CN109214841B

CN109214841B - Method and device for obtaining advertisement predicted value and terminal

Info

Publication number: CN109214841B
Application number: CN201710525325.5A
Authority: CN
Inventors: 刘海旺
Original assignee: Beijing Kingsoft Internet Security Software Co Ltd
Current assignee: Beijing Kingsoft Internet Security Software Co Ltd
Priority date: 2017-06-30
Filing date: 2017-06-30
Publication date: 2021-10-22
Anticipated expiration: 2037-06-30
Also published as: CN109214841A

Abstract

The embodiment of the application discloses a method, a device and a terminal for obtaining an advertisement predicted value, wherein the method for obtaining the advertisement predicted value comprises the following steps: determining the initial selection characteristic combination of the advertisements in the advertisement candidate queue according to the initial selection model; searching a key value corresponding to the initially selected feature combination of the advertisement in a feature cache pool; if the advertisement is found, inquiring the primary selection model to obtain a primary selection predicted value of the advertisement; selecting a preset number of advertisements in the advertisement candidate queue to form a carefully selected candidate queue; obtaining a predictive value of the candidate advertisements in the culling candidate queue through a culling model. According to the method and the device, the key values corresponding to the feature combinations of the advertisements can be cached at the request level, and the method and the device are convenient and quick to use, so that when each prediction model is queried, the key values corresponding to the feature combinations do not need to be acquired independently for each prediction model, the query efficiency of the advertisement prediction values is improved, the system overhead is reduced, and the system performance is improved.

Description

Method and device for obtaining advertisement predicted value and terminal

Technical Field

The application relates to the technical field of internet, in particular to a method, a device and a terminal for obtaining an advertisement predicted value.

Background

For advertisement or personalized recommendation, even general internet products, the most concerned index is click-through rate, and whether manual operation or machine decision, a prejudgment on the possible click-through rate of a certain advertisement or recommended content is needed to judge which items should be placed at more important positions.

In the related art, for each advertisement, the steps of evaluating the predicted value to be passed include: the method comprises the steps of initially selecting truncation, click rate prediction and conversion prediction, wherein each prediction needs to use an independent prediction model, so that for each advertisement, the prediction models need to be inquired one by one to obtain the predicted value of the advertisement, and when each prediction model is inquired, a feature combination needed by the prediction model needs to be independently generated, a weight value corresponding to the feature combination needs to be inquired, and the predicted value of the advertisement is finally obtained.

In the above scheme for obtaining the predicted value of the advertisement provided in the prior art, when each prediction model is queried, a feature combination required by the prediction model needs to be generated separately, so that the query efficiency of the predicted value is reduced, and the process of generating the feature combination inevitably occupies system resources, thereby increasing the system overhead and reducing the system performance.

Disclosure of Invention

Embodiments of the present application provide a method, an apparatus, and a terminal for obtaining an advertisement predicted value, which can implement request-level caching of key values corresponding to feature combinations of advertisements, and are convenient and fast to use, so that when querying each prediction model, it is not necessary to separately obtain key values corresponding to feature combinations for each prediction model, thereby improving query efficiency of advertisement predicted values, reducing system overhead, and improving system performance.

In order to achieve the above purpose, the embodiment of the present application adopts the following technical solutions:

in a first aspect, an embodiment of the present application provides a method for obtaining an advertisement prediction value, including:

determining a primary selection feature combination of the advertisements in the advertisement candidate queue according to a primary selection model, wherein the primary selection feature combination is used for predicting the advertisements in the advertisement candidate queue;

searching a key value corresponding to the initially selected feature combination of the advertisement in a feature cache pool;

if the advertisement is found, inquiring the primary selection model according to a key value corresponding to the primary selection feature combination of the advertisement to obtain a primary selection predicted value of the advertisement;

selecting a preset number of advertisements in the advertisement candidate queue according to the sequence from high to low of the initial selection predicted value to form a fine selection candidate queue;

obtaining a predictive value of the candidate advertisements in the culling candidate queue through a culling model.

In a specific embodiment, before determining the combination of the primary selection features required for predicting the advertisements in the advertisement candidate queue according to the primary selection model, the method further includes:

after receiving an interactive request sent by a terminal in the application using process, acquiring attribute information of the terminal and flow information corresponding to the terminal;

and searching advertisements with advertisement putting conditions matched with the attribute information of the terminal and the flow information corresponding to the terminal in an advertisement library, and putting the searched advertisements into an advertisement candidate queue.

In a specific implementation manner, after searching for a key value corresponding to the initially selected feature combination of the advertisement in the feature cache pool, the method further includes:

if the key value corresponding to the primary selection feature combination of the advertisement is not found in the feature cache pool, obtaining the key value corresponding to the primary selection feature combination of the advertisement from the attribute information of the terminal, the traffic information corresponding to the terminal and/or the advertisement library, and storing the key value corresponding to the primary selection feature combination of the advertisement in the feature cache pool.

In one embodiment, the obtaining the predictive value of the candidate advertisement in the culled candidate queue through the culling model comprises:

determining a cull feature combination of candidate advertisements in the cull candidate queue according to the cull model, the cull feature combination being used to predict candidate advertisements in the cull candidate queue;

searching key values corresponding to the selected feature combinations of the candidate advertisements in the feature cache pool;

and if the candidate advertisements are found, inquiring the selection model according to key values corresponding to the selection feature combinations of the candidate advertisements to obtain the predicted values of the candidate advertisements.

In a specific implementation manner, after the key value corresponding to the selected feature combination of the candidate advertisement is searched in the feature cache pool, the method further includes:

and if the key value corresponding to the selected feature combination of the candidate advertisement is not found in the feature cache pool, obtaining the key value corresponding to the selected feature combination of the candidate advertisement from the attribute information of the terminal, the flow information corresponding to the terminal and/or the advertisement library, and storing the key value corresponding to the selected feature combination of the candidate advertisement in the feature cache pool.

