步地，根据所述转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例包括：将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率；根据转化总概率和去除目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对目标推广信息转化为预设指标的贡献比例。
进一步地，将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率包括；
其中，Y表示去除目标渠道后的转化总概率，Sir表示去除目标渠道后的源头矩阵，Cir表示去除目标渠道后的转化矩阵，Mir
j表示去除目标渠道后从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
进一步地，根据转化总概率和去除所述目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对目标推广信息转化为预设指标的贡献比例包括：
A＝X-Y
其中，A为去掉目标渠道后目标推广信息转化为预设指标的概率。
为了实现上述目的，根据本申请的另一方面，提供了一种归因装置。该装置包括： 采集单元，设置为采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；确定单元，设置为根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目标推广信息转化为预设指标的概率；获取单元，设置为根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，其中，转化总概率为通过多个渠道将目标推广信息转化为预设指标的概率；计算单元，设置为根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例。
进一步地，获取单元包括：
其中，X表示转化总概率，S表示源头矩阵，C表示转化矩阵，M
j表示从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
进一步地，计算单元包括：第一计算子模块，设置为将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率；第二计算子模块，设置为根据转化总概率和去除所述目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对所述目标推广信息转化为预设指标的贡献比例。
为了实现上述目的，根据本申请的另一方面，提供了一种存储介质，该存储介质包括存储的程序，其中，程序执行上述任意一种归因方法。
为了实现上述目的，根据本申请的另一方面，提供了一种处理器，该处理器用于运行程序，其中，程序运行时执行上述任意一种归因方法。
通过本申请，采用以下步骤：采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目 标推广信息转化为预设指标的概率；根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，其中，转化总概率为通过多个渠道将目标推广信息转化为预设指标的概率；根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例，解决了相关技术中难以快速计算出每个渠道的归因比例的问题。通过将首次访问概率、跳转概率以及转化概率矩阵化，通过矩阵之间的运算，得到个渠道对目标推广信息转化为预设指标的贡献比例，进而达到了快速计算出每个渠道的归因比例的效果。
构成本申请的一部分的附图用来提供对本申请的进一步理解，本申请的示意性实施例及其说明用于解释本申请，并不构成对本申请的不当限定。在附图中：
图1是根据本申请实施例提供的归因方法的流程图；以及
图2是根据本申请实施例提供的归因装置的示意图。
需要说明的是，在不冲突的情况下，本申请中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本申请。
为了使本技术领域的人员更好地理解本申请方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分的实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本申请保护的范围。
需要说明的是，本申请的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便这里描述的本申请的实施例。此外，术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或单元，而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或单元。
为了便于描述，以下对本申请实施例涉及的部分名词或术语进行说明：
下面结合优选的实施步骤对本发明进行说明，图1是根据本发明实施例的归因方法的流程图，如图1所示，该方法包括如下步骤：
步骤S101，采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；
例如，预设时间段为3个月，目标推广信息为A品牌的手机广告，商家在b浏览器，和g浏览器均投放了A品牌的手机广告；
采集3个月内各个用户通过b浏览器和g浏览器访问A品牌的手机广告的访问数据，得到访问数据集合，其中，访问数据集合中包括多条数据，每条数据中至少包括b浏览器和g浏览器的ID以及通过b浏览器和g浏览器访问A品牌的手机广告的访问时间等。
步骤S102，根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目标推广信息转化为预设指标的概率；
