CN112258207A - Advertisement flow determination method, device, equipment and storage medium - Google Patents
Advertisement flow determination method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN112258207A CN112258207A CN202010550505.0A CN202010550505A CN112258207A CN 112258207 A CN112258207 A CN 112258207A CN 202010550505 A CN202010550505 A CN 202010550505A CN 112258207 A CN112258207 A CN 112258207A
- Authority
- CN
- China
- Prior art keywords
- advertisement
- traffic
- flow
- time period
- webpage type
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000012549 training Methods 0.000 claims description 34
- 230000001419 dependent effect Effects 0.000 claims description 18
- 230000015654 memory Effects 0.000 claims description 15
- 238000003062 neural network model Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 5
- 230000000875 corresponding effect Effects 0.000 description 39
- 230000008685 targeting Effects 0.000 description 34
- 230000008569 process Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 10
- 238000005516 engineering process Methods 0.000 description 9
- 238000012360 testing method Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 6
- 238000012417 linear regression Methods 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 5
- 210000002569 neuron Anatomy 0.000 description 4
- 210000004027 cell Anatomy 0.000 description 3
- 230000006403 short-term memory Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 2
- 230000007787 long-term memory Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004883 computer application Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 235000013350 formula milk Nutrition 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000000843 powder Substances 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 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
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Transfer Between Computers (AREA)
Abstract
The embodiment of the invention discloses a method, a device, equipment and a storage medium for determining advertisement traffic. The method comprises the following steps: acquiring the orientation condition and the playing time period of the advertisement to be played, resolving the webpage type and the user attribute from the orientation condition, and determining the counted advertisement flow corresponding to the playing time period under the webpage type; inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, and obtaining the total advertisement flow of the advertisement to be played under the webpage type and the playing time period according to the output result of the time sequence prediction model; and determining the targeted advertisement flow of the advertisement to be played under the targeted condition and the playing time period according to the flow ratio of the user attribute under the webpage type and the total advertisement flow. According to the technical scheme of the embodiment of the invention, the effects of accurate prediction and efficient prediction of the advertisement flow are realized through effective matching of the time sequence prediction model and the flow ratio.
Description
Technical Field
The embodiment of the invention relates to the technical field of computer application, in particular to a method, a device, equipment and a storage medium for determining advertisement flow.
Background
In recent years, with the rapid development of internet advertising technology, various related technologies capable of improving advertising effects have been developed, and crowd targeting is one of the technologies. Specifically, different advertisers often own respective potential user groups, and each potential user group often has some obvious characteristics, for example, an advertiser of high-end imported infant milk powder wants to play advertisements to a user group of 25-34 years old, school calendar above university, and mobile phone price above 3000 yuan, while an advertiser of middle and low-end electric vehicles wants to play advertisements to a user group of male, 35-44 years old, high and low, and mobile phone price within 2000 yuan.
Currently, most advertising platforms do not provide predictive data of advertisement traffic to advertisers after the advertisers have set targeting conditions, and thus, advertisers can only set budgets and plan by virtue of prior experience. To solve this problem, the prior art proposes a technical solution that takes advertisement traffic of the same period and satisfying the targeting condition set by the advertiser as prediction data. For example, a game advertiser sets targeting conditions that are "science-type web pages" and "gender is male" and "age is 18-24 years" and "academic history is major or subject" in which the "science-type web pages" are web page types, and "gender is male", "age is 18-24 years" and "academic history is major or subject" are user attributes, and the game advertiser intends to predict the advertisement traffic that the targeting condition can bring in the next week, and in this case, the current technology is to provide the game advertiser with statistical data of the advertisement traffic that satisfies the targeting condition in the same week as the previous month as prediction data.
In the process of implementing the invention, the inventor finds that the advertisement traffic predicted according to a single time factor in the prior art is likely to have large discrepancy with the actual advertisement traffic, and the accuracy of advertisement traffic prediction is low.
Disclosure of Invention
The embodiment of the invention provides a method, a device, equipment and a storage medium for determining advertisement traffic, which are used for improving the accuracy of advertisement traffic prediction.
In a first aspect, an embodiment of the present invention provides an advertisement traffic determining method, which may include:
acquiring the orientation condition and the playing time period of the advertisement to be played, resolving the webpage type and the user attribute from the orientation condition, and determining the counted advertisement flow corresponding to the playing time period under the webpage type;
inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, and obtaining the total advertisement flow of the advertisement to be played under the webpage type and the playing time period according to the output result of the time sequence prediction model;
and determining the targeted advertisement flow of the advertisement to be played under the targeted condition and the playing time period according to the flow ratio of the user attribute under the webpage type and the total advertisement flow.
In a second aspect, an embodiment of the present invention further provides an advertisement traffic determination apparatus, which may include:
a module for determining the statistic advertisement flow rate, which is used for obtaining the targeting condition and the playing time period of the advertisement to be played, resolving the webpage type and the user attribute from the targeting condition, and determining the statistic advertisement flow rate corresponding to the playing time period in the webpage type
The total advertisement flow determining module is used for inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type and obtaining the total advertisement flow of the advertisement to be played under the webpage type and the playing time period according to the output result of the time sequence prediction model;
and the targeted advertisement flow determining module is used for determining the targeted advertisement flow of the advertisement to be played under the targeted condition and the playing time period according to the flow ratio of the user attribute under the webpage type and the total advertisement flow.
In a third aspect, an embodiment of the present invention further provides an apparatus, where the apparatus may include:
one or more processors;
a memory for storing one or more programs;
when executed by one or more processors, cause the one or more processors to implement the advertisement traffic determination methods provided by any of the embodiments of the present invention.
In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the advertisement traffic determination method provided in any embodiment of the present invention.