In a specific implementation manner, the querying the primary selection model according to a key value corresponding to the primary selection feature combination of the advertisement to obtain a primary selection predicted value of the advertisement includes:

inquiring the primary selection model according to the key value corresponding to each primary selection feature combination of the advertisement to obtain the weight corresponding to each primary selection feature combination of the advertisement;

and obtaining the initial selection predicted value of the advertisement according to the weight corresponding to each initial selection feature combination of the advertisement.

In a second aspect, an embodiment of the present application provides an apparatus for obtaining an advertisement prediction value, including:

the determining module is used for determining a primary selection feature combination of the advertisements in the advertisement candidate queue according to a primary selection model, wherein the primary selection feature combination is used for predicting the advertisements in the advertisement candidate queue;

the searching module is used for searching the key value corresponding to the initially selected feature combination of the advertisement obtained by the determining module in a feature cache pool;

the query module is used for querying the primary selection model according to the key value corresponding to the primary selection feature combination of the advertisement to obtain a primary selection predicted value of the advertisement when the search module finds the key value corresponding to the primary selection feature combination of the advertisement;

the selection module is used for selecting a preset number of advertisements in the advertisement candidate queue according to the sequence of the initial selection predicted value from high to low to form a fine selection candidate queue;

an obtaining module for obtaining a predictive value of the candidate advertisements in the culling candidate queue through a culling model.

In a specific implementation manner, the obtaining module is further configured to obtain attribute information of the terminal and traffic information corresponding to the terminal after receiving an interaction request sent by the terminal in a process of using an application;

the searching module is further configured to search an advertisement library for an advertisement with an advertisement delivery condition matching the attribute information of the terminal and the traffic information corresponding to the terminal, which are obtained by the obtaining module, and place the searched advertisement in an advertisement candidate queue.

In a specific embodiment, the device for obtaining the advertisement prediction value further comprises: a storage module;

the obtaining module is further configured to obtain, from the attribute information of the terminal, the traffic information corresponding to the terminal, and/or the advertisement library, a key value corresponding to the initially selected feature combination of the advertisement when the key value corresponding to the initially selected feature combination of the advertisement is not found in the feature cache pool by the searching module;

the saving module is configured to save the key values corresponding to the initially selected feature combinations of the advertisements obtained by the obtaining module into the feature cache pool.

In a specific embodiment, the obtaining module includes:

a combination determination sub-module for determining a refined feature combination of the candidate advertisements in the refined candidate queue according to the refined model, the refined feature combination being used for predicting the candidate advertisements in the refined candidate queue;

the key value searching submodule is used for searching the key values corresponding to the carefully selected feature combinations of the candidate advertisements in the feature cache pool;

and the predicted value obtaining sub-module is used for inquiring the selection model according to the key values corresponding to the selection feature combinations of the candidate advertisements to obtain the predicted values of the candidate advertisements when the key value searching sub-module finds the key values corresponding to the selection feature combinations of the candidate advertisements.

the key value searching sub-module is further configured to, when a key value corresponding to a selected feature combination of the candidate advertisement is not found in the feature cache pool, obtain a key value corresponding to the selected feature combination of the candidate advertisement from the attribute information of the terminal, the traffic information corresponding to the terminal, and/or the advertisement library;

the storage module is configured to store the selected feature combination of the candidate advertisement obtained by the obtaining module into the feature cache pool.

In a specific implementation manner, the query module is specifically configured to query the primary selection model according to a key value corresponding to each primary selection feature combination of the advertisement, and obtain a weight corresponding to each primary selection feature combination of the advertisement; and obtaining the initial selection predicted value of the advertisement according to the weight corresponding to each initial selection feature combination of the advertisement.

In a third aspect, an embodiment of the present application further provides a terminal device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the method is implemented.

In a fourth aspect, the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method as described above.

In a fifth aspect, the present application further provides a computer program product, wherein when the instructions in the computer program product are executed by a processor, the method as described above is implemented.

According to the method, the device and the terminal for obtaining the advertisement predicted value, the primary selection feature combination of the advertisements in the advertisement candidate queue is determined according to the primary selection model, then the key value corresponding to the primary selection feature combination of the advertisements is searched in the feature cache pool, if the key value is searched, the primary selection model is inquired according to the key value corresponding to the primary selection feature combination of the advertisements, and the primary selection predicted value of the advertisements is obtained, so that the request-level cache of the key value corresponding to the feature combination of the advertisements can be realized, the convenience and the high speed in use are realized, and when each prediction model is inquired, the key value corresponding to the feature combination does not need to be independently obtained for each prediction model, and the inquiry efficiency of the advertisement predicted value is improved; then, a predetermined number of advertisements in the advertisement candidate queue are selected according to the sequence of the initial selection predicted value from high to low to form a selected candidate queue, so that the advertisement candidate queue can be initially selected and cut off, the advertisement sorting queue is reduced, and the predicted value of the candidate advertisements in the selected candidate queue is obtained through a selected model, so that the system overhead can be greatly reduced, the system performance is improved, and the query efficiency of the advertisement predicted value is further improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.

FIG. 1 is a flowchart of an embodiment of a method for obtaining an advertisement prediction value according to the present application;

FIG. 2 is a flowchart illustrating another embodiment of a method for obtaining an advertisement prediction value according to the present application;

FIG. 3 is a flowchart of a method for obtaining an advertisement prediction value according to yet another embodiment of the present application;

FIG. 4 is a flowchart of a method for obtaining an advertisement prediction value according to yet another embodiment of the present application;

FIG. 5 is a flowchart of a method for obtaining an advertisement prediction value according to yet another embodiment of the present application;

FIG. 6 is a schematic structural diagram illustrating an apparatus for obtaining advertisement prediction values according to an embodiment of the present application;

FIG. 7 is a schematic structural diagram of an apparatus for obtaining advertisement prediction values according to another embodiment of the present application;

FIG. 8 is a schematic structural diagram of an embodiment of a terminal of the present application;

fig. 9 is a schematic structural diagram of an embodiment of an internal portion of the mobile phone 10 according to the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The term "and/or" herein is merely an association describing an associated object, meaning that three relationships may exist, e.g., a and/or B, may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" herein generally indicates that the former and latter related objects are in an "or" relationship.

