CN112445992B - Information processing method and device - Google Patents

Information processing method and device Download PDF

Info

Publication number
CN112445992B
CN112445992B CN201910828416.5A CN201910828416A CN112445992B CN 112445992 B CN112445992 B CN 112445992B CN 201910828416 A CN201910828416 A CN 201910828416A CN 112445992 B CN112445992 B CN 112445992B
Authority
CN
China
Prior art keywords
model
information
module
processed
information carriers
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
Application number
CN201910828416.5A
Other languages
Chinese (zh)
Other versions
CN112445992A (en
Inventor
孙佳佳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Alibaba Group Holding Ltd
Original Assignee
Alibaba Group Holding Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Alibaba Group Holding Ltd filed Critical Alibaba Group Holding Ltd
Priority to CN201910828416.5A priority Critical patent/CN112445992B/en
Publication of CN112445992A publication Critical patent/CN112445992A/en
Application granted granted Critical
Publication of CN112445992B publication Critical patent/CN112445992B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/958Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION 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
    • G06Q30/00Commerce
    • G06Q30/02Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241Advertisements
    • G06Q30/0242Determining effectiveness of advertisements
    • G06Q30/0243Comparative campaigns
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION 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
    • G06Q30/00Commerce
    • G06Q30/02Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241Advertisements
    • G06Q30/0277Online advertisement
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Theoretical Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Databases & Information Systems (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

The embodiment of the application provides an information processing method and device. Wherein the method comprises the following steps: obtaining an information carrier from the history; wherein the information carrier carries object recommendation information; determining quality comparison results of any two information carriers according to effect evaluation indexes of the two information carriers; setting model labels of the two information carriers based on the quality comparison result; training an evaluation model by using the two information carriers and the model tag; wherein the evaluation model is used for calculating the quality comparison result of two information carriers to be processed. The technical scheme provided by the embodiment of the application improves the evaluation accuracy of the information carrier.