例如，源头矩阵用于表示在访问数据集合中首次通过b浏览器和g浏览器访问A品牌的手机广告的概率，该情况下，源头矩阵记为(sb sg)，为1*2的矩阵，只有一行代表的是起始点s，2列代表的是两个渠道，源头矩阵里对应的值sb表示的是起始点s到b浏览器的概率，即为首次通过b浏览器访问A品牌的手机广告的概率，sg表示的是起始点s到g浏览器的概率，即为首次通过g浏览器访问A品牌的手机广告的概率；
转化矩阵用于表示在访问数据集合中通过b浏览器和g浏览器查看A品牌的手机广告后直接转化为预设指标的概率，转化矩阵记为
为2*1的矩阵，2行代表的是两个渠道，只有一列代表的是转化点c，转化矩阵里对应的值bc表示的是b浏览器到转化点c的概率，即为通过b浏览器查看A品牌的手机广告后直接转化为预设指标的概率，例如预设指标可以为查看A品牌的手机广告后产生的下订单购买该品牌手机的行为，gc表示的是g浏览器到转化点c的概率，即为通过g浏览器查看A品牌的手机广告后直接下订单购买该品牌手机的概率；
跳转矩阵集合用于表示在访问数据集合中访问目标广告时在多个渠道之间进行跳转的概率，从起始点s直接到转化点c，或从始点s到转化点c只由经一个渠道，这两种情况均未发生渠道之间的跳转，未生成跳转矩阵；即跳转矩阵包含从起始点s到转 化点c，经过一次渠道跳转的概率构成的跳转矩阵，为
其中，bb表示查看A品牌的手机广告时从b浏览器跳转到b浏览器自身的概率，计算规则为找出所有用户会话路径链上，由路径数除以总跳转路径数，如a->b->b->c->b->g->b->b->c，抽取b->b的路径出现次数作为分子，所有用户路径的总路径跳转次数作为分母，bb表示查看A品牌的手机广告时从b浏览器跳转到b浏览器自身的概率，具体地，通过b浏览器访问A品牌的手机广告的用户数量为m，离开b浏览器后，在这m个用户中，有m
1个用户随后又再次通过b浏览器访问了A品牌的手机广告，
即为查看A品牌的手机广告时从b浏览器跳转到b浏览器自身的概率；bg表示查看A品牌的手机广告时从b浏览器跳转到g浏览器的概率，具体地，通过b浏览器访问A品牌的手机广告的用户数量为m，离开b浏览器后，在这m个用户中，有m
2个用户随后通过g浏览器访问了A品牌的手机广告，
即为查看A品牌的手机广告时从b浏览器跳转到g浏览器的概率；gb表示查看A品牌的手机广告时从g浏览器跳转到b浏览器的概率，具体地，通过g浏览器访问A品牌的手机广告的用户数量为n，离开g浏览器后，在这n个用户中，有n
1个用户随后通过b浏览器访问了A品牌的手机广告，
即为查看A品牌的手机广告时从g浏览器跳转到b浏览器的概率；gg表示查看A品牌的手机广告时从g浏览器跳转到g浏览器自身的概率，具体地，通过g浏览器访问A品牌的手机广告的用户数量为n，离开g浏览器后，在这n个用户中，有n
2个用户随后又再次通过g浏览器访问了A品牌的手机广告，
即为查看A品牌的手机广告时从g浏览器跳转到g浏览器自身的概率；相对于图归因方法，本申请的归因方法允许从b浏览器跳转到b浏览器或从g浏览器跳转到g浏览器的自回路，计算复杂程度未因此增加；包含从起始点s到转化点c，途中经过两次渠道跳转的概率构成的跳转矩阵，为
包含从起始点s到转化点c，途中经过三次渠道跳转的概率构成的跳转矩阵，为
以此类推，直到从起始点s到转化点c，途中经过n次渠道跳转的概率构成的跳转矩阵，为
上述矩阵构成跳转矩阵集合。
步骤S103，根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，上述情 况下，转化总概率即为通过b浏览器和g浏览器将A品牌的手机推广信息转化为预设指标的概率；
例如，设定推广信息转化为的预设指标为用户查看A品牌的手机广告后登录官网下订单购买了该品牌手机的行为，根据源头矩阵(sb sg)、跳转矩阵集合和转化矩阵
获取转化总概率，即为通过b浏览器和g浏览器访问A品牌的手机广告后登录官网下订单购买了该品牌手机的概率。
步骤S104，根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例；
例如，根据转化总概率，分别计算在b浏览器和g浏览器投放A品牌的手机广告对用户访问该广告后登录官网下订单购买了该品牌手机的概率。
可选地，在本申请实施例提供的归因方法中，根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率包括：
其中，X表示转化总概率，S表示源头矩阵，C表示转化矩阵，M
j表示从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
具体地，转化总概率为从起始点s到转化点c经过不同次渠道跳转的概率构成的跳转矩阵与源头矩阵和转化矩阵相乘后累加求和得到的结果：设P0为未经渠道跳转从起始点s到转化点c的概率；P1为经过一次渠道跳转从起始点s到转化点c，的概率，P2为经过两次渠道跳转从起始点s到转化点c的概率，Pn为经过n次渠道跳转从起始点s到转化点c的概率；
……
Pn＝s<b|g>{n}c，其中，<·|·>表示或的关系，<b|g>表示b渠道或g渠道，{n}是正则的写法，表示出现n次；
将P0到Pn累加求和，得到转化总概率X，即为通过b浏览器和g浏览器访问A品牌的手机广告后登录官网下订单购买了该品牌手机的概率。
可选地，在本申请实施例提供的归因方法中，根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例包括：将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率；根据转化总概率和去除目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对目标推广信息转化为预设指标的贡献比例。
