CN110189173B - Advertisement generation method and device - Google Patents
Advertisement generation method and device Download PDFInfo
- Publication number
- CN110189173B CN110189173B CN201910451519.4A CN201910451519A CN110189173B CN 110189173 B CN110189173 B CN 110189173B CN 201910451519 A CN201910451519 A CN 201910451519A CN 110189173 B CN110189173 B CN 110189173B
- Authority
- CN
- China
- Prior art keywords
- advertisement
- material segment
- primary
- pointer network
- native
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0276—Advertisement creation
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides an advertisement generation method and device, wherein the method comprises the following steps: acquiring a plurality of advertisement material segments; generating a plurality of candidate advertisements according to the plurality of advertisement material segments and a preset pointer network; and determining the target advertisement from the candidate advertisements according to a preset rule. Therefore, a plurality of advertisement material segments are arranged and combined based on a preset pointer network to automatically generate a plurality of candidate advertisements with good smoothness, a risk control mechanism is introduced to filter the plurality of candidate advertisements to generate target advertisements, the advertisement pool of an advertiser is expanded, and the advertisements delivered by the advertiser can have a larger chance to be exposed or clicked, and the like.
Description
Technical Field
The present invention relates to the field of advertisement technologies, and in particular, to an advertisement generation method and apparatus.
Background
At present, information flow advertisements are mainly composed of advertisement creatives by advertisers, and then advertisement accurate pushing is carried out through an algorithm. However, the advertiser is not able to write the document or understand the user's preference for the advertisement, so that the written advertisement is not attractive and the written advertisement is not abundant. As an advertisement recommender, it can analyze and summarize mass data, common simple methods such as replacement of synonyms including setting of dynamic creative word packets, and can expand a single advertisement to a certain extent to obtain a relatively high advertisement recommendation ranking to help an advertiser optimize and enrich its advertisement creative idea, such as writing a primary advertisement "buy car 0 first payment, buy by stages, buy car easily", and expand the advertisement "buy car 0 first payment, buy by stages, buy car easily, you can't see soon" by algorithm. However, such methods have limited scalability and have a small margin of rewriting native advertisements written by advertisers.
Disclosure of Invention
The present invention is directed to solving, at least to some extent, one of the technical problems in the related art.
To this end, a first object of the present invention is to propose an advertisement generation method.
A second object of the present invention is to provide an advertisement generating apparatus.
A third object of the invention is to propose a computer device.
A fourth object of the invention is to propose a computer-readable storage medium.
In order to achieve the above object, an embodiment of a first aspect of the present invention provides an advertisement generating method, including:
acquiring a plurality of advertisement material segments;
generating a plurality of candidate advertisements according to the plurality of advertisement material segments and a preset pointer network;
and determining the target advertisement from the candidate advertisements according to a preset rule.
In one possible implementation, the method further includes:
obtaining a sample set, wherein the sample set comprises material segment sets corresponding to a plurality of primary advertisement samples, and the material segment set corresponding to each primary advertisement sample comprises at least one material segment of each primary advertisement sample and at least one corresponding expansion material segment;
and performing multiple rounds of training on the initial pointer network according to the material segment set corresponding to each primary advertisement sample to obtain the preset pointer network.
In a possible implementation manner, the performing multiple rounds of training on an initial pointer network according to a material segment set corresponding to each native advertisement sample to obtain the preset pointer network includes:
after the current round of training is finished, calculating a loss function corresponding to the current round of native advertisement samples according to the current round of native advertisement samples and a current round of actual output results, wherein the current round of actual output results are obtained by inputting material fragment sets corresponding to the current round of native advertisement samples into the initial pointer network;
updating the initial pointer network according to the loss function and a random gradient descent method to obtain an updated pointer network, performing next round of training by using the updated pointer network, calculating a loss function corresponding to a next round of native advertisement samples according to a next round of native advertisement samples and a next round of actual output results, and repeating the step until all the native advertisement samples in the sample set are trained, wherein the next round of actual output results are obtained by inputting a material fragment set corresponding to the next round of native advertisement samples into the updated pointer network;
and taking the updated pointer network as the preset pointer network.
In one possible implementation, the obtaining a plurality of advertisement material segments includes:
obtaining at least one material segment of the primary advertisement and at least one expanded material segment obtained by expanding the primary advertisement;
and scattering at least one material segment of the primary advertisement and at least one expansion material segment to obtain a plurality of advertisement material segments.
In one possible implementation, the means for obtaining at least one augmented material segment by augmenting the native advertisement includes one or more of the following:
acquiring at least one material segment matched with each material segment of the primary advertisement from an advertisement material library, and determining the at least one material segment matched with each material segment of the primary advertisement as at least one expanded material segment obtained by expanding the primary advertisement;
or the like, or, alternatively,
acquiring historical material segments corresponding to an advertiser who writes the primary advertisement, acquiring at least one historical material segment matched with each material segment of the primary advertisement from the historical material segments, and determining the at least one historical material segment matched with each material segment of the primary advertisement as at least one expanded material segment obtained by expanding the primary advertisement;
or the like, or, alternatively,
and acquiring a general segment corresponding to the category to which the primary advertisement belongs, and determining the general segment as at least one expansion material segment obtained by expanding the primary advertisement.
