CN113706211A - Advertisement click rate prediction method and system based on neural network - Google Patents
Advertisement click rate prediction method and system based on neural network Download PDFInfo
- Publication number
- CN113706211A CN113706211A CN202111016975.XA CN202111016975A CN113706211A CN 113706211 A CN113706211 A CN 113706211A CN 202111016975 A CN202111016975 A CN 202111016975A CN 113706211 A CN113706211 A CN 113706211A
- Authority
- CN
- China
- Prior art keywords
- advertisement
- features
- user
- target
- embedding
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 71
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 37
- 238000012545 processing Methods 0.000 claims abstract description 42
- 238000007499 fusion processing Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 14
- 230000004927 fusion Effects 0.000 claims description 37
- 230000006870 function Effects 0.000 claims description 22
- 238000004590 computer program Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 18
- 238000012384 transportation and delivery Methods 0.000 claims description 16
- 238000013507 mapping Methods 0.000 claims description 15
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 230000004913 activation Effects 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000011176 pooling Methods 0.000 claims description 7
- 238000005516 engineering process Methods 0.000 abstract description 20
- 238000013473 artificial intelligence Methods 0.000 abstract description 7
- 239000013598 vector Substances 0.000 description 19
- 238000010586 diagram Methods 0.000 description 11
- 230000008901 benefit Effects 0.000 description 6
- 238000007726 management method Methods 0.000 description 6
- 239000000463 material Substances 0.000 description 6
- 238000003062 neural network model Methods 0.000 description 6
- 238000009826 distribution Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- RTZKZFJDLAIYFH-UHFFFAOYSA-N Diethyl ether Chemical compound CCOCC RTZKZFJDLAIYFH-UHFFFAOYSA-N 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 238000010606 normalization Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000004880 explosion Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
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
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06Q30/0249—Advertisements based upon budgets or funds
Abstract
The invention relates to an artificial intelligence technology, and provides an advertisement click rate prediction method and system based on a neural network, wherein the method comprises the following steps: acquiring historical advertisement characteristics of a target user and attribute characteristics of the target advertisement, wherein the historical advertisement characteristics comprise basic information of the target user and historical purchase information between the target user and a target product, and the target advertisement is used for popularizing the target product; inputting the historical advertisement characteristics and the attribute characteristics into a click prediction model, sequentially performing embedding density processing and cross fusion processing, predicting the click rate of the target user on the target advertisement, and training the click prediction model for a user sample, an advertisement sample and a score label to obtain the target advertisement. The embodiment of the invention integrates the user characteristics and the advertisement characteristics, and excavates the user information and the advertisement information from different layers, thereby overcoming the problem of reduced prediction precision caused by high-dimensional sparse characteristic data loss in the prior art and improving the prediction precision of the user on the advertisement.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to an advertisement click rate prediction method and system based on a neural network.
Background
With the rapid popularization and development of the internet, the product promotion by online advertisement using the internet becomes a very important marketing mode. The advertisement click rate refers to the ratio of the number of times of advertisement click to the number of times of advertisement display, the higher the advertisement click rate is, the more interested the user is in the advertisement, and the correct prediction of the advertisement click rate is an important means for realizing accurate marketing of products and also an important index for reflecting the advertisement putting effect and evaluating the advertisement conversion rate. Therefore, the method has very important practical significance in realizing the accurate prediction of the advertisement click rate.
The existing advertisement click prediction basically inputs detected data into a deep learning model for probability prediction, the probability prediction is often related according to the specific structure of deep learning, the model framework basically needs to map low-dimensional features into a high-dimensional space, data loss and improper processing often exist in the processing process, the comprehensiveness of data feature information cannot be met, and therefore the reliability of advertisement click rate prediction is greatly reduced.
Therefore, a method for predicting the advertisement click rate with high reliability is needed.
Disclosure of Invention
The invention provides an advertisement click rate prediction method and system based on a neural network, and mainly aims to improve the prediction precision of the advertisement click rate so that a user can reasonably plan an advertisement putting strategy.
In a first aspect, an embodiment of the present invention provides a method for predicting an advertisement click rate based on a neural network, including:
acquiring historical advertisement characteristics of a target user and attribute characteristics of a target advertisement, wherein the historical advertisement characteristics comprise basic information of the target user and historical purchase information between the target user and a target product, and the attribute characteristics are acquired according to the basic information of the target advertisement;
inputting the historical advertisement features and the attribute features into a click prediction model, sequentially embedding the historical advertisement features and the attribute features to be densely processed, performing cross fusion processing on the processed historical advertisement features and the processed attribute features, and predicting the click rate of the target user on the target advertisement, wherein the click prediction model is obtained by training user samples, advertisement samples and score labels.
The inputting the historical advertisement features and the attribute features into a click prediction model, sequentially embedding the historical advertisement features and the attribute features to be dense, performing cross fusion processing on the processed historical advertisement features and the processed attribute features, and predicting the click rate of the target user on the target advertisement specifically comprises:
embedding the historical advertisement features and the attribute features to be dense, so as to obtain user embedding features and advertisement embedding features respectively;
performing cross fusion and average pooling on the user embedding characteristics and the advertisement embedding characteristics to obtain fusion characteristics;
and predicting the click rate of the target user on the target advertisement according to the fusion characteristics.
Preferably, the embedding density processing is performed on the historical advertisement features and the attribute features respectively to obtain user embedded features and advertisement embedded features respectively, and the embedding density processing includes:
mapping high-dimensional sparse features in the historical advertisement features to dense features of the users; and/or mapping high-dimensional sparse features in the attribute features of the target advertisements to advertisement dense features;
normalizing the user continuous features in the historical advertisement features to obtain normalized user continuous features; and/or normalizing the advertisement continuous features in the attribute features to obtain normalized advertisement continuous features;
splicing the dense user features and the normalized continuous user features to obtain user splicing features; and/or splicing the advertisement dense features and the normalized advertisement continuous features to obtain advertisement splicing features;
and respectively embedding the user splicing characteristics and/or the advertisement splicing characteristics in multiple layers to obtain the user embedding characteristics and the advertisement embedding characteristics.