Description

Information processing method and device
Technical Field
The embodiment of the application relates to the technical field of computer application, in particular to an information processing method and device
Background
With the development of network technology, object promotion through network media has become a popular trend, wherein objects may refer to tangible products or intangible services, etc.
One common popularization way is to display information carriers such as popularization pictures carrying recommendation information of objects in popularization media such as web pages, wherein the popularization pictures can be linked to description pages of the objects, and the like, and users are interested in the popularization pictures when browsing the web pages and can enter the description pages of the objects by clicking the popularization pictures.
At present, the information carrier is usually edited and generated by a user based on object recommendation information, and the quality of the information carrier directly influences the popularization effect, so that how to measure the quality of the information carrier, help the user to screen, and become a technical problem to be solved.
Disclosure of Invention
The embodiment of the application provides an information processing method and device, which are used for solving the technical problem of low evaluation accuracy of an information carrier.
In a first aspect, an embodiment of the present application provides an information processing method, including:
obtaining an information carrier from the history; wherein the information carrier carries object recommendation information;
determining quality comparison results of any two information carriers according to effect evaluation indexes of the two information carriers;
setting model labels of the two information carriers based on the quality comparison result;
training an evaluation model by using the two information carriers and the model tag; wherein the evaluation model is used for calculating the quality comparison result of two information carriers to be processed.
In a second aspect, an embodiment of the present application provides an information processing method, including:
Acquiring two information carriers to be processed;
calculating quality comparison results of the two information carriers to be processed by using an evaluation model;
wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model tag is determined based on the effect evaluation index of the two information carriers.
In a third aspect, an embodiment of the present application provides an information processing method, including:
acquiring popularization pictures from the history record; wherein, the promotion picture bears commodity recommendation information;
determining quality comparison results of any two popularization pictures according to effect evaluation indexes of the two popularization pictures;
setting model labels of the two popularization pictures based on the quality comparison result;
and training an evaluation model by using the two popularization pictures and the model label.
In a fourth aspect, an embodiment of the present application provides an information processing method, including:
acquiring two popularization pictures to be processed;
calculating quality comparison results of the two popularization pictures to be processed by using an evaluation model;
the evaluation model is obtained in advance based on any two popularization pictures in the history record and model labels of the two popularization pictures; and determining the model label based on the effect evaluation indexes of the two popularization pictures.
In a fifth aspect, an embodiment of the present application provides an information processing method, including:
acquiring a target popularization picture uploaded by a user;
acquiring a plurality of candidate popularization pictures similar to the target popularization picture from a picture database;
calculating a quality comparison result of the target popularization picture and any candidate popularization picture by using an evaluation model;
determining the quality grade of the target popularization picture based on the quality comparison results of the target popularization picture and the candidate popularization pictures respectively;
and outputting the quality grade prompt information of the target popularization picture.
In a sixth aspect, an embodiment of the present application provides an information processing apparatus, including:
a first acquisition unit for acquiring the information carrier from the history; wherein the information carrier carries object recommendation information;
a determining unit for determining a quality comparison result of any two information carriers according to the effect evaluation index of the two information carriers;
a label setting unit for setting model labels of the two information carriers based on the quality comparison result;
the model training unit is used for training an evaluation model by utilizing the two information carriers and the model label; wherein the evaluation model is used for calculating the quality comparison result of two information carriers to be processed.
In a seventh aspect, an embodiment of the present application provides an information processing apparatus, including:
the second acquisition unit is used for acquiring two information carriers to be processed;
the quality comparison unit is used for calculating quality comparison results of the two information carriers to be processed by using the evaluation model;
wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model tag is determined based on the effect evaluation index of the two information carriers.
In an eighth aspect, in an embodiment of the present application, there is provided a computing device including a processing component and a storage component;
the storage component stores one or more computer instructions; the one or more computer instructions are to be invoked for execution by the processing component;
the processing assembly is configured to:
obtaining an information carrier from the history; wherein the information carrier carries object recommendation information;
determining quality comparison results of any two information carriers according to effect evaluation indexes of the two information carriers;
setting model labels of the two information carriers based on the quality comparison result;
training an evaluation model by using the two information carriers and the model tag; wherein the evaluation model is used for calculating the quality comparison result of two information carriers to be processed.
In a ninth aspect, embodiments of the present application provide a computing device, including a processing component and a storage component;
the storage component stores one or more computer instructions; the one or more computer instructions are to be invoked for execution by the processing component;
the processing assembly is configured to:
acquiring two information carriers to be processed;
calculating quality comparison results of the two information carriers to be processed by using an evaluation model;
wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model tag is determined based on the effect evaluation index of the two information carriers.
In the embodiment of the application, an information carrier is obtained from a history record; determining quality comparison results of any two information carriers according to effect evaluation indexes of the two information carriers; determining model labels of the two information carriers based on the quality comparison result; training an evaluation model based on the two information carriers and the model tag; wherein the evaluation model can be used for calculating the quality comparison result of two information carriers to be processed. The effect evaluation index of the information carrier in the history record is utilized to realize the training of the evaluation model, the training data is more objective, the evaluation model can objectively realize the evaluation of the information carrier, and the evaluation accuracy of the information carrier is improved.
These and other aspects of the present application will be more readily apparent from the following description of the embodiments.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief description will be given below of the drawings that are needed in the embodiments or the prior art descriptions, and it is obvious that the drawings in the following description are some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart illustrating one embodiment of an information processing method provided herein;
FIG. 2 is a schematic diagram of a network structure of an evaluation model in a practical application according to an embodiment of the present application;
FIG. 3 is a flow chart illustrating yet another embodiment of an information processing method provided herein;
FIG. 4 is a flow chart illustrating yet another embodiment of an information processing method provided herein;
FIG. 5 is a flow chart illustrating yet another embodiment of an information processing method provided herein;
FIG. 6 is a flow chart illustrating yet another embodiment of an information processing method provided herein;
FIG. 7 is a schematic view showing the structure of an embodiment of an information processing apparatus provided in the present application;
FIG. 8 illustrates a schematic diagram of one embodiment of a computing device provided herein;
fig. 9 is a schematic diagram showing the structure of a further embodiment of an information processing apparatus provided in the present application;
fig. 10 illustrates a schematic diagram of a configuration of yet another embodiment of a computing device provided herein.
Detailed Description
In order to enable those skilled in the art to better understand the present application, the following description will make clear and complete descriptions of the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application.
In some of the flows described in the specification and claims of this application and in the foregoing figures, a number of operations are included that occur in a particular order, but it should be understood that the operations may be performed in other than the order in which they occur or in parallel, that the order of operations such as 101, 102, etc. is merely for distinguishing between the various operations, and that the order of execution is not by itself represented by any order of execution. In addition, the flows may include more or fewer operations, and the operations may be performed sequentially or in parallel. It should be noted that, the descriptions of "first" and "second" herein are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, and are not limited to the "first" and the "second" being different types.
The technical solution of the embodiment of the application can be used in an application scene for evaluating any form of information carrier, for example, the information carrier can be a popularization picture for displaying in popularization media, such as a banner (banner advertisement) picture, and of course, the information carrier can also be other media such as texts and videos. The information carrier carries object recommendation information, such as picture content in a promotion picture, so that the object recommendation information can be transmitted, and the object promotion purpose is achieved. The information carrier can also index to description pages of the objects, etc., so that the object conversion rate can also be improved by the information carrier. In an actual application, the object may specifically refer to a commodity for online selling, so that the information carrier is used for commodity propaganda and linked to a description page of the commodity, and therefore the information carrier with good quality can effectively improve the purchasing rate of the commodity.
Taking an information carrier as an example of a promotion picture, at present, the promotion picture is edited and generated by a user, and the promotion effect of the promotion picture can be directly influenced by the display form or style of object recommendation information in the promotion picture, for example, the promotion picture is used for promoting and promoting goods on a certain line so as to improve the commodity purchasing rate, and clicking the promotion picture can enter a commodity description page to purchase the commodity, if the promotion picture cannot attract the user to click, the promotion effect can be greatly discounted. However, at present, after the information carrier is generated, the quality of the information carrier is still manually evaluated to decide whether to use. Human assessment is highly subjective, thus resulting in lower assessment accuracy.
In order to improve the evaluation accuracy of the information carrier, the inventor provides a technical scheme of the application through a series of researches, and in the embodiment of the application, the information carrier is obtained from a history record; wherein the information carrier carries object recommendation information; determining quality comparison results of any two information carriers according to effect evaluation indexes of the two information carriers; determining model labels of the two information carriers based on the quality comparison result; training an evaluation model based on the two information carriers and the model tag; wherein the evaluation model can be used for calculating the quality comparison result of two information carriers to be processed. The effect evaluation index of the information carrier in the history record is utilized to realize the training of the evaluation model, the training data is objective, the evaluation model can objectively realize the evaluation of the information carrier, and the evaluation accuracy is improved.
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
Fig. 1 is a flowchart of an embodiment of an information processing method provided in an embodiment of the present application, where the method may include the following steps:
101: obtaining an information carrier from the history; wherein the information carrier carries object recommendation information.
Since information carriers carrying object recommendation information are released for use on-line, one or more information carriers are also typically released for the same object. Wherein the history, i.e. the use of the information carrier distributed on-line of the record, is recorded. The embodiment of the application uses online data for model training, and the online data is generated by a large number of user behaviors, so that the online data is more objective.
The information carrier may have an indexing property that may be linked to a description page of the object.
102: and determining the quality comparison result of any two information carriers according to the effect evaluation index of the two information carriers.
Wherein the effect evaluation index may be obtained from a history.
For an information carrier that can be linked to an object description page, the effect evaluation index may refer to, for example, the number of displays of the information carrier, or the number of clicks of the information carrier, and of course, in order to further improve accuracy, the effect evaluation index may be a Click-Through-Rate (CTR). The click through rate may refer to the ratio of the number of clicks of the information carrier to the number of displays of the information carrier.
Thus, optionally, step 102 may include:
and determining the comparison result of any two information carriers according to the click through rate of the two information carriers.
Based on the effect evaluation index of any two information carriers, the quality comparison result of the two information carriers can be determined, and the quality comparison result comprises the following three general information carriers:
1) The quality of the first information carrier is better than that of said second information carrier;
2) The first information carrier and the second information carrier have the same mass;
3) The quality of the second information carrier is better than that of said first information carrier.
Specifically, based on the effect evaluation index of the first information carrier and the second information carrier, a difference value of the effect evaluation index of the first information carrier compared with the second information carrier may be calculated, for example, when the effect evaluation index is the click through rate, the difference value of the click through rates of the first information carrier and the second information carrier may be calculated, if the difference value is smaller than a first threshold value, the quality of the first information carrier and the quality of the second information carrier may be considered to be the same, and if the difference value is larger than the first threshold value, the larger the click through rate may be considered to be the better, so if the click through rate of the first information carrier is larger than the click through rate of the second information carrier, the quality of the first information carrier may be considered to be better than the second information carrier, otherwise, the quality of the second information carrier is better than the first information carrier.
Wherein the first threshold may be set in connection with a service requirement. Wherein, the smaller the first threshold value is, the more accurate the recognition result will be. In some practical applications, when screening is required based on the quality comparison result of two information carriers to be processed, for example, when one is used as a screening reference and the other is screened, in order to ensure the screening result and to avoid that the two information carriers to be processed are very similar or even completely consistent, a larger value can be selected as the first threshold in a certain range, for example, when the information carriers are advertisement pictures, the first threshold can be improved in order to avoid that the content of the two advertisement pictures are very similar.
It should be noted that "first" and "second" of the first information carrier and the second information carrier are merely for distinguishing between two different information carriers from the description, and do not represent that the two information carriers have a relationship such as progressive or sequential.
Optionally, to further improve the evaluation accuracy, this step 102 may include:
and determining quality comparison results of the two information carriers according to effect evaluation indexes of the two different information carriers corresponding to the same object.
The two information carriers being different may refer to differences in layout, content or format of the object recommendation information.
Two different information carriers of the same object are used as a training sample, so that the model training accuracy can be further improved, and the evaluation accuracy can be further improved.
103: and setting model labels of the two information carriers based on the quality comparison result.
104: and training an evaluation model by using the two information carriers and the model label.
Wherein the evaluation model is used for calculating the quality comparison result of two information carriers to be processed.
The two information carriers are used as training samples, training of the evaluation model can be achieved based on the model labels, specifically, the two information carriers are used as model input data, the model labels are used as model output results, and training of the evaluation model is achieved.
In this embodiment, training of the evaluation model is achieved by using any two information carriers and effect evaluation indexes of any two information carriers obtained from the history record, so that the evaluation model can be used for evaluating the quality of the information carrier to be processed.
As is apparent from the above description, the quality comparison results of any two information carriers may include three types, depending on the effect evaluation index of the two information carriers.
In certain embodiments, said setting model labels of said two information carriers based on said quality comparison result comprises:
and taking the identifier corresponding to the quality comparison result as a model label based on the identifier corresponding to the quality comparison result.
The identification symbols corresponding to the different quality comparison results can be preset, so that the corresponding identification symbols can be obtained by searching the identification symbols based on the quality comparison results of any two information carriers, and the identification symbols can be used as model labels.
For example, for convenience of description, the arbitrary two information carriers may include a first information carrier and a second information carrier, where the first information carrier and the second information carrier may be either one information carrier or two different information carriers corresponding to the same object, respectively. The identification symbol may be represented by an arabic numeral. As a result of the three quality comparisons described above, the identification symbol may be set to "0" if the quality of the first information carrier is better than that of the second information carrier; if the first information carrier and the second information carrier are of the same quality, the identification symbol may be set to "1"; the identification symbol may be set to "2" if the quality of the second information carrier is better than said first information carrier.
The evaluation model can be implemented by a neural network model, and therefore, the evaluation model can be composed of an input layer, an output layer and at least one intermediate layer between the input layer and the output layer.
In some embodiments, the training the assessment model based on the two information carriers and the model tag may comprise:
taking the two information carriers as model input data and taking the model labels as model output results;
and respectively extracting model features of the two information carriers through the evaluation model, calculating difference features of the two model features, and mapping the difference features into the model output result to realize training of the evaluation model.
When the evaluation model is a neural network model, that is, two information carriers are used as input of the input layer, and the model output result is used as output of the output layer. At least one intermediate layer is responsible for extracting model features of the two information carriers, calculating difference features of the two model features, and mapping the difference features to the model output result, so as to realize training of an evaluation model and obtain model parameters of the evaluation model. The difference feature is mapped to a model output result, that is, the difference feature is taken as an input of the output layer and corresponds to the output of the output layer.
Wherein the evaluation model may be a VGG (Visual Geometry Group Network, visual geometry network, a neural network) model, at least one intermediate layer in the VGG model typically consists of multiple convolution layers, multiple fully connected layers, and the layers are separated by pooling layers. VGG uses small convolution kernels, small pooling kernels, and gradually increases in depth. The accuracy of the evaluation can be improved by VGG, in particular in the case of a picture of the information carrier.
In some embodiments, the evaluation model may include a plurality of convolution modules connected in sequence, a fusion module connected to the plurality of convolution modules, a plurality of full connection modules, and a difference processing module.
That is, the evaluation model may include, in addition to the input layer and the output layer, at least one middle layer including a plurality of convolution modules connected in sequence, a fusion module connected to the plurality of convolution modules, a plurality of full connection modules, and a difference processing module, respectively. Each module may be made up of one or more neural network layers.
The evaluation model can be implemented by adopting a VGG architecture, wherein, as two information carriers are input, the convolution module, the fusion module and the full connection module can be all sharing modules and are used for respectively processing the two information carriers.
Thus, said extracting model features of said two information carriers via said evaluation model, calculating difference features of the two model features comprises:
extracting convolution features of the two information carriers via the plurality of convolution modules, respectively;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each information carrier through the fusion module to obtain the fusion characteristics of each information carrier;
and processing the fusion characteristics of the two information carriers through the plurality of fully-connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module.
In practical application, the information carrier can be a picture, the convolution modules are sequentially connected to extract the convolution characteristics of the picture, the convolution neural network is formed by the convolution modules, and the picture characteristics of different layers of the picture can be obtained through convolution processing one by one, so that the convolution characteristics of the convolution modules are combined together through the fusion module, the formed fusion characteristics can comprise the visual characteristics of the picture at a low layer and the visual characteristics of the picture at a high layer, and the model accuracy can be further improved.
In addition, to further improve model accuracy, in some embodiments, the assessment model may further include a first pooling module and a second pooling module disposed between the fusion module and a first fully-connected module, and between two adjacent fully-connected modules, respectively;
the arbitrary two information carriers comprise a first information carrier and a second information carrier;
the first pooling module is specifically configured to normalize, randomly inactivate and/or excite the output result of the first information carrier by the previous module;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on the output module of the second information carrier by using the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
Wherein the first pooling module and the second pooling module may be composed of one or more pooling layers.
The two information carriers are independently processed by different pooling modules respectively, so that the difference of the two information carriers can be reserved, and the accuracy of a model is improved.
In some embodiments, the difference processing module may be specifically disposed between the last two fully connected modules, and configured to subtract the output result of the first pooling module from the output result of the second pooling module to obtain a difference feature, and input the difference feature to the last fully connected module. And the last full-connection module is used for carrying out full-connection processing and then inputting and outputting the layer.
In practical application, the information carrier may be specifically a picture, the evaluation model may be implemented based on VGG architecture, for convenience of understanding, as shown in fig. 2, a network structure schematic diagram of an evaluation model is shown, and it is assumed that two information carriers are pictures, in the evaluation model shown in fig. 2, the evaluation model may include an input layer 201, 5 convolution modules 202 sequentially connected, a fusion module 203, 4 full connection modules 204, a difference processing module 205, a first pooling module 206 and a second pooling module 207 between the fusion module 203 and the first full connection module 204, and disposed between two adjacent full connection modules 204, and an output layer 208; the difference processing module 205 is disposed between the last two fully connected modules 204, and is configured to subtract the output result of the first pooling module 206 from the output result of the second pooling module 207 to obtain a difference feature, and input the difference feature to the last fully connected module 204.
The first pooling module 206 is configured to process a corresponding result corresponding to the first picture, and the second pooling module 207 is configured to process a corresponding result corresponding to the second picture.
The first picture and the second picture are input through the input layer 201, then the first picture and the second picture are respectively subjected to feature extraction one by one through 5 convolution modules, and the latter convolution module continues to perform feature extraction after the feature basis extracted by the former convolution module, so that visual features of different layers of the pictures can be mined. The convolution features respectively extracted by the 5 convolution modules 202 and the fusion module 203,5 are combined by the fusion module 203, alternatively, the convolution features are usually vector representations, and a plurality of feature vectors can be spliced together by combining a plurality of convolution features. The fusion module 203 fuses the 5 convolution features to obtain fusion features, so that the fusion features can comprise different layers of visual features, and feature accuracy is ensured.