例如，将b浏览器设定为目标渠道，计算去除b浏览器后用户访问A品牌的手机广告后登录官网下订单购买了该品牌手机的概率；根据转化总概率和去除b浏览器后用户购买该品牌手机的概率，得到b浏览器对用户访问A品牌的手机广告后登录官网下订单购买了该品牌手机的贡献比例。
可选地，在本申请实施例提供的归因方法中，将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率包括；
其中，Y表示去除目标渠道后的转化总概率，Sir表示去除目标渠道后的源头矩阵，Cir表示去除目标渠道后的转化矩阵，Mir
j表示去除目标渠道后从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数；
例如，将b浏览器设定为目标渠道，只剩g浏览器这一渠道，去除b浏览器后的源头矩阵Sir＝(sg)，去除b浏览器后的转化矩阵Cir＝(gc)，去除b浏览器后的经过一次渠道跳转的概率构成的跳转矩阵Mir
1＝(gg)，根据Sir、Cir和Mir
j计算去除b浏览器后的转化总概率Y。
可选地，在本申请实施例提供的归因方法中，根据转化总概率和去除所述目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对目标推广信息转化为预设指标的贡献比例包括：
A＝X-Y
其中，A为去掉目标渠道后目标推广信息转化为预设指标的概率。
例如，商家在b浏览器，和g浏览器均投放了A品牌的手机广告，计算b浏览器对用户访问A品牌的手机广告后登录官网下订单购买了该品牌手机的贡献比例为用户通过b浏览器和g浏览器访问A品牌的手机广告后登录官网下订单购买了该品牌手机的概率与去除b浏览器后用户通过g浏览器访问A品牌的手机广告后登录官网下订单购买了该品牌手机的概率的差值。
需要说明的是，本申请的归因方法，去除掉了图数据库的依赖，全部数据在内存里通过数学矩阵方式参与运算，快速出具归因结果，归因性能大幅提升。
本申请以上任一实施例提供的方案，在通过计算机进行各个渠道投放贡献比例计算时，由于采用数学矩阵进行计算，有效简化了计算过程，可以在计算机内存中执行，不必采用外部的图数据库进行多层复杂递归计算，能够有效提升计算机的处理速度和内存资源利用效率，也即，大幅度提升计算机的处理性能，进而快速得到归因结果。
本申请实施例提供的归因方法，通过采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在 多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目标推广信息转化为预设指标的概率；根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，其中，转化总概率为通过多个渠道将目标推广信息转化为预设指标的概率；根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例，解决了相关技术中难以快速计算出每个渠道的归因比例的问题。通过将首次访问概率、跳转概率以及转化概率矩阵化，通过矩阵之间的运算，得到个渠道对目标推广信息转化为预设指标的贡献比例，进而达到了快速计算出每个渠道的归因比例的效果。
需要说明的是，在附图的流程图示出的步骤可以在诸如一组计算机可执行指令的计算机系统中执行，并且，虽然在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于此处的顺序执行所示出或描述的步骤。
本申请实施例还提供了一种归因装置，需要说明的是，本申请实施例的归因装置可以用于执行本申请实施例所提供的用于归因方法。以下对本申请实施例提供的归因装置进行介绍。
图2是根据本申请实施例的归因装置的示意图。如图2所示，该装置包括：采集单元10、确定单元20、获取单元30和计算单元40。
具体地，采集单元10，设置为采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；
确定单元20，设置为根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目标推广信息转化为预设指标的概率；
获取单元30，设置为根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，其中，转化总概率为通过多个渠道将目标推广信息转化为预设指标的概率；
计算单元40，设置为根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例。
可选地，在本申请实施例提供的归因装置中，获取单元30包括：
其中，X表示转化总概率，S表示源头矩阵，C表示转化矩阵，M
j表示从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
可选地，在本申请实施例提供的归因装置中，计算单元40包括：第一计算子模块，设置为将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率；第二计算子模块，设置为根据转化总概率和去除所述目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对所述目标推广信息转化为预设指标的贡献比例。
可选地，在本申请实施例提供的归因装置中，第一计算子模块包括；
其中，Y表示去除目标渠道后的转化总概率，Sir表示去除目标渠道后的源头矩阵，Cir表示去除目标渠道后的转化矩阵，Mir
j表示去除目标渠道后从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
可选地，在本申请实施例提供的归因装置中，第二计算子模块包括：
A＝X-Y
其中，A为去掉目标渠道后目标推广信息转化为预设指标的概率。