In one possible implementation manner, the determining a target advertisement from the plurality of candidate advertisements according to a preset rule includes:
displaying the plurality of candidate advertisements;
determining the target advertisement from the plurality of candidate advertisements according to the selection information of the plurality of candidate advertisements;
or the like, or, alternatively,
and screening the plurality of candidate advertisements according to a preset screening model, and determining the screened candidate advertisements as the target advertisements.
According to the advertisement generation method provided by the embodiment of the invention, a plurality of advertisement material segments are obtained; generating a plurality of candidate advertisements according to the plurality of advertisement material segments and a preset pointer network; and determining the target advertisement from the candidate advertisements according to a preset rule. Therefore, a plurality of advertisement material segments are arranged and combined based on a preset pointer network to automatically generate a plurality of candidate advertisements with good smoothness, a risk control mechanism is introduced to filter the plurality of candidate advertisements to generate target advertisements, the advertisement pool of an advertiser is expanded, and the advertisements delivered by the advertiser can have a larger chance to be exposed or clicked, and the like.
In order to achieve the above object, a second embodiment of the present invention provides an advertisement generating apparatus, including:
the acquisition module is used for acquiring a plurality of advertisement material segments;
the generating module is used for generating a plurality of candidate advertisements according to the plurality of advertisement material segments and a preset pointer network;
and the determining module is used for determining the target advertisement from the candidate advertisements according to a preset rule.
In one possible implementation, the apparatus further includes: a training module;
the obtaining module is further configured to obtain a sample set, where the sample set includes a material segment set corresponding to a plurality of primary advertisement samples, and the material segment set corresponding to each primary advertisement sample includes at least one material segment of each primary advertisement sample and at least one corresponding augmented material segment;
and the training module is used for performing multi-round training on the initial pointer network according to the material segment set corresponding to each primary advertisement sample to obtain the preset pointer network.
In a possible implementation manner, the training module is specifically configured to:
after the current round of training is finished, calculating a loss function corresponding to the current round of native advertisement samples according to the current round of native advertisement samples and a current round of actual output results, wherein the current round of actual output results are obtained by inputting material fragment sets corresponding to the current round of native advertisement samples into the initial pointer network;
updating the initial pointer network according to the loss function and a random gradient descent method to obtain an updated pointer network, performing next round of training by using the updated pointer network, calculating a loss function corresponding to a next round of native advertisement samples according to a next round of native advertisement samples and a next round of actual output results, and repeating the step until all the native advertisement samples in the sample set are trained, wherein the next round of actual output results are obtained by inputting a material fragment set corresponding to the next round of native advertisement samples into the updated pointer network;
and taking the updated pointer network as the preset pointer network.
In one possible implementation manner, the obtaining module includes a obtaining unit and a breaking unit:
the acquisition unit is used for acquiring at least one material segment of the primary advertisement and at least one expanded material segment obtained by expanding the primary advertisement;
the scattering unit is used for scattering at least one material segment of the primary advertisement and at least one expansion material segment to obtain a plurality of advertisement material segments.
In a possible implementation manner, the obtaining unit is specifically configured to:
acquiring at least one material segment matched with each material segment of the primary advertisement from an advertisement material library, and determining the at least one material segment matched with each material segment of the primary advertisement as at least one expanded material segment obtained by expanding the primary advertisement;
or the like, or, alternatively,
acquiring historical material segments corresponding to an advertiser who writes the primary advertisement, acquiring at least one historical material segment matched with each material segment of the primary advertisement from the historical material segments, and determining the at least one historical material segment matched with each material segment of the primary advertisement as at least one expanded material segment obtained by expanding the primary advertisement;
or the like, or, alternatively,
and acquiring a general segment corresponding to the category to which the primary advertisement belongs, and determining the general segment as at least one expansion material segment obtained by expanding the primary advertisement.
In a possible implementation manner, the determining module is specifically configured to:
displaying the plurality of candidate advertisements;
determining the target advertisement from the plurality of candidate advertisements according to the selection information of the plurality of candidate advertisements;
or the like, or, alternatively,
and screening the plurality of candidate advertisements according to a preset screening model, and determining the screened candidate advertisements as the target advertisements.
The advertisement generating device provided by the embodiment of the invention obtains a plurality of advertisement material segments; generating a plurality of candidate advertisements according to the plurality of advertisement material segments and a preset pointer network; and determining the target advertisement from the candidate advertisements according to a preset rule. Therefore, a plurality of advertisement material segments are arranged and combined based on a preset pointer network to automatically generate a plurality of candidate advertisements with good smoothness, a risk control mechanism is introduced to filter the plurality of candidate advertisements to generate target advertisements, the advertisement pool of an advertiser is expanded, and the advertisements delivered by the advertiser can have a larger chance to be exposed or clicked, and the like.
To achieve the above object, a third embodiment of the present invention provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor executes the computer program to implement the advertisement generation method as described above.
In order to achieve the above object, a fourth aspect of the present invention provides a computer-readable storage medium, wherein instructions in the storage medium, when executed by a processor, implement the advertisement generation method as described above.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The foregoing and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
fig. 1 is a schematic flowchart of an advertisement generation method according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of another advertisement generation method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an advertisement generating apparatus according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of another advertisement generating apparatus according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a computer device according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
An advertisement generation method and apparatus according to an embodiment of the present invention will be described below with reference to the drawings.
Fig. 1 is a schematic flowchart of an advertisement generation method according to an embodiment of the present invention. The embodiment provides an advertisement generating method, the execution subject of which is an advertisement generating device, and the execution subject is composed of hardware and/or software. The advertisement generating device may specifically be a hardware device, such as a terminal device, a backend server, or the like, or software or an application installed on the hardware device, or the like.
As shown in fig. 1, the advertisement generating method includes the following steps:
s101, acquiring a plurality of advertisement material segments.
In this embodiment, in order to greatly expand the richness of the target advertisement, the advertisement material segment is derived from at least one expanded material segment obtained by expanding the primary advertisement, in addition to the primary advertisement written by the advertiser.
Specifically, after the primary advertisement written by the advertiser is obtained, the primary advertisement is cut by taking the segments as granularity, and at least one material segment of the primary advertisement is obtained.
For example, the primary advertisement is "world of wind" Mmorpg, AR pet catch, hand catch pet adventure ", and the 3 material segments obtained by cutting using the punctuation mark in the primary advertisement as the separator between the segments are" here, "world of wind" Mmorpg "," AR pet catch ", and hand catch pet adventure", respectively.
In particular, the at least one augmented material segment from the native advertisement may be augmented in one or more of the following ways, but is not limited thereto.
The first example:
the "manner of obtaining at least one augmented material segment by augmenting a native advertisement" is: and acquiring at least one material segment matched with each material segment of the primary advertisement from an advertisement material library, and determining at least one material segment matched with each material segment of the primary advertisement as at least one expanded material segment obtained by expanding the primary advertisement.
Taking the above-mentioned primary advertisement, "mainland of wind," Mmorpg, AR catching pets, hand-in-hand catching pet adventure ", as an example, at least one augmentation material segment extracted from the advertisement material library is a japanese magic hand tour, butcher, boy knife click-and-go, and the like.
In this embodiment, a large amount of advertisement materials are stored in the advertisement material library. Abundant expansion material segments can be obtained through the advertisement material library, synonymous expansion of the primary advertisement is realized, the richness of the target advertisement is greatly expanded, and the problems of service tampering, improper collocation and the like are avoided as much as possible.
Specifically, according to different application scenes, at least one material segment matched with each material segment of the primary advertisement is obtained from an advertisement material library in different modes.
For example, after the raw advertisement is cut into material segments, semantic analysis may be performed on each material segment of the raw advertisement, and one or more material segments having the same semantics may be extracted from the advertisement material library according to the semantic analysis result of each material segment.
For another example, after the material segments are obtained by cutting the primary advertisement, semantic analysis may be performed on each word of each material segment of the primary advertisement, and one or more material segments having synonyms may be extracted from the advertisement material library according to the semantic analysis result of each word of each material segment.
The second example is:
the "manner of obtaining at least one augmented material segment by augmenting a native advertisement" is: the method comprises the steps of obtaining historical material segments corresponding to an advertiser writing the primary advertisement, obtaining at least one historical material segment matched with each material segment of the primary advertisement from the historical material segments, and determining the at least one historical material segment matched with each material segment of the primary advertisement as at least one expanded material segment obtained by expanding the primary advertisement.
In the embodiment, the expansion material segments are obtained from the historical material segments compiled by the same advertiser, so that the target advertisement has rich material segments capable of describing the business scene under the same business scene as much as possible, the practicability of the target advertisement is improved, and the risk of the target advertisement is reduced. Note that the advertisement generation means stores history material segments of the primary advertisement written by the history of each advertiser in advance.
For example, the advertiser 1 manufactures the product 1, the advertiser 2 manufactures the product 2, the expansion material segments extracted from the history material segments of the advertiser 1 are all related to the product 1, and the expansion material segments extracted from the history material segments of the advertiser 2 are all related to the product 2.
Specifically, according to different application scenes, at least one material segment matched with each material segment of the primary advertisement is obtained from historical material segments in different modes.
For example, after the raw advertisement is cut into material segments, semantic analysis may be performed on each material segment of the raw advertisement, and one or more material segments having the same semantic meaning may be extracted from the historical material segments according to the semantic analysis result of each material segment.
For another example, after the material segments are obtained by cutting the primary advertisement, semantic analysis may be performed on each word of each material segment of the primary advertisement, and one or more material segments having synonyms may be extracted from the historical material segments according to the semantic analysis result of each word of each material segment.
The third example:
the "manner of obtaining at least one augmented material segment by augmenting a native advertisement" is: and acquiring a general segment corresponding to the category to which the primary advertisement belongs, and determining the general segment as at least one expansion material segment obtained by expanding the primary advertisement.
In this embodiment, the common segments of the same category as the primary advertisement can provide some high-frequency applicable document materials of the same category advertisement for the target advertisement. Note that the advertisement generation device collects general segments for each advertisement category in advance, and stores the general segments in a sorted manner according to the advertisement category. The advertisement category is, for example, game category, e-commerce category, news category, and the like.
S102, generating a plurality of candidate advertisements according to the plurality of advertisement material segments and a preset pointer network.