Preferably, the user splicing features and/or the advertisement splicing features are embedded in multiple layers respectively to obtain the user embedding features and the advertisement embedding features, and a specific calculation formula is as follows:
wherein x isuserRepresenting said user splicing feature, xitemThe characteristics of the splicing of the advertisements are represented,
representing the ith element in the user-embedded feature,
representing the ith element in the ad embedding feature,
represents the ith element of the preset user embedding matrix,
represents the ith element in the preset advertisement embedding matrix, and m represents the number of embedding into the subspace.
Preferably, the predicting the click rate of the target user on the target advertisement according to the fusion feature includes:
and inputting the fusion characteristics into a full connection layer, and obtaining the click rate by the output of the full connection layer through an activation function.
The method for acquiring the user sample and the advertisement sample comprises the following steps:
obtaining an initial user sample and an initial advertisement sample;
and respectively processing the initial user sample and the initial advertisement sample by adopting a preset rule to obtain a user sample and an advertisement sample, wherein the preset rule comprises missing value filling and layered sampling of a nearest neighbor algorithm in sequence.
Preferably, the method further comprises the following steps:
acquiring a comprehensive score of the target advertisement according to the click rate and the delivery price of the target advertisement;
and judging whether the target advertisement is delivered or not according to the comprehensive score.
In a second aspect, an embodiment of the present invention provides an advertisement click-through rate prediction system based on a neural network, including:
the characteristic acquisition module is used for acquiring historical advertisement characteristics of a target user and attribute characteristics of a target advertisement, wherein the historical advertisement characteristics comprise basic information of the target user and historical purchase information between the target user and a target product, and the attribute characteristics are acquired according to the basic information of the target advertisement;
and the prediction module is used for inputting the historical advertisement features and the attribute features into a click prediction model, sequentially embedding the historical advertisement features and the attribute features into a dense process, performing cross fusion processing on the processed historical advertisement features and the processed attribute features, and predicting the click rate of the target user on the target advertisement, wherein the click prediction model is obtained by training a user sample, an advertisement sample and a score tag.
In a third aspect, an embodiment of the present invention provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the steps of the neural network-based advertisement click rate prediction method when executing the computer program.
In a fourth aspect, an embodiment of the present invention provides a computer storage medium, where a computer program is stored, and when the computer program is executed by a processor, the computer program implements the steps of the advertisement click rate prediction method based on a neural network.
According to the advertisement click rate prediction method and system based on the neural network, the neural network model is adopted to carry out embedding dense processing on historical advertisement features and attribute features, and user embedding features and advertisement embedding features represented by dense vectors are obtained through the embedding dense processing; and on the basis of performing dense processing on the historical advertisement characteristics and the attribute characteristics of the user independently, the user characteristics and the advertisement characteristics are fused and embedded in multiple layers, and the information of the user and the advertisement can be mined from different layers, so that the comprehensiveness of data processing is improved, the information of the user and the advertisement is fully fused, and the prediction precision of the user on the advertisement is further improved.
Drawings
Fig. 1 is an application scenario diagram of an advertisement click rate prediction method based on a neural network according to an embodiment of the present invention;
FIG. 2 is a flowchart of a method for predicting advertisement click-through rate based on a neural network according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a click prediction model according to an embodiment of the present invention;
fig. 4 is an architecture diagram of an advertisement bidding delivery system based on the neural network-based advertisement click rate prediction method provided by the embodiment of the present invention;
FIG. 5 is a schematic structural diagram of an advertisement click-through rate prediction system based on a neural network according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of a computer device according to an embodiment of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Fig. 1 is an application scenario diagram of an advertisement click rate prediction method based on a neural network according to an embodiment of the present invention, as shown in fig. 1, a client acquires a historical advertisement feature of a target user and an attribute feature of a target advertisement, and sends the historical advertisement feature of the target user and the attribute feature of the target advertisement to a server, and the server receives the historical advertisement feature of the target user and the attribute feature of the target advertisement, and then executes the advertisement click rate prediction method based on the neural network, so as to predict a click rate of the target user on the target advertisement.
The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, Artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.
The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.
It should be noted that the server may be implemented by an independent server or a server cluster composed of a plurality of servers, or may be a cloud server that provides basic cloud computing services such as cloud service, a cloud database, cloud computing, a cloud function, cloud storage, Network service, cloud communication, middleware service, domain name service, security service, Content Delivery Network (CDN), big data, and an artificial intelligence platform.
The client may be, but is not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, and the like. The client and the server may be connected through bluetooth, USB (Universal Serial Bus), or other communication connection manners, which is not limited in this embodiment of the present invention.
Fig. 2 is a flowchart of an advertisement click-through rate prediction method based on a neural network according to an embodiment of the present invention, as shown in fig. 2, the method includes:
s1, acquiring historical advertisement characteristics of a target user and attribute characteristics of a target advertisement, wherein the historical advertisement characteristics comprise basic information of the target user and historical purchase information between the target user and a target product, and the attribute characteristics are acquired according to the basic information of the target advertisement;
aiming at the target users and the target advertisements, when the probability that the target users click the target advertisements needs to be judged, firstly, basic information of a target user and a target advertisement is input, then information related to the target user and attribute information related to the advertisement are extracted, the information is referred to as the historical advertisement characteristics of the target user, which describes some basic information of the target user and some information related to the advertisement in the past, the basic information of the user includes information of user age, gender, occupation, region, etc., and the information related to the advertisement is different according to the advertisement category, for example, if the target product of the advertisement is car insurance, the historical purchase information of the user related to the car insurance comprises the information of the brand, the age, the historical vehicle claim settlement information, the vehicle insurance, the insurance amount, the insurance time, the vehicle mileage and the like of the owned vehicle; if the target product of the advertisement is insurance, the user's historical purchase information related to insurance includes information on the user's marital status, insurance products, premium, start of insurance period, and end of insurance period. It can be seen that the historical advertisement characteristics of the target users corresponding to different target advertisements are different, and may be determined specifically according to the actual situation, and the embodiments of the present invention are not described herein one by one.
The attribute characteristics of the target advertisement include some basic information of the target advertisement, such as advertisement duration, advertisement category, advertisement quality, advertisement audience and crowd, selling price of advertisement products, historical click rate of advertisement products, and the like, and can be specifically determined according to actual conditions.
And S2, inputting the historical advertisement characteristics and the attribute characteristics into a click prediction model, sequentially embedding the historical advertisement characteristics and the attribute characteristics in a dense mode, performing cross fusion processing on the processed historical advertisement characteristics and the processed attribute characteristics, and predicting the click rate of the target user on the target advertisement, wherein the click prediction model is obtained by training a user sample, an advertisement sample and a score label.