Wherein, the first pooling module 206 and the second pooling module 207 between the fusion module 203 and the first fully-connected module 204 may perform random inactivation (dropout) processing on the two fusion features respectively; the first and second pooling modules 206 and 207 between the first and second full-connection modules 204 and between the second and third full-connection modules 204 and 204, respectively, perform normalization, random inactivation, and excitation function processing; obtaining model features of the first picture and model features of the second picture via the third fully connected module 204; the model features of the first picture are subjected to activation function processing through a first pooling module 206, the model features of the second picture are subjected to activation function processing through a second pooling module 207, and the difference processing module 205 subtracts the output result of the first pooling module from the output result of the second pooling module to obtain a difference feature; after the difference features are processed by the fourth fully-connected module 204, that is, the input-output layer 208, model training is implemented in a training stage by mapping the model output results of the output layer 208, that is, model labels.
Based on the evaluation model obtained by training in the technical solution shown in fig. 1, the quality comparison can be performed on two information carriers to be processed, as shown in fig. 3, which is a flowchart of another embodiment of an information processing method provided in an embodiment of the present application, where the method may include the following steps:
301: two information carriers to be processed are acquired.
302: and calculating the quality comparison result of the two information carriers to be processed by using an evaluation model.
Wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model tag is determined based on the effect evaluation index of the two information carriers.
The specific training manner of the evaluation model may be shown in the embodiment of fig. 1, and will not be described herein.
In practical application, for the same object, multiple information carriers are usually involved for screening, in the prior art, the quality comparison results of two information carriers to be processed can be calculated by using an evaluation model through the technical scheme of the application, and the target information carrier with better quality can be obtained through the quality comparison results of the two information carriers to be processed.
Wherein the method may further comprise:
And outputting comparison result prompt information which can comprise the quality comparison result so as to be convenient for a user to check and the like.
In addition, in an actual application, the technical scheme of the embodiment of the application can be applied to the information carrier evaluation platform, the information carrier manufactured by the user can be uploaded to the server through the client of the information carrier evaluation platform, and the server of the information carrier evaluation platform is combined with the existing information carrier to evaluate.
Thus, in certain embodiments, said determining two information carriers to be processed may comprise:
acquiring a target information carrier uploaded by a user;
obtaining a plurality of candidate information carriers similar to the target information carrier from a target database;
respectively taking the target information carrier and any one candidate information carrier as two information carriers to be processed;
the calculating the comparison result of the two information carriers to be processed by using the evaluation model comprises:
calculating a quality comparison result of the target information carrier and any one of the candidate information carriers by using an evaluation model;
the method further comprises the steps of:
and determining a quality grade of the target information carrier based on quality comparison results of the target information carrier and the plurality of candidate information carriers.
The target database stores a large number of information carriers, and in one practical application, the target database can be the information carrier in the history record, namely the information carrier used for evaluation model training.
The method may acquire a plurality of candidate information carriers similar to the target information carrier through an image recognition technology, for example, by extracting image features and calculating similarity, which is the same as the prior art, and will not be described herein.
The information carriers in the target database may have quality levels, which may be determined based on an effect evaluation index, such as click through rate, for each information carrier.
Optionally, based on the quality comparison of the target information carrier with the plurality of candidate information carriers, determining the quality level of the target information carrier may be:
determining a quality level of the target information carrier based on a quality comparison of the target information carrier with the plurality of candidate information carriers and the quality levels of the plurality of candidate information carriers.
Optionally, the target information carrier and the plurality of candidate information carriers may be ordered based on a quality comparison of the target information carrier and the plurality of candidate information carriers; based on the ranking result, a quality level of the target information carrier is determined.
Wherein the quality level of the target information carrier may be determined in combination with the quality levels of the plurality of candidate information carriers based on the ranking result.
If the quality of the target information carrier is the same as that of a candidate information carrier, the quality grade of the candidate information carrier can be directly used as the quality grade of the target information carrier.
Wherein the quality level may be represented by a quality score.
For example, the target information carrier X corresponds to 3 candidate information carriers, which are respectively denoted as A, B and C, the quality comparison result of X and a is that the quality of X is better than that of a, the quality comparison result of X and B is that the quality of X is better than that of B, the quality comparison result of X and C is that the quality of C is better than that of X, if the quality grade of a is higher than that of B, the ranking result may be C, X, A, B, and the quality grade of X may be determined based on the quality grades of C and a, that is, the average value of the quality grades of C and a may be the like.
In some embodiments, after determining the quality level of the target information carrier, the method may further comprise:
and outputting the quality grade prompt information of the target information carrier.
The quality level cues are used to assist the user in deciding whether to use the target information carrier, etc.
Alternatively, the quality level prompt information of the target information carrier may be output at the client.
Furthermore, in certain embodiments, the method may further comprise:
generating content rating information for the target information carrier;
the outputting the quality grade prompt information of the target information carrier comprises the following steps:
and outputting quality grade prompt information comprising the content evaluation information.
The content evaluation information may include one or more of prompt information of content meeting quality requirements, prompt information of content not meeting quality requirements, and modification advice in the target information carrier.
In a further practical application, the quality evaluation of the other information carrier can also be performed on the basis of a standard information carrier, so that the acquisition of two information carriers to be processed comprises:
acquiring a target information carrier uploaded by a user;
determining a standard information carrier; the standard information carrier may be a higher quality level information carrier;
and taking the target information carrier and the standard information carrier as two information carriers to be processed.
The method may further comprise:
and outputting a comparison result prompt message of the target information carrier based on the quality comparison result of the target information carrier and the standard information carrier.
The quality comparison result can be included in the comparison result prompt message.
Of course, based on the standard information carrier, content rating information of the target information carrier may also be generated.
Thus, outputting the comparison result prompt information of the target information carrier may be outputting quality prompt information including the content evaluation information, so that the user knows how to improve the target information carrier based on the comparison result prompt information.
In one practical application, the information carrier may refer to a popularization picture, such as a canner picture, an advertisement picture displayed in a popularization medium, etc., and the technical scheme of the embodiment of the application may be applied to an online transaction scene, the popularization picture may bear commodity recommendation information, the purpose of commodity propaganda may be achieved through the popularization picture, and the popularization picture may be linked to a description page of a commodity, and commodity purchase may be achieved in the description page of the commodity, so that quality of the popularization picture may affect commodity purchase rate. Taking a generalized picture as an example, as shown in fig. 4, a flowchart of still another embodiment of an information processing method provided in the present application may include the following steps:
401: and obtaining popularization pictures from the history record.
The promotion picture carries commodity recommendation information and can be linked to a description page of the commodity.
402: and determining quality comparison results of any two popularization pictures according to effect evaluation indexes of the two popularization pictures.
403: setting model labels of the two popularization pictures based on the quality comparison result;
404: and training an evaluation model by using the two popularization pictures and the model label.
The embodiment shown in fig. 4 differs from the embodiment shown in fig. 1 in that the information carrier is a promotional picture, and other corresponding or identical steps may be referred to in the above embodiments, and a detailed description thereof will not be repeated.
Fig. 5 is a flowchart of another embodiment of an information processing method provided in the present application, where the method may include the following steps:
501: and obtaining two popularization pictures to be processed.
502: calculating quality comparison results of the two popularization pictures to be processed by using an evaluation model;
the evaluation model is obtained in advance based on any two popularization pictures in the history record and model labels of the two popularization pictures; and determining the model label based on the effect evaluation indexes of the two popularization pictures.
The embodiment shown in fig. 5 is different from the embodiment shown in fig. 3 in that the information carrier is a promotional picture, and other corresponding or identical steps can be referred to in the above embodiments, and a detailed description thereof will not be repeated.
In addition, in an actual application, the technical solution of the present application may be applied to a picture evaluation platform, where a user uploads a target popularization picture to the picture evaluation platform, and the picture evaluation platform performs quality evaluation on the target popularization picture, as in the information processing method shown in fig. 6, the method may be applied to the server 60, and the method may include the following steps:
601: and obtaining the target popularization picture uploaded by the user.
Alternatively, the target promotion picture 62 uploaded by the user through the client 61 may be acquired.
The target promotional picture is generated by user editing.
602: and acquiring a plurality of candidate popularization pictures similar to the target popularization picture from a picture database.
603: and calculating the quality comparison result of the target popularization picture and any candidate popularization picture by using an evaluation model.
The evaluation model is obtained in advance based on training of model labels of any two popularization pictures in the history record; the model tag is determined based on the effect evaluation indexes of the two popularization pictures, which can be described in detail in the embodiment shown in fig. 4.
604: and determining the quality grade of the target popularization picture based on the quality comparison results of the target popularization picture and the candidate popularization pictures.
605: and outputting the quality grade prompt information of the target popularization picture.
Alternatively, based on the quality level of the target promotional picture, quality level prompt information may be generated, which may be sent to the client 61 for presentation by the client 61.
The quality level prompt information may include quality levels, such as an evaluation score, a grading identifier, a text description, etc., and may further include a schematic diagram of the target promotion picture, such as a thumbnail of the target promotion picture, and may further include a schematic diagram of a reference promotion picture, so as to facilitate a user to refer to, etc., specifically may refer to the quality level prompt information 63 shown in fig. 6, including the quality level 631, the schematic diagram 632 of the target promotion picture, and may further include the schematic diagram 633 of the reference promotion picture, etc.
In addition, content evaluation information for the target popularization picture can be generated;
the quality class cue information may further include the content rating information.
The content evaluation information can comprise one or more of prompt information of picture content meeting quality requirements, prompt information of picture content not meeting quality requirements, modification suggestions and the like in the target popularization picture, so that a user can conveniently improve the target popularization picture through the quality evaluation information.
In still another practical application, obtaining two popularization pictures to be processed may include:
acquiring a target popularization picture uploaded by a user;
determining a standard popularization picture; the standard evaluation picture can be a popularization picture with higher quality;
and taking the target popularization picture and the standard popularization picture as two popularization pictures to be processed.
The method may further comprise:
and outputting comparison result prompt information based on the quality comparison result of the target popularization picture and the standard evaluation picture.
In addition, based on the standard popularization picture, content evaluation information of the target popularization picture can be generated. The content evaluation information can comprise one or more of prompt information of picture content meeting quality requirements, prompt information of picture content not meeting quality requirements, modification suggestions and the like in the target popularization picture.
And outputting the quality prompt information can be outputting comparison result prompt information comprising the content evaluation information, so that a user can know how to improve the target popularization picture based on the comparison result prompt information.
Fig. 7 is a schematic structural diagram of an embodiment of an information processing apparatus according to an embodiment of the present application, where the apparatus may include:
A first acquisition unit 701 for acquiring an information carrier from a history; wherein the information carrier carries object recommendation information;
a determining unit 702, configured to determine a quality comparison result of any two information carriers according to an effect evaluation index of the two information carriers;
a label setting unit 703 for setting model labels of the two information carriers based on the quality comparison result;
a model training unit 704 for training an evaluation model using the two information carriers and the model tag; wherein the evaluation model is used for calculating the quality comparison result of two information carriers to be processed.
In some embodiments, the determining unit is specifically configured to determine the quality comparison result of any two different information carriers corresponding to the same object according to the effect evaluation index of the two information carriers.
In some embodiments, the information carrier is configured to link to a description page of the object;
the determining unit is specifically configured to determine a quality comparison result of any two information carriers according to click through rates of the two information carriers.
In certain embodiments, the two information carriers comprise a first information carrier and a second information carrier;
The quality comparison result comprises that the quality of the first information carrier is better than the quality of the second information carrier, that the quality of the first information carrier is the same as the quality of the second information carrier, or that the quality of the second information carrier is better than the quality of the first information carrier;
the label setting unit is specifically configured to use the identifier corresponding to the quality comparison result as a model label based on the identifier corresponding to the quality comparison result.
In some embodiments, the model training unit is specifically configured to take the two information carriers as model input data and take the model tag as a model output result;
and respectively extracting model features of the two information carriers through the evaluation model, calculating difference features of the two model features, and mapping the difference features into the model output result to realize training of the evaluation model.
In some embodiments, the evaluation model includes a plurality of convolution modules, a fusion module, a plurality of full connection modules, and a difference processing module that are connected in sequence, respectively, with the plurality of convolution modules;
the model training module extracts model features of the two information carriers via the evaluation model, and calculating difference features of the two model features comprises:
Extracting convolution features of the two information carriers via the plurality of convolution modules, respectively;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each information carrier through the fusion module to obtain the fusion characteristics of each information carrier;
and processing the fusion characteristics of the two information carriers through the plurality of fully-connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module.
In some embodiments, the assessment model further includes a first pooling module and a second pooling module disposed between the fusion module and the first fully-connected module, and between two adjacent fully-connected modules, respectively;
the arbitrary two information carriers comprise a first information carrier and a second information carrier;
the first pooling module is specifically configured to normalize, randomly inactivate and/or excite the output result of the first information carrier by the previous module;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on the output module of the second information carrier by using the previous module;
The processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
In some embodiments, the difference processing module is disposed between the last two fully connected modules, and is configured to subtract the output result of the first pooling module from the output result of the second pooling module to obtain a difference feature, and input the difference feature to the last fully connected module.
In one practical application, the information carrier may be a promotional picture;
the first obtaining unit may be specifically configured to obtain a promotion picture from the history record; wherein, the promotion picture bears commodity recommendation information;
the determining unit is specifically configured to determine a quality comparison result of any two popularization pictures according to an effect evaluation index of the two popularization pictures;
the label setting unit is specifically configured to set model labels of the two popularization pictures based on the quality comparison result;
the model training unit is specifically configured to train an evaluation model by using the two popularization pictures and the model tag.
The information processing apparatus shown in fig. 7 may perform the information processing method described in the embodiment shown in fig. 1, and its implementation principle and technical effects are not repeated. The specific manner in which the respective modules, units, and operations of the information processing apparatus in the above embodiments are performed has been described in detail in the embodiments concerning the method, and will not be described in detail here.
In one possible design, the information processing apparatus shown in fig. 7 may be implemented as a computing device, which may include a storage component 801 and a processing component 802, as shown in fig. 8;
the storage component 801 stores one or more computer instructions for execution by the processing component 802.
The processing component 802 is configured to:
obtaining an information carrier from the history; wherein the information carrier carries object recommendation information;
determining quality comparison results of any two information carriers according to effect evaluation indexes of the two information carriers;
setting model labels of the two information carriers based on the quality comparison result;
training an evaluation model by using the two information carriers and the model tag; wherein the evaluation model is used for calculating the quality comparison result of two information carriers to be processed.
Wherein the processing component 802 may include one or more processors to execute computer instructions to perform all or part of the steps in the methods described above. Of course, the processing component may also be implemented as one or more Application Specific Integrated Circuits (ASICs), digital Signal Processors (DSPs), digital Signal Processing Devices (DSPDs), programmable Logic Devices (PLDs), field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic elements for executing the methods described above.
The storage component 801 is configured to store various types of data to support this operation at the computing device. The memory component may be implemented by any type or combination of volatile or nonvolatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk.
Of course, the computing device may necessarily include other components, such as input/output interfaces, communication components, and the like.
The input/output interface provides an interface between the processing component and a peripheral interface module, which may be an output device, an input device, etc.
The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, and the like.
The computing device may be a physical device or an elastic computing host provided by the cloud computing platform, and at this time, the computing device may be a cloud server, and the processing component, the storage component, and the like may be a base server resource rented or purchased from the cloud computing platform.
In actual practice, the computing device may be a remote web server, a computer networking device, a chipset, a desktop computer, a notebook computer, a workstation, or any other processing device or equipment.
The embodiment of the application further provides a computer readable storage medium storing a computer program, where the computer program can implement the information processing method of the embodiment shown in fig. 1 when executed by a computer.
Fig. 9 is a schematic structural diagram of another embodiment of an information processing apparatus according to an embodiment of the present application, where the apparatus may include:
a second acquisition unit 901 for acquiring two information carriers to be processed;
a quality comparison unit 902 for calculating a quality comparison result of the two information carriers to be processed using an evaluation model;
wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model tag is determined based on the effect evaluation index of the two information carriers.
In some embodiments, the second obtaining unit is specifically configured to obtain a target information carrier uploaded by a user; obtaining a plurality of candidate information carriers similar to the target information carrier from a target database; respectively taking the target information carrier and any one candidate information carrier as two information carriers to be processed;
The quality comparison unit is specifically configured to calculate a quality comparison result between the target information carrier and any one of the candidate information carriers by using an evaluation model;
the apparatus may further include:
and the quality rating unit is used for determining the quality grade of the target information carrier based on the quality comparison result of the target information carrier and the candidate information carriers.
In some embodiments, the apparatus may further comprise:
and the information output unit is used for outputting the quality grade prompt information of the target information carrier.
In a practical application, the information carrier may be a promotional picture. The second obtaining unit is specifically configured to determine two popularization pictures to be processed;
the quality comparison unit is specifically used for calculating quality comparison results of the two popularization pictures to be processed by using an evaluation model.
The information processing apparatus shown in fig. 9 may perform the information processing method described in the embodiment shown in fig. 3, and its implementation principle and technical effects are not described again. The detailed description of the information processing apparatus in the above embodiments, in which each unit, the specific manner in which the unit performs the operation, has been described in detail in the embodiments concerning the method, will not be explained in detail here.
In one possible design, the information processing apparatus of the embodiment shown in fig. 9 may be implemented as a computing device, which may include a storage component 1001 and a processing component 1002 as shown in fig. 10;
the storage component 1001 stores one or more computer instructions for execution by the processing component 1002.
The processing component 1002 is configured to:
acquiring two information carriers to be processed;
calculating quality comparison results of the two information carriers to be processed by using an evaluation model;
wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model tag is determined based on the effect evaluation index of the two information carriers.
Wherein the processing component 1002 can include one or more processors to execute computer instructions to perform all or part of the steps in the methods described above. Of course, the processing component may also be implemented as one or more Application Specific Integrated Circuits (ASICs), digital Signal Processors (DSPs), digital Signal Processing Devices (DSPDs), programmable Logic Devices (PLDs), field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic elements for executing the methods described above.
The storage component 1001 is configured to store various types of data to support operations in a computing device. The memory component may be implemented by any type or combination of volatile or nonvolatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk.
Of course, the computing device may necessarily include other components, such as input/output interfaces, communication components, and the like.
The input/output interface provides an interface between the processing component and a peripheral interface module, which may be an output device, an input device, etc.
The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, and the like.
The computing device may be a physical device or an elastic computing host provided by the cloud computing platform, and at this time, the computing device may be a cloud server, and the processing component, the storage component, and the like may be a base server resource rented or purchased from the cloud computing platform.
In actual practice, the computing device may be a remote web server, a computer networking device, a chipset, a desktop computer, a notebook computer, a workstation, or any other processing device or equipment.
In addition, the embodiment of the present application further provides a computer readable storage medium storing a computer program, where the computer program when executed by a computer may implement the information processing method of the embodiment shown in fig. 3.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting thereof; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.

Claims (14)

1. An information processing method, characterized by comprising:
obtaining an information carrier from the history; wherein the information carrier carries object recommendation information;
determining quality comparison results of any two information carriers according to effect evaluation indexes of the two information carriers;
setting model labels of the two information carriers based on the quality comparison result;
training an evaluation model by using the two information carriers and the model tag; the evaluation model is used for calculating quality comparison results of two information carriers to be processed;
the determining the quality comparison result of the two information carriers according to the effect evaluation index of any two information carriers comprises the following steps:
determining quality comparison results of any two different information carriers corresponding to the same object according to effect evaluation indexes of the two information carriers;
the information carrier is used for linking to a description page of the object;
the determining the quality comparison result of the two information carriers according to the effect evaluation index of any two information carriers comprises the following steps:
determining quality comparison results of any two information carriers according to click through rates of the two information carriers;
Said training an assessment model using said two information carriers and said model tag comprises:
taking the two information carriers as model input data and taking the model labels as model output results;
respectively extracting model features of the two information carriers through the evaluation model, calculating difference features of the two model features, and mapping the difference features into model output results to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
said extracting model features of the two information carriers via the evaluation model, computing difference features of the two model features comprising:
extracting convolution features of the two information carriers via the plurality of convolution modules, respectively;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each information carrier through the fusion module to obtain the fusion characteristics of each information carrier;
processing the fusion characteristics of two information carriers through the plurality of fully connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module;
The evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
the two information carriers comprise a first information carrier and a second information carrier, and the first information carrier and the second information carrier are used for calculating the evaluation effect index difference value of the first information carrier compared with the second information carrier;
the first pooling module is specifically configured to normalize, randomly inactivate and/or excite the output result of the first information carrier by the previous module;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on the output module of the second information carrier by using the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
2. Method according to claim 1, characterized in that the two information carriers comprise a first information carrier and a second information carrier;
the quality comparison result comprises that the quality of the first information carrier is better than the quality of the second information carrier, that the quality of the first information carrier is the same as the quality of the second information carrier, or that the quality of the second information carrier is better than the quality of the first information carrier;
Said setting of model labels of said two information carriers based on said quality comparison result comprises:
and taking the identifier corresponding to the quality comparison result as a model label based on the identifier corresponding to the quality comparison result.
3. The method according to claim 1, wherein the difference processing module is disposed between the last two fully connected modules, and is configured to subtract the output result of the first pooling module from the output result of the second pooling module to obtain a difference feature, and input the difference feature to the last fully connected module.
4. Method according to claim 1, characterized in that the information carrier is a picture.
5. An information processing method, characterized by comprising:
acquiring two information carriers to be processed;
calculating quality comparison results of the two information carriers to be processed by using an evaluation model;
wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model labels are determined based on effect evaluation indexes of the two information carriers;
the model tag is specifically determined by:
Determining quality comparison results of any two different information carriers to be processed according to effect evaluation indexes of the two information carriers to be processed corresponding to the same object;
the information carrier to be processed is used for being linked to a description page of an object;
according to the effect evaluation index of any two information carriers to be processed, determining the quality comparison result of the two information carriers to be processed comprises the following steps:
determining quality comparison results of any two information carriers to be processed according to click passing rates of the two information carriers to be processed;
the evaluation model is trained by:
taking the two information carriers to be processed as model input data, and taking the model labels as model output results;
respectively extracting model features of the two information carriers to be processed through the evaluation model, calculating difference features of the two model features, and mapping the difference features into model output results to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
Said extracting model features of the two information carriers via the evaluation model, computing difference features of the two model features comprising:
extracting convolution characteristics of the two information carriers to be processed respectively through the convolution modules;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each information carrier to be processed through the fusion module to obtain the fusion characteristics of each information carrier to be processed;
processing the fusion characteristics of two information carriers to be processed through the plurality of fully connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module;
the evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
the two information carriers to be processed comprise a first information carrier to be processed and a second information carrier to be processed, and the first information carrier to be processed and the second information carrier to be processed are used for calculating the evaluation effect index difference value of the first information carrier to be processed compared with the second information carrier to be processed;
The first pooling module is specifically configured to normalize, randomly inactivate and/or excite a function of an output result of the previous module for the first information carrier to be processed;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on an output module of the second information carrier to be processed by the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
6. The method according to claim 5, wherein said acquiring two information carriers to be processed comprises:
acquiring a target information carrier uploaded by a user;
obtaining a plurality of candidate information carriers similar to the target information carrier from a target database;
respectively taking the target information carrier and any one candidate information carrier as two information carriers to be processed;
the calculating the comparison result of the two information carriers to be processed by using the evaluation model comprises:
calculating a quality comparison result of the target information carrier and any one of the candidate information carriers by using an evaluation model;
the method further comprises the steps of:
and determining a quality grade of the target information carrier based on quality comparison results of the target information carrier and the plurality of candidate information carriers.
7. An information processing method, characterized by comprising:
acquiring popularization pictures from the history record; wherein, the promotion picture bears commodity recommendation information;
determining quality comparison results of any two popularization pictures according to effect evaluation indexes of the two popularization pictures;
setting model labels of the two popularization pictures based on the quality comparison result;
training an evaluation model by using the two popularization pictures and the model label;
according to the effect evaluation index of any two popularization pictures, determining the quality comparison result of the two popularization pictures comprises:
determining quality comparison results of any two different popularization pictures corresponding to the same commodity according to effect evaluation indexes of the two popularization pictures;
the promotion picture is used for being linked to a description page of the commodity;
according to the effect evaluation index of any two popularization pictures, determining the quality comparison result of the two popularization pictures comprises:
determining quality comparison results of any two popularization pictures according to click passing rate of the two popularization pictures;
the training and evaluating model by using the two popularization pictures and the model label comprises the following steps:
Taking the two popularization pictures as model input data and taking the model labels as model output results;
respectively extracting model features of the two popularization pictures through the evaluation model, calculating difference features of the two model features, and mapping the difference features into the model output result to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
extracting model features of the two popularization pictures through the evaluation model, and calculating difference features of the two model features comprises:
respectively extracting convolution characteristics of the two popularization pictures through the convolution modules;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each popularization picture through the fusion module to obtain the fusion characteristics of each popularization picture;
processing the fusion characteristics of the two popularization pictures through the plurality of fully connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module;
The evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
the two promotion pictures comprise a first promotion picture and a second promotion picture, and the first promotion picture and the second promotion picture are used for calculating an evaluation effect index difference value of the first promotion picture compared with the second promotion picture;
the first pooling module is specifically configured to normalize, randomly inactivate and/or excite a function of the output result of the previous module for the first popularization picture;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on an output module of the second popularization picture aiming at the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
8. An information processing method, characterized by comprising:
acquiring two popularization pictures to be processed;
calculating quality comparison results of the two popularization pictures to be processed by using an evaluation model;
the evaluation model is obtained in advance based on any two popularization pictures in the history record and model labels of the two popularization pictures; the model labels are determined based on effect evaluation indexes of the two popularization pictures;
The model tag is specifically determined by:
determining quality comparison results of any two different popularization pictures to be processed according to effect evaluation indexes of the two different popularization pictures to be processed corresponding to the same commodity;
the popularization picture to be processed is used for being linked to a description page of the commodity;
according to the effect evaluation index of any two to-be-processed popularization pictures, determining the quality comparison result of the two to-be-processed popularization pictures comprises:
determining quality comparison results of any two popularization pictures to be processed according to click passing rate of the two popularization pictures to be processed;
the evaluation model is trained by:
taking the two popularization pictures to be processed as model input data, and taking the model labels as model output results;
respectively extracting model features of the two popularization pictures to be processed through the evaluation model, calculating difference features of the two model features, and mapping the difference features into the model output result to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
Extracting model features of the two popularization pictures through the evaluation model, and calculating difference features of the two model features comprises:
respectively extracting convolution characteristics of the two popularization pictures to be processed through the convolution modules;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each popularization picture to be processed through the fusion module to obtain the fusion characteristics of each popularization picture to be processed;
processing the fusion characteristics of the two popularization pictures to be processed through the plurality of fully connected modules to obtain two model characteristics, and calculating the difference characteristics of the two model characteristics through the difference processing module;
the evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
the two random popularization pictures to be processed comprise a first popularization picture to be processed and a second popularization picture to be processed, and the first popularization picture to be processed and the second popularization picture to be processed are used for calculating an evaluation effect index difference value of the first popularization picture to be processed compared with the second popularization picture to be processed;
The first pooling module is specifically configured to normalize, randomly inactivate and/or excite the output result of the first to-be-processed popularization picture by using the previous module;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on an output module of the second popularization picture to be processed by the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
9. An information processing method, characterized by comprising:
acquiring a target popularization picture uploaded by a user;
acquiring a plurality of candidate popularization pictures similar to the target popularization picture from a picture database;
calculating a quality comparison result of the target popularization picture and any candidate popularization picture by using an evaluation model;
determining the quality grade of the target popularization picture based on the quality comparison results of the target popularization picture and the candidate popularization pictures respectively;
outputting quality grade prompt information of the target popularization picture;
the promotion picture is used for being linked to a description page of the object;
the calculating the quality comparison result of the target popularization picture and any candidate popularization picture by using the evaluation model comprises the following steps:
Determining a quality comparison result of the target popularization picture and any candidate popularization picture according to click passing rate of the target popularization picture and any candidate popularization picture;
the evaluation model is trained by:
taking the two popularization pictures as model input data and taking a model label as a model output result;
respectively extracting model features of the two popularization pictures through the evaluation model, calculating difference features of the two model features, and mapping the difference features into the model output result to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
extracting model features of the two popularization pictures through the evaluation model, and calculating difference features of the two model features comprises:
respectively extracting convolution characteristics of the two popularization pictures through the convolution modules;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each popularization picture through the fusion module to obtain the fusion characteristics of each popularization picture;
Processing the fusion characteristics of the two popularization pictures through the plurality of fully connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module;
the evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
the target popularization picture and the candidate popularization picture are used for calculating an evaluation effect index difference value of the target popularization picture compared with the candidate popularization picture;
the first pooling module is specifically configured to normalize, randomly inactivate and/or excite the output result of the previous module for the first target popularization picture;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on an output module of the candidate popularization picture by using the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
10. The method as recited in claim 9, further comprising:
generating content evaluation information for the target popularization picture;
The outputting the quality grade prompt information of the target popularization picture comprises the following steps:
and outputting quality grade prompt information comprising the content evaluation information.
11. An information processing apparatus, characterized by comprising:
a first acquisition unit for acquiring the information carrier from the history; wherein the information carrier carries object recommendation information;
a determining unit for determining a quality comparison result of any two information carriers according to the effect evaluation index of the two information carriers; the determining unit is specifically configured to: determining quality comparison results of any two different information carriers corresponding to the same object according to effect evaluation indexes of the two information carriers; the information carrier is used for linking to a description page of the object; determining quality comparison results of any two information carriers according to click through rates of the two information carriers;
a label setting unit for setting model labels of the two information carriers based on the quality comparison result;
the model training unit is used for training an evaluation model by utilizing the two information carriers and the model label; the evaluation model is used for calculating quality comparison results of two information carriers to be processed; the model training unit is specifically used for: taking the two information carriers as model input data and taking the model labels as model output results;
Respectively extracting model features of the two information carriers through the evaluation model, calculating difference features of the two model features, and mapping the difference features into model output results to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
said extracting model features of the two information carriers via the evaluation model, computing difference features of the two model features comprising:
extracting convolution features of the two information carriers via the plurality of convolution modules, respectively;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each information carrier through the fusion module to obtain the fusion characteristics of each information carrier;
processing the fusion characteristics of two information carriers through the plurality of fully connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module;
the evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
The two information carriers comprise a first information carrier and a second information carrier, and the first information carrier and the second information carrier are used for calculating the evaluation effect index difference value of the first information carrier compared with the second information carrier;
the first pooling module is specifically configured to normalize, randomly inactivate and/or excite the output result of the first information carrier by the previous module;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on the output module of the second information carrier by using the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
12. An information processing apparatus, characterized by comprising:
the second acquisition unit is used for acquiring two information carriers to be processed;
the quality comparison unit is used for calculating quality comparison results of the two information carriers to be processed by using the evaluation model;
wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model labels are determined based on effect evaluation indexes of the two information carriers;
The model tag is specifically determined by:
determining quality comparison results of any two different information carriers to be processed according to effect evaluation indexes of the two information carriers to be processed corresponding to the same object;
the information carrier to be processed is used for being linked to a description page of an object;
according to the effect evaluation index of any two information carriers to be processed, determining the quality comparison result of the two information carriers to be processed comprises the following steps:
determining quality comparison results of any two information carriers to be processed according to click passing rates of the two information carriers to be processed;
the evaluation model is trained by:
taking the two information carriers to be processed as model input data, and taking the model labels as model output results;
respectively extracting model features of the two information carriers to be processed through the evaluation model, calculating difference features of the two model features, and mapping the difference features into model output results to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
Said extracting model features of the two information carriers via the evaluation model, computing difference features of the two model features comprising:
extracting convolution characteristics of the two information carriers to be processed respectively through the convolution modules;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each information carrier to be processed through the fusion module to obtain the fusion characteristics of each information carrier to be processed;
processing the fusion characteristics of two information carriers to be processed through the plurality of fully connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module;
the evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
the two information carriers to be processed comprise a first information carrier to be processed and a second information carrier to be processed, and the first information carrier to be processed and the second information carrier to be processed are used for calculating the evaluation effect index difference value of the first information carrier to be processed compared with the second information carrier to be processed;
The first pooling module is specifically configured to normalize, randomly inactivate and/or excite a function of an output result of the previous module for the first information carrier to be processed;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on an output module of the second information carrier to be processed by the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
13. A computing device comprising a processing component and a storage component;
the storage component stores one or more computer instructions; the one or more computer instructions are to be invoked for execution by the processing component;
the processing assembly is configured to:
obtaining an information carrier from the history; wherein the information carrier carries object recommendation information;
determining quality comparison results of any two information carriers according to effect evaluation indexes of the two information carriers;
setting model labels of the two information carriers based on the quality comparison result;
training an evaluation model by using the two information carriers and the model tag; the evaluation model is used for calculating quality comparison results of two information carriers to be processed;
The determining the quality comparison result of the two information carriers according to the effect evaluation index of any two information carriers comprises the following steps:
determining quality comparison results of any two different information carriers corresponding to the same object according to effect evaluation indexes of the two information carriers;
the information carrier is used for linking to a description page of the object;
the determining the quality comparison result of the two information carriers according to the effect evaluation index of any two information carriers comprises the following steps:
determining quality comparison results of any two information carriers according to click through rates of the two information carriers;
said training an assessment model using said two information carriers and said model tag comprises:
taking the two information carriers as model input data and taking the model labels as model output results;
respectively extracting model features of the two information carriers through the evaluation model, calculating difference features of the two model features, and mapping the difference features into model output results to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
Said extracting model features of the two information carriers via the evaluation model, computing difference features of the two model features comprising:
extracting convolution features of the two information carriers via the plurality of convolution modules, respectively;
combining the convolution characteristics respectively extracted by the convolution modules aiming at each information carrier through the fusion module to obtain the fusion characteristics of each information carrier;
processing the fusion characteristics of two information carriers through the plurality of fully connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module;
the evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
the two information carriers comprise a first information carrier and a second information carrier, and the first information carrier and the second information carrier are used for calculating the evaluation effect index difference value of the first information carrier compared with the second information carrier;
the first pooling module is specifically configured to normalize, randomly inactivate and/or excite the output result of the first information carrier by the previous module;
The second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on the output module of the second information carrier by using the previous module;
the processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
14. A computing device comprising a processing component and a storage component;
the storage component stores one or more computer instructions; the one or more computer instructions are to be invoked for execution by the processing component;
the processing assembly is configured to:
acquiring two information carriers to be processed;
calculating quality comparison results of the two information carriers to be processed by using an evaluation model;
wherein the evaluation model is obtained based on any two information carriers in a history record and model labels of the two information carriers in a training way; the model labels are determined based on effect evaluation indexes of the two information carriers;
the model tag is specifically determined by:
determining quality comparison results of any two different information carriers to be processed according to effect evaluation indexes of the two information carriers to be processed corresponding to the same object;
The information carrier to be processed is used for being linked to a description page of an object;
according to the effect evaluation index of any two information carriers to be processed, determining the quality comparison result of the two information carriers to be processed comprises the following steps:
determining quality comparison results of any two information carriers to be processed according to click passing rates of the two information carriers to be processed;
the evaluation model is trained by:
taking the two information carriers to be processed as model input data, and taking the model labels as model output results;
respectively extracting model features of the two information carriers to be processed through the evaluation model, calculating difference features of the two model features, and mapping the difference features into model output results to realize training of the evaluation model;
the evaluation model comprises a plurality of convolution modules, a fusion module, a plurality of full-connection modules and a difference processing module which are sequentially connected, wherein the fusion module is respectively connected with the convolution modules;
said extracting model features of the two information carriers via the evaluation model, computing difference features of the two model features comprising:
extracting convolution characteristics of the two information carriers to be processed respectively through the convolution modules;
Combining the convolution characteristics respectively extracted by the convolution modules aiming at each information carrier to be processed through the fusion module to obtain the fusion characteristics of each information carrier to be processed;
processing the fusion characteristics of two information carriers to be processed through the plurality of fully connected modules to obtain two model characteristics and calculating the difference characteristics of the two model characteristics through the difference processing module;
the evaluation model further comprises a first pooling module and a second pooling module which are respectively arranged between the fusion module and the first fully-connected module and between two adjacent fully-connected modules;
the two information carriers to be processed comprise a first information carrier to be processed and a second information carrier to be processed, and the first information carrier to be processed and the second information carrier to be processed are used for calculating the evaluation effect index difference value of the first information carrier to be processed compared with the second information carrier to be processed;
the first pooling module is specifically configured to normalize, randomly inactivate and/or excite a function of an output result of the previous module for the first information carrier to be processed;
the second pooling module is specifically configured to perform normalization, random inactivation and/or excitation function processing on an output module of the second information carrier to be processed by the previous module;
The processing modes of the first pooling module and the second pooling module with the same deployment position are the same.
CN201910828416.5A 2019-09-03 2019-09-03 Information processing method and device Active CN112445992B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910828416.5A CN112445992B (en) 2019-09-03 2019-09-03 Information processing method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910828416.5A CN112445992B (en) 2019-09-03 2019-09-03 Information processing method and device

Publications (2)

Publication Number Publication Date
CN112445992A CN112445992A (en) 2021-03-05
CN112445992B true CN112445992B (en) 2024-02-20

Family

ID=74735172

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910828416.5A Active CN112445992B (en) 2019-09-03 2019-09-03 Information processing method and device

Country Status (1)

Country Link
CN (1) CN112445992B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108197532A (en) * 2017-12-18 2018-06-22 深圳云天励飞技术有限公司 The method, apparatus and computer installation of recognition of face
CN108520196A (en) * 2018-02-01 2018-09-11 平安科技(深圳)有限公司 Luxury goods discriminating conduct, electronic device and storage medium
CN109102029A (en) * 2018-08-23 2018-12-28 重庆科技学院 Information, which maximizes, generates confrontation network model synthesis face sample quality appraisal procedure
CN109858770A (en) * 2019-01-02 2019-06-07 口口相传(北京)网络技术有限公司 Object quality appraisal procedure and device

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108197532A (en) * 2017-12-18 2018-06-22 深圳云天励飞技术有限公司 The method, apparatus and computer installation of recognition of face
CN108520196A (en) * 2018-02-01 2018-09-11 平安科技(深圳)有限公司 Luxury goods discriminating conduct, electronic device and storage medium
CN109102029A (en) * 2018-08-23 2018-12-28 重庆科技学院 Information, which maximizes, generates confrontation network model synthesis face sample quality appraisal procedure
CN109858770A (en) * 2019-01-02 2019-06-07 口口相传(北京)网络技术有限公司 Object quality appraisal procedure and device

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
A CTR prediction method based on feature engineering and online learning;Chen Jie-Hao等;2017 17th International Symposium on Communications and Information Technologies (ISCIT);全文 *
基于信息融合的自然灾害等级评估方法研究;龚日朝;王爱平;刘玲;;中国安全科学学报(第11期);全文 *
基于卷积神经网络的诗词隐写检测方法;金鹏;杨忠良;黄永峰;;电子技术应用(第10期);全文 *

Also Published As

Publication number Publication date
CN112445992A (en) 2021-03-05

Similar Documents

Publication Publication Date Title
CN110363213B (en) Method and system for cognitive analysis and classification of garment images
CN109697629B (en) Product data pushing method and device, storage medium and computer equipment
CN105989004B (en) Information delivery preprocessing method and device
CN109460512B (en) Recommendation information processing method, device, equipment and storage medium
US20190156395A1 (en) System and Method for Analyzing and Searching for Features Associated with Objects
US7813965B1 (en) Method, system, and computer readable medium for ranking and displaying a pool of links identified and aggregated from multiple customer reviews pertaining to an item in an electronic catalog
CN108805598B (en) Similarity information determination method, server and computer-readable storage medium
US20170004569A1 (en) Visually generated consumer product presentation
KR20200030252A (en) Apparatus and method for providing artwork
US11803872B2 (en) Creating meta-descriptors of marketing messages to facilitate in delivery performance analysis, delivery performance prediction and offer selection
US10290032B2 (en) Blacklisting based on image feature analysis and collaborative filtering
CN108921398A (en) Shop quality evaluating method and device
CN111666275A (en) Data processing method and device, electronic equipment and storage medium
US20190333077A1 (en) Purchase information utilization system, purchase information utilization method, and program
CN113689144A (en) Quality assessment system and method for product description
Ilakkuvan et al. Cameras for public health surveillance: a methods protocol for crowdsourced annotation of point-of-sale photographs
Guan et al. When images backfire: The effect of customer-generated images on product rating dynamics
CN111680213B (en) Information recommendation method, data processing method and device
CN115018588A (en) Product recommendation method and device, electronic equipment and readable storage medium
Musa et al. Marketing research trends using technology acceptance model (TAM): A comprehensive review of researches (2002–2022)
Karimi et al. Purchase decision processes in the internet age
KR101764361B1 (en) Method of providing shopping mall service based sns and apparatus for the same
CN110209944B (en) Stock analyst recommendation method and device, computer equipment and storage medium
CN112445992B (en) Information processing method and device
CN111352624A (en) Interface display method, information processing method and device

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