本申请实施例提供的归因装置，通过采集单元10，采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；确定单元20，根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目标推广信息转化为预设指标的概率；获取单元30，根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，其中，转化总概率为通过多个渠道将目标推广信息转化为预设指标的概率；计算单元40，根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例，解决了相关技术中难以快速计算出每个渠道的归因比例的问题。通过将首次访问概率、跳转概率以及转化概率矩阵化，通过矩阵之间的运算，得到个渠道对目标推广信息转化为预设指标的贡献比例，进而达到了快速计算出每个渠道的归因比例的效果。
所述归因装置包括处理器和存储器，上述采集单元10、确定单元20、获取单元 30和计算单元40等均作为程序单元存储在存储器中，由处理器执行存储在存储器中的上述程序单元来实现相应的功能。
处理器中包含内核，由内核去存储器中调取相应的程序单元。内核可以设置一个或以上，通过调整内核参数来快速计算出每个渠道的归因比例。
存储器可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)，存储器包括至少一个存储芯片。
本发明实施例提供了一种存储介质，其上存储有程序，该程序被处理器执行时实现所述归因方法。
本发明实施例提供了一种处理器，所述处理器用于运行程序，其中，所述程序运行时执行所述归因方法。
本发明实施例提供了一种设备，设备包括处理器、存储器及存储在存储器上并可在处理器上运行的程序，处理器执行程序时实现以下步骤：采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目标推广信息转化为预设指标的概率；根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，其中，转化总概率为通过多个渠道将目标推广信息转化为预设指标的概率；根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例。
进一步地，根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率包括：
其中，X表示转化总概率，S表示源头矩阵，C表示转化矩阵，M
j表示从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
进一步地，根据所述转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例包括：将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率；根据转化总概率和去除目标渠道后目标推广信息转化为预设指 标的概率，得到目标渠道对目标推广信息转化为预设指标的贡献比例。
进一步地，将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率包括；
其中，Y表示去除目标渠道后的转化总概率，Sir表示去除目标渠道后的源头矩阵，Cir表示去除目标渠道后的转化矩阵，Mir
j表示去除目标渠道后从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
进一步地，根据转化总概率和去除所述目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对目标推广信息转化为预设指标的贡献比例包括：
A＝X-Y
其中，A为去掉目标渠道后目标推广信息转化为预设指标的概率。
本文中的设备可以是服务器、PC、PAD、手机等。
本申请还提供了一种计算机程序产品，当在数据处理设备上执行时，适于执行初始化有如下方法步骤的程序：采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，多个渠道为所述目标推广信息投放的渠道，访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问目标推广信息的访问时间；根据访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，源头矩阵用于表示在访问数据集合中首次通过各个渠道访问目标推广信息的概率；跳转矩阵集合用于表示在访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；转化矩阵用于表示在访问数据集合中通过各个渠道直接将目标推广信息转化为预设指标的概率；根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率，其中，转化总概率为通过多个渠道将目标推广信息转化为预设指标的概率；根据转化总概率，计算各个渠道对目标推广信息转化为预设指标的贡献比例。
进一步地，根据源头矩阵、跳转矩阵集合和转化矩阵，获取转化总概率包括：
其中，X表示转化总概率，S表示源头矩阵，C表示转化矩阵，M
j表示从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。
进一步地，根据所述转化总概率，计算各个渠道对目标推广信息转化为预设指标 的贡献比例包括：将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率；根据转化总概率和去除目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对目标推广信息转化为预设指标的贡献比例。
进一步地，将一个渠道设定为目标渠道，计算去除目标渠道后目标推广信息转化为预设指标的概率包括；