According to the technical scheme of the embodiment of the invention, the webpage type and the user attribute can be decomposed from the directional condition by acquiring the directional condition and the playing time period of the advertisement to be played, and the counted advertisement flow corresponding to the playing time period under the webpage type is determined; inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, so as to obtain the total advertisement flow of all users of the advertisement to be played under the webpage type and the playing time period; furthermore, according to the total advertisement flow and the flow ratio of the user attribute of the targeted user in the webpage type, the targeted advertisement flow of the targeted user of the advertisement to be played in the webpage type and the playing time period can be obtained. According to the technical scheme, the time sequence prediction model and the traffic proportion are effectively matched, so that the effects of accurate prediction and efficient prediction of the advertisement traffic are achieved.
Drawings
Fig. 1 is a flowchart of an advertisement traffic determination method according to a first embodiment of the present invention;
fig. 2 is a schematic diagram of a targeting condition in an advertisement traffic determination method according to a first embodiment of the present invention;
fig. 3a is a schematic diagram illustrating a timing prediction model trained according to a web page type in an advertisement traffic determination method according to an embodiment of the present invention;
fig. 3b is a first schematic diagram of the operation process of the long-term and short-term memory network unit in the advertisement traffic determination method according to the first embodiment of the present invention;
fig. 3c is a second schematic diagram of the operation process of the long-term and short-term memory network unit in the advertisement traffic determination method according to the first embodiment of the present invention;
fig. 4 is a flowchart of an advertisement traffic determination method according to a second embodiment of the present invention;
fig. 5a is a schematic application diagram of an advertisement traffic determination method in the second embodiment of the present invention.
FIG. 5b is a diagram illustrating the operation of the targeting condition querier in the advertisement traffic determination method according to the second embodiment of the present invention;
fig. 6 is a block diagram of an advertisement traffic determination apparatus according to a third embodiment of the present invention;
fig. 7 is a schematic structural diagram of an apparatus according to a fourth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Example one
Fig. 1 is a flowchart of an advertisement traffic determination method according to an embodiment of the present invention. The embodiment can be applied to the situation of determining the targeted advertisement flow of the advertisement to be played under the targeted condition and the playing time period. The method can be executed by the advertisement flow determination device provided by the embodiment of the invention, the device can be realized by software and/or hardware, and the device can be integrated on various user terminals or servers.
Referring to fig. 1, the method of the embodiment of the present invention specifically includes the following steps:
s110, acquiring the directional condition and the playing time period of the advertisement to be played, resolving the webpage type and the user attribute from the directional condition, and determining the counted advertisement flow corresponding to the playing time period and under the webpage type.
The targeting condition is a condition set by an advertiser for realizing advertisement targeting, and a targeting attribute value of each targeting attribute can be determined according to the targeting condition. For example, as shown in fig. 2, the targeting condition may be "gender ═ male" and "age ═ unlimited" and "school calendar ═ university and above" and "network ═ mobile network", and "model price-. In practical application, the targeting attribute may be a user attribute, or a web page type, where the user attribute is attribute information of a targeting user, such as gender, age, academic calendar, network, model price, and the like, and the web page type is a web page type of a web page where an advertisement is to be played, such as news, sports, games, science and technology, and therefore, the web page type and the user attribute may be resolved from the targeting condition.
The playing time period is a time period expected to be played by the advertisement to be played, and the playing time period of the advertisement to be played can be determined according to the playing time period, and is the total duration of the playing time period, such as 1 day, 1 week, one month and the like. For example, assuming that the play period is 6/8/2020-6/14/2020, the play time period is 1 week. The counted advertisement traffic is the advertisement traffic of all users that have occurred in the past corresponding to a certain playing time period under a certain webpage type, the advertisement traffic is the advertisement display times, namely the number of independent visitors (UV) of the webpage where the advertisement is located, the webpage type is the webpage type decomposed from the targeting condition, the playing time period is the playing time period of the advertisement to be played, and all users mean that all user attributes do not have any limitation.
The reason for this is that the web page type of each web page is set at the beginning of the establishment, and the flow-time variation trends of the web pages with the same web page type are similar, for example, the flow peak of the related web page of the news type is generally on the working day, the flow peak of the related web page of the sports type is generally on some sports competition days, and the flow peak of the related web page of the travel type is generally before the holiday, etc. As known from practical experience, the total number of types of web pages is usually within 100. Thus, the counted ad traffic for the type of web page matching the targeting conditions may be used to predict the targeted ad traffic. In addition, the counted advertisement traffic corresponding to the playing time period is the counted advertisement traffic corresponding to the playing time period, and the specific selection is related to the training process of the timing prediction model described later, which will be described in detail later.
S120, inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, and obtaining the total advertisement flow of the advertisement to be played under the webpage type and the playing time period according to the output result of the time sequence prediction model.
Each webpage type corresponds to a respective trained time sequence prediction model, because the flow-time change trends of the webpages with the same webpage type are similar, the time sequence prediction model can predict the total advertisement flow of the advertisement to be played under the corresponding webpage type and the corresponding playing time period according to the counted advertisement flow, and the total advertisement flow is the advertisement flow under the condition that the attributes of various users are not limited. On this basis, the time sequence prediction Model may be a trained Long Short Term Memory network Model (LSTM), an Autoregressive Moving Average Model (ARMA), an Autoregressive Integrated Moving Average Model (ARIMA), or the like.
It should be noted that the time sequence prediction model closely links the advertisement traffic of the webpage type and the time factor, and the effective connection of the advertisement traffic and the time factor improves the prediction accuracy of the advertisement traffic. In addition, the timing sequence prediction model predicts the total advertisement flow of all users under a certain webpage type, and does not relate to specific user attributes, which means that respective timing sequence prediction models do not need to be trained for various user attributes, and the timing sequence prediction models corresponding to the webpage types with limited quantity can be completely trained in an off-line state relative to various user attributes with huge quantity, so that the corresponding total advertisement flow can be directly predicted based on the trained timing sequence prediction models during on-line application, and the advertisement flow prediction efficiency is improved.