Specifically, the historical advertisement features of the target user and the attribute features of the target advertisement are input into a click prediction model, the click prediction model carries out embedding dense processing and cross fusion processing on the historical advertisement features and the attribute features, and the embedding dense processing refers to the fact that high-dimensional sparse features in the historical advertisement features and the attribute features are mapped into dense vectors and are spliced with continuous features, so that the problem that prediction accuracy is reduced due to the fact that high-dimensional sparse vector data are lost in the prior art is solved.
The cross fusion processing is to fuse the user characteristics and the advertisement characteristics only on the basis of performing dense processing on the historical advertisement characteristics and the attribute characteristics of the user independently, so that the user information and the advertisement information can be mined from different layers, and the user information and the advertisement information are fully fused, thereby further improving the prediction precision of the user on the advertisement.
It should be noted that the click prediction model is a neural network model, and common structures of the neural network model include a BP neural network model, a convolutional neural network, and the like, which can be specifically selected according to actual needs, and this embodiment is not specifically limited herein.
Generally, before a click prediction model is used, the click prediction model needs to be trained, the training of the neural network needs to define the structure of the neural network and the output result of forward propagation, then define a loss function and select a back propagation optimization algorithm, and repeatedly run the back propagation optimization algorithm on training data, wherein the training data are historical advertisement characteristics of a user sample, attribute characteristics of the advertisement sample and a label, and the label refers to the click rate of the user sample on the advertisement sample.
The training process is to determine the most appropriate parameters of the click prediction model, after the parameters of each structure in the click prediction model are determined, the click prediction model has the capability of predicting the click rate of the advertisement, and when prediction is needed, the historical advertisement characteristics of the target user and the attribute characteristics of the target advertisement are directly input into the click prediction model, so that the click rate of the target user on the target advertisement can be obtained.
The embodiment of the invention provides an advertisement click prediction method, which adopts a neural network model to carry out embedding dense processing on historical advertisement characteristics and attribute characteristics, and obtains user embedding characteristics and advertisement embedding characteristics represented by dense vectors through the embedding dense processing, wherein for dense vectors which are relatively high-dimensional sparse vectors, most dimensional samples are effective, so that the problem of prediction precision reduction caused by data loss of high-dimensional sparse vectors in the prior art is solved; and on the basis of performing dense processing on the historical advertisement characteristics and the attribute characteristics of the user independently, the user characteristics and the advertisement characteristics are fused and embedded in multiple layers, and the information of the user and the advertisement can be mined from different layers, so that the comprehensiveness of data processing is improved, the information of the user and the advertisement is fully fused, and the prediction precision of the user on the advertisement is further improved.
The inputting the historical advertisement features and the attribute features into a click prediction model, sequentially embedding the historical advertisement features and the attribute features to be dense, performing cross fusion processing on the processed historical advertisement features and the processed attribute features, and predicting the click rate of the target user on the target advertisement specifically comprises:
embedding the historical advertisement features and the attribute features to be dense, so as to obtain user embedding features and advertisement embedding features respectively;
performing cross fusion and average pooling on the user embedding characteristics and the advertisement embedding characteristics to obtain fusion characteristics;
and predicting the click rate of the target user on the target advertisement according to the fusion characteristics.
Specifically, embedding density processing is carried out on historical advertisement features, and user embedding features are obtained; embedding the attribute characteristics to obtain advertisement embedding characteristics; performing cross fusion and average pooling on the user embedding characteristics and the advertisement embedding characteristics to obtain fusion characteristics; and predicting the click rate of the target user on the target advertisement according to the fusion characteristics.
And carrying out embedding density processing on the historical advertisement features to obtain user embedding features, and carrying out embedding density processing on the attribute features to obtain advertisement embedding features.
It should be noted that, embedding dense processing refers to mapping features into a dense vector, embedding refers to converting (dimensionality reduction) data into a feature representation (vector) of a fixed size for processing and calculation (e.g., distance calculation), the user embedded features and advertisement embedded features obtained after the embedding dense processing are dense vectors, the dense vectors are relatively sparse vectors, sample values of most dimensions in the sparse vectors are 0, only a small part of valid sample values, and correspondingly, sample values of most dimensions in the dense vectors are valid, so that the problem of data missing in the prior art does not exist when the features are mapped into the dense vectors. General class features are sparser after being subjected to single-hot coding, and are subjected to embedding dense processing to obtain denser user embedding features and advertisement embedding features, so that the problem of dimension explosion caused by directly inputting sparse features is avoided.
And then, carrying out cross fusion and average pooling on the user embedding characteristics and the advertisement embedding characteristics to obtain fusion characteristics, wherein the cross fusion of the characteristics can fully combine the advantages of the user embedding characteristics and the advertisement embedding characteristics, mine user information and advertisement information from different layers, fully fuse the user information and the advertisement information, and improve the prediction precision of the user on the advertisement.
And then predicting the click rate of the target user on the target advertisement according to the fusion characteristics.
The embodiment of the invention carries out embedding density processing on the original historical advertisement characteristics and attribute characteristics, and obtains the denser user embedding characteristics and advertisement embedding characteristics because the category characteristics are sparser after being subjected to unique hot coding and are subjected to embedding density processing, thereby avoiding the problem of data loss in the process of mapping low-dimensional vectors to high-dimensional spaces in the prior art and simultaneously overcoming the problem of dimension explosion caused by directly inputting sparse characteristics; in addition, the advantages of the user information and the advertisement information can be fully combined by performing cross fusion processing on the user embedding characteristics and the advertisement embedding characteristics, and the prediction precision is improved.
Specifically, mapping high-dimensional sparse features in the historical advertisement features to dense features of the users; normalizing the user continuous features in the advertisement features to obtain normalized user continuous features; splicing the dense user features and the normalized continuous user features to obtain user splicing features; and carrying out multilayer embedding on the user splicing characteristics to obtain the user embedding characteristics.
The high-dimensional sparse feature can be understood as that the low-dimensional dense feature is mapped to a high-dimensional space, a large number of 0 exists in the high-dimensional sparse feature, the number of 0 is far greater than that of other values, and the high-dimensional sparse feature is generally a class feature, is usually expressed by one-hot and is a vector in mathematical form.