Fig. 1 is a flowchart of an embodiment of a method for obtaining an advertisement prediction value according to the present application, and as shown in fig. 1, the method for obtaining an advertisement prediction value may include:

step 101, determining a primary selection feature combination of the advertisements in the advertisement candidate queue according to a primary selection model, wherein the primary selection feature combination is used for predicting the advertisements in the advertisement candidate queue.

The primary selection model is a prediction model for predicting behavior probability of advertisement occurrence, such as click probability and/or installation probability, and the primary selection model may include: one or a combination of a Click Through Rate (CTR) prediction model and a CVR prediction model, although the above-mentioned initial selection model may also include other prediction models, which is not limited in this embodiment; the above features are combined into factors that affect the behavior of the advertisement, such as the location, time, and/or gender of the user at which the advertisement is displayed.

In this embodiment, a feature combination required for predicting the advertisement in the advertisement candidate queue may be determined according to the primary selection model, for example, the primary selection model requires two feature combinations, namely, a country _ time _ creative ID and an ID _ scene ID _ network type allocated by APP when the advertisement is predicted, so that two feature combinations, namely, a country _ time _ creative ID and an ID _ scene ID _ network type allocated by APP, may be determined according to the primary selection model.

And step 102, searching a key value corresponding to the initially selected feature combination of the advertisement in a feature cache pool.

For example, for two feature combinations of country _ time _ creative ID and ID _ scene ID _ network type allocated to the APP, the key value corresponding to the feature combination of country _ time _ creative ID of the advertisement to be predicted in the advertisement candidate queue may be US _05_100, and the key value corresponding to the feature combination of ID _ scene ID _ network type allocated to the APP of the advertisement to be predicted may be 104_4201_ 2.

And 103, if the advertisement is found, inquiring the primary selection model according to the key value corresponding to the primary selection feature combination of the advertisement to obtain the primary selection predicted value of the advertisement.

In this embodiment, after the primary selection feature combination required for predicting the advertisement is determined, the key value corresponding to the primary selection feature combination of the advertisement may be searched in the feature cache pool, and then the primary selection model may be queried using the key value corresponding to the searched primary selection feature combination of the advertisement, so that when each prediction model is queried, the key value corresponding to the feature combination does not need to be separately obtained for each prediction model, and the query efficiency of the advertisement prediction value is improved.

And 104, selecting a preset number of advertisements in the advertisement candidate queue according to the sequence of the initial selection predicted value from high to low to form a fine selection candidate queue.

The predetermined number may be set according to system performance and/or implementation requirements during specific implementation, and the size of the predetermined number is not limited in this embodiment, for example, the predetermined number may be 100.

And 105, obtaining the predicted value of the candidate advertisement in the selected candidate queue through the selection model.

In this embodiment, after the advertisement candidate queue is obtained, the advertisement candidate queue is initially selected and truncated, the logic is to reduce the advertisement sorting queue, and then the advertisement in the selected candidate queue is predicted by using a selection model, which is also a prediction model, and the selection model is different from the initial selection model in that the combination of features required by the initial selection model is small, and the combination of features required by the selection model is large, for example: the initial selection model may only require 5 feature combinations, while the refined model requires 10 feature combinations.

Therefore, the initial selection truncation is performed through the initial selection model, the advertisement sorting queue is reduced, the candidate advertisements in the selected candidate queue are predicted through the selected model, the system overhead can be greatly reduced, and the query efficiency of the advertisement predicted value is further improved.

According to the method for obtaining the advertisement predicted value, the primary selection feature combination of the advertisements in the advertisement candidate queue is determined according to the primary selection model, then the key value corresponding to the primary selection feature combination of the advertisements is searched in the feature cache pool, if the key value is searched, the primary selection model is inquired according to the key value corresponding to the primary selection feature combination of the advertisements, and the primary selection predicted value of the advertisements is obtained, so that the request-level cache of the key value corresponding to the feature combination of the advertisements can be realized, the method is convenient and rapid to use, and when each prediction model is inquired, the key value corresponding to the feature combination does not need to be separately obtained for each prediction model, and the inquiry efficiency of the advertisement predicted value is improved; then, a predetermined number of advertisements in the advertisement candidate queue are selected according to the sequence of the initial selection predicted value from high to low to form a selected candidate queue, so that the advertisement candidate queue can be initially selected and cut off, the advertisement sorting queue is reduced, and the predicted value of the candidate advertisements in the selected candidate queue is obtained through a selected model, so that the system overhead can be greatly reduced, the system performance is improved, and the query efficiency of the advertisement predicted value is further improved.

Fig. 2 is a flowchart of another embodiment of a method for obtaining an advertisement prediction value in the present application, as shown in fig. 2, in the embodiment shown in fig. 1 in the present application, before step 101, the method may further include:

step 201, after receiving an interaction request sent by a terminal in the process of using an APP, obtaining attribute information of the terminal and traffic information corresponding to the terminal.

The attribute information of the terminal may be attribute information of a user using the terminal. Specifically, after the terminal sends an interaction request to the server during the APP usage process of the user, the server may obtain attribute information of the terminal and traffic information corresponding to the terminal according to the user ID, where the attribute information of the terminal may include gender, interest, age, and/or the like of the user using the terminal, and the traffic information corresponding to the terminal may include the following information:

1) information of the terminal: the brand of the terminal, the model of the terminal, the internet surfing condition, the version of an operating system and/or the resolution of the terminal and the like;

2) region information: country, city and/or language information used by the user using the terminal;

3) APP information: APP version number and/or scene ID, etc.

Step 202, searching for the advertisement with the advertisement putting condition matched with the attribute information of the terminal and the flow information corresponding to the terminal in an advertisement library, and putting the searched advertisement into an advertisement candidate queue.

In this embodiment, when the advertiser places the advertisement, the advertiser may set advertisement placement conditions, such as placement positions (whether the scanning result page or the screen locking interface is used), placement time, and/or attribute information (gender, age, and/or interest) of people to be placed.

After receiving the interaction request sent by the terminal, the server can search the advertisement with the advertisement delivery condition matched with the attribute information of the terminal and the flow information corresponding to the terminal in an advertisement library according to the obtained attribute information of the terminal and the flow information corresponding to the terminal, and then place the searched advertisement in an advertisement candidate queue.

In addition, as shown in fig. 2, in the embodiment shown in fig. 1 of the present application, after step 102, the method may further include:

step 203, if the key value corresponding to the initially selected feature combination of the advertisement is not found in the feature cache pool, obtaining the key value corresponding to the initially selected feature combination of the advertisement from the attribute information of the terminal, the traffic information corresponding to the terminal, and/or the advertisement library, and storing the key value corresponding to the initially selected feature combination of the advertisement in the feature cache pool.

In this embodiment, if the key value corresponding to the first selected feature combination of the advertisement is not found in the feature cache pool, the key value corresponding to the first selected feature combination of the advertisement may be obtained from the attribute information of the terminal, the traffic information corresponding to the terminal, and/or the advertisement library, and the key value corresponding to the first selected feature combination of the advertisement may be stored in the feature cache pool, so that multiple generation of key values corresponding to feature combinations may be effectively reduced, query time may be saved, and query efficiency of advertisement prediction values may be improved.

Fig. 3 is a flowchart of a further embodiment of the method for obtaining an advertisement prediction value, as shown in fig. 3, in the embodiment shown in fig. 1 of the present application, step 105 may include:

step 301, determining a refined feature combination of the candidate advertisements in the refined candidate queue according to the refined model, wherein the refined feature combination is used for predicting the candidate advertisements in the refined candidate queue.

The choice model is also a prediction model, and the choice model is different from the initial choice model in that the initial choice model requires a small number of feature combinations, and the choice model requires a large number of feature combinations, for example: the initial selection model may only require 5 feature combinations, while the refined model requires 10 feature combinations. The selection model may include: one or a combination of the selected CTR predictive model and the selected CVR predictive model, although the selected CTR predictive model may also include other predictive models, and the embodiment is not limited thereto.

Step 302, searching key values corresponding to the selected feature combinations of the candidate advertisements in the feature cache pool. If so, executing step 303; if the key value corresponding to the selected feature combination of the candidate advertisement is not found in the feature cache pool, step 304 is executed.

Step 303, querying the refined model according to the key values corresponding to the refined feature combinations of the candidate advertisements to obtain the predicted values of the candidate advertisements.

And 304, obtaining key values corresponding to the selected feature combinations of the candidate advertisements from the attribute information of the terminal, the flow information corresponding to the terminal and/or the advertisement library, and storing the key values corresponding to the selected feature combinations of the candidate advertisements in the feature cache pool.

Similarly, if the key value corresponding to the selected feature combination of the candidate advertisement is not found in the feature cache pool, the key value corresponding to the selected feature combination of the candidate advertisement can be obtained from the attribute information of the terminal, the traffic information corresponding to the terminal and/or the advertisement library, and the key value corresponding to the selected feature combination of the candidate advertisement is stored in the feature cache pool, so that multiple generation of the key value corresponding to the feature combination is effectively reduced, the query time is saved, and the query efficiency of the advertisement predicted value is improved.

Fig. 4 is a flowchart of a further embodiment of the method for obtaining an advertisement prediction value, as shown in fig. 4, in the embodiment shown in fig. 1 of the present application, step 103 may be:

step 401, if the advertisement is found, querying the primary selection model according to the key value corresponding to each primary selection feature combination of the advertisement to obtain the weight corresponding to each primary selection feature combination of the advertisement.

And 402, obtaining the initial selection predicted value of the advertisement according to the weight corresponding to each initial selection feature combination of the advertisement.

In this embodiment of the application, taking the initially selected CTR prediction model as an example, assuming that the initially selected CTR prediction model includes N initially selected feature combinations, where N is a positive integer, querying the initially selected CTR prediction model according to key values corresponding to the N initially selected feature combinations of an advertisement, obtaining weights W1, W2, and … WN corresponding to the N initially selected feature combinations, and adding up the N weights, where the obtained sum is a predicted value of the initially selected CTR prediction model for the CTR of the advertisement, and similarly, obtaining a predicted value of the initially selected CVR prediction model for the CVR of the advertisement, where the initially selected predicted value of the advertisement can be calculated according to the following formula:

the initial selection prediction value of the advertisement (bid) × the prediction value of CTR of the advertisement × the prediction value of CVR of the advertisement × constant (1)

In the formula (1), the size of the constant may be set according to system performance and/or implementation requirements during implementation, and the size of the constant is not limited in this embodiment, for example, the constant may be 1000.

Similarly, in step 303 of the embodiment shown in fig. 3 of the present application, the process of querying the refined model according to the key values corresponding to the refined feature combinations of the candidate advertisements to obtain the predicted values of the candidate advertisements is the same as the process described above, and is not repeated here.

With reference to the above description, the present application may further provide a method for obtaining an advertisement predicted value, where fig. 5 is a flowchart of another embodiment of the method for obtaining an advertisement predicted value, as shown in fig. 5, the method specifically includes:

step 501, after receiving an interaction request sent by a terminal in the process of using an APP, obtaining attribute information of the terminal and traffic information corresponding to the terminal.

Step 502, searching for an advertisement in which the advertisement delivery condition is matched with the attribute information of the terminal and the traffic information corresponding to the terminal in an advertisement library, and placing the searched advertisement in an advertisement candidate queue.

Step 503, determining the initial selection feature combination of the advertisements in the advertisement candidate queue according to the initial selection model, wherein the initial selection feature combination is used for predicting the advertisements in the advertisement candidate queue.

Step 504, searching the key value corresponding to the initially selected feature combination of the advertisement in the feature cache pool. If so, executing step 505; if the key value corresponding to the initially selected feature combination of the advertisement is not found in the feature cache pool, step 512 is executed.

And 505, querying the primary selection model according to the key value corresponding to each primary selection feature combination of the advertisement to obtain the weight corresponding to each primary selection feature combination of the advertisement.

Step 506, obtaining the initial selection predicted value of the advertisement according to the weight corresponding to each initial selection feature combination of the advertisement.

And 507, selecting a preset number of advertisements in the advertisement candidate queue according to the sequence of the initial selection predicted value from high to low to form a selected candidate queue.

Step 508, determining a refined feature combination of the candidate advertisements in the refined candidate queue according to the refined model, the refined feature combination being used to predict the candidate advertisements in the refined candidate queue.

Step 509, search the key values corresponding to the selected feature combinations of the candidate advertisements in the feature cache pool. If so, go to step 510; if the key value corresponding to the selected feature combination of the candidate advertisement is not found in the feature cache pool, step 511 is executed.

And 510, inquiring the selection model according to key values corresponding to the selection feature combinations of the candidate advertisements to obtain the predicted values of the candidate advertisements.

And 511, obtaining key values corresponding to the selected feature combinations of the candidate advertisements from the attribute information of the terminal, the flow information corresponding to the terminal and/or the advertisement library, and storing the key values corresponding to the selected feature combinations of the candidate advertisements in the feature cache pool.

And 512, obtaining a key value corresponding to the primary selection feature combination of the advertisement from the attribute information of the terminal, the flow information corresponding to the terminal and/or the advertisement library, and storing the key value corresponding to the primary selection feature combination of the advertisement in the feature cache pool.

The method for obtaining the advertisement predicted value can realize the request level caching of the key values corresponding to the characteristic combinations of the advertisements, is convenient and quick to use, so that when each prediction model is queried, the key values corresponding to the characteristic combinations do not need to be separately obtained for each prediction model, and the query efficiency of the advertisement predicted value is improved; then, a predetermined number of advertisements in the advertisement candidate queue are selected according to the sequence of the initial selection predicted value from high to low to form a selected candidate queue, so that the advertisement candidate queue can be initially selected and cut off, the advertisement sorting queue is reduced, and the predicted value of the candidate advertisements in the selected candidate queue is obtained through a selected model, so that the system overhead can be greatly reduced, the system performance is improved, and the query efficiency of the advertisement predicted value is further improved.

Fig. 6 is a schematic structural diagram of an embodiment of an apparatus for obtaining an advertisement predicted value in the present application, where the apparatus for obtaining an advertisement predicted value in the present application may be used as a terminal or a part of a terminal to implement the method for obtaining an advertisement predicted value provided in the present application.

The terminal may be an intelligent terminal device such as a mobile phone, a tablet Computer, a notebook Computer, or a Personal Computer (PC), and the specific form of the terminal is not limited in the present application.

As shown in fig. 6, the means for obtaining the advertisement prediction value may include: a determining module 61, a searching module 62, a querying module 63, a selecting module 64 and an obtaining module 65;

the determining module 61 is configured to determine a primary selection feature combination of the advertisements in the advertisement candidate queue according to a primary selection model, where the primary selection feature combination is used to predict the advertisements in the advertisement candidate queue; the primary selection model is a prediction model for predicting behavior probability of advertisement occurrence, such as click probability and/or installation probability, and the primary selection model may include: one or a combination of the initially selected CTR prediction model and the initially selected CVR prediction model, although the initially selected CTR prediction model may also include other prediction models, which is not limited in this embodiment; the above features are combined into factors that affect the behavior of the advertisement, such as the location, time, and/or gender of the user at which the advertisement is displayed.

In this embodiment, the determining module 61 may determine, according to the initial selection model, a feature combination required for predicting the advertisement in the advertisement candidate queue, for example, the initial selection model requires two feature combinations, namely, a country _ time _ creative ID and an ID _ scene ID _ network type allocated by APP when performing advertisement prediction, so the determining module 61 may determine, according to the initial selection model, two feature combinations, namely, a country _ time _ creative ID and an ID _ scene ID _ network type allocated by APP, required for predicting the advertisement in the advertisement queue.

And the searching module 62 is configured to search the key value corresponding to the initially selected feature combination of the advertisement obtained by the determining module 61 in the feature cache pool.

And the query module 63 is configured to query the primary selection model according to the key value corresponding to the primary selection feature combination of the advertisement when the search module 62 finds the key value corresponding to the primary selection feature combination of the advertisement, so as to obtain a primary selection predicted value of the advertisement.

In this embodiment, after the determining module 61 determines the primary selection feature combination required for predicting the advertisement, the searching module 62 may search the key value corresponding to the primary selection feature combination of the advertisement in the feature cache pool, and then the querying module 63 queries the primary selection model using the key value corresponding to the primary selection feature combination of the advertisement searched by the searching module 62, so that when querying each prediction model, it is not necessary to separately obtain the key value corresponding to the feature combination for each prediction model, and the query efficiency of the advertisement prediction value is improved.

A selecting module 64, configured to select a predetermined number of advertisements in the candidate advertisement queue according to the sequence from high to low of the initial selection predicted value, so as to form a selected candidate queue; the predetermined number may be set according to system performance and/or implementation requirements during specific implementation, and the size of the predetermined number is not limited in this embodiment, for example, the predetermined number may be 100.

An obtaining module 65 is configured to obtain the predicted value of the candidate advertisement in the refined candidate queue through a refining model.

In this embodiment, after obtaining the candidate queue of advertisements, the extracting module 64 performs a preliminary selection truncation on the candidate queue of advertisements, which is a logic for reducing the queue of ordered advertisements, and then the obtaining module 65 predicts the advertisements in the candidate queue of advertisements by using a selection model, which is also a prediction model, and the selection model is different from the preliminary selection model in that the combination of features required by the preliminary selection model is small, and the combination of features required by the selection model is large, for example: the initial selection model may only require 5 feature combinations, while the refined model requires 10 feature combinations.

In the device for obtaining the advertisement predicted value, the determining module 61 determines the primary selection feature combination of the advertisements in the advertisement candidate queue according to the primary selection model, then the searching module 62 searches the key value corresponding to the primary selection feature combination of the advertisements in the feature cache pool, if the key value is found, the searching module 63 queries the primary selection model according to the key value corresponding to the primary selection feature combination of the advertisements to obtain the primary selection predicted value of the advertisements, so that the request-level cache of the key value corresponding to the feature combination of the advertisements can be realized, the device is convenient and quick to use, when each prediction model is queried, the key value corresponding to the feature combination does not need to be separately obtained for each prediction model, and the query efficiency of the advertisement predicted value is improved; then, the selecting module 64 selects a predetermined number of advertisements in the advertisement candidate queue according to the order of the initial selection predicted value from high to low to form a refined candidate queue, so as to perform initial selection truncation on the advertisement candidate queue, thereby reducing the advertisement sorting queue, and the obtaining module 65 obtains the predicted value of the candidate advertisement in the refined candidate queue through a refined model, thereby greatly reducing the system overhead, improving the system performance, and further improving the query efficiency of the advertisement predicted value.

Fig. 7 is a schematic structural diagram of another embodiment of an apparatus for obtaining an advertisement prediction value according to the present application.

In this embodiment, the obtaining module 65 is further configured to obtain the attribute information of the terminal and the traffic information corresponding to the terminal after receiving an interaction request sent by the terminal in the process of using the application.

3) APP information: APP version number and/or scene ID, etc.

The searching module 62 is further configured to search the advertisement library for the advertisement whose advertisement delivery condition matches the attribute information of the terminal and the traffic information corresponding to the terminal, which are obtained by the obtaining module 65, and place the searched advertisement in the advertisement candidate queue.

After receiving the interactive request sent by the terminal, the searching module 62 may search for an advertisement whose advertisement delivery condition matches the attribute information of the terminal and the traffic information corresponding to the terminal in an advertisement library according to the attribute information of the terminal and the traffic information corresponding to the terminal, which are obtained by the obtaining module 65, and then place the searched advertisement in an advertisement candidate queue.

Further, compared with the apparatus for obtaining an advertisement prediction value shown in fig. 6, the apparatus for obtaining an advertisement prediction value shown in fig. 7 may further include: a save module 66;

the obtaining module 65 is further configured to, when the key value corresponding to the initially selected feature combination of the advertisement is not found in the feature cache pool by the searching module 62, obtain the key value corresponding to the initially selected feature combination of the advertisement from the attribute information of the terminal, the traffic information corresponding to the terminal, and/or the advertisement library;

a saving module 66, configured to save the key value corresponding to the initially selected feature combination of the advertisement obtained by the obtaining module 65 into the feature cache pool.

In this embodiment, if the search module 62 does not search for the key value corresponding to the first selected feature combination of the advertisement in the feature cache pool, the obtaining module 65 may obtain the key value corresponding to the first selected feature combination of the advertisement from the attribute information of the terminal, the traffic information corresponding to the terminal, and/or the advertisement library, and the storage module 66 stores the key value corresponding to the first selected feature combination of the advertisement in the feature cache pool, so that multiple times of generation of the key value corresponding to the feature combination are effectively reduced, the query time is saved, and the query efficiency of the advertisement prediction value is improved.

In this embodiment, the obtaining module 65 may include: a combination determination sub-module 651, a key value searching sub-module 652 and a predicted value obtaining sub-module 653;

wherein, the combination determining submodule 651 is configured to determine, according to the culling model, a culling feature combination of the candidate advertisements in the culling candidate queue, where the culling feature combination is used to predict the candidate advertisements in the culling candidate queue; the choice model is also a prediction model, and the choice model is different from the initial choice model in that the initial choice model requires a small number of feature combinations, and the choice model requires a large number of feature combinations, for example: the initial selection model may only require 5 feature combinations, while the refined model requires 10 feature combinations. The selection model may include: one or a combination of the selected CTR predictive model and the selected CVR predictive model, although the selected CTR predictive model may also include other predictive models, and the embodiment is not limited thereto.

The key value searching submodule 652 is configured to search, in the feature cache pool, a key value corresponding to a selected feature combination of the candidate advertisement;

the predicted value obtaining sub-module 653 is configured to, when the key value searching sub-module 652 finds the key value corresponding to the selected feature combination of the candidate advertisement, query the selected model according to the key value corresponding to the selected feature combination of the candidate advertisement, and obtain the predicted value of the candidate advertisement.

Further, the apparatus for obtaining the advertisement prediction value may further include: a save module 66;

the key value searching sub-module 652 is further configured to, when a key value corresponding to the selected feature combination of the candidate advertisement is not found in the feature cache pool, obtain a key value corresponding to the selected feature combination of the candidate advertisement from the attribute information of the terminal, the traffic information corresponding to the terminal, and/or the advertisement library;

a saving module 66, configured to save the selected feature combinations of the candidate advertisements obtained by the key value search sub-module 652 into the feature cache pool.

Similarly, if the key value corresponding to the selected feature combination of the candidate advertisement is not found in the feature cache pool by the key value search sub-module 652, the key value corresponding to the selected feature combination of the candidate advertisement may be obtained from the attribute information of the terminal, the traffic information corresponding to the terminal, and/or the advertisement library, and the key value corresponding to the selected feature combination of the candidate advertisement may be stored in the feature cache pool by the storage module 66, so that multiple generation of key values corresponding to feature combinations may be effectively reduced, the query time may be saved, and the query efficiency of the advertisement prediction value may be improved.

In this embodiment, the query module 63 is specifically configured to query the primary selection model according to a key value corresponding to each primary selection feature combination of the advertisement, and obtain a weight corresponding to each primary selection feature combination of the advertisement; and obtaining the initial selection predicted value of the advertisement according to the weight corresponding to each initial selection characteristic combination of the advertisement.

Specifically, taking the initially selected CTR prediction model as an example, assuming that the initially selected CTR prediction model includes N initially selected feature combinations, where N is a positive integer, the query module 63 queries the initially selected CTR prediction model according to key values corresponding to the N initially selected feature combinations of the advertisement, so as to obtain weights W1, W2, and … WN corresponding to the N initially selected feature combinations, and add the N weights, where the obtained sum is a predicted value of the initially selected CTR prediction model for the CTR of the advertisement, and similarly, a predicted value of the initially selected CVR prediction model for the CVR of the advertisement may be obtained, and thus, the initially selected predicted value of the advertisement may be calculated according to the following formula:

Similarly, the process of obtaining the predicted value of the candidate advertisement by the predicted value obtaining sub-module 653 according to the key value corresponding to the selected feature combination of the candidate advertisement is the same as the process described above, and is not repeated here.

The device for obtaining the advertisement predicted value can realize the request level caching of the key values corresponding to the characteristic combinations of the advertisements, is convenient and quick to use, so that when each prediction model is queried, the key values corresponding to the characteristic combinations do not need to be separately obtained for each prediction model, and the query efficiency of the advertisement predicted value is improved; then, the selecting module 64 selects a predetermined number of advertisements in the advertisement candidate queue according to the order of the initial selection predicted value from high to low to form a refined candidate queue, so as to perform initial selection truncation on the advertisement candidate queue, thereby reducing the advertisement sorting queue, and the obtaining module 65 obtains the predicted value of the candidate advertisement in the refined candidate queue through a refined model, thereby greatly reducing the system overhead, improving the system performance, and further improving the query efficiency of the advertisement predicted value.

Fig. 8 is a schematic structural diagram of an embodiment of a terminal device according to the present application, where the terminal device may include: the processor executes the computer program, and the method for obtaining the advertisement prediction value provided by the embodiment of the application can be realized.

The terminal device may be an intelligent terminal device such as a mobile phone, a tablet computer, a notebook computer, or a PC, and the specific form of the terminal device is not limited in the present application, and the terminal device is taken as an example to be explained in the present embodiment.

It should be understood that the handset 10 shown in fig. 8 is merely one example of the terminal device described above, and that the handset 10 may have more or fewer components than shown in fig. 8, may combine two or more components, or may have a different configuration of components. The various components shown in fig. 8 may be implemented in hardware, software, or a combination of hardware and software, including one or more signal processing and/or application specific integrated circuits.

The mobile phone 10 will be specifically described as an example. As shown in fig. 8, the mobile phone 10 may include a memory 11, a Central Processing Unit (CPU) 12, a peripheral interface 13, a Radio Frequency (RF) circuit 14, an audio circuit 15, a speaker 16, a power supply system 17, an Input Output (I/O) subsystem 18, other Input/control devices 19, and an external port 20, which communicate via one or more communication buses or signal lines 21.

It should be noted that the mobile phone provided in this embodiment is only one example of the terminal device, and the terminal device related to this embodiment may have more or fewer components than those shown in fig. 8, may combine two or more components, or may have different configurations or arrangements of components, and each component may be implemented in hardware, software, or a combination of hardware and software including one or more signal processing and/or application specific integrated circuits.

The following describes the mobile phone provided in this embodiment in detail.

The memory 11: the memory 11 may be accessed by the CPU12, the peripheral interface 13, and the like, and the memory 11 may include high speed random access memory and may also include non-volatile memory, such as one or more magnetic disk storage devices, flash memory devices, or other volatile solid state storage devices.

A peripheral interface 13 which may connect input and output peripherals of the handset 10 to the CPU12 and the memory 11.

I/O subsystem 18: the I/O subsystem 18 may connect input and output peripherals on the handset 10, such as a touch screen 22 and other input/control devices 19, to the peripheral interface 13. The I/O subsystem 18 may include a display controller 181 and one or more input controllers 182 for controlling other input/control devices 19. Where one or more input controllers 182 receive electrical signals from or transmit electrical signals to other input/control devices 19, the other input/control devices 19 may include physical buttons (e.g., push buttons or rocker buttons, etc.), dials, slide switches, joysticks, or click wheels. It is noted that the input controller 182 may be connected to any of: a keyboard, an infrared port, a USB interface, and a pointing device such as a mouse.

The touch screen 22: the touch screen 22 is an input and output interface between the handset 10 and the user that displays visual output to the user, which may include graphics, text, icons, video, and the like.

The display controller 181 in the I/O subsystem 18 receives electrical signals from the touch screen 22 or transmits electrical signals to the touch screen 22. The touch screen 22 detects a contact on the touch screen, and the display controller 181 converts the detected contact into an interaction with a user interface object displayed on the touch screen 22, i.e., implements a human-computer interaction, which may be an icon for running a game, an icon networked to a corresponding network, etc., displayed on the touch screen 22. It is noted that the handset 10 may also include a light mouse, which is a touch sensitive surface that does not display visual output, or an extension of the touch sensitive surface formed by a touch screen.

The RF circuit 14 is mainly used to establish communication between the mobile phone 10 and a wireless network (i.e., a network side), so as to implement data transmission and reception between the mobile phone 10 and the wireless network. Such as sending and receiving short messages, e-mails, etc. Specifically, the RF circuitry 14 receives and transmits RF signals, also referred to as electromagnetic signals, by which the RF circuitry 14 converts electrical signals to or from electromagnetic signals and communicates with communication networks and other devices. The RF circuitry 14 may include known circuitry for performing these functions including, but not limited to, an antenna system, an RF transceiver, one or more amplifiers, a tuner, one or more oscillators, a digital signal processor, a CODEC (CODEC) chipset, a Subscriber Identity Module (SIM), and so forth.

The audio circuit 15 is mainly used to receive audio data from the peripheral interface 13, convert the audio data into an electric signal, and transmit the electric signal to the speaker 16.

A speaker 16 for reproducing the voice signal received by the handset 10 from the wireless network through the RF circuit 14 into sound and playing the sound to the user.

And the power supply system 17 is used for supplying power and managing the power for the hardware connected with the CPU12, the I/O subsystem 18 and the peripheral interface 13. The power system 17 may include a power management system, one or more power sources (e.g., batteries or ac), a recharging system, power failure detection circuitry, a power converter or inverter, a power status indicator (e.g., light emitting diodes), and any other components associated with power generation, management, and distribution in the portable device.

Fig. 9 is a schematic structural diagram of an embodiment of an internal portion of the mobile phone 10 according to the present application. In the present embodiment, the software components stored in the memory 11 may include an operating system 1001, a communication module 1002, a contact/movement module 1003, a graphic module 1004, and a function module 1005.

The operating system 1001 (e.g., Darwin, RTXC, LINUX, UNIX, OS X, WINDOWS, or an embedded operating system such as VxWorks) includes various software components and/or drivers for controlling and managing general system tasks (e.g., memory management, storage device control, power management, etc.) and facilitates communication between various hardware and software components.

The communications module 1002 facilitates communications with other devices through one or more external ports 20, and also includes various software components for processing data received by the RF circuitry 14 and/or the external ports 20.

The contact/movement module 1003 may detect contact with the touch screen 22 (in conjunction with the display controller 181) and other touch sensitive devices (e.g., a touchpad or a physical click wheel). The contact/movement module 1003 includes various software components for performing various operations related to detecting contact, such as determining whether contact has occurred, determining whether the contact has moved and tracked the movement on the touch screen 22, and determining whether the contact has been broken (i.e., whether contact has ceased). Determining movement of the point of contact may include determining velocity (magnitude), velocity (magnitude and direction), and/or acceleration (change in magnitude and/or direction) of the point of contact. These operations may be applied to a single contact (e.g., one finger contact) or to multiple simultaneous contacts (e.g., "multi-touch"/multi-finger contacts). In some embodiments, the contact/movement module 1003 and the display controller 181 also detect contact on the touch panel.

The graphics module 1004 includes various known software components for displaying graphics on the touch screen 22, including components for changing the darkness of the displayed graphics. For example, a graphic user interface of various software is displayed on the touch panel 22 in response to an instruction from the CPU 12.

The function module 1005 executes various functional applications and data processing by executing the program stored in the memory 11, for example, to implement the method for obtaining an advertisement prediction value provided by the embodiment of the present application.

The RF circuit 14 receives a message sent by a network side or other devices, where the message includes an email or a short message or an instant message, and the message may specifically be the message in the embodiments shown in fig. 1 to fig. 5. It is understood that the message received by the RF circuit 14 may be other types of messages, and is not limited in the embodiments of the present application. Those skilled in the art will appreciate that the received message may carry data of a variety of data types. There may be only one data type of data, or there may be two or more data types of data.

The CPU12 realizes the method of obtaining an advertisement prediction value provided by the embodiment of the present application when executing the program stored in the memory 11. In the above embodiment, the CPU12 may be embodied as a pentium series processor or an itanium processor manufactured by intel corporation.

The embodiment of the present application further provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method for obtaining an advertisement prediction value provided in the embodiment of the present application may be implemented.

The non-transitory computer readable storage medium described above may take any combination of one or more computer readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a Read Only Memory (ROM), an Erasable Programmable Read Only Memory (EPROM), a flash Memory, an optical fiber, a portable compact disc Read Only Memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for carrying out operations for aspects of the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of Network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).

The embodiment of the present application further provides a computer program product, and when instructions in the computer program product are executed by a processor, the method for obtaining the advertisement prediction value provided by the embodiment of the present application can be implemented.

The terminology used in the embodiments of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in the examples of this application and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

It should be understood that although the terms "first", "second", "third", etc. may be used to describe various connection ports and identification information, etc. in the embodiments of the present application, these connection ports and identification information, etc. should not be limited to these terms. These terms are only used to distinguish the connection port and the identification information and the like from each other. For example, the first connection port may also be referred to as a second connection port, and similarly, the second connection port may also be referred to as a first connection port, without departing from the scope of embodiments of the present application.

The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination" or "in response to a detection", depending on the context. Similarly, the phrases "if determined" or "if detected (a stated condition or event)" may be interpreted as "when determined" or "in response to a determination" or "when detected (a stated condition or event)" or "in response to a detection (a stated condition or event)", depending on the context.

Through the above description of the embodiments, it is clear to those skilled in the art that, for convenience and simplicity of description, the foregoing division of the functional modules is merely used as an example, and in practical applications, the above function distribution may be completed by different functional modules according to needs, that is, the internal structure of the device may be divided into different functional modules to complete all or part of the above described functions. For the specific working processes of the system, the apparatus and the unit described above, reference may be made to the corresponding processes in the foregoing method embodiments, and details are not described here again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules or units is only one logical division, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, a network device, or the like) or a processor (processor) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for obtaining advertisement predicted values is characterized by comprising the following steps:

obtaining a predicted value of the candidate advertisement in the refined candidate queue through a refined model; the querying the primary selection model according to the key values corresponding to the primary selection feature combinations of the advertisements to obtain the primary selection predicted values of the advertisements comprises the following steps:

2. The method of claim 1, wherein prior to determining a combination of initially selected features required to predict the advertisements in the candidate queue of advertisements based on the initial selection model, further comprising:

3. The method of claim 2, wherein after searching for a key value corresponding to the initially selected feature combination of the advertisement in the feature cache pool, the method further comprises:

4. The method of claim 2 or 3, wherein obtaining a predictive value of candidate advertisements in the culling candidate queue through a culling model comprises:

5. The method of claim 4, wherein after searching the feature cache pool for key values corresponding to the selected feature combinations of the candidate advertisements, further comprising:

6. An apparatus for obtaining advertisement prediction values, comprising:

an obtaining module, configured to obtain, through a culling model, a predicted value of a candidate advertisement in the culling candidate queue;

the query module is specifically configured to query the primary selection model according to a key value corresponding to each primary selection feature combination of the advertisement, and obtain a weight corresponding to each primary selection feature combination of the advertisement; and obtaining the initial selection predicted value of the advertisement according to the weight corresponding to each initial selection feature combination of the advertisement.

7. The apparatus of claim 6,

the obtaining module is further configured to obtain attribute information of the terminal and traffic information corresponding to the terminal after receiving an interaction request sent by the terminal in an application using process;

8. The apparatus of claim 7, further comprising: a storage module;

9. The apparatus of any of claims 7-8, wherein the means for obtaining comprises:

10. The apparatus of claim 9, further comprising: a storage module;

11. A terminal device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method according to any one of claims 1-5 when executing the computer program.

12. A non-transitory computer-readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the method of any one of claims 1-5.