S130, determining the targeted advertisement flow of the advertisement to be played under the targeted condition and in the playing time period according to the flow ratio of the user attribute under the webpage type and the total advertisement flow.
The total advertisement flow is the advertisement flow of all users which are expected to visit the webpage where the advertisement to be played is located, the all users comprise the targeted users and the non-targeted users of the advertisement to be played, the targeted users are users which accord with the user attribute in the targeted condition, and the non-targeted users are users which do not accord with the user attribute in the targeted condition. Considering that there is a difference in the degree of tendency of users having different user attributes to each web page type, that is, there is a difference in traffic occupancy of users having different user attributes in each web page type, for example, a male user has a higher traffic occupancy in a game type and a lower traffic occupancy in a shopping type relative to a female user. Therefore, the flow attribute of the historical advertisement flow under the webpage type can be determined, the flow attribute can present various user attributes of all users, and the flow ratio of each user attribute under the webpage type can be obtained by counting the flow attribute.
Illustratively, as shown in table 1, taking the traffic attribute of the historical advertisement traffic under the financial type as an example, it may be composed of a large number of user access logs, each of which may include the access time of the user, the access web page, the model price, the location, and the like, and may further include the gender, age, academic history, and the like. Therefore, the traffic ratio of each user attribute in each webpage type can be obtained according to the traffic attributes shown in table 1, such as the traffic ratio of a male user in a financial type, the traffic ratio of a high-school user in a financial type, and the like. In practical applications, some user attributes of some users may be unknown, and such user attributes may be counted "without limitation", for example, the model price in row 2 of table 1 is unknown, which may be counted "without limitation".
TABLE 1 flow Attribute
Further, the number of web page types resolved from the targeting conditions may be one, two or more, and similarly, the number of user attributes resolved from the targeting conditions may also be one, two or more. Taking a certain webpage type as an example, when the number of the user attributes is one, the traffic proportion of the user attributes in the webpage type can be directly obtained according to the statistical result. When the number of the user attributes is at least two, the traffic proportion of the at least two user attributes under the webpage type can be directly obtained according to the statistical result, namely the traffic proportion of the user with the at least two user attributes under the webpage type; the traffic proportion of each user attribute in the at least two user attributes under the webpage type can be obtained according to the statistical result, and then the traffic proportion of the at least two user attributes under the webpage type can be obtained according to each traffic proportion; the method can also group at least two user attributes, obtain the traffic proportion of each group of user attributes under the webpage type according to the statistical result, and obtain the traffic proportion of the at least two user attributes under the webpage type according to each traffic proportion; etc., and are not specifically limited herein.
And further, after the flow rate of the user attribute in the targeting condition under the webpage type is calculated, the targeting advertisement flow rate of the advertisement to be played under the targeting condition and the playing time period can be determined according to the product result of the flow rate ratio and the total advertisement flow rate, wherein the targeting advertisement flow rate is the advertisement flow rate of the user which is expected to access the webpage where the advertisement to be played is located and accords with the user attribute in the targeting condition.
According to the technical scheme of the embodiment of the invention, the webpage type and the user attribute can be decomposed from the directional condition by acquiring the directional condition and the playing time period of the advertisement to be played, and the counted advertisement flow corresponding to the playing time period under the webpage type is determined; inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, so as to obtain the total advertisement flow of all users of the advertisement to be played under the webpage type and the playing time period; furthermore, according to the total advertisement flow and the flow ratio of the user attribute of the targeted user in the webpage type, the targeted advertisement flow of the targeted user of the advertisement to be played in the webpage type and the playing time period can be obtained. According to the technical scheme, the time sequence prediction model and the traffic proportion are effectively matched, so that the effects of accurate prediction and efficient prediction of the advertisement traffic are achieved.
An optional technical solution is that the time sequence prediction model may be obtained by pre-training through the following steps: acquiring historical advertisement traffic of a webpage type corresponding to the time sequence prediction model, extracting multiple groups of training samples from the historical advertisement traffic based on a preset sliding window strategy, and training the original neural network model based on the multiple groups of training samples to obtain the time sequence prediction model; each group of training samples comprises historical advertisement traffic of a plurality of preset time periods and historical advertisement traffic of the next preset time period of the plurality of preset time periods. It should be noted that the execution subject of the training process of the timing prediction model and the execution subject of the application process may be the same or different, and are not specifically limited herein.
The number of the historical advertisement traffic is multiple, each historical advertisement traffic is the advertisement traffic of all users of the webpage type corresponding to the time sequence prediction model in a certain time period in the past time period, such as the advertisement traffic of a certain day, the advertisement traffic of a certain week, the advertisement traffic of a certain month and the like in the past time period, and the past time period is a preset time period, such as the past 1 year, the past 2 years and the like. In practical application, optionally, historical traffic data of each webpage may be aggregated according to a webpage type based on a map-reduce distributed computing model (MapReduce), where the historical traffic data is advertisement traffic of all users of a certain webpage within a certain time period in the past time period, and the historical traffic data of each webpage belonging to the same webpage type are aggregated together to obtain respective historical advertisement traffic of each webpage type.
Extracting multiple groups of training samples from historical advertisement traffic based on a preset sliding window strategy, wherein actual input data in each group of training samples are historical advertisement traffic of multiple preset time periods, the preset time periods are certain preset time periods in the past time periods, the preset time periods are different from one another, the total duration of the preset time periods is consistent with the time period of the historical advertisement traffic, if the historical advertisement traffic is a statistical value of the advertisement traffic of a certain day in the past time periods, the total duration of the preset time periods is a day, and if the historical advertisement traffic is a statistical value of the advertisement traffic of a certain week in the past time periods, the total duration of the preset time periods is a week; the expected output data in each set of training samples is historical advertisement traffic of a next preset time period of the plurality of preset time periods, namely historical advertisement traffic of a next preset time period of a last time period of the plurality of preset time periods, for example, the plurality of preset time periods are today, yesterday and the day before, the next preset time period is tomorrow, and for example, the plurality of preset time periods are this week, last week and last week, the next preset time period is the next week. Thus, training the original neural network model, which may be untrained LSTM, ARMA, ARIMA, or the like, based on sets of training samples, may result in a timing prediction model.