Firstly, mapping high-dimensional sparse features in historical advertisement features into dense user features, normalizing continuous features in the historical advertisement features, splicing the dense user features and the normalized continuous user features to obtain spliced user features, and recording the spliced user features as xuser。
It should be further noted that the dense user features can be obtained by multiplying the high-dimensional sparse features by a preset transformation matrix. By taking 1024 independent words as an example for explanation, the high-dimensional sparse feature is represented by one-hot, namely 1024 dimensions, and the method has the advantages that the dimensions are mutually orthogonal, and the cosine similarity is 0. The dense vector representation dimension is lower than 1024 dimensions, and is 10 dimensions at minimum, i.e. binary coded representation.
In order to further mine the characteristics of the user, a multi-layer embedding needs to be carried out on the user splicing characteristics, the main idea is to project the user embedding characteristics into m different subspaces to obtain different representations of user embedding, and the calculation formula is as follows:
wherein x isuserRepresenting the user's splice characteristics and,
representing the ith element in the user-embedded feature,
representing the ith element of the preset user embedding matrix.
By carrying out multi-layer embedding processing on the historical advertisement characteristics of the user, the information contained in the user is fully mined, and the user information existing in the historical advertisement characteristics is mined from different layers, so that the prediction precision of the click rate is improved.
Specifically, mapping high-dimensional sparse features in attribute features of the target advertisements to dense advertisement features; normalizing the advertisement continuous features in the attribute features to obtain normalized advertisement continuous features; splicing the advertisement dense features and the normalized advertisement continuous features to obtain advertisement splicing features; and embedding the advertisement splicing characteristics in multiple layers to obtain the advertisement embedding characteristics.
In the same way, the high-dimensional sparse features in the attribute features are mapped into the advertisement dense features, the continuous features in the attribute features are normalized, and the advertisement dense features and the normalized advertisement continuous features are splicedObtaining the advertisement splicing characteristics which are marked as xitemThe feature is represented by a vector.
In order to further mine the characteristics of the advertisement, a multi-layer embedding needs to be carried out on the advertisement splicing characteristics, the main idea is to project the advertisement embedding characteristics into m different subspaces to obtain different expressions of advertisement embedding, and the calculation formula is as follows:
wherein x isitemThe characteristics of the splicing of the advertisements are represented,
representing the ith element in the ad embedding feature,
representing the ith element of the preset ad embedding matrix.
By carrying out multi-layer embedding processing on the attribute features of the advertisements, the advertisement information existing in the attribute features is mined from different layers, and the information contained in the advertisements is fully mined, so that the prediction precision of the click rate is improved.
After the user embedding characteristic and the advertisement embedding characteristic are obtained by the method, the characteristics are embedded into the user
And advertisement embedding features
The fusion is carried out in the following specific fusion mode:
wherein the content of the first and second substances,
represents the ith element of the fused user embedded feature,
the ith element of the embedded characteristic of the fused advertisement is shown, j, m and k are positive integers, e is an identity matrix,
representing the ith element of the user's embedded feature,
representing the ith element in the ad embedding feature.
Then, performing an average pooling (i.e. averaging) on the fused features to obtain fused features:
wherein the content of the first and second substances,
representing the average pooled user fusion features,
representing the average pooled ad fusion characteristics.
On the basis of the foregoing embodiment, preferably, the predicting the click rate of the target user for the target advertisement according to the fusion feature includes:
and inputting the fusion characteristics into a full connection layer, and obtaining the click rate by the output of the full connection layer through an activation function.
In the specific implementation process, the fusion features are input into an output layer, the output layer is a full-connection network, then an activation function is connected, and the purpose of introducing the activation function is as follows: because the data distribution is mostly nonlinear, while the calculation of the general neural network is linear, the neural network can simulate the nonlinear data distribution by introducing the activation function, and the learning capability of the network is enhanced. The largest feature of the activation function is nonlinearity. Several activation functions are common: the Sigmoid function, the tanh function and the ReLU function are selected according to the distribution characteristics of the simulation data, and in the embodiment of the invention, the Sigmoid function is selected as the activation function.
And obtaining the click rate of the user to the advertisement by the data output by the output layer through a sigmoid activation function, wherein the calculation formula of the output layer is as follows:
where y represents click through rate, FCN represents full convolution network, xuserRepresenting a user splice feature, xitemThe characteristics of splicing of the advertisements are represented,
representing the average pooled user fusion features,
representing the average pooled ad fusion characteristics.
Fig. 3 is a schematic structural diagram of a click prediction model in an embodiment of the present invention, and as shown in fig. 3, in a specific implementation process, the click prediction model is composed of three layers, namely an embedding layer 310, an interacting layer 320 and an output layer 330, the embedding layer further includes two parallel independent user embedding units 311 and an advertisement embedding unit 312, the user embedding units are used for embedding historical advertisement features of users, and the advertisement embedding units are used for embedding attribute features of advertisements. And the embedding layer, the interaction layer and the output layer are sequentially connected, and finally, the embedding layer, the interaction layer and the output layer are connected and transmitted to the sigmoid layer.
Specifically, the embedding layer comprises a user embedding unit and an advertisement embedding unit, historical advertisement characteristics are input into the user embedding unit to obtain user embedding characteristics, attribute characteristics of the target advertisement are input into the advertisement embedding unit to obtain advertisement embedding characteristics, and the embedding layer separates and processes the user and the advertisement so as to fully mine user information and advertisement information.
And inputting the user embedding characteristics and the advertisement embedding characteristics into the interaction layer to obtain fusion characteristics, and inputting the fusion characteristics into the output layer to obtain the click rate of the target user on the target advertisement.
On the basis of the above embodiment, preferably, the method for obtaining the user sample and the advertisement sample includes:
obtaining an initial user sample and an initial advertisement sample;
and respectively processing the initial user sample and the initial advertisement sample by adopting a preset rule to obtain a user sample and an advertisement sample, wherein the preset rule comprises missing value filling and layered sampling of a nearest neighbor algorithm in sequence.
The obtained initial user sample and the initial advertisement sample are the most original data samples, the sample data of the initial user sample and the initial advertisement sample has a lot of noises, and in the application of subsequent click prediction model training, the problems of partial feature data loss and extreme imbalance of positive and negative samples can be faced, which causes very bad influence on the subsequent click prediction model training and causes low precision of the trained click prediction model.