其中，Y表示去除目标渠道后的转化总概率，Sir表示去除目标渠道后的源头矩阵，Cir表示去除目标渠道后的转化矩阵，Mir
j表示去除目标渠道后从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数，。
进一步地，根据转化总概率和去除所述目标渠道后目标推广信息转化为预设指标的概率，得到目标渠道对目标推广信息转化为预设指标的贡献比例包括：
A＝X-Y
其中，A为去掉目标渠道后目标推广信息转化为预设指标的概率。
本领域内的技术人员应明白，本申请的实施例可提供为方法、系统、或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本申请以上任一实施例提供的方案，在通过计算机进行各个渠道投放贡献比例计算时，由于采用数学矩阵进行计算，有效简化了计算过程，可以在计算机内存中执行，不必采用外部的图数据库进行多层复杂递归计算，能够有效提升计算机的处理速度和内存资源利用效率，也即，大幅度提升计算机的处理性能，进而快速得到归因结果。
本申请是参照根据本申请实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定 方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
在一个典型的配置中，计算设备包括一个或多个处理器(CPU)、输入/输出接口、网络接口和内存。
存储器可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)。存储器是计算机可读介质的示例。
计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带，磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。按照本文中的界定，计算机可读介质不包括暂存电脑可读媒体(transitory media)，如调制的数据信号和载波。
还需要说明的是，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、商品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括要素的过程、方法、商品或者设备中还存在另外的相同要素。
本领域技术人员应明白，本申请的实施例可提供为方法、系统或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
以上仅为本申请的实施例而已，并不用于限制本申请。对于本领域技术人员来说， 本申请可以有各种更改和变化。凡在本申请的精神和原理之内所作的任何修改、等同替换、改进等，均应包含在本申请的权利要求范围之内。
Claims (10)
- 一种归因方法，包括：采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，所述多个渠道为所述目标推广信息投放的渠道，所述访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问所述目标推广信息的访问时间；根据所述访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，所述源头矩阵用于表示在所述访问数据集合中首次通过各个渠道访问目标推广信息的概率；所述跳转矩阵集合用于表示在所述访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；所述转化矩阵用于表示在所述访问数据集合中通过各个渠道直接将所述目标推广信息转化为预设指标的概率；根据所述源头矩阵、所述跳转矩阵集合和所述转化矩阵，获取转化总概率，其中，所述转化总概率为通过所述多个渠道将所述目标推广信息转化为预设指标的概率；根据所述转化总概率，计算各个渠道对所述目标推广信息转化为预设指标的贡献比例。
- 根据权利要求1所述的方法，其中，根据所述源头矩阵、所述跳转矩阵集合和所述转化矩阵，获取转化总概率包括：
其中，X表示所述转化总概率，S表示所述源头矩阵，C表示所述转化矩阵，M j表示从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。 - 根据权利要求2所述的方法，其中，根据所述转化总概率，计算各个渠道对所述目标推广信息转化为预设指标的贡献比例包括：将一个渠道设定为目标渠道，计算去除所述目标渠道后所述目标推广信息转化为预设指标的概率；根据所述转化总概率和去除所述目标渠道后所述目标推广信息转化为预设指标的概率，得到所述目标渠道对所述目标推广信息转化为预设指标的贡献比例。
- 根据权利要求3所述的方法，其中，将一个渠道设定为目标渠道，计算去除所述目标渠道后所述目标推广信息转化为预设指标的概率包括；
其中，Y表示去除所述目标渠道后的转化总概率，Sir表示去除所述目标渠道后的源头矩阵，Cir表示去除所述目标渠道后的转化矩阵，Mir j表示去除所述目标渠道后从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。 - 根据权利要求4所述的方法，其中，根据所述转化总概率和去除所述目标渠道后所述目标推广信息转化为预设指标的概率，得到所述目标渠道对所述目标推广信息转化为预设指标的贡献比例包括：A＝X-Y其中，A为去掉目标渠道后目标推广信息转化为预设指标的概率。
- 一种归因装置，包括：采集单元，设置为采集预设时间段内通过多个渠道访问目标推广信息的访问数据，得到访问数据集合，其中，所述多个渠道为所述目标推广信息投放的渠道，所述访问数据集合中包括多条数据，每条数据中至少包括渠道ID，通过每个渠道访问所述目标推广信息的访问时间；确定单元，设置为根据所述访问数据集合中的数据确定源头矩阵、跳转矩阵集合和转化矩阵，其中，所述源头矩阵用于表示在所述访问数据集合中首次通过各个渠道访问目标推广信息的概率；所述跳转矩阵集合用于表示在所述访问数据集合中访问目标推广信息时在多个渠道之间进行跳转的概率；所述转化矩阵用于表示在所述访问数据集合中通过各个渠道直接将所述目标推广信息转化为预设指标的概率；获取单元，设置为根据所述源头矩阵、所述跳转矩阵集合和所述转化矩阵，获取转化总概率，其中，所述转化总概率为通过所述多个渠道将所述目标推广信息转化为预设指标的概率；计算单元，设置为根据所述转化总概率，计算各个渠道对所述目标推广信息转化为预设指标的贡献比例。
- 根据权利要求6所述的装置，其中，所述获取单元包括：
其中，X表示所述转化总概率，S表示所述源头矩阵，C表示所述转化矩阵， M j表示从访问目标推广信息到转化为预设指标在渠道之间经过j次跳转的概率形成的矩阵，n表示最高跳转次数。 - 根据权利要求7所述的装置，其中，所述计算单元包括：第一计算子模块，设置为将一个渠道设定为目标渠道，计算去除所述目标渠道后所述目标推广信息转化为预设指标的概率；第二计算子模块，设置为根据所述转化总概率和去除所述目标渠道后所述目标推广信息转化为预设指标的概率，得到所述目标渠道对所述目标推广信息转化为预设指标的贡献比例。
- 一种存储介质，其中，所述存储介质包括存储的程序，其中，所述程序执行权利要求1至5中任意一项所述的归因方法。
- 一种处理器，其中，所述处理器用于运行程序，其中，所述程序运行时执行权利要求1至5中任意一项所述的归因方法。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810155690.6A CN110189153B (zh) | 2018-02-23 | 2018-02-23 | 归因方法和装置 |
| CN201810155690.6 | 2018-02-23 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2019161718A1 true WO2019161718A1 (zh) | 2019-08-29 |
Family
ID=67687489
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2019/072098 WO2019161718A1 (zh) | 2018-02-23 | 2019-01-17 | 归因方法和装置 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110189153B (zh) |
| WO (1) | WO2019161718A1 (zh) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111324842B (zh) * | 2020-02-13 | 2022-06-07 | 贝壳技术有限公司 | 用于实现页面优化的方法、装置、介质和电子设备 |
| CN112529634A (zh) * | 2020-12-18 | 2021-03-19 | 恩亿科(北京)数据科技有限公司 | 基于大数据的转化链路分析方法、系统和计算机设备 |
| CN113420261B (zh) * | 2021-08-23 | 2021-11-09 | 平安科技（深圳）有限公司 | 基于归因分析的课程推荐方法、装置、设备及介质 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663616A (zh) * | 2012-03-19 | 2012-09-12 | 北京国双科技有限公司 | 一种基于多触点归因模型的网络广告效果衡量方法和系统 |
| US20130246167A1 (en) * | 2012-03-15 | 2013-09-19 | Microsoft Corporation | Cost-Per-Action Model Based on Advertiser-Reported Actions |
| CN103562946A (zh) * | 2011-05-27 | 2014-02-05 | 谷歌公司 | 对于广告开支回报的多个归因模型 |
| CN106447396A (zh) * | 2016-09-22 | 2017-02-22 | 晶赞广告（上海）有限公司 | 一种广告在线预算的分配方法 |
| CN107563781A (zh) * | 2016-06-30 | 2018-01-09 | 阿里巴巴集团控股有限公司 | 一种信息投放效果归因方法及装置 |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9420970B2 (en) * | 2013-10-22 | 2016-08-23 | Mindstrong, LLC | Method and system for assessment of cognitive function based on mobile device usage |
| US9852439B2 (en) * | 2013-12-05 | 2017-12-26 | Google Llc | Methods and systems for measuring conversion probabilities of paths for an attribution model |
| CN104834675B (zh) * | 2015-04-02 | 2018-02-23 | 浪潮集团有限公司 | 一种基于用户行为分析的查询性能优化方法 |
| US10452724B2 (en) * | 2016-05-18 | 2019-10-22 | Google Llc | Attribution model for content item conversions |
-
2018
- 2018-02-23 CN CN201810155690.6A patent/CN110189153B/zh active Active
-
2019
- 2019-01-17 WO PCT/CN2019/072098 patent/WO2019161718A1/zh active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103562946A (zh) * | 2011-05-27 | 2014-02-05 | 谷歌公司 | 对于广告开支回报的多个归因模型 |
| US20130246167A1 (en) * | 2012-03-15 | 2013-09-19 | Microsoft Corporation | Cost-Per-Action Model Based on Advertiser-Reported Actions |
| CN102663616A (zh) * | 2012-03-19 | 2012-09-12 | 北京国双科技有限公司 | 一种基于多触点归因模型的网络广告效果衡量方法和系统 |
| CN107563781A (zh) * | 2016-06-30 | 2018-01-09 | 阿里巴巴集团控股有限公司 | 一种信息投放效果归因方法及装置 |
| CN106447396A (zh) * | 2016-09-22 | 2017-02-22 | 晶赞广告（上海）有限公司 | 一种广告在线预算的分配方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110189153A (zh) | 2019-08-30 |
| CN110189153B (zh) | 2021-09-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2019161731A1 (zh) | 渠道的归因方法和装置 | |
| US8983930B2 (en) | Facet group ranking for search results | |
| CN109040844B (zh) | 一种获取视频热度的方法、装置及电子设备 | |
| Khan et al. | Hadoop performance modeling for job estimation and resource provisioning | |
| US10417650B1 (en) | Distributed and automated system for predicting customer lifetime value | |
| US9858587B2 (en) | Methods and systems for creating a data-driven attribution model for assigning attribution credit to a plurality of events | |
| WO2019161718A1 (zh) | 归因方法和装置 | |
| Pachilakis et al. | No more chasing waterfalls: a measurement study of the header bidding ad-ecosystem | |
| TW201931256A (zh) | 營銷資訊的推送方法及裝置 | |
| WO2018121700A1 (zh) | 基于已安装应用来推荐应用信息的方法、装置、终端设备及存储介质 | |
| TW201737176A (zh) | 業務對象的分時推薦方法和系統 | |
| TWI615723B (zh) | 網路搜尋方法及設備 | |
| US9779406B2 (en) | User feature identification method and apparatus | |
| WO2019169964A1 (zh) | 一种资源和营销推荐方法、装置及电子设备 | |
| CN107451918B (zh) | 资产数据管理方法及装置 | |
| US20200387926A1 (en) | Methods and apparatus to determine informed holdouts for an advertisement campaign | |
| CN111125376B (zh) | 知识图谱生成方法、装置、数据处理设备及存储介质 | |
| US20160196579A1 (en) | Dynamic deep links based on user activity of a particular user | |
| CN111612560A (zh) | 用于促销对象的推荐方法、系统、存储介质及电子设备 | |
| CN111260388A (zh) | 一种商品生命周期的确定、展示方法和装置 | |
| CN110766225A (zh) | 一种基于神经网络的电力日前交易收益预测方法及装置 | |
| TWI587228B (zh) | 自使用者活動產生線上使用者與網站評價之系統與方法 | |
| CN110209687B (zh) | 多维度归因的查询方法和装置 | |
| KR20120136503A (ko) | 광고 시스템 및 광고 성과 평가 방법 | |
| CN110020118B (zh) | 一种计算用户之间相似度的方法及装置 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 19757856 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 19757856 Country of ref document: EP Kind code of ref document: A1 |