On this basis, optionally, determining the counted advertisement traffic corresponding to the playing time period in the webpage type includes: determining a historical time period corresponding to the playing time period according to a plurality of preset time periods; and taking the historical advertisement flow under the webpage type and the historical time period as the counted advertisement flow. The number of the historical time periods and the number of the preset time periods may be consistent, and the total duration of each historical time period and the total duration of each preset time period may also be consistent. For example, if the preset time periods are adjacent 3 days and the play time period is tomorrow, the historical time periods are today, yesterday and the previous day, respectively, and then the counted ad traffic may include today's historical ad traffic, yesterday's historical ad traffic and the previous day's historical ad traffic in the corresponding webpage type.
In order to better understand the specific implementation process of the above steps, the following describes an exemplary training process of the timing prediction model according to this embodiment with reference to a specific example. For example, taking the original neural network model as an untrained LSTM and the historical advertisement traffic as the advertisement traffic of a certain day in the past N days as an example, as shown in fig. 3a, the historical traffic data of each web page is aggregated according to the web page type, so as to obtain the respective N-day traffic data of each web page type (i.e., the historical advertisement traffic of N days, where N is 730 is an example), and then, taking the historical advertisement traffic of a news type as an example, the training process of the LSTM is explained.
Obtaining N-day historical advertisement traffic for news types [ A0，A1，......AN-1]Wherein A is0Is the historical advertisement traffic of day 1 within N days, A1Is the historical advertisement traffic of day 2 within N days, and so on. From [ A ] based on a preset sliding window strategy0，A1，......AN-1]Extracting continuous historical advertisement traffic [ B ] with length of m0，B1，......Bm-1]And m +1 th historical advertisement traffic BmWherein [ B ]0，B1，......Bm-1]Is the historical advertisement traffic of m preset time periods in the training sample, BmIs the historical advertisement traffic for the next preset time period of the m preset time periods in the training sample. If the historical advertisement traffic of 1 st to 30 th days in N days is actual input data, the historical advertisement traffic of 31 th day is expected output data; if the historical advertisement traffic of days 2-31 in N days is the actual input data, the historical advertisement traffic of day 32 is the expected output data. On the basis, optionally, in order to improve the model prediction accuracy, the training samples may be normalized, that is, [ B0，B1，......Bm-1]Is processed into [1, B ]1/B0，......Bm-1/B0]And B ismIs processed into Bm/B0. Thus, the advertisement traffic predicted by the time series prediction model is the advertisement traffic trend from day m +1 to day m, rather than the actual advertisement traffic.
The specific result of the LSTM cells in the LSTM is shown in FIG. 3b, which is to compare the input data h of the first LSTM cell0And cell state c0Initialization is 0; weight matrix W of hidden layercAnd bias bcWeight matrix W of input gatesiAnd bias biWeight matrix W of output gatesOAnd bias boAnd weight matrix W of forgetting gatefAnd biasbfInitializing to a random number between 0 and 1; setting the number of neurons in an input layer of the LSTM as z +1, wherein the number of neurons in a hidden layer, an input gate, a forgetting gate and an output gate is z; weighting matrix W of linear regression networklAnd bias blThe initialization is random number between 0 and 1, the number of neurons in the input layer of the linear regression network is z, and the number of neurons in the output layer is 1. It should be noted that the linear regression network only matches with the last 1 LSTM units, as shown in FIG. 3c, and the output data h of the last 1 LSTM unitstCan be input into the linear regression network because h (t) and the output Y of the linear regression networkpre-trainThere is a correlation between, Ypre-trainIs the prediction result output by the time sequence prediction model. In fig. 3B and 3c, h (t-1) is the output value of t-1-day LSTM, x (t) is the input value of t-day LSTM, and x (t) is Bt，B0Input to the 1 st LSTM unit … … Bm-1Inputting to the mth LSTM unit; σ denotes sigmoid function transformation, and tanh denotes hyperbolic tangent function transformation.
80% of the multiple sets of training samples are taken as training set (X)trainAnd Ytrain) And the other 20% as test set (X)testAnd Ytest) Where X is the actual input data and Y is the desired output data. Setting an optimization goal to LosstargetT-round training is performed on LSTM, assuming that the current iteration number K is 1. Thus, a) XtrainAnd h0Respectively inputting the data into a hidden layer, an input gate, a forgetting gate and an output gate of the LSTM to obtain an output value of the hidden layer of the LSTM on the t day and an output value i of the input gatetOutput value f of forgetting gatetAnd the output value o of the output gatet. b) Calculating to obtain cell states c (t) and output values h (t) of the last 1 LSTM unit, and inputting h (t) into a linear regression network to obtain Ypre-trainAnd according to Ypre-trainAnd YtrainThe Mean Square Error (MSE) is calculated. c) And (4) adjusting the network parameters of the LSTM by taking the MSE as a loss function, and updating the network parameters according to the adjustment result. d) And K is K +1, if K is more than T, the model training is initially finished, otherwise, the steps a) to d) are repeatedly executed. 4) Mixing XtestInputting into LSTM after preliminary training to obtain Ypre-testMSE as evaluation criterion, according to Ypre-testAnd YtestCalculate LosstestIf Losstest＞LosstargetThe model continues to be optimized, otherwise, the model training is completely finished. The training process of the timing prediction models of the other web page types is similar to that described above, and is not described herein again.
Example two
Fig. 4 is a flowchart of an advertisement traffic determination method provided in the second embodiment of the present invention. The present embodiment is optimized based on the above technical solutions. In this embodiment, optionally, the method for determining advertisement traffic may further include: the independent user attributes and the dependent user attributes are separated from the user attributes, and the dependent user attributes are combined to obtain combined user attributes; respectively acquiring a combined flow rate ratio of the combined user attribute in the webpage type and an independent flow rate ratio of the independent user attribute in the webpage type; and determining the traffic ratio of the user attribute under the webpage type according to the combined traffic ratio and the independent traffic ratio. The same or corresponding terms as those in the above embodiments are not explained in detail herein.
Referring to fig. 4, the method of this embodiment may specifically include the following steps:
s210, acquiring the directional condition and the playing time period of the advertisement to be played, resolving the webpage type and the user attribute from the directional condition, and determining the counted advertisement flow corresponding to the playing time period under the webpage type;
s220, inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, and obtaining the total advertisement flow of the advertisement to be played under the webpage type and the playing time period according to the output result of the time sequence prediction model.
And S230, resolving the independent user attribute and the dependent user attribute from the user attributes, and combining the dependent user attributes to obtain a combined user attribute.
The number of the user attributes is multiple, independent user attributes and non-independent user attributes can be decomposed from the multiple user attributes, each independent user attribute is a user attribute which is independent of the traffic ratio of historical advertisement traffic, and taking sex and age as examples, the traffic ratio of male users is alpha%, the traffic ratio of 20-year-old users is beta%, and the traffic ratio of 20-year-old male users is alpha beta/100%, so that sex and age are independent user attributes; the dependent user attribute is a user attribute in which traffic occupancy of historical advertisement traffic is correlated, and taking "scholars" and "revenues" as an example, traffic occupancy of the subject user is γ%, traffic occupancy of the 7000k user is δ%, and traffic occupancy of the subject 7000k user is greater than γ δ/100%, then "scholars" and "revenues" are dependent user attributes correlated with each other.
On this basis, the dependent user attributes may be combined to obtain a combined user attribute, which is a user attribute that needs to satisfy the dependent user attributes at the same time, such as combining the dependent user attributes "age is 30" and "income is 7000 k" into a combined user attribute "age is 30" and "income is 7000 k". Therefore, any independent user attribute and any combined user attribute are mutually independent, namely any two independent user attributes are mutually independent, any two combined user attributes are mutually independent, and any independent user attribute and any combined user attribute are mutually independent.
Further, the directional attribute value of the combined user attribute may be obtained by: taking two mutually associated non-independent user attributes A and B as an example, the directional attribute value of A comprises
The directional property value of B comprises
Then, the directional property value of the combined directional property Φ may be
And
cartesian product of (a)1&b1，a1&b2...a1&bv，a2&b1...a2&bv...au&bv]One of them has u × v values in total. Illustratively, the directional attribute values of age include "15 years" and "16 years", the directional attribute values of the academic calendar include "junior middle school", "senior high school", and "college student", and the combined user attributes that the two can constitute include "15 years junior middle school", "15 years senior middle school", "16 years junior middle school", "16 years senior middle school", and "16 years college student". Accordingly, the processing of the three interrelated dependent user attributes is similar, and the directional attribute value of the combined user attribute may be a Cartesian product of the directional attribute values of the three dependent user attributes. In practical applications, optionally, two or three mutually associated non-independent user attributes may be combined into a combined user attribute, and for at least four mutually associated non-independent user attributes, the three non-independent user attributes with the smallest directional attribute value may be selected to generate the combined user attribute, for example, the combined user attribute is generated based on "academic history", "income", and "gender" in "age", "academic history", "income", and "gender", thereby avoiding an excessive number of combined user attributes.
S240, respectively obtaining the combined flow rate of the combined user attribute in the webpage type and the independent flow rate of the independent user attribute in the webpage type.
The independent flow rate ratio of each independent user attribute in the webpage type and the combined flow rate ratio of each combined user attribute in the webpage type are respectively obtained. For example, the traffic attributes of the historical advertisement traffic in the web page type may be obtained first, where the traffic attributes include various user attributes of all users in the web page type, and the traffic occupation ratio of each independent user attribute and each combined user attribute in the web page type may be obtained according to the traffic attributes. For example, needlesFor the independent user attribute, the ratio of the flow count of the independent user attribute to the total flow calculation of the flow attribute can be used as an independent flow ratio, and the total flow count is the total user flow for accessing the webpage type; for the combined user attribute, the ratio of the traffic count of the combined user attribute to the total traffic calculation of the traffic attribute may be used as the combined traffic ratio, that is, the traffic ratio of the combined user attribute under the historical advertisement traffic is determined according to the traffic attribute, and the traffic ratio is used as the combined traffic ratio of the combined user attribute under the webpage type. Illustratively, taking table 2 as an example, for example, the combined user attribute "age 22-30" and "scholarly home" and "income 5000-1/T1The independent traffic ratio of the independent user attribute "region ═ Sichuan" under "house type" is F5/T4。
TABLE 2 flow Rate
S250, determining the flow rate ratio of the user attribute in the webpage type according to the combined flow rate ratio and the independent flow rate ratio, and determining the directional advertisement flow rate of the advertisement to be played under the directional condition and the playing time period according to the flow rate ratio of the user attribute in the webpage type and the total advertisement flow rate.
After the combined traffic proportion and the independent traffic proportion are obtained, the traffic proportion of the user attribute in the webpage type can be determined. For example, considering that any independent user attribute and any combined user attribute are independent from each other, the product of each combined traffic ratio and each independent traffic ratio can be used as the traffic ratio of the user attribute in the web page type.
According to the technical scheme of the embodiment of the invention, the independent user attributes and the dependent user attributes are separated from the user attributes, and the dependent user attributes are combined to obtain the combined user attributes, so that the flow ratio of any user attribute in the webpage type can be rapidly calculated according to the combined flow ratio of the combined user attributes in the webpage type and the independent flow ratio of the independent user attributes in the webpage type, and the prediction efficiency of the advertisement flow is improved.
In order to better understand the specific implementation process of the above steps, the advertisement traffic determination method of the present embodiment is exemplarily described below with reference to specific examples. For example, as shown in fig. 5a, the advertisement traffic determination method according to the embodiment of the present invention may be divided into an offline part and an online part, where the offline part is a timing prediction model obtained by training based on historical advertisement traffic in a certain webpage type and an independent probability model obtained by statistics, and the total advertisement traffic may be predicted according to the timing prediction model, and the traffic ratio may be calculated according to the independent probability model; the online part is a directional condition querier which can obtain the directional advertisement flow of the advertisement to be played under the directional condition and the playing time period according to the directional condition and the playing time period of the advertisement to be played and by combining the output results of the time sequence prediction model and the independent probability model.
Specifically, the operation process of the targeting condition querier is as shown in fig. 5b, and a web page type (e.g., "technology type") and a user attribute (e.g., "sex is male" and "age is 18-24 years" and "academic calendar is major" and "region is sichuan") are resolved from the targeting condition of the advertisement to be played set by the advertiser. Further, predicting the total advertisement flow gamma of the advertisement to be played in the playing time period based on a time sequence prediction model corresponding to the science and technology type; in each user attribute, "sex is male" and "region is Sichuan" are independent user attributes, and "age is 18-24 years" and "academic calendar is major" are dependent user attributes, thus, the combined user attribute of age & scholarly & age & scholarly & 18-24 years is combined with the age & scholarly & 18-24 years, therefore, the flow rate proportion phi of the user attribute under the science and technology type can be obtained based on the independent probability model corresponding to the science and technology type, therefore, the targeted advertisement flow E ═ Γ Φ of the advertisement to be played under the targeted condition and the playing time period, wherein, Φ is P (male, 18-24 years old, college, sichuan) is P (male, 18-24 years old & college, sichuan) is P (male) P (18-24 years old & college).
According to the technical scheme, the prediction accuracy of the advertisement flow is improved based on the time sequence prediction model, the flow ratio of any user attribute can be calculated through the independent probability model, the prediction efficiency of the advertisement flow is improved, the two are matched with each other, and the targeted advertisement flow under any directional condition and any playing time period can be predicted efficiently and accurately according to the historical advertisement flows of all users under the corresponding webpage types. Accurate advertisement flow prediction is helpful for an advertiser to judge whether the promotion target in the playing time period can be reached, and the method has more visual and accurate feeling on the advertisement coverage and the drainage capacity and has greater economic significance.
EXAMPLE III
Fig. 6 is a block diagram of an advertisement traffic determination apparatus according to a third embodiment of the present invention, which is configured to execute an advertisement traffic determination method according to any of the above embodiments. The device and the advertisement traffic determination method of each embodiment belong to the same inventive concept, and details which are not described in detail in the embodiment of the advertisement traffic determination device can refer to the embodiment of the advertisement traffic determination method. Referring to fig. 6, the apparatus may specifically include: a counted ad traffic determination module 310, an aggregate ad traffic determination module 320, and a targeted ad traffic determination module 330.
The counted advertisement traffic determining module 310 is configured to obtain a targeting condition and a playing time period of an advertisement to be played, resolve a webpage type and a user attribute from the targeting condition, and determine a counted advertisement traffic corresponding to the playing time period and in the webpage type;
the aggregate advertisement traffic determining module 320 is configured to input the counted advertisement traffic into a trained timing sequence prediction model corresponding to the web page type, and obtain the aggregate advertisement traffic of the advertisement to be played in the web page type and the playing time period according to an output result of the timing sequence prediction model;
the targeted advertisement traffic determining module 330 is configured to determine, according to the traffic proportion of the user attribute in the webpage type and the aggregate advertisement traffic, a targeted advertisement traffic of the advertisement to be played under the targeted condition and in the playing time period.
Optionally, on the basis of the above apparatus, the apparatus may further include:
the combined user attribute obtaining module is used for resolving the independent user attributes and the dependent user attributes from the user attributes and combining the dependent user attributes to obtain the combined user attributes;
the flow ratio acquisition module is used for respectively acquiring the combined flow ratio of the combined user attribute in the webpage type and the independent flow ratio of the independent user attribute in the webpage type;
and the flow ratio determining module is used for determining the flow ratio of the user attribute under the webpage type according to the combined flow ratio and the independent flow ratio.
Optionally, the traffic ratio obtaining module may specifically include:
the flow ratio determining unit is used for acquiring the flow attribute of the historical advertisement flow of the webpage type and determining the flow ratio of the combined user attribute under the historical advertisement flow according to the flow attribute;
and the combined flow ratio determining unit is used for taking the flow ratio of the combined user attribute under the historical advertisement flow as the combined flow ratio of the combined user attribute under the webpage type.
Optionally, the flow ratio determining module may be specifically configured to:
and taking the product result of the combined traffic ratio and the independent traffic ratio as the traffic ratio of the user attribute under the webpage type.
Optionally, on the basis of the above apparatus, the apparatus may further include:
the timing prediction model training module is used for acquiring historical advertisement traffic of a webpage type corresponding to the timing prediction model, extracting multiple groups of training samples from the historical advertisement traffic based on a preset sliding window strategy, and training an original neural network model based on the multiple groups of training samples to obtain a timing prediction model;
each group of training samples comprises historical advertisement traffic of a plurality of preset time periods and historical advertisement traffic of the next preset time period of the plurality of preset time periods.
Optionally, the statistical advertisement traffic determining module 310 may include:
and the counted advertisement flow determining unit is used for determining a historical time period corresponding to the playing time period according to a plurality of preset time periods, and taking the historical advertisement flow under the webpage type and the historical time period as the counted advertisement flow.
Alternatively, the raw neural network model may comprise an untrained long-short term memory network model.
According to the advertisement traffic determination device provided by the third embodiment of the invention, the statistical advertisement traffic determination module is used for acquiring the targeting condition and the playing time period of the advertisement to be played, so that the webpage type and the user attribute can be decomposed from the targeting condition, and the statistical advertisement traffic corresponding to the playing time period in the webpage type is determined; the total advertisement flow determining module inputs the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, so that the total advertisement flow of all users of the advertisements to be played under the webpage type and the playing time period can be obtained; furthermore, the targeted advertisement flow rate determining module can obtain the targeted advertisement flow rate of the targeted user of the advertisement to be played in the webpage type and the playing time period according to the total advertisement flow rate and the flow rate ratio of the user attribute of the targeted user in the webpage type. By means of the device, the effects of accurate prediction and efficient prediction of the advertisement flow are achieved through effective matching of the time sequence prediction model and the flow ratio.
The advertisement traffic determination device provided by the embodiment of the invention can execute the advertisement traffic determination method provided by any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.
It should be noted that, in the embodiment of the advertisement traffic determination apparatus, the units and modules included in the embodiment are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be implemented; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.
Example four
Fig. 7 is a schematic structural diagram of an apparatus according to a fourth embodiment of the present invention, as shown in fig. 7, the apparatus includes a memory 410, a processor 420, an input device 430, and an output device 440. The number of processors 420 in the device may be one or more, and one processor 420 is taken as an example in fig. 7; the memory 410, processor 420, input device 430, and output device 440 of the apparatus may be connected by a bus or other means, such as by bus 450 in fig. 7.
The memory 410, which is a computer-readable storage medium, may be used for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the advertisement traffic determination method in the embodiment of the present invention (e.g., the counted advertisement traffic determination module 310, the aggregate advertisement traffic determination module 320, and the targeted advertisement traffic determination module 330 in the advertisement traffic determination device). The processor 420 executes various functional applications of the device and data processing by executing software programs, instructions, and modules stored in the memory 410, that is, implements the advertisement traffic determination method described above.
The memory 410 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to use of the device, and the like. Further, the memory 410 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, memory 410 may further include memory located remotely from processor 420, which may be connected to devices through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input device 430 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function controls of the device. The output device 440 may include a display device such as a display screen.
EXAMPLE five
An embodiment of the present invention provides a storage medium containing computer-executable instructions, which when executed by a computer processor, are configured to perform a method for advertisement traffic determination, where the method includes:
acquiring the orientation condition and the playing time period of the advertisement to be played, resolving the webpage type and the user attribute from the orientation condition, and determining the counted advertisement flow corresponding to the playing time period under the webpage type;
inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, and obtaining the total advertisement flow of the advertisement to be played under the webpage type and the playing time period according to the output result of the time sequence prediction model;
and determining the targeted advertisement flow of the advertisement to be played under the targeted condition and the playing time period according to the flow ratio of the user attribute under the webpage type and the total advertisement flow.
Of course, the storage medium provided by the embodiment of the present invention contains computer-executable instructions, and the computer-executable instructions are not limited to the method operations described above, and may also perform related operations in the advertisement traffic determination method provided by any embodiment of the present invention.
From the above description of the embodiments, it is obvious for those skilled in the art that the present invention can be implemented by software and necessary general hardware, and certainly, can also be implemented by hardware, but the former is a better embodiment in many cases. With this understanding, the technical solutions of the present invention may be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, and includes instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute the methods according to the embodiments of the present invention.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (10)
1. An advertisement traffic determination method, comprising:
acquiring a directional condition and a playing time period of an advertisement to be played, resolving a webpage type and a user attribute from the directional condition, and determining a counted advertisement flow corresponding to the playing time period and under the webpage type;
inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, and obtaining the total advertisement flow of the advertisement to be played under the webpage type and the playing time period according to the output result of the time sequence prediction model;
and determining the targeted advertisement flow of the advertisement to be played under the targeted condition and the playing time period according to the flow ratio of the user attribute under the webpage type and the total advertisement flow.
2. The method of claim 1, further comprising:
resolving independent user attributes and dependent user attributes from the user attributes, and combining the dependent user attributes to obtain combined user attributes;
respectively acquiring the combined traffic ratio of the combined user attribute under the webpage type and the independent traffic ratio of the independent user attribute under the webpage type;
and determining the traffic ratio of the user attribute under the webpage type according to the combined traffic ratio and the independent traffic ratio.
3. The method of claim 2, wherein the obtaining the combined traffic fraction of the combined user attribute under the webpage type comprises:
acquiring the flow attribute of the historical advertisement flow of the webpage type, and determining the flow ratio of the combined user attribute under the historical advertisement flow according to the flow attribute;
and taking the traffic proportion of the combined user attribute under the historical advertisement traffic as the combined traffic proportion of the combined user attribute under the webpage type.
4. The method of claim 2, wherein determining the traffic fraction of the user attribute in the web page type according to the combined traffic fraction and the independent traffic fraction comprises:
and taking the product result of the combined traffic ratio and the independent traffic ratio as the traffic ratio of the user attribute under the webpage type.
5. The method of claim 1, further comprising:
acquiring historical advertisement traffic of the webpage type corresponding to the time sequence prediction model, extracting multiple groups of training samples from the historical advertisement traffic based on a preset sliding window strategy, and training an original neural network model based on the multiple groups of training samples to obtain the time sequence prediction model;
wherein each set of training samples includes the historical advertisement traffic for a plurality of preset time periods and the historical advertisement traffic for a next preset time period of the plurality of preset time periods.
6. The method of claim 5, wherein the determining the counted advertisement traffic for the webpage type and corresponding to the playing time period comprises:
determining a historical time period corresponding to the playing time period according to the plurality of preset time periods;
and taking the historical advertisement flow under the webpage type and the historical time period as the counted advertisement flow.
7. The method of claim 5, wherein the raw neural network model comprises an untrained long-short term memory network model.
8. An advertisement traffic determination apparatus, comprising:
the system comprises a counted advertisement flow determining module, a playing time period determining module and a statistical advertisement flow determining module, wherein the counted advertisement flow determining module is used for acquiring a directional condition and a playing time period of an advertisement to be played, resolving a webpage type and a user attribute from the directional condition, and determining the counted advertisement flow corresponding to the playing time period in the webpage type;
the total advertisement flow determining module is used for inputting the counted advertisement flow into a trained time sequence prediction model corresponding to the webpage type, and obtaining the total advertisement flow of the advertisement to be played under the webpage type and the playing time period according to the output result of the time sequence prediction model;
and the targeted advertisement flow determining module is used for determining the targeted advertisement flow of the advertisement to be played under the targeted condition and the playing time period according to the flow ratio of the user attribute under the webpage type and the total advertisement flow.
9. An apparatus, comprising:
one or more processors;
a memory for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the advertisement traffic determination method of any of claims 1-7.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the advertisement traffic determination method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010550505.0A CN112258207A (en) | 2020-06-16 | 2020-06-16 | Advertisement flow determination method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010550505.0A CN112258207A (en) | 2020-06-16 | 2020-06-16 | Advertisement flow determination method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112258207A true CN112258207A (en) | 2021-01-22 |
Family
ID=74223770
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010550505.0A Pending CN112258207A (en) | 2020-06-16 | 2020-06-16 | Advertisement flow determination method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112258207A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101079063A (en) * | 2007-06-25 | 2007-11-28 | 腾讯科技（深圳）有限公司 | Method, system and apparatus for transmitting advertisement based on scene information |
| US20090043648A1 (en) * | 2007-08-08 | 2009-02-12 | Yahoo! Inc. | Pay-per-action system for selling advertisements |
| CN101399848A (en) * | 2008-09-01 | 2009-04-01 | 腾讯科技（深圳）有限公司 | System and method for bandwidth flow prediction for webpage advertisement |
| CA2846353A1 (en) * | 2011-08-23 | 2013-02-28 | Audience Partners, Llc | Targeting online ads based on political demographics |
| CN109784978A (en) * | 2018-12-19 | 2019-05-21 | 平安科技（深圳）有限公司 | Advertisement competition power calculation method, device, medium and equipment based on big data |
| CN111062746A (en) * | 2019-12-06 | 2020-04-24 | 微梦创科网络科技（中国）有限公司 | Advertisement flow estimation method and device and electronic equipment |
-
2020
- 2020-06-16 CN CN202010550505.0A patent/CN112258207A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101079063A (en) * | 2007-06-25 | 2007-11-28 | 腾讯科技（深圳）有限公司 | Method, system and apparatus for transmitting advertisement based on scene information |
| US20090043648A1 (en) * | 2007-08-08 | 2009-02-12 | Yahoo! Inc. | Pay-per-action system for selling advertisements |
| CN101399848A (en) * | 2008-09-01 | 2009-04-01 | 腾讯科技（深圳）有限公司 | System and method for bandwidth flow prediction for webpage advertisement |
| CA2846353A1 (en) * | 2011-08-23 | 2013-02-28 | Audience Partners, Llc | Targeting online ads based on political demographics |
| CN109784978A (en) * | 2018-12-19 | 2019-05-21 | 平安科技（深圳）有限公司 | Advertisement competition power calculation method, device, medium and equipment based on big data |
| CN111062746A (en) * | 2019-12-06 | 2020-04-24 | 微梦创科网络科技（中国）有限公司 | Advertisement flow estimation method and device and electronic equipment |
Non-Patent Citations (1)
| Title |
|---|
| 纪文迪;王晓玲;周傲英;: "广告点击率估算技术综述", 华东师范大学学报(自然科学版), no. 03, pages 2 - 14 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11531867B2 (en) | User behavior prediction method and apparatus, and behavior prediction model training method and apparatus | |
| CN110457589B (en) | Vehicle recommendation method, device, equipment and storage medium | |
| Chen et al. | MOOC dropout prediction using a hybrid algorithm based on decision tree and extreme learning machine | |
| WO2021081962A1 (en) | Recommendation model training method, recommendation method, device, and computer-readable medium | |
| WO2020135535A1 (en) | Recommendation model training method and related apparatus | |
| WO2021135562A1 (en) | Feature validity evaluation method and apparatus, and electronic device and storage medium | |
| CN111798280B (en) | Multimedia information recommendation method, device and equipment and storage medium | |
| CN112765480B (en) | Information pushing method and device and computer readable storage medium | |
| CN111967971B (en) | Bank customer data processing method and device | |
| CN112183818A (en) | Recommendation probability prediction method and device, electronic equipment and storage medium | |
| WO2020258773A1 (en) | Method, apparatus, and device for determining pushing user group, and storage medium | |
| WO2021208535A1 (en) | Recommendation method and device based on automatic feature grouping | |
| CN112819523B (en) | Marketing prediction method combining inner/outer product feature interaction and Bayesian neural network | |
| CN111177579A (en) | Integrated diversity enhanced ultra-deep factorization machine model and construction method and application thereof | |
| CN110796513A (en) | Multitask learning method and device, electronic equipment and storage medium | |
| US20230342797A1 (en) | Object processing method based on time and value factors | |
| CN111523035B (en) | Recommendation method, device, server and medium for APP browsing content | |
| CN108734499A (en) | Promotion message effect analysis method and device, computer-readable medium | |
| CN112883265A (en) | Information recommendation method and device, server and computer readable storage medium | |
| CN115730125A (en) | Object identification method and device, computer equipment and storage medium | |
| CN112258207A (en) | Advertisement flow determination method, device, equipment and storage medium | |
| WO2022247671A1 (en) | User recall method and apparatus, and computer device and storage medium | |
| CN108388652B (en) | Method and device for sending song list identification | |
| CN112581177B (en) | Marketing prediction method combining automatic feature engineering and residual neural network | |
| Mardanlou et al. | Statistical plant modeling and simulation in online advertising |
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 |