Processing the initial user sample in sequence by adopting a missing value filling technology and a layered sampling technology of a nearest neighbor algorithm to obtain a user sample;
and processing the initial advertisement sample in sequence by adopting a missing value filling technology and a layered sampling technology of a nearest neighbor algorithm to obtain the advertisement sample.
In the embodiment of the invention, a missing value filling technology of a nearest neighbor algorithm (KNN) is adopted to fill data in an initial user sample and an initial advertisement, an auxiliary variable (namely a variable without a missing value) is utilized to define a distance function between samples, K samples without the missing value, which are closest to the missing value sample, are searched, and the missing data is filled by utilizing an average value or a weighted average value of the K samples without the missing value.
Then, a hierarchical sampling technology is adopted to respectively cluster initial user samples and initial advertisement sample data, a large number of category samples (negative samples) are clustered, a certain number of samples are randomly extracted from each category according to a proportion to form a new negative sample set, and the sampling can keep the balanced distribution of the samples to a great extent, so that the prediction effect of the model is ensured on the premise of the balance of the positive and negative samples.
It should be noted that if a photo classification of a car is made, the positive sample is the correct picture of the car, and the negative sample is whether the picture of the car is the correct picture of the car. Through model training, the user can tell the click to predict the model, and the model is right and wrong, and the right is a positive sample and the wrong is a negative sample.
In the embodiment of the invention, missing value filling is carried out through a nearest neighbor algorithm in a sample collection stage, and the balance degree of a positive sample and a negative sample in the sample is ensured through a hierarchical sampling technology, so that the prediction precision of a click prediction model is ensured.
On the basis of the above embodiment, it is preferable to further include:
acquiring a comprehensive score of the target advertisement according to the click rate and the delivery price of the target advertisement;
and judging whether the target advertisement is delivered or not according to the comprehensive score.
Specifically, the price weight of the target advertisement is determined according to the placement price of the target advertisement and other advertisement bids, and the placement price of the target advertisement is expressed in a weight form, so that the scores of all advertisements are unified conveniently.
And obtaining the comprehensive score of the target advertisement by multiplying the target advertisement delivery price by the click rate.
Fig. 4 is an architecture diagram of an advertisement bidding delivery system based on an advertisement click-through rate prediction method of a neural network according to an embodiment of the present invention, and as shown in fig. 4, the system includes an OBM410(Own Branding and Manufacturing), a DSP420(Demand-Side Platform), a DPP430(Data Product Platform), and an AEther 440.
The OBM comprises 4 advertisement strategies for a user to select, wherein the strategy 1 is a default advertisement, namely a default advertisement interface is opened; strategy 2 is a model advertisement, namely, an advertisement which is most likely to be clicked by a user is selected according to the user characteristics and the advertisement characteristics, and strategy 3 is a company customized strategy, wherein the company customization refers to that a customized interface is displayed for a specific enterprise or a specific user.
The method comprises the steps that a model advertisement strategy and a company customization strategy are directly adopted for specific customers (including new users) (the default advertisements or the customized advertisements are customized in advance and are directly displayed), and a strategy 2 adopted for general users needs to enter a DSP advertisement putting platform subsequently, and a proper advertisement is selected through a model by combining user characteristics and advertisement characteristics.
The DSP platform is used for managing the advertisements and transmitting corresponding information to the DPP and the OBM according to the selection of the advertiser.
The DPP is used for looking up the crowd packet to which the user belongs, acquiring the user ID, the material information and the advertisement information transmitted by the DSP, and packaging the input parameters of the prediction model;
the Ather modeling platform is used for predicting the advertisement click rate according to the user ID, the material information and the advertisement information.
The terminal 1 is a certain vehicle management APP, the terminal 2 is a certain advertisement management APP, and the vehicle insurance delivery is taken as an example for explanation, and the specific implementation process in the advertisement flow of the accurate pushing seen by the user accessing the vehicle management APP and the algorithm model involved in the process are described in detail below.
Firstly, a user accesses a vehicle management APP, and the system sends out an advertisement putting request
When a user accesses the vehicle management APP, the OBM is triggered to select an advertising strategy, generally, a strategy 2 is selected, and the OBM simultaneously executes two action instructions, sends an advertising request of the strategy 2 to the DSP, and simultaneously communicates with the inside of the DPP system to inquire historical advertising characteristics of a crowd represented by a logged-in user.
Second, DSP advertisement putting platform request prediction result
An advertiser logs in an advertisement management APP and triggers a DSP platform to acquire the attribute characteristics of a target advertisement; and after receiving the OBM strategy 2 advertisement request, the DSP packages the advertisement and material information to the DPP, namely the attribute characteristics, the attribute characteristics are combined with the historical advertisement characteristics and sent to the DPP, and the DPP requests a click prediction model deployed by an Aether data analysis modeling platform to obtain a prediction result.
And (3) advertising materials: the method is characterized in that some pictures, videos and the like are needed when advertisements are made, namely, one advertisement is composed of a plurality of advertisement materials;
and (3) advertisement operation: overall Process for advertisement initiation, planning and execution
Bidding and releasing: the novel network advertisement form is characterized by being autonomously put by a user, autonomously managed, ranked by adjusting price and paid according to the advertisement effect.
And thirdly, obtaining a real-time click rate prediction result by the AEther according to the prediction request by combining information such as user ID, advertisement materials, crowd characteristics and the like.
Fourthly, putting in combination with bidding of advertisers
And the DSP receives the click rate and advertiser bidding price to carry out weight calculation to obtain the comprehensive score of the target advertisement.
The advertisement bidding delivery system provided by the embodiment has the following advantages:
1. in the existing advertisement industry, a plurality of advertisement putting platforms currently support advertisers to select advertisement slots and bid on the advertisement slots, but do not support advertisement click rate prediction and selection of advertisement audience groups, which results in poor advertisement putting effect. The advertisement bidding delivery system provided by the embodiment of the invention supports advertisement bidding on the basis of advertisement click rate prediction, and can select suitable audience populations of the advertisement according to the comprehensive score to realize accurate delivery of the advertisement.
2. In the utilization of the existing advertisement resources, the use of the same resource position in the same time interval cannot reasonably distribute the use of the resource position according to scientific basis, and the launching effect cannot be scientifically estimated, so that the marketing resource utilization efficiency is low; the use of the advertising marketing resource slot completely belongs to planned delivery, and the advertising marketing resource slot is used successively according to resource requirements, lacks marketable competition, and cannot be used as a commodity to maximize the benefit of the commodity. The advertisement bidding and releasing system comprehensively considers the rights and interests of an advertiser, a user and a traffic party, takes an advertisement click rate prediction model of thousands of people and thousands of faces as a core, and pushes a proper advertisement to interested crowds on the premise of ensuring the releasing requirements of the traffic party and the advertiser as much as possible.
3. The advertisement click rate is not estimated by combining a scientific algorithm model, and advertisement audience users are not selected and delivered by combining crowd images, so that the advertisement demand seen by the users is low in matching, the click rate is low, the conversion rate is low, and a large amount of waste and invalid information of exposure resource positions are caused. The advertisement bidding and releasing system builds a DSP label system through a project, and accurate recommendation capability of thousands of people and thousands of faces is provided for a user; building a click rate prediction model and a development model of an insurance system, and providing an effect-oriented intelligent delivery capacity for an advertiser; and a calculation advertisement data report system and an optimization strategy are built, and an automatic advertisement operation decision service capability is provided for the platform.
Fig. 5 is a schematic structural diagram of an advertisement click-through rate prediction system based on a neural network according to an embodiment of the present invention, as shown in fig. 5, the system includes a feature obtaining module 510 and a prediction module 520, where:
the feature obtaining module 510 is configured to obtain historical advertisement features of a target user and attribute features of a target advertisement, where the historical advertisement features include basic information of the target user and historical purchase information between the target user and a target product, and the attribute features are obtained according to the basic information of the target advertisement;
the prediction module 520 is configured to input the historical advertisement features and the attribute features into a click prediction model, sequentially perform embedding density processing on the historical advertisement features and the attribute features, perform cross fusion processing on the processed historical advertisement features and the processed attribute features, and predict a click rate of the target user on the target advertisement, where the click prediction model is obtained by training a user sample, an advertisement sample, and a score tag.
The embodiment is a system embodiment corresponding to the advertisement click prediction method, the implementation process of which is the same as the method embodiment, please refer to the method embodiment for details, and the system embodiment is not described herein again.
According to the advertisement click rate prediction system based on the neural network, the neural network model is adopted to carry out embedded dense processing and cross fusion processing on historical advertisement characteristics and attribute characteristics, and the historical advertisement characteristics and the attribute characteristics are spliced with continuous characteristics, so that the problem that prediction accuracy is reduced due to high-dimensional sparse characteristic data loss in the prior art is solved; and on the basis of performing dense processing on historical advertisement characteristics of the user and attribute characteristics of the target advertisement, the user characteristics and the advertisement characteristics are fused, so that information of the user and the advertisement can be mined from different layers, the information of the user and the advertisement is fully fused, and the prediction accuracy of the user on the advertisement is further improved.
Specifically, the prediction module further comprises a user embedding module, an advertisement embedding module, a fusion module and an output module, wherein:
the user embedding module is used for carrying out embedding density processing on the historical advertisement characteristics to obtain user embedding characteristics;
the advertisement embedding module is used for embedding the attribute characteristics to obtain advertisement embedding characteristics;
the fusion module is used for performing cross fusion and average pooling on the user embedding characteristics and the advertisement embedding characteristics to obtain fusion characteristics;
and the output module is used for predicting the click rate of the target user on the target advertisement according to the fusion characteristics.
Specifically, the user embedding module includes a user mapping unit, a user normalization unit, a user splicing unit and a user embedding unit, wherein:
the user mapping unit is used for mapping the high-dimensional sparse features in the historical advertisement features into dense user features;
the user normalization unit is used for normalizing the user continuous features in the historical advertisement features to obtain normalized user continuous features;
the user splicing unit is used for splicing the dense user features and the normalized continuous user features to obtain user splicing features;
and the user embedding unit is used for carrying out multilayer embedding on the user splicing characteristics to obtain the user embedding characteristics.
Specifically, the advertisement embedding module comprises an advertisement mapping unit, an advertisement normalizing unit, an advertisement splicing unit and an advertisement embedding unit, wherein:
the advertisement mapping unit is used for mapping high-dimensional sparse features in the attribute features of the target advertisements into advertisement dense features;
the advertisement normalization unit is used for normalizing the advertisement continuous features in the attribute features to obtain normalized advertisement continuous features;
the advertisement splicing unit is used for splicing the advertisement dense characteristics and the normalized advertisement continuous characteristics to obtain advertisement splicing characteristics;
the advertisement embedding unit is used for embedding the advertisement splicing characteristics in a multi-layer mode to obtain advertisement embedding characteristics.
Specifically, the output module includes:
and inputting the fusion characteristics into a full connection layer, and obtaining the click rate by the output of the full connection layer through a sigmoid function.
Specifically, the user sample and the advertisement sample in the prediction module are obtained by the following steps:
obtaining an initial user sample and an initial advertisement sample;
processing the initial user sample in sequence by adopting a missing value filling technology and a layered sampling technology of a nearest neighbor algorithm to obtain the user sample;
and processing the initial advertisement sample in sequence by adopting a missing value filling technology and a hierarchical sampling technology of a nearest neighbor algorithm to obtain the advertisement sample.
Specifically, the advertisement click prediction system further comprises a synthesis module and a delivery module, wherein:
the comprehensive module is used for acquiring the comprehensive score of the target advertisement according to the click rate and the delivery price of the target advertisement;
and the delivery module is used for judging whether to deliver the target advertisement according to the comprehensive score.
The present embodiment is a system embodiment corresponding to the advertisement click rate prediction method based on the neural network, and the specific implementation process is consistent with the above method embodiment, please refer to the above method embodiment for details, which is not described herein again.
The modules in the neural network-based advertisement click-through rate prediction system may be implemented in whole or in part by software, hardware, and combinations thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
Fig. 6 is a schematic structural diagram of a computer device according to an embodiment of the present invention, as shown in fig. 6, the computer device may be a server, and its internal structural diagram may be as shown in fig. 6. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a computer storage medium and an internal memory. The computer storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the computer storage media. The database of the computer device is used for storing data generated or acquired in the process of executing the neural network-based advertisement click rate prediction method, such as a process node number and a target service node. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a neural network-based advertisement click-through rate prediction method.
In one embodiment, a computer device is provided, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor executes the computer program to implement the steps of the neural network-based advertisement click rate prediction method in the above embodiments. Alternatively, the processor, when executing the computer program, implements the functions of the modules/units in the embodiment of the neural network-based advertisement click-through rate prediction system.
In an embodiment, a computer storage medium is provided, on which a computer program is stored, which when executed by a processor implements the steps of the neural network-based advertisement click rate prediction method in the above embodiments. Alternatively, the computer program may be executed by a processor to implement the functions of the modules/units in the embodiment of the neural network-based advertisement click-through rate prediction system described above.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.
Claims (10)
1. An advertisement click rate prediction method based on a neural network is characterized by comprising the following steps:
acquiring historical advertisement characteristics of a target user and attribute characteristics of a target advertisement, wherein the historical advertisement characteristics comprise basic information of the target user and historical purchase information between the target user and a target product, and the attribute characteristics are acquired according to the basic information of the target advertisement;
inputting the historical advertisement features and the attribute features into a click prediction model, and sequentially embedding the historical advertisement features and the attribute features to obtain user embedded features and advertisement embedded features respectively, wherein the click prediction model is obtained by training a user sample, an advertisement sample and a score tag;
and performing cross fusion processing on the user embedding characteristics and the advertisement embedding characteristics to predict the click rate of the target user on the target advertisement.
2. The method according to claim 1, wherein the step of inputting the historical advertisement features and the attribute features into a click prediction model, performing dense embedding processing on the historical advertisement features and the attribute features in sequence, performing cross fusion processing on the processed historical advertisement features and the processed attribute features, and predicting the click rate of the target user on the target advertisement specifically comprises:
embedding the historical advertisement features and the attribute features to be dense, so as to obtain user embedding features and advertisement embedding features respectively;
performing cross fusion and average pooling on the user embedding characteristics and the advertisement embedding characteristics to obtain fusion characteristics;
and predicting the click rate of the target user on the target advertisement according to the fusion characteristics.
3. The method of claim 2, wherein the embedding density processing is performed on the historical advertisement features and the attribute features to obtain user embedded features and advertisement embedded features, respectively, and the method comprises:
mapping high-dimensional sparse features in the historical advertisement features to dense features of the users; and/or mapping high-dimensional sparse features in the attribute features of the target advertisements to advertisement dense features;
normalizing the user continuous features in the historical advertisement features to obtain normalized user continuous features; and/or normalizing the advertisement continuous features in the attribute features to obtain normalized advertisement continuous features;
splicing the dense user features and the normalized continuous user features to obtain user splicing features; and/or splicing the advertisement dense features and the normalized advertisement continuous features to obtain advertisement splicing features;
and respectively embedding the user splicing characteristics and/or the advertisement splicing characteristics in multiple layers to obtain the user embedding characteristics and the advertisement embedding characteristics.
4. The method according to claim 2, wherein the user splicing features and/or the advertisement splicing features are embedded in multiple layers to obtain the user embedding features and the advertisement embedding features, and the specific calculation formula is as follows:
wherein x isuserRepresenting said user splicing feature, xitemThe characteristics of the splicing of the advertisements are represented,
representing the ith element in the user-embedded feature,
representing the ith element in the ad embedding feature,
represents the ith element of the preset user embedding matrix,
represents the ith element in the preset advertisement embedding matrix, and m represents the number of embedding into the subspace.
5. The method of claim 2, wherein predicting the click-through rate of the target user for the target advertisement according to the fusion feature comprises:
and inputting the fusion characteristics into a full connection layer, and obtaining the click rate by the output of the full connection layer through an activation function.
6. The method of any one of claims 1 to 5, wherein the method of obtaining the user sample and the advertisement sample comprises:
obtaining an initial user sample and an initial advertisement sample;
and respectively processing the initial user sample and the initial advertisement sample by adopting a preset rule to obtain a user sample and an advertisement sample, wherein the preset rule comprises missing value filling and layered sampling of a nearest neighbor algorithm in sequence.
7. The neural network-based advertisement click-through rate prediction method according to any one of claims 1 to 5, further comprising:
acquiring a comprehensive score of the target advertisement according to the click rate and the delivery price of the target advertisement;
and judging whether the target advertisement is delivered or not according to the comprehensive score.
8. An advertisement click-through rate prediction system based on a neural network, comprising:
the characteristic acquisition module is used for acquiring historical advertisement characteristics of a target user and attribute characteristics of a target advertisement, wherein the historical advertisement characteristics comprise basic information of the target user and historical purchase information between the target user and a target product, and the attribute characteristics are acquired according to the basic information of the target advertisement;
and the prediction module is used for inputting the historical advertisement features and the attribute features into a click prediction model, sequentially embedding the historical advertisement features and the attribute features into a dense process, performing cross fusion processing on the processed historical advertisement features and the processed attribute features, and predicting the click rate of the target user on the target advertisement, wherein the click prediction model is obtained by training a user sample, an advertisement sample and a score tag.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor when executing the computer program performs the steps of the neural network based advertisement click rate prediction method according to any one of claims 1 to 7.
10. A computer storage medium storing a computer program, wherein the computer program when executed by a processor implements the steps of the neural network-based advertisement click-through rate prediction method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111016975.XA CN113706211B (en) | 2021-08-31 | 2021-08-31 | Advertisement click rate prediction method and system based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111016975.XA CN113706211B (en) | 2021-08-31 | 2021-08-31 | Advertisement click rate prediction method and system based on neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113706211A true CN113706211A (en) | 2021-11-26 |
| CN113706211B CN113706211B (en) | 2024-04-02 |
Family
ID=78658430
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111016975.XA Active CN113706211B (en) | 2021-08-31 | 2021-08-31 | Advertisement click rate prediction method and system based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113706211B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023140781A3 (en) * | 2022-01-19 | 2023-08-24 | Lemon Inc. | Embedding optimization for a machine learning model |
| CN117455366A (en) * | 2023-12-19 | 2024-01-26 | 杭州网易云音乐科技有限公司 | Inventory allocation method, device, medium and equipment based on neural network model |
| CN117522479A (en) * | 2023-11-07 | 2024-02-06 | 北京创信合科技有限公司 | Accurate Internet advertisement delivery method and system |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070112840A1 (en) * | 2005-11-16 | 2007-05-17 | Yahoo! Inc. | System and method for generating functions to predict the clickability of advertisements |
| US20140108094A1 (en) * | 2012-06-21 | 2014-04-17 | Data Ventures, Inc. | System, method, and computer program product for forecasting product sales |
| CN110619540A (en) * | 2019-08-13 | 2019-12-27 | 浙江工业大学 | Click stream estimation method of neural network |
| CN110825957A (en) * | 2019-09-17 | 2020-02-21 | 中国平安人寿保险股份有限公司 | Deep learning-based information recommendation method, device, equipment and storage medium |
| CN111325579A (en) * | 2020-02-25 | 2020-06-23 | 华南师范大学 | Advertisement click rate prediction method |
| CN111339415A (en) * | 2020-02-25 | 2020-06-26 | 中国科学技术大学 | Click rate prediction method and device based on multi-interactive attention network |
| CN111506811A (en) * | 2020-03-19 | 2020-08-07 | 上海理工大学 | Click rate prediction method based on deep residual error network |
| US20200327603A1 (en) * | 2019-04-10 | 2020-10-15 | The Climate Corporation | Leveraging feature engineering to boost placement predictability for seed product selection and recommendation by field |
| CN112365297A (en) * | 2020-12-04 | 2021-02-12 | 东华理工大学 | Advertisement click rate estimation method |
| CN112381581A (en) * | 2020-11-17 | 2021-02-19 | 东华理工大学 | Advertisement click rate estimation method based on improved Transformer |
| KR102232393B1 (en) * | 2020-03-26 | 2021-03-25 | 한국해양대학교 산학협력단 | Method and apparatus for developing economic operation solution of autonomous ship |
| CN112700274A (en) * | 2020-12-29 | 2021-04-23 | 华南理工大学 | Advertisement click rate estimation method based on user preference |
-
2021
- 2021-08-31 CN CN202111016975.XA patent/CN113706211B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070112840A1 (en) * | 2005-11-16 | 2007-05-17 | Yahoo! Inc. | System and method for generating functions to predict the clickability of advertisements |
| US20140108094A1 (en) * | 2012-06-21 | 2014-04-17 | Data Ventures, Inc. | System, method, and computer program product for forecasting product sales |
| US20200327603A1 (en) * | 2019-04-10 | 2020-10-15 | The Climate Corporation | Leveraging feature engineering to boost placement predictability for seed product selection and recommendation by field |
| CN110619540A (en) * | 2019-08-13 | 2019-12-27 | 浙江工业大学 | Click stream estimation method of neural network |
| CN110825957A (en) * | 2019-09-17 | 2020-02-21 | 中国平安人寿保险股份有限公司 | Deep learning-based information recommendation method, device, equipment and storage medium |
| CN111325579A (en) * | 2020-02-25 | 2020-06-23 | 华南师范大学 | Advertisement click rate prediction method |
| CN111339415A (en) * | 2020-02-25 | 2020-06-26 | 中国科学技术大学 | Click rate prediction method and device based on multi-interactive attention network |
| CN111506811A (en) * | 2020-03-19 | 2020-08-07 | 上海理工大学 | Click rate prediction method based on deep residual error network |
| KR102232393B1 (en) * | 2020-03-26 | 2021-03-25 | 한국해양대학교 산학협력단 | Method and apparatus for developing economic operation solution of autonomous ship |
| CN112381581A (en) * | 2020-11-17 | 2021-02-19 | 东华理工大学 | Advertisement click rate estimation method based on improved Transformer |
| CN112365297A (en) * | 2020-12-04 | 2021-02-12 | 东华理工大学 | Advertisement click rate estimation method |
| CN112700274A (en) * | 2020-12-29 | 2021-04-23 | 华南理工大学 | Advertisement click rate estimation method based on user preference |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023140781A3 (en) * | 2022-01-19 | 2023-08-24 | Lemon Inc. | Embedding optimization for a machine learning model |
| CN117522479A (en) * | 2023-11-07 | 2024-02-06 | 北京创信合科技有限公司 | Accurate Internet advertisement delivery method and system |
| CN117455366A (en) * | 2023-12-19 | 2024-01-26 | 杭州网易云音乐科技有限公司 | Inventory allocation method, device, medium and equipment based on neural network model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113706211B (en) | 2024-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111368210B (en) | Information recommendation method and device based on artificial intelligence and electronic equipment | |
| WO2020135535A1 (en) | Recommendation model training method and related apparatus | |
| CN113706211B (en) | Advertisement click rate prediction method and system based on neural network | |
| CN109783730A (en) | Products Show method, apparatus, computer equipment and storage medium | |
| WO2022252363A1 (en) | Data processing method, computer device and readable storage medium | |
| US20210056458A1 (en) | Predicting a persona class based on overlap-agnostic machine learning models for distributing persona-based digital content | |
| CN111008335B (en) | Information processing method, device, equipment and storage medium | |
| CN112287238B (en) | User characteristic determination method and device, storage medium and electronic equipment | |
| CN109801101A (en) | Label determines method, apparatus, computer equipment and storage medium | |
| CN112348592A (en) | Advertisement recommendation method and device, electronic equipment and medium | |
| US20240104159A1 (en) | Creating an effective product using an attribute solver | |
| US20230316106A1 (en) | Method and apparatus for training content recommendation model, device, and storage medium | |
| Singhal et al. | An E‐commerce prediction system for product allocation to bridge the gap between cultural analytics and data science | |
| CN112115354A (en) | Information processing method, information processing apparatus, server, and storage medium | |
| CN117057855A (en) | Data processing method and related device | |
| CN114881712B (en) | Intelligent advertisement putting method, device, equipment and storage medium | |
| CN116756281A (en) | Knowledge question-answering method, device, equipment and medium | |
| CN113570422B (en) | Creative guide information generation method and device, computer equipment and storage medium | |
| WO2023050143A1 (en) | Recommendation model training method and apparatus | |
| CN115222112A (en) | Behavior prediction method, behavior prediction model generation method and electronic equipment | |
| CN114693409A (en) | Product matching method, device, computer equipment, storage medium and program product | |
| CN116069959A (en) | Method and device for determining display data | |
| CN115329183A (en) | Data processing method, device, storage medium and equipment | |
| CN114417944B (en) | Recognition model training method and device, and user abnormal behavior recognition method and device | |
| KR102651114B1 (en) | Methods, devices and systems for generating artificial intelligence information through real-time analysis and distributed processing of ad exchange-related big data and optimizing the targeting of ad space offerings |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |