A promotion information processing method, device, equipment, medium, and program product

By using the overlapping feature vector combination in the feature library to build the promotion information identification model, the problem of repeated vector processing in the construction of the promotion information identification model is solved, and the saving of computing resources and the improvement of model construction efficiency is achieved.

CN113988949BActive Publication Date: 2024-12-06BEIJING YOUZHUJU NETWORK TECH CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202111348147.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-15
Publication Date
2024-12-06
Estimated Expiration
2041-11-15

AI Technical Summary

Technical Problem

When building multiple different types of promotional information identification models, the prior art needs to repeatedly vectorize sample promotion information, resulting in wasted computing resources and reducing the efficiency of building models.

Method used

By obtaining the combination of feature vectors in the feature library, using overlapping feature vectors to build a variety of promotion information identification models, avoiding repeated vector processing of sample promotion information, saving computing resources and improving model construction efficiency.

Benefits of technology

This method effectively saves computing resources, improves the efficiency of building models, and reduces storage space requirements by storing feature vectors with high similarity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113988949B_ABST
    Figure CN113988949B_ABST
Patent Text Reader

Abstract

The present disclosure provides a promotion information processing method, device, equipment, medium, and program product. The method includes: obtaining a feature library, the feature library includes a feature vector obtained by vectorizing at least one sample promotion information; obtaining a first feature vector combination and a second feature vector combination from the feature library, wherein the first feature vector combination and the second feature vector combination overlap; using the first feature vector combination to construct a first promotion information recognition model, and using the second feature vector combination to construct a second promotion information recognition model. The method can save computing resources and improve the efficiency of model construction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a promotion information processing method, device, equipment, computer-readable storage medium, and computer program product. Background Art

[0002] Merchants usually publish promotional information (such as advertisements) with target objects (such as promoted digital products, tourist attractions, literary works, etc.) to increase users' attention to the target objects and guide users to consume the target objects. With the development of artificial intelligence technology, it is possible to construct a variety of different types of promotional information recognition models to identify promotional information from multiple angles (such as the perspective of visual effects and auditory effects), and then obtain modification suggestions from multiple angles. In this way, by modifying the promotional information based on modification suggestions from multiple angles, it is possible to obtain promotional information that is more friendly to both visual and auditory effects, thereby achieving better promotional effects.

[0003] At present, in the process of building a promotion information recognition model, it is necessary to first obtain the identifier of the sample promotion information from the database, such as a uniform resource locator (URL), obtain the sample promotion information through the identifier, and then vectorize the sample promotion information to obtain a feature vector (such as a text vector, an image vector), and then train the model based on the feature vector to obtain a promotion information recognition model.

[0004] However, when it is necessary to build multiple different types of promotion information recognition models, it is necessary to repeat the above process of vectorizing the sample promotion information to obtain the feature vector for model training. Therefore, the above method of repeatedly vectorizing the sample promotion information not only wastes computing resources, but also reduces the efficiency of model building. Summary of the invention

[0005] The purpose of the present disclosure is to provide a promotion information processing method, device, equipment, computer-readable storage medium, and computer program product that can save computing resources and improve the efficiency of building models.

[0006] In a first aspect, the present disclosure provides a promotion information processing method, comprising:

[0007] Acquire a feature library, wherein the feature library includes a feature vector obtained by vectorizing at least one sample promotion information;

[0008] Acquire a first feature vector combination and a second feature vector combination from the feature library, where the first feature vector combination and the second feature vector combination overlap;

[0009] A first promotion information recognition model is constructed by using the first feature vector combination, and a second promotion information recognition model is constructed by using the second feature vector combination.

[0010] In a second aspect, the present disclosure provides a promotion information processing device, comprising:

[0011] A vector unit, used to obtain a feature library, wherein the feature library includes a feature vector obtained by vectorizing at least one sample promotion information;

[0012] An acquiring unit, configured to acquire a first feature vector combination and a second feature vector combination from the feature library, wherein the first feature vector combination and the second feature vector combination overlap;

[0013] A construction unit is used to construct a first promotion information recognition model by using the first feature vector combination, and to construct a second promotion information recognition model by using the second feature vector combination.

[0014] In a third aspect, the present disclosure provides a computer-readable medium having a computer program stored thereon, which, when executed by a processing device, implements the steps of any one of the methods described in the first aspect of the present disclosure.

[0015] In a fourth aspect, the present disclosure provides an electronic device, including:

[0016] a storage device having a computer program stored thereon;

[0017] A processing device is used to execute the computer program in the storage device to implement the steps of any one of the methods described in the first aspect of the present disclosure.

[0018] In a fifth aspect, the present disclosure provides a computer program product comprising instructions, which, when executed on a device, enables the device to execute the method described in any one of the implementations of the first aspect.

[0019] It can be seen from the above technical solution that the present disclosure has the following advantages:

[0020] The method obtains multiple feature vector combinations, such as a first feature vector combination and a second feature vector combination, based on feature vectors obtained by vectorizing sample promotion information in a feature library, and the first feature vector combination and the second feature vector combination overlap. In the process of constructing a first promotion information recognition model using the first feature vector combination and constructing a second promotion information recognition model using the second feature vector combination, for feature vectors that overlap between the first feature vector combination and the second feature vector combination, the overlapping feature vectors are obtained based on a single vectorization of the sample promotion information, and there is no need to vectorize the sample promotion information again. In this way, the method not only saves computing resources, but also improves the efficiency of model construction.

[0021] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical method of the embodiments of the present disclosure, the drawings required for use in the embodiments are briefly introduced below.

[0023] Figure 1 A flow chart of a method for processing promotion information provided by an embodiment of the present disclosure;

[0024] Figure 2 A schematic diagram of feature classification provided in an embodiment of the present disclosure;

[0025] Figure 3 A schematic diagram of a sample promotion information vectorization process provided by an embodiment of the present disclosure;

[0026] Figure 4 A schematic diagram of a device for processing promotion information provided by an embodiment of the present disclosure;

[0027] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0028] The terms "first" and "second" in the embodiments of the present disclosure are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0029] First, some technical terms involved in the embodiments of the present disclosure are introduced.

[0030] Promotional information refers to information used to promote or publicize a target object. For example, promotional information can be an advertisement. By publishing advertisements, users' attention to the target object in the advertisement can be increased. In some examples, advertisements can be used to call on users to save water resources; advertisements can also be used to guide users to consume digital products in the advertisement. In order to achieve better promotional effects for promotional information, a variety of different types of promotional information recognition models are constructed based on artificial intelligence technology, and then promotional information is identified from multiple angles to obtain modification suggestions from multiple angles.

[0031] When the sample promotion information is a video advertisement, directly storing the sample promotion information requires a large storage space. In order to reduce the storage space occupied by the sample promotion information, it is usually adopted to store only the URL of the sample promotion information in the database. When the sample promotion information is needed (for example, to build a promotion information recognition model based on the sample promotion information), the sample promotion information is first obtained according to the URL. When building multiple different types of promotion information recognition models, it is also necessary to repeatedly perform vectorization processing on the same sample promotion information.

[0032] Taking the construction of the first promotion information recognition model and the second promotion information recognition model as an example, in the process of constructing the first promotion information recognition model, it is necessary to first obtain sample promotion information, then vectorize the sample promotion information, obtain the feature vector of the sample promotion information, and then construct the first promotion information recognition model based on the feature vector. Since the feature vector obtained after the vectorization of the sample promotion information is not stored in the process of constructing the first promotion information recognition model, it is necessary to vectorize the sample promotion information again in the subsequent process of constructing the second promotion information recognition model. Repeatedly vectorizing the same sample promotion information not only leads to a waste of computing resources, but also reduces the efficiency of model construction.

[0033] In view of this, an embodiment of the present disclosure provides a promotion information processing method. Specifically, the method includes obtaining a feature library, the feature library includes a feature vector obtained by vectorizing at least one sample promotion information, then obtaining a first feature vector combination and a second feature vector combination from the feature library, the first feature vector combination and the second feature vector combination overlap, then using the first feature vector combination to construct a first promotion information recognition model, and using the second feature vector combination to construct a second promotion information recognition model.

[0034] In the method, in the process of constructing the first promotion information recognition model using the first feature vector combination and constructing the second promotion information recognition model using the second feature vector combination, for the overlapping feature vectors in the first feature vector combination and the second feature vector combination, the overlapping feature vectors are obtained based on a vectorization process of the sample promotion information, and the overlapping feature vectors are used to construct multiple promotion information recognition models. It can be seen that in the process of constructing multiple models, the method does not need to vectorize the sample promotion information again by directly obtaining the feature vectors from the feature library, which not only saves computing resources but also improves the efficiency of model construction.

[0035] Furthermore, in the process of storing the feature vector in the feature library, the method can also calculate the similarity between the feature vector stored in the feature library and the feature vector to be stored, and when the similarity is lower than a preset threshold, the feature vector to be stored is stored in the feature library. In this way, the method can also reduce the feature vectors with high similarity in the feature library, further saving the storage space required for storing the feature vectors.

[0036] In some possible implementations, the above-mentioned promotion information processing method can be executed by a processing device. Among them, the processing device can be a device with data processing functions. The processing device can be a terminal or a server. Terminals include but are not limited to smart phones, tablet computers, laptops, personal digital assistants (PDAs) or smart wearable devices. The server can be a cloud server, such as a central server in a central cloud computing cluster, or an edge server in an edge cloud computing cluster. Of course, the server can also be a server in a local data center. A local data center refers to a data center directly controlled by a user.

[0037] In order to make the technical solution of the present disclosure clearer and easier to understand, the promotion information processing method provided by the embodiment of the present disclosure is introduced from the perspective of a processing device.

[0038] like Figure 1 As shown, this figure is a flow chart of a promotion information processing method provided by an embodiment of the present disclosure, and the method can be executed by a processing device. Specifically, the method includes:

[0039] S101: A processing device obtains a feature library.

[0040] The feature library includes a feature vector obtained by vectorizing at least one sample promotion information, and the feature vector of at least one sample promotion information is stored in the feature library.

[0041] In some embodiments, the processing device may pre-construct the feature library. The processing device may vectorize the sample promotion information from multiple dimensions, and thereby obtain feature vectors corresponding to the multiple dimensions. The multiple dimensions may include any combination of audio, video, pictures, and text. For example, the processing device may vectorize the sample promotion information from the picture dimension and the text dimension, and thereby obtain a feature vector corresponding to the picture dimension and a feature vector corresponding to the text dimension, and then store the feature vector corresponding to the picture dimension and the feature vector corresponding to the text dimension in the feature library, thereby realizing the construction of the feature library. The disclosed embodiment provides a method for vectorizing sample promotion information, which will be described in detail later.

[0042] It should be noted that the above is only an introduction to the method of vectorizing a sample promotion information. In the actual process, the processing device adopts the above method of vectorizing a sample promotion information to perform vectorization processing on each sample promotion information in at least one sample promotion information to obtain the feature vector of each sample promotion information, thereby enriching the feature vector in the feature library.

[0043] In other embodiments, the processing device may also directly obtain a pre-built feature library, which includes a feature vector obtained by vectorizing at least one sample promotion information. The method of pre-building the feature library can be described in the above content and will not be repeated here.

[0044] In some possible implementations, the sample promotion information may be a sample advertisement. Based on this, the processing device may classify the features represented by the feature vectors stored in the feature library based on the sample advertisement. For example, the feature vectors stored in the feature library may include at least one of the publisher feature of the sample promotion information, the material content feature of the sample promotion information, the material remark feature of the sample promotion information, the modification suggestion feature of the sample promotion information, and the version maintenance feature of the sample promotion information.

[0045] like Figure 2 As shown in the figure, a schematic diagram of feature classification is shown. Taking the sample promotion information as a sample advertisement as an example, the sample advertisement can be divided into advertisement publisher features, advertisement plan or creative features, advertisement material features, advertisement material note features, advertisement modification suggestion features, and advertisement version maintenance features. Among them, advertisement plan or creative features and advertisement material features can be collectively referred to as advertisement material content features.

[0046] Furthermore, the processing device can also subdivide the characteristics of the advertiser into advertiser category, advertiser avatar and advertiser source; it should be noted that this information requires the user's authorization in advance, and the processing device can obtain data such as the advertiser characteristics after obtaining the user's authorization to use the corresponding data.

[0047] The material content features of the advertisements are subdivided into copy, pictures, videos, landing pages, industry, number of creative ideas, number of copy, number of pictures and number of videos; the material note features of the advertisements are subdivided into qualification authorization information, validity period, trademark and logo; the version maintenance features of the advertisements are subdivided into the number of modifications and the number of submissions.

[0048] It should be noted that the above Figure 2The schematic diagram of feature classification shown is only a possible classification method. Those skilled in the art can choose other classification methods according to actual needs, and the embodiments of the present disclosure are not limited to this.

[0049] In some embodiments, the processing device may also update the sample promotion information, and then store the feature vector obtained by vectorizing the updated sample promotion information in the feature library. For example, the processing device may update the sample advertisement, and then store the feature vector obtained by vectorizing the updated sample advertisement in the feature library.

[0050] In some possible implementations, the processing device may obtain an update request for the sample advertisement, the update request carrying the updated information of the sample advertisement, and then the processing device updates the sample advertisement according to the updated information carried in the update request. The updated information of the sample advertisement may be the above-mentioned publisher avatar, publisher category, copy, picture, etc. In some examples, when the publisher avatar needs to be updated, the new publisher avatar may be carried in the update request, and then the processing device may update the original sample advertisement based on the new publisher avatar.

[0051] Next, the processing device stores the feature vector obtained by vectorizing the updated publisher avatar into the feature library, and adds a version identifier (e.g., version number) to the feature vector of the updated publisher avatar. In some examples, the processing device can store the feature vector of the publisher avatar before the update and the feature vector of the updated publisher avatar at the same time by assigning version identifiers to the feature vector of the publisher avatar before the update and the feature vector of the updated publisher avatar. For example, the version identifier of the feature vector of the publisher avatar before the update is "V1.1", and the version identifier of the feature vector of the updated publisher avatar is "V1.2".

[0052] It should be noted that the above version identification is only an example, and the embodiment of the present disclosure does not specifically limit the form of the version identification.

[0053] In other embodiments, the processing device may also update the publisher category, publisher avatar, and publisher source in the publisher characteristics, and may also update the copy, picture, video, landing page, industry, number of creatives, number of copy, number of pictures, and number of videos in the material content feature details, and may also update the qualification authorization information, validity period, trademark and logo in the material note features of the advertisement, and may also update the number of modifications and submissions in the version maintenance features of the advertisement.

[0054] In some possible implementations, the processing device may classify the features of the sample advertisements into long-term features and short-term features based on the frequency of updating the sample advertisements. Among them, the long-term features are features whose frequency of updates is lower than a preset degree, and the short-term features are features whose frequency of updates is higher than or equal to the preset degree. The frequency of updates can be characterized by the number of updates within a preset time, for example, the frequency can be 3 updates within a week. For example, the preset degree can be updated once within a month, the long-term features are features whose frequency of updates is lower than once within a month, and the short-term features are features whose frequency of updates is higher than or equal to once within a month.

[0055] In actual scenarios, the processing device updates the advertisement publisher characteristics and advertisement modification suggestion characteristics less frequently, for example, both are lower than the above preset degree; the processing device updates the advertisement material characteristics, advertisement plan or creative characteristics, advertisement material note characteristics and advertisement version maintenance characteristics more frequently, for example, both are higher than or equal to the above preset degree. Based on this, long-term characteristics may include advertisement publisher characteristics and advertisement modification suggestion characteristics, and short-term characteristics may include advertisement material characteristics, advertisement plan or creative characteristics, advertisement material note characteristics and advertisement version maintenance characteristics.

[0056] S102: The processing device obtains a first feature vector combination and a second feature vector combination from a feature library, and the first feature vector combination and the second feature vector combination overlap.

[0057] The first feature vector combination includes multiple feature vectors, such as feature vector 1, feature vector 2, and feature vector 3; the second feature vector combination includes multiple feature vectors, such as feature vector 2, feature vector 3, and feature vector 4. In some examples, the overlap between the first feature vector combination and the second feature vector combination means that both the first feature vector combination and the second feature vector combination include the feature vector 2 and the feature vector 3.

[0058] It should be noted that the embodiments of the present disclosure do not specifically limit the number of overlapping feature vectors, and the above example is only introduced by taking the number of overlapping feature vectors as 2 as an example. Of course, in actual scenarios, the number of overlapping feature vectors may also be 1, or more, such as 4 or 5.

[0059] For the above-mentioned overlapping feature vectors (e.g., feature vector 2 and feature vector 3), the processing device does not need to perform vectorization processing on the sample promotion information again, but directly obtains the overlapping feature vector from the feature library. In this way, the method reduces the number of times the processing device performs vectorization processing on the sample promotion information, saving computing resources.

[0060] In some possible implementations, the processing device may obtain a feature vector corresponding to the feature identifier from a feature library based on the feature identifier of the feature vector (e.g., feature number, feature sequence number, etc.). Continuing with the above example, the processing device may obtain feature vector 1, feature vector 2, and feature vector 3 from the feature library based on the feature identifier of feature vector 1, the feature identifier of feature vector 2, and the feature identifier of feature vector 3, respectively, and thereby obtain a first feature vector combination. Similarly, the processing device may obtain feature vector 2, feature vector 3, and feature vector 4 from the feature library based on the feature identifier of feature vector 2, the feature identifier of feature vector 3, and the feature identifier of feature vector 4, respectively, and thereby obtain a second feature vector combination.

[0061] In some other possible implementations, in the feature library, a feature identifier may correspond to multiple different versions of feature vectors. Based on this, the processing device may also determine the multiple different versions of feature vectors corresponding to the feature identifier based on the feature identifier of the feature vector and the version identifier of the feature vector, and then obtain the feature vector corresponding to the version identifier based on the version identifier. See Table 1 below:

[0062] Table 1:

[0063]

[0064] Among them, the feature identifier "001" corresponds to "feature vector 1-1", "feature vector 1-2" and "feature vector 1-3", and their version identifiers are "V1.1", "V1.2" and "V1.3" respectively. The feature identifier "002" corresponds to "feature vector 2-1" and "feature vector 2-2", and their version identifiers are "V1.1" and "V1.2" respectively. In some examples, when the feature identifier is "001" and the version identifier is "V1.2", the processing device can obtain "feature vector 1-2" from the feature library based on the corresponding relationship recorded in Table 1 above.

[0065] S103: The processing device constructs a first promotion information recognition model using the first feature vector combination, and constructs a second promotion information recognition model using the second feature vector combination.

[0066] Then, the processing device constructs a first promotion information recognition model using the first feature vector combination obtained from the feature library, and constructs a second promotion information recognition model using the second feature vector combination. The first promotion information recognition model and the second promotion information recognition model are used to recognize the promotion information to be recognized from different angles, thereby obtaining recognition results for different angles.

[0067] In some embodiments, after constructing the first promotion information recognition model and the second promotion information recognition model, the processing device can also use the first promotion information recognition model and the second promotion information recognition model to identify the promotion information. Specifically, the processing device can obtain the promotion information to be identified, input the promotion information to be identified into the first promotion information recognition model to obtain a first recognition result, and input the promotion information to be identified into the second promotion information recognition model to obtain a second recognition result. The first recognition result and the second recognition result include different modification suggestions for the above-mentioned promotion information to be identified. For example, the first recognition result includes a first modification suggestion for the above-mentioned promotion information to be identified, and the second recognition result includes a second modification suggestion for the above-mentioned promotion information to be identified, and the first modification suggestion is different from the second modification suggestion.

[0068] In some examples, the promotion information to be identified may be unpublished promotion information. Thus, the unpublished promotion information may be modified from multiple perspectives based on the first identification result and the second identification result, thereby improving the promotion effect of the unpublished promotion information after it is published.

[0069] In other examples, the promotion information to be identified may also be the promotion information that has been published. In this way, it is possible to judge whether the published promotion information meets the requirements from multiple perspectives based on the above-mentioned first identification result and the second identification result. When the first identification result or the second identification result indicates that the published promotion information does not meet the requirements, taking the first identification result not meeting the requirements as an example, the modification suggestion included in the first identification result may be "The promotion information contains words such as 'XX', which may be related to gambling, and it is recommended to be removed from the shelves", and then the published promotion information can be removed from the shelves according to the modification suggestion, thereby reducing the number of low-quality promotion information.

[0070] Based on the above description, an embodiment of the present disclosure provides a promotion information processing method. The method first obtains a feature library based on the feature vector obtained by vectorizing the sample promotion information, and then obtains multiple feature vector combinations from the feature library, such as a first feature vector combination and a second feature vector combination, and the first feature vector combination and the second feature vector combination overlap. In the process of constructing a first promotion information recognition model using the first feature vector combination and constructing a second promotion information recognition model using the second feature vector combination, for the overlapping feature vectors in the first feature vector combination and the second feature vector combination, the overlapping feature vectors are obtained based on a vectorization of the sample promotion information, and the overlapping feature vectors are used to construct a variety of promotion information recognition models. It can be seen that in the process of constructing multiple models, the method does not need to vectorize the sample promotion information again by directly obtaining feature vectors from the feature library, which not only saves computing resources but also improves the efficiency of model construction.

[0071] The embodiment of the present disclosure provides a method for vectorizing sample promotion information, which is described in detail below.

[0072] like Figure 3 As shown, the figure shows a schematic diagram of vectorizing sample promotion information. Taking the sample promotion information as an advertisement as an example, the sample promotion information may include picture materials, video materials, landing page materials, etc. Among them, the picture materials further include pictures, picture optical character recognition (OCR) text; the video materials further include video frame pictures, video frame OCR text and video automatic speech recognition (ASR) text; the landing page materials further include pictures, video frames, picture OCR text, video frame OCR text, landing page text, video ASR text. The sample promotion information may also include discrete features and copywriting, wherein the discrete features may be the industry to which the sample promotion information belongs, etc., and the copywriting may be the title of the sample promotion information, the title of the material, etc.

[0073] For data in different forms, the processing device can use different vectorization methods. In some embodiments, for discrete features, the processing device can use the one-hot encoding One-Hot method for vectorization to obtain a discrete feature vector. For short texts such as copywriting and video ASR text, the processing device can use the transformer method for vectorization to obtain a text feature vector, wherein the short text can be a text with fewer than a preset number of characters. For long texts such as landing page text, the processing device can use the pre-trained language model XLNet method for vectorization to obtain a text feature vector. For image OCR text and video frame OCR text, it can be vectorized by convolution and pooling to obtain a text feature vector. For pictures and video frame pictures, the processing device can use the residual network ResNet method for vectorization to obtain a picture feature vector.

[0074] Then, the processing device can store the above discrete feature vectors, text feature vectors and image feature vectors in a feature repository. In some possible implementations, the processing device can also respectively calculate the similarity between the target feature vectors in the feature vectors of the above multiple dimensions and the historical feature vectors in the feature repository. Taking the target feature vector as a text feature vector as an example, the processing device can calculate the similarity between the text feature vector and the historical feature vector in the feature repository. When the similarity is lower than a preset threshold, the processing device can store the text feature vector in the feature repository; when the similarity is higher than or equal to the preset threshold, the processing device does not need to store the text feature vector in the feature repository, thereby reducing the density of the feature vectors stored in the feature repository, thereby saving the storage space required to store the feature vectors.

[0075] Based on the above description, the promotion information processing method provided by the embodiment of the present disclosure stores the feature vector obtained after vectorization processing in the feature library. When the similarity between the feature vector to be stored and the historical feature vector in the feature library is high, the feature vector to be stored is no longer stored in the feature library, thereby reducing the storage resources required for storing the feature vector. In this way, when it is necessary to build the model again in the future, it is only necessary to directly obtain the feature vector from the feature library, and there is no need to vectorize the sample customs clearance information again, which not only saves computing resources, but also improves the efficiency of building the model. In addition, after the feature vector obtained by vectorizing the sample promotion information is stored in the feature library, when the feature vector in the feature library needs to be used in other scenarios, the feature vector can be directly obtained from the feature library and used, which expands the application scope of the feature library.

[0076] Figure 4 is a schematic diagram of a promotion information processing device according to an exemplary disclosed embodiment. Figure 4 As shown, the promotion information processing device includes:

[0077] A vector unit 401 is used to obtain a feature library, wherein the feature library includes a feature vector obtained by vectorizing at least one sample promotion information;

[0078] An acquiring unit 402 is configured to acquire a first feature vector combination and a second feature vector combination from the feature library, wherein the first feature vector combination and the second feature vector combination overlap;

[0079] The construction unit 403 is used to construct a first promotion information recognition model by using the first feature vector combination, and to construct a second promotion information recognition model by using the second feature vector combination.

[0080] Optionally, the vector unit 401 is specifically used to vectorize each sample promotion information in the at least one sample promotion information from multiple dimensions to obtain feature vectors corresponding to the multiple dimensions respectively; and construct a feature library based on the feature vectors corresponding to the multiple dimensions respectively.

[0081] Optionally, the multiple dimensions include any combination of audio, video, picture and text.

[0082] Optionally, the feature vector includes at least one of a publisher feature of the sample promotion information, a material content feature of the sample promotion information, a material note feature of the sample promotion information, a modification suggestion feature of the sample promotion information, and a version maintenance feature of the sample promotion information.

[0083] Optionally, the vector unit 401 is specifically configured to store the target feature vector in the feature library when the similarity between a target feature vector in the feature vectors corresponding to the multiple dimensions and a historical feature vector in the feature library is lower than a preset threshold.

[0084] Optionally, the device further includes: an updating unit; the updating unit is used to update the at least one sample promotion information; and store a feature vector obtained by vectorizing each sample promotion information in the at least one updated sample promotion information into the feature library.

[0085] Optionally, the device also includes: an identification unit; the identification unit is used to obtain the promotional information to be identified; input the promotional information to be identified into the first promotional information identification model and the second promotional information identification model respectively to obtain a first identification result and a second identification result, the first identification result and the second identification result include different modification suggestions for the promotional information to be identified.

[0086] The functions of the above modules have been described in detail in the method steps in the previous embodiment and will not be repeated here.

[0087] Reference below Figure 5 , which shows a schematic diagram of the structure of an electronic device 500 suitable for implementing the embodiment of the present disclosure, the electronic device can be used to implement the following Figure 5 The functions corresponding to the promotion information processing device shown. Figure 5 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0088] like Figure 5As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 are also stored. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0089] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 500 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 5 The electronic device 500 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0090] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0091] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0092] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0093] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0094] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains a feature library, the feature library includes feature vectors obtained by vectorizing at least one sample promotion information; obtains a first feature vector combination and a second feature vector combination from the feature library, and the first feature vector combination and the second feature vector combination overlap; uses the first feature vector combination to construct a first promotion information recognition model, and uses the second feature vector combination to construct a second promotion information recognition model.

[0095] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving 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 may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0096] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0097] The modules involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a module does not limit the module itself in some cases. For example, the first acquisition module may also be described as a "module for acquiring at least two Internet Protocol addresses".

[0098] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0099] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0100] According to one or more embodiments of the present disclosure, Example 1 provides a promotion information processing method, including:

[0101] Acquire a feature library, wherein the feature library includes a feature vector obtained by vectorizing at least one sample promotion information;

[0102] Acquire a first feature vector combination and a second feature vector combination from the feature library, where the first feature vector combination and the second feature vector combination overlap;

[0103] A first promotion information recognition model is constructed by using the first feature vector combination, and a second promotion information recognition model is constructed by using the second feature vector combination.

[0104] According to one or more embodiments of the present disclosure, Example 2 provides the method of Example 1, wherein obtaining a feature library includes:

[0105] Performing vectorization processing on each sample promotion information in the at least one sample promotion information from multiple dimensions to obtain feature vectors corresponding to the multiple dimensions respectively;

[0106] A feature library is constructed according to the feature vectors respectively corresponding to the multiple dimensions.

[0107] According to one or more embodiments of the present disclosure, Example 3 provides the method of Example 2, wherein the multiple dimensions include any combination of audio, video, pictures, and text.

[0108] According to one or more embodiments of the present disclosure, Example 4 provides the method of Example 2, wherein the feature vector includes at least one of a publisher feature of the sample promotion information, a material content feature of the sample promotion information, a material note feature of the sample promotion information, a modification suggestion feature of the sample promotion information, and a version maintenance feature of the sample promotion information.

[0109] According to one or more embodiments of the present disclosure, Example 5 provides the method of Example 2, wherein constructing a feature library according to feature vectors corresponding to the multiple dimensions respectively includes:

[0110] When the similarity between a target feature vector in the feature vectors corresponding to the multiple dimensions and a historical feature vector in the feature library is lower than a preset threshold, the target feature vector is stored in the feature library.

[0111] According to one or more embodiments of the present disclosure, Example 6 provides the method of Example 1, wherein the method further includes:

[0112] Updating the at least one sample promotion information;

[0113] A feature vector obtained by vectorizing each sample promotion information in the at least one updated sample promotion information is stored in the feature library.

[0114] According to one or more embodiments of the present disclosure, Example 7 provides the method of Examples 1-6, wherein the method further includes:

[0115] Obtain promotional information to be identified;

[0116] The promotion information to be identified is input into the first promotion information identification model and the second promotion information identification model respectively to obtain a first identification result and a second identification result, wherein the first identification result and the second identification result include different modification suggestions for the promotion information to be identified.

[0117] According to one or more embodiments of the present disclosure, Example 8 provides a promotion information processing device, including:

[0118] A vector unit, used to obtain a feature library, wherein the feature library includes a feature vector obtained by vectorizing at least one sample promotion information;

[0119] An acquiring unit, configured to acquire a first feature vector combination and a second feature vector combination from the feature library, wherein the first feature vector combination and the second feature vector combination overlap;

[0120] A construction unit is used to construct a first promotion information recognition model by using the first feature vector combination, and to construct a second promotion information recognition model by using the second feature vector combination.

[0121] According to one or more embodiments of the present disclosure, Example 9 provides the device of Example 8, wherein the vector unit is specifically used to vectorize each sample promotion information in the at least one sample promotion information from multiple dimensions to obtain feature vectors corresponding to the multiple dimensions respectively; and construct a feature library based on the feature vectors corresponding to the multiple dimensions respectively.

[0122] According to one or more embodiments of the present disclosure, Example 10 provides the apparatus of Example 9, wherein the multiple dimensions include any combination of audio, video, pictures, and text.

[0123] According to one or more embodiments of the present disclosure, Example 11 provides the device of Example 9, wherein the feature vector includes at least one of a publisher feature of the sample promotion information, a material content feature of the sample promotion information, a material note feature of the sample promotion information, a modification suggestion feature of the sample promotion information, and a version maintenance feature of the sample promotion information.

[0124] According to one or more embodiments of the present disclosure, Example 12 provides the device of Example 9, wherein the vector unit is specifically used to store the target feature vector in the feature library when the similarity between the target feature vector in the feature vectors corresponding to the multiple dimensions and the historical feature vector in the feature library is lower than a preset threshold.

[0125] According to one or more embodiments of the present disclosure, Example 13 provides the device of Example 8, which further includes: an updating unit; the updating unit is used to update the at least one sample promotion information; and a feature vector obtained by vectorizing each sample promotion information in the at least one updated sample promotion information is stored in the feature library.

[0126] According to one or more embodiments of the present disclosure, Example 14 provides the device of Examples 8-13, which further includes: an identification unit; the identification unit is used to obtain the promotional information to be identified; the promotional information to be identified is input into the first promotional information identification model and the second promotional information identification model respectively to obtain a first identification result and a second identification result, wherein the first identification result and the second identification result include different modification suggestions for the promotional information to be identified.

[0127] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0128] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0129] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Claims

1. A promotion information processing method, characterized in that: include: Acquire a feature library, wherein the feature library includes a feature vector obtained by vectorizing at least one sample promotion information, wherein the feature vector includes at least one of a publisher feature of the sample promotion information, a material content feature of the sample promotion information, a material remark feature of the sample promotion information, a modification suggestion feature of the sample promotion information, and a version maintenance feature of the sample promotion information; Acquire a first feature vector combination and a second feature vector combination from the feature library, wherein the first feature vector combination and the second feature vector combination partially overlap; The first feature vector combination is used to construct a first promotion information recognition model, and the second feature vector combination is used to construct a second promotion information recognition model. The first promotion information recognition model and the second promotion information recognition model are used to recognize the promotion information to be recognized from different angles.

2. The method according to claim 1, characterized in that The acquiring of the feature library comprises: Performing vectorization processing on each sample promotion information in the at least one sample promotion information from multiple dimensions to obtain feature vectors corresponding to the multiple dimensions respectively; A feature library is constructed according to the feature vectors respectively corresponding to the multiple dimensions.

3. The method according to claim 2, characterized in that The multiple dimensions include any combination of audio, video, picture and text.

4. The method according to claim 2, characterized in that: The constructing a feature library according to the feature vectors corresponding to the multiple dimensions respectively includes: When the similarity between a target feature vector in the feature vectors corresponding to the multiple dimensions and a historical feature vector in the feature library is lower than a preset threshold, the target feature vector is stored in the feature library.

5. The method according to claim 1, characterized in that The method further comprises: Updating the at least one sample promotion information; A feature vector obtained by vectorizing each sample promotion information in the at least one updated sample promotion information is stored in the feature library.

6. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: Obtain promotional information to be identified; The promotion information to be identified is input into the first promotion information identification model and the second promotion information identification model respectively to obtain a first identification result and a second identification result, wherein the first identification result and the second identification result include different modification suggestions for the promotion information to be identified.

7. A promotion information processing device, characterized in that: include: a vector unit, used to obtain a feature library, wherein the feature library includes a feature vector obtained by vectorizing at least one sample promotion information, wherein the feature vector includes at least one of a publisher feature of the sample promotion information, a material content feature of the sample promotion information, a material note feature of the sample promotion information, a modification suggestion feature of the sample promotion information, and a version maintenance feature of the sample promotion information; An acquiring unit, configured to acquire a first feature vector combination and a second feature vector combination from the feature library, wherein the first feature vector combination and the second feature vector combination partially overlap; A construction unit is used to construct a first promotion information recognition model using the first feature vector combination, and to construct a second promotion information recognition model using the second feature vector combination, wherein the first promotion information recognition model and the second promotion information recognition model are used to recognize the promotion information to be recognized from different angles.

8. An electronic device, characterized in that: include: a storage device having a computer program stored thereon; A processing device, configured to execute the computer program in the storage device to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processing device, the steps of the method described in any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that When the computer program product is run on a computer, the computer is caused to execute the method according to any one of claims 1 to 6.

Citation Information

Patent Citations

  • Drug recommendation method based on artificial intelligence, and related device

    WO2021159640A1