In this embodiment, because the problem of combinatorial optimization can be simply and effectively solved by a Pointer network (Pointer Networks), for this reason, a preset Pointer network obtained by training a large number of samples is used, and a plurality of input advertisement material segments are arranged and combined based on the preset Pointer network to form a plurality of candidate advertisements with good compliance, so that the advertisement pool of an advertiser is expanded, and advertisements delivered by the advertiser can have a greater chance to be exposed or clicked, and the like.
It should be noted that, when the plurality of advertisement material segments are at least from the primary advertisement, the advertisement material library, the historical material segments compiled by the same advertiser, and the general segments of the same category as the primary advertisement, the plurality of candidate advertisements generated by the preset pointer network permutation and combination have certain guarantees on the aspects of the compliance, the consistency with the service product, and the like, and the problems of service tampering and mismatching are avoided as much as possible.
S103, determining a target advertisement from the candidate advertisements according to a preset rule.
In this embodiment, although a plurality of candidate advertisements with good popularity can be obtained through the preset pointer network, some unnecessary risks may be brought by introducing part of the material segments, and in order to avoid some unnecessary risks as much as possible, a risk control mechanism is introduced to filter out candidate advertisements with risks, so as to ensure the quality of the output target advertisements.
According to different application scenes, the target advertisement is determined from the candidate advertisements according to preset rules in different modes.
For example, the "determining a target advertisement from the plurality of candidate advertisements according to a preset rule" is implemented by: displaying the plurality of candidate advertisements; determining the target advertisement from the plurality of candidate advertisements according to the selection information of the plurality of candidate advertisements.
In the embodiment, unnecessary risks are reduced as much as possible in a manual intervention mode. Specifically, a plurality of candidate advertisements are directly displayed to an advertiser, and the advertiser selects a target advertisement to be delivered from the plurality of candidate advertisements.
Taking an advertisement putting platform with an advertisement generating device as a content provider as an example, displaying a selection interface on the advertisement putting platform, wherein the selection interface comprises a plurality of candidate advertisements; and receiving selection information of a user on a plurality of candidate advertisements, and determining a target advertisement from the plurality of candidate advertisements.
For another example, the "determining the target advertisement from the plurality of candidate advertisements according to the preset rule" is implemented by: and screening the plurality of candidate advertisements according to a preset screening model, and determining the screened candidate advertisements as the target advertisements.
In this embodiment, the target advertisement is automatically screened from the multiple candidate advertisements through a preset screening model. Wherein, the preset screening model is set according to the actual requirement. For example, the preset screening model is an advertisement classifier constructed by massive risk advertisements, and whether the advertisements to be identified are the risk advertisements can be judged. The risk advertisement is, for example, a common error collocation advertisement, an advertisement with sensitive words, and the like, but is not limited thereto.
Table 1 shows targeted advertising augmented based on native advertising.
TABLE 1
The advertisement generating method provided by the embodiment of the invention obtains a plurality of advertisement material segments; generating a plurality of candidate advertisements according to the plurality of advertisement material segments and a preset pointer network; and determining the target advertisement from the candidate advertisements according to a preset rule. Therefore, a plurality of advertisement material segments are arranged and combined based on a preset pointer network to automatically generate a plurality of candidate advertisements with good smoothness, a risk control mechanism is introduced to filter the plurality of candidate advertisements to generate target advertisements, the advertisement pool of an advertiser is expanded, and the advertisements delivered by the advertiser can have a larger chance to be exposed or clicked, and the like.
Fig. 2 is a schematic flow chart of another advertisement generation method according to an embodiment of the present invention. The present embodiment describes a training phase of a predetermined pointer network. With reference to fig. 2, on the basis of the embodiment shown in fig. 1, before step S101, the advertisement generation method further includes the following steps:
and S104, acquiring a sample set.
Specifically, the sample set includes a material segment set corresponding to a plurality of primary advertisement samples, and the material segment set corresponding to each primary advertisement sample includes at least one material segment of each primary advertisement sample and at least one corresponding augmented material segment.
The at least one expansion material segment corresponding to each of the primary advertisement samples may be derived from an advertisement material library, historical materials of an advertiser who writes the primary advertisement samples, and a common segment of the same category as the primary advertisement, but is not limited thereto.
It should be noted that at least one material segment of each of the raw advertisement samples is obtained by performing a cutting process on the raw advertisement samples with the granularity of the segment.
It should be noted that, the implementation manner of the at least one augmented material segment obtained by the native advertisement sample may refer to the manner of "extending the at least one augmented material segment obtained by the native advertisement" described in the above embodiment, and is not described herein again.
For example, the sample set includes a material segment set of the primary advertisement sample a, a material segment set of the primary advertisement sample B, and the like; the material segment set of the primary advertisement sample A comprises at least one material segment obtained by cutting the primary advertisement sample A and at least one expanded material segment obtained by expanding the primary advertisement sample A; the material segment set of the primary advertisement sample B comprises at least one material segment obtained by cutting the primary advertisement sample B and at least one expanded material segment obtained by expanding the primary advertisement sample B.
And S105, performing multiple rounds of training on the initial pointer network according to the material segment set corresponding to each primary advertisement sample to obtain the preset pointer network.
Specifically, a material segment set corresponding to one primary advertisement sample is adopted to train the pointer network once, parameters of the pointer network are updated and optimized after each training, and an expected output result of each training is an input primary advertisement sample.
In this embodiment, each of the native advertisement samples includes not only cutting the native advertisement sample to obtain one or more material segments, but also expanding the native advertisement sample to obtain one or more expanded material segments, so that the trained preset pointer network has the capability of arranging and combining the plurality of input advertisement material segments to form a plurality of candidate advertisements with good compliance, thereby expanding the advertisement pool of the advertiser, and enabling the advertisements delivered by the advertiser to have a greater chance to be exposed or clicked.
In this embodiment, the specific implementation process of S105 includes the following steps:
and S1, after the current round of training is finished, calculating a loss function corresponding to the current round of primary advertisement samples according to the current round of primary advertisement samples and the current round of actual output results, wherein the current round of actual output results are obtained by inputting the material fragment sets corresponding to the current round of primary advertisement samples into the initial pointer network.
Specifically, the degree of the gap between the characterization prediction and the actual data of the loss function is an index for measuring the quality of the model prediction, and more introduction on the loss function is described in the related art.
And S2, updating the initial pointer network according to the loss function and a random gradient descent method to obtain an updated pointer network, performing next round of training by adopting the updated pointer network, calculating a loss function corresponding to the next round of native advertisement samples according to the next round of native advertisement samples and a next round of actual output result, and repeating the steps until all the native advertisement samples in the sample set are trained, wherein the next round of actual output result is obtained by inputting the material fragment set corresponding to the next round of native advertisement samples into the updated pointer network.
Specifically, the random gradient descent method randomly gives a set of values of the parameters, and then updates the parameters, so that the loss function of the structure after each update can be reduced, and finally the minimum value is reached. More on the stochastic gradient descent method is detailed in the related art, and more on the updating of the parameters of the model according to the loss function and the stochastic gradient descent method is detailed in the related art.
In this embodiment, after the first round of training, parameters of the initial pointer network are adjusted according to the current loss function and a random gradient descent method, so as to obtain an updated pointer network; after each subsequent round of training, the updated pointer network obtained in the previous round is updated according to the current loss function.
And S3, taking the updated pointer network as the preset pointer network.
In this embodiment, a sample set is obtained by obtaining the sample set, where the sample set includes a material segment set corresponding to a plurality of primary advertisement samples, and the material segment set corresponding to each primary advertisement sample includes at least one material segment of each primary advertisement sample and at least one corresponding extended material segment; and performing multiple rounds of training on the initial pointer network according to the material segment set corresponding to each primary advertisement sample to obtain the preset pointer network. Therefore, the trained preset pointer network has the capability of arranging and combining a plurality of input advertisement material segments to form a plurality of candidate advertisements with good smoothness, the advertisement pool of the advertiser is expanded, and the advertisements delivered by the advertiser can have a larger chance to be exposed or clicked, and the like.
Fig. 3 is a schematic structural diagram of an advertisement generating apparatus according to an embodiment of the present invention. The present embodiment provides an advertisement generation apparatus, which is an execution subject of the advertisement generation method, and the execution subject is composed of hardware and/or software. As shown in fig. 3, the advertisement generating apparatus includes: the device comprises an acquisition module 11, a generation module 12 and a determination module 13.
An obtaining module 11, configured to obtain a plurality of advertisement material segments;
a generating module 12, configured to generate multiple candidate advertisements according to the multiple advertisement material segments and a preset pointer network;
and the determining module 13 is configured to determine a target advertisement from the plurality of candidate advertisements according to a preset rule.
In one possible implementation, the obtaining module 11 includes an obtaining unit and a breaking unit:
the acquisition unit is used for acquiring at least one material segment of the primary advertisement and at least one expanded material segment obtained by expanding the primary advertisement;
the scattering unit is used for scattering at least one material segment of the primary advertisement and at least one expansion material segment to obtain a plurality of advertisement material segments.
In a possible implementation manner, the obtaining unit 11 is specifically configured to:
acquiring at least one material segment matched with each material segment of the primary advertisement from an advertisement material library, and determining the at least one material segment matched with each material segment of the primary advertisement as at least one expanded material segment obtained by expanding the primary advertisement;
or the like, or, alternatively,
acquiring historical material segments corresponding to an advertiser who writes the primary advertisement, acquiring at least one historical material segment matched with each material segment of the primary advertisement from the historical material segments, and determining the at least one historical material segment matched with each material segment of the primary advertisement as at least one expanded material segment obtained by expanding the primary advertisement;
or the like, or, alternatively,
and acquiring a general segment corresponding to the category to which the primary advertisement belongs, and determining the general segment as at least one expansion material segment obtained by expanding the primary advertisement.
In a possible implementation manner, the determining module 13 is specifically configured to:
displaying the plurality of candidate advertisements;
determining the target advertisement from the plurality of candidate advertisements according to the selection information of the plurality of candidate advertisements;
or the like, or, alternatively,
and screening the plurality of candidate advertisements according to a preset screening model, and determining the screened candidate advertisements as the target advertisements.
It should be noted that the foregoing explanation on the embodiment of the advertisement generation method is also applicable to the advertisement generation apparatus of this embodiment, and is not repeated here.
The advertisement generating device provided by the embodiment of the invention obtains a plurality of advertisement material segments; generating a plurality of candidate advertisements according to the plurality of advertisement material segments and a preset pointer network; and determining the target advertisement from the candidate advertisements according to a preset rule. Therefore, a plurality of advertisement material segments are arranged and combined based on a preset pointer network to automatically generate a plurality of candidate advertisements with good smoothness, a risk control mechanism is introduced to filter the plurality of candidate advertisements to generate target advertisements, the advertisement pool of an advertiser is expanded, and the advertisements delivered by the advertiser can have a larger chance to be exposed or clicked, and the like.
Fig. 4 is a schematic structural diagram of another advertisement generating apparatus according to an embodiment of the present invention. With reference to fig. 4 in combination, on the basis of the embodiment shown in fig. 3, the apparatus further includes: a training module 14;
the obtaining module 11 is further configured to obtain a sample set, where the sample set includes a material segment set corresponding to a plurality of primary advertisement samples, and the material segment set corresponding to each primary advertisement sample includes at least one material segment of each primary advertisement sample and at least one corresponding augmented material segment;
the training module 14 is configured to perform multiple rounds of training on the initial pointer network according to the material segment set corresponding to each primary advertisement sample, so as to obtain the preset pointer network.
In a possible implementation manner, the training module 14 is specifically configured to:
after the current round of training is finished, calculating a loss function corresponding to the current round of native advertisement samples according to the current round of native advertisement samples and a current round of actual output results, wherein the current round of actual output results are obtained by inputting material fragment sets corresponding to the current round of native advertisement samples into the initial pointer network;
updating the initial pointer network according to the loss function and a random gradient descent method to obtain an updated pointer network, performing next round of training by using the updated pointer network, calculating a loss function corresponding to a next round of native advertisement samples according to a next round of native advertisement samples and a next round of actual output results, and repeating the step until all the native advertisement samples in the sample set are trained, wherein the next round of actual output results are obtained by inputting a material fragment set corresponding to the next round of native advertisement samples into the updated pointer network;
and taking the updated pointer network as the preset pointer network.
It should be noted that the foregoing explanation on the embodiment of the advertisement generation method is also applicable to the advertisement generation apparatus of this embodiment, and is not repeated here.
The advertisement generating device provided by the embodiment of the invention obtains the sample set by obtaining the sample set, wherein the sample set comprises the material segment sets corresponding to a plurality of primary advertisement samples, and the material segment set corresponding to each primary advertisement sample comprises at least one material segment of each primary advertisement sample and at least one corresponding expansion material segment; and performing multiple rounds of training on the initial pointer network according to the material segment set corresponding to each primary advertisement sample to obtain the preset pointer network. Therefore, the trained preset pointer network has the capability of arranging and combining a plurality of input advertisement material segments to form a plurality of candidate advertisements with good smoothness, the advertisement pool of the advertiser is expanded, and the advertisements delivered by the advertiser can have a larger chance to be exposed or clicked, and the like.
Fig. 5 is a schematic structural diagram of a computer device according to an embodiment of the present invention. The computer device includes:
The processor 1002, when executing the program, implements the advertisement generation method provided in the above-described embodiments.
Further, the computer device further comprises:
a communication interface 1003 for communicating between the memory 1001 and the processor 1002.
A memory 1001 for storing computer programs that may be run on the processor 1002.
The processor 1002 is configured to implement the advertisement generating method according to the foregoing embodiment when executing the program.
If the memory 1001, the processor 1002, and the communication interface 1003 are implemented independently, the communication interface 1003, the memory 1001, and the processor 1002 may be connected to each other through a bus and perform communication with each other. The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended ISA (EISA) bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown in FIG. 5, but this is not intended to represent only one bus or type of bus.
Optionally, in a specific implementation, if the memory 1001, the processor 1002, and the communication interface 1003 are integrated on one chip, the memory 1001, the processor 1002, and the communication interface 1003 may complete communication with each other through an internal interface.
The processor 1002 may be a Central Processing Unit (CPU), an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits configured to implement embodiments of the present invention.
The present embodiment also provides a computer-readable storage medium having stored thereon a computer program, characterized in that the program, when executed by a processor, implements the advertisement generation method as described above.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present invention.
The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable storage medium, and when the program is executed, the program includes one or a combination of the steps of the method embodiments.
In addition, functional units in the embodiments of the present invention may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a stand-alone product, may also be stored in a computer readable storage medium.
The storage medium mentioned above may be a read-only memory, a magnetic or optical disk, etc. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Claims (12)
1. An advertisement generation method, comprising:
the method comprises the steps of obtaining a plurality of advertisement material segments, wherein the advertisement material segments comprise a primary advertisement written by an advertiser and at least one expansion material segment obtained after the primary advertisement is expanded, obtaining at least one material segment matched with each material segment of the primary advertisement from an advertisement material library, and determining at least one material segment matched with each material segment of the primary advertisement as at least one expansion material segment obtained by expanding the primary advertisement;
combining the native advertisements and the extended material segments according to a preset pointer network to generate a plurality of candidate advertisements;
and determining the target advertisement from the candidate advertisements according to a preset rule according to different application scenes, wherein the target advertisement determining modes from the candidate advertisements according to the preset rule are different corresponding to different application scenes.
2. The method of claim 1, further comprising:
obtaining a sample set, wherein the sample set comprises material segment sets corresponding to a plurality of primary advertisement samples, and the material segment set corresponding to each primary advertisement sample comprises at least one material segment of each primary advertisement sample and at least one corresponding expansion material segment;
and performing multiple rounds of training on the initial pointer network according to the material segment set corresponding to each primary advertisement sample to obtain the preset pointer network.
3. The method of claim 2, wherein the training the initial pointer network for multiple rounds according to the material segment sets corresponding to the respective native advertisement samples to obtain the preset pointer network comprises:
after the current round of training is finished, calculating a loss function corresponding to the current round of native advertisement samples according to the current round of native advertisement samples and a current round of actual output results, wherein the current round of actual output results are obtained by inputting material fragment sets corresponding to the current round of native advertisement samples into the initial pointer network;
updating the initial pointer network according to the loss function and a random gradient descent method to obtain an updated pointer network, performing next round of training by using the updated pointer network, calculating a loss function corresponding to a next round of native advertisement samples according to a next round of native advertisement samples and a next round of actual output results, and repeating the step until all the native advertisement samples in the sample set are trained, wherein the next round of actual output results are obtained by inputting a material fragment set corresponding to the next round of native advertisement samples into the updated pointer network;
and taking the updated pointer network as the preset pointer network.
4. The method of claim 1, wherein obtaining a plurality of advertising material segments comprises:
obtaining at least one material segment of the primary advertisement and at least one expanded material segment obtained by expanding the primary advertisement;
and scattering at least one material segment of the primary advertisement and at least one expansion material segment to obtain a plurality of advertisement material segments.
5. The method of claim 1, wherein the determining the target advertisement from the candidate advertisements according to the preset rule comprises:
displaying the plurality of candidate advertisements;
determining the target advertisement from the plurality of candidate advertisements according to the selection information of the plurality of candidate advertisements;
or the like, or, alternatively,
and screening the plurality of candidate advertisements according to a preset screening model, and determining the screened candidate advertisements as the target advertisements.
6. An advertisement generation apparatus, comprising:
the system comprises an acquisition module, a display module and a display module, wherein the acquisition module is used for acquiring a plurality of advertisement material segments, the advertisement material segments comprise a primary advertisement written by an advertiser and at least one expansion material segment obtained after the primary advertisement is expanded, at least one material segment matched with each material segment of the primary advertisement is acquired from an advertisement material library, and at least one material segment matched with each material segment of the primary advertisement is determined as at least one expansion material segment obtained by expanding the primary advertisement;
the generating module is used for combining the native advertisements and the expansion material segments according to a preset pointer network so as to generate a plurality of candidate advertisements;
and the determining module is used for determining the target advertisement from the candidate advertisements according to the different application scenes and the preset rules, and the different application scenes correspond to different target advertisement determining modes from the candidate advertisements.
7. The apparatus of claim 6, further comprising: a training module;
the obtaining module is further configured to obtain a sample set, where the sample set includes a material segment set corresponding to a plurality of primary advertisement samples, and the material segment set corresponding to each primary advertisement sample includes at least one material segment of each primary advertisement sample and at least one corresponding augmented material segment;
and the training module is used for performing multi-round training on the initial pointer network according to the material segment set corresponding to each primary advertisement sample to obtain the preset pointer network.
8. The apparatus of claim 7, wherein the training module is specifically configured to:
after the current round of training is finished, calculating a loss function corresponding to the current round of native advertisement samples according to the current round of native advertisement samples and a current round of actual output results, wherein the current round of actual output results are obtained by inputting material fragment sets corresponding to the current round of native advertisement samples into the initial pointer network;
updating the initial pointer network according to the loss function and a random gradient descent method to obtain an updated pointer network, performing next round of training by using the updated pointer network, calculating a loss function corresponding to a next round of native advertisement samples according to a next round of native advertisement samples and a next round of actual output results, and repeating the step until all the native advertisement samples in the sample set are trained, wherein the next round of actual output results are obtained by inputting a material fragment set corresponding to the next round of native advertisement samples into the updated pointer network;
and taking the updated pointer network as the preset pointer network.
9. The apparatus of claim 6, wherein the acquisition module comprises an acquisition unit and a breakup unit:
the acquisition unit is used for acquiring at least one material segment of the primary advertisement and at least one expanded material segment obtained by expanding the primary advertisement;
the scattering unit is used for scattering at least one material segment of the primary advertisement and at least one expansion material segment to obtain a plurality of advertisement material segments.
10. The apparatus of claim 6, wherein the determining module is specifically configured to:
displaying the plurality of candidate advertisements;
determining the target advertisement from the plurality of candidate advertisements according to the selection information of the plurality of candidate advertisements;
or the like, or, alternatively,
and screening the plurality of candidate advertisements according to a preset screening model, and determining the screened candidate advertisements as the target advertisements.
11. A computer device, comprising:
memory, processor and computer program stored on the memory and executable on the processor, characterized in that the processor implements the advertisement generation method according to any of claims 1-5 when executing the program.
12. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the advertisement generation method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910451519.4A CN110189173B (en) | 2019-05-28 | 2019-05-28 | Advertisement generation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910451519.4A CN110189173B (en) | 2019-05-28 | 2019-05-28 | Advertisement generation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110189173A CN110189173A (en) | 2019-08-30 |
| CN110189173B true CN110189173B (en) | 2021-11-30 |
Family
ID=67718286
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910451519.4A Active CN110189173B (en) | 2019-05-28 | 2019-05-28 | Advertisement generation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110189173B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112862516B (en) * | 2021-01-14 | 2024-03-12 | 北京达佳互联信息技术有限公司 | Resource release method and device, electronic equipment and storage medium |
| CN112767024B (en) * | 2021-01-15 | 2024-02-20 | 北京搜狗科技发展有限公司 | Advertisement creative generation method and related device |
| CN113099267B (en) * | 2021-06-04 | 2021-11-12 | 武汉卓尔数字传媒科技有限公司 | Video generation method and device, electronic equipment and storage medium |
| CN115471263A (en) * | 2022-09-16 | 2022-12-13 | 深圳掌屿软件有限公司 | Advertisement generation method, device, equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1321278A (en) * | 1998-10-01 | 2001-11-07 | 国际商业机器公司 | Systems, methods and computer program products for assigning, generating and delivering content to intranet users |
| US7920689B2 (en) * | 2005-05-25 | 2011-04-05 | Radziewicz Clifford J | Ringback replacement insertion system |
| CN108280112A (en) * | 2017-06-22 | 2018-07-13 | 腾讯科技（深圳）有限公司 | Abstraction generating method, device and computer equipment |
| CN109544201A (en) * | 2018-10-11 | 2019-03-29 | 北京奇虎科技有限公司 | A kind of generation method of Advertising Copy, device and electronic equipment |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109241536B (en) * | 2018-09-21 | 2020-11-06 | 浙江大学 | Deep learning self-attention mechanism-based sentence sequencing method |
| CN109493116B (en) * | 2018-10-15 | 2021-07-23 | 上海趣蕴网络科技有限公司 | Method and system for automatically generating advertisement |
-
2019
- 2019-05-28 CN CN201910451519.4A patent/CN110189173B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1321278A (en) * | 1998-10-01 | 2001-11-07 | 国际商业机器公司 | Systems, methods and computer program products for assigning, generating and delivering content to intranet users |
| US7920689B2 (en) * | 2005-05-25 | 2011-04-05 | Radziewicz Clifford J | Ringback replacement insertion system |
| CN108280112A (en) * | 2017-06-22 | 2018-07-13 | 腾讯科技（深圳）有限公司 | Abstraction generating method, device and computer equipment |
| CN109544201A (en) * | 2018-10-11 | 2019-03-29 | 北京奇虎科技有限公司 | A kind of generation method of Advertising Copy, device and electronic equipment |
Non-Patent Citations (2)
| Title |
|---|
| End-to-End Neural Sentence Ordering Using Pointer Network;Jingjing Gong;《https://arxiv.org/abs/1611.04953》;20161125;1-3 * |
| 基于注意力机制的句子排序方法;栾克鑫;《中文信息学报》;20180131;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110189173A (en) | 2019-08-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110189173B (en) | Advertisement generation method and device | |
| CN109710841B (en) | Comment recommendation method and device | |
| CN108733706B (en) | Method and device for generating heat information | |
| CN109511015B (en) | Multimedia resource recommendation method, device, storage medium and equipment | |
| CN107590102B (en) | Random forest model generation method and device | |
| CN110968767B (en) | Ranking engine training method and device, and business card ranking method and device | |
| CN108304426B (en) | Identification obtaining method and device | |
| CN113422988B (en) | Live broadcast cover generation method and device, electronic equipment and storage medium | |
| US20130318021A1 (en) | Information processing apparatus, information processing method, and program | |
| CN112995690B (en) | Live content category identification method, device, electronic equipment and readable storage medium | |
| CN111898037A (en) | Method and device for updating ranking list data and server | |
| CN109582904A (en) | Amending method, device, server, terminal and the storage medium of content are issued | |
| CN112507218A (en) | Business object recommendation method and device, electronic equipment and storage medium | |
| CN112199582A (en) | Content recommendation method, device, equipment and medium | |
| CN106354841B (en) | Method and device for testing recommendation algorithm | |
| CN111371605A (en) | Network service scoring method and device | |
| CN110362751A (en) | Business recommended method, apparatus, computer equipment and storage medium | |
| CN114520931A (en) | Video generation method and device, electronic equipment and readable storage medium | |
| CN110245070B (en) | Test user screening method, device, computer equipment and storage medium | |
| CN111475409B (en) | System test method, device, electronic equipment and storage medium | |
| CN114943005A (en) | Picture display processing method and device | |
| CN110995529B (en) | Configuration data testing method and device of recommendation system and electronic equipment | |
| CN113779403A (en) | Book information acquisition method, book information display method, server and user terminal | |
| CN110730342A (en) | Video quality analysis method and device, server and terminal | |
| CN111950821A (en) | Test method, test device and server |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |