CN112053200A

CN112053200A - Information recommendation method and device

Info

Publication number: CN112053200A
Application number: CN202011059853.4A
Authority: CN
Inventors: 黄雅楠; 徐晨敏
Original assignee: Bank of China Ltd
Current assignee: Bank of China Ltd
Priority date: 2020-09-30
Filing date: 2020-09-30
Publication date: 2020-12-08

Abstract

The application provides an information recommendation method and device, the method is applied to a block chain system, and the method comprises the following steps: determining a target advertisement and acquiring advertisement characteristic information of the target advertisement. And classifying the advertisement characteristic information through an advertisement characteristic classification model to obtain first classification characteristic information and second classification characteristic information. And screening at least one first user according to the first classification characteristic information to obtain a second user. And obtaining an information matching model, and obtaining an information matching result output by the information matching model and corresponding to the second user. And weighting the information matching result corresponding to the second user to obtain the information matching score corresponding to the second user, sorting the second user according to the information matching score, and selecting a preset number of second users according to the sorting result to determine the second users as target users. And recommending the advertisement information of the target advertisement to the target user. Through the technical scheme, accurate recommendation of the advertisement information is achieved.

Description

Information recommendation method and device

Technical Field

The application relates to the technical field of internet, in particular to an information recommendation method and device.

Background

Currently, users may download some applications from the application marketplace. The advertiser can well recommend the advertisement information to the user in the application programs. However, the recommendation method of the advertisement information generally aims at all users of the application program, the user range is large, and an advertiser does not know which users are interested in the recommended advertisement information, so that the accuracy and efficiency of information recommendation are low, and accurate recommendation cannot be realized.

Disclosure of Invention

In order to solve the technical problem, the application provides an information recommendation method and device, which are used for realizing accurate recommendation of advertisement information in an application program.

In order to achieve the above purpose, the technical solutions provided in the embodiments of the present application are as follows:

the embodiment of the application provides an information recommendation method, which is applied to a block chain system and comprises the following steps:

determining a target advertisement;

acquiring advertisement characteristic information of the target advertisement;

classifying the advertisement characteristic information through an advertisement characteristic classification model to obtain first classification characteristic information and second classification characteristic information;

screening at least one first user according to the first classification characteristic information and first user characteristic information of the first user to obtain a second user;

acquiring an information matching model;

inputting each second classification characteristic information and second user characteristic information of the second user into the information matching model to obtain an information matching result output by the information matching model and corresponding to the second user; the information matching result corresponding to the second user is used for indicating the degree of correlation between each second classification characteristic information and the second user characteristic information of the second user;

weighting the information matching result corresponding to the second user to obtain an information matching score corresponding to the second user;

sorting the second users according to the information matching scores to obtain a sorting result, and selecting a preset number of second users according to the sorting result to determine the second users as target users;

recommending the advertisement information of the target advertisement to the target user.

Optionally, the obtaining the information matching model includes:

determining labels whether the historical second classification characteristic information, the second user characteristic information of the historical second user and the second user characteristic information of the historical second user are related to each other as target data, and dividing the target data into training data and verification data;

adjusting model parameters of an information matching model according to the training data and the verification result after the last iteration to generate the information matching model after the current iteration, wherein the verification result after the last iteration is zero when the model parameters of the information matching model are adjusted for the first time;

inputting each historical second classification characteristic information in the verification data and second user characteristic information of the historical second user into the information matching model after the current iteration to obtain an information matching result corresponding to the second user output by the information matching model after the current iteration;

calculating to obtain a verification result after the iteration according to the label whether the historical second classification characteristic information and the second user characteristic information of the historical second user are related and the information matching result corresponding to the second user;

and re-executing the verification result after the last iteration according to the training data, adjusting the model parameters of the information matching model, generating the information matching model after the current iteration and subsequent steps until a preset stop condition is reached, and training to obtain the information matching model.

Optionally, the determining the target advertisement includes:

acquiring advertisement information and certification information of preselected advertisements;

acquiring an inspection standard model, inputting the advertisement information of the preselected advertisements into the inspection standard model, and acquiring a first inspection standard corresponding to the preselected advertisements output by the inspection standard model;

judging whether the advertisement information and the certification information of the preselected advertisements meet a first inspection standard corresponding to the preselected advertisements or not;

if the first inspection standard is met, judging whether the advertisement information and the certification information of the preselected advertisements meet a second inspection standard;

determining the preselected advertisement meeting the second inspection criteria as a targeted advertisement.

Optionally, the method may be characterized in that,

when the type of the check standard corresponding to the preselected advertisement is automatic check, the first check standard is a professional detection standard, and the judging whether the advertisement information and the certification information of the preselected advertisement meet the first check standard corresponding to the preselected advertisement includes: judging whether the advertisement information and the certification information of the preselected advertisements meet the professional detection standard or not;

when the type of the check criteria corresponding to the preselected advertisement is manual check, the first check criteria is manual check criteria, and the judging whether the advertisement information and the certification information of the preselected advertisement meet the first check criteria corresponding to the preselected advertisement includes:

receiving a manual inspection result sent by the first user, wherein the manual inspection result is obtained by inspecting the advertisement information and the certification information of the preselected advertisement by the first user according to the manual inspection standard;

judging whether the advertisement information and the certification information of the preselected advertisements meet the manual inspection standard or not according to the manual inspection result;

when the type of the check criteria corresponding to the preselected advertisement is a certification check, the first check criteria is certification check criteria, and the determining whether the advertisement information and the certification information of the preselected advertisement meet the first check criteria corresponding to the preselected advertisement includes:

determining a proof check user from the first user;

receiving a certification inspection result sent by the inspection user, wherein the certification inspection result is obtained by the inspection user inspecting the advertisement information and the certification information of the preselected advertisement according to the certification inspection standard;

and judging whether the advertisement information and the certification information of the preselected advertisement meet the certification inspection standard or not according to the certification inspection result.

Optionally, when the type of the check criterion corresponding to the preselected advertisement is automatic check, the meeting the first check criterion includes: the advertisement information and the certification information of the preselected advertisements meet the professional detection standard;

when the type of the check standard corresponding to the preselected advertisement is manual check, the manual check result is divided into check passing and check failing; the compliance with the first inspection criterion includes: the manual inspection result is that the ratio of passing inspection is greater than a first ratio threshold value;

when the type of the check standard corresponding to the preselected advertisement is certification check, the certification check result is divided into a certification pass and a certification fail; the compliance with the first inspection criterion includes: the ratio of the proof checking result to the proof passing is larger than a second ratio threshold value.

The embodiment of the present application further provides an information recommendation device, where the information recommendation device is applied to a blockchain system, and the device includes:

a determining unit for determining a target advertisement;

the first acquisition unit is used for acquiring advertisement characteristic information of the target advertisement;

the classification unit is used for classifying the advertisement characteristic information through an advertisement characteristic classification model to obtain first classification characteristic information and second classification characteristic information;

the screening unit is used for screening at least one first user according to the first classification characteristic information to obtain a second user;

a second obtaining unit configured to obtain an information matching model;

a third obtaining unit, configured to input each piece of second classification feature information and second user feature information of the second user into the information matching model, so as to obtain an information matching result corresponding to the second user output by the information matching model; the information matching result corresponding to the second user is used for indicating the degree of correlation between each second classification characteristic information and the second user characteristic information of the second user;

the weighting processing unit is used for weighting the information matching result corresponding to the second user by the user to obtain the information matching score corresponding to the second user;

the sorting unit is used for sorting the second users according to the information matching scores to obtain a sorting result, and selecting a preset number of second users according to the sorting result to determine the second users as target users;

and the recommending unit is used for recommending the advertisement information of the target advertisement to the target user.

Optionally, the second obtaining unit includes:

the first determining subunit is configured to determine, as target data, the second classification feature information of each historical second user, the second user feature information of the historical second user, and a label indicating whether each historical second classification feature information and the second user feature information of the historical second user are related, and divide the target data into training data and verification data;

the adjusting subunit is used for adjusting model parameters of the information matching model according to the training data and the verification result after the last iteration to generate the information matching model after the current iteration, and when the model parameters of the information matching model are adjusted for the first time, the verification result after the last iteration is zero;

the first obtaining subunit is configured to input each piece of historical second classification feature information in the verification data and second user feature information of the historical second user into the information matching model after the current iteration, and obtain an information matching result corresponding to the second user output by the information matching model after the current iteration;

the calculation subunit is used for calculating to obtain a verification result after the iteration according to the label whether the historical second classification characteristic information and the second user characteristic information of the historical second user are related and the information matching result corresponding to the second user;

and the circulation subunit is used for re-executing the model parameters of the information matching model according to the training data and the verification result after the last iteration, generating the information matching model after the current iteration and subsequent steps until a preset stop condition is reached, and training to obtain the information matching model.

Optionally, the determining unit includes:

the second acquisition subunit is used for acquiring advertisement information and certification information of the preselected advertisements;

the third acquisition subunit is used for acquiring an inspection standard model, inputting the advertisement information of the preselected advertisement into the inspection standard model and obtaining a first inspection standard corresponding to the preselected advertisement output by the inspection standard model;

the first judgment subunit is used for judging whether the advertisement information and the certification information of the preselected advertisements meet the first check standard corresponding to the preselected advertisements or not;

the second judging subunit is used for judging whether the advertisement information and the certification information of the preselected advertisements meet the second inspection standard or not when the judgment result of the first judging subunit is that the advertisement information and the certification information meet the first inspection standard;

a second determining subunit, configured to determine the preselected advertisement meeting the second check criterion as a target advertisement.

Optionally, when the type of the check criterion corresponding to the preselected advertisement is automatic check, the first check criterion is a professional detection criterion, and the first determining subunit includes:

the third judging subunit is used for judging whether the advertisement information and the certification information of the preselected advertisements meet the professional detection standard or not;

when the type of the check criteria corresponding to the preselected advertisement is manual check, the first check criteria is manual check criteria, and the first judging subunit includes:

the first receiving subunit is configured to receive a manual inspection result sent by the first user, where the manual inspection result is obtained by inspecting, by the first user, advertisement information and certification information of the preselected advertisement according to the manual inspection standard;

a fourth judging subunit, configured to judge, according to the manual inspection result, whether the advertisement information and the certification information of the preselected advertisement meet the manual inspection standard;

when the type of the check criteria corresponding to the preselected advertisement is a proof check, the first check criteria is a proof check criteria, and the first judging subunit includes: the method comprises the following steps:

a third determining subunit configured to determine a proof check user from the first user;

the second receiving subunit is configured to receive a certification inspection result sent by the inspection user, where the certification inspection result is obtained by inspecting the advertisement information and the certification information of the preselected advertisement by the inspection user according to the certification inspection standard;

and the fifth judging subunit is used for judging whether the advertisement information and the certification information of the preselected advertisements meet the certification inspection standard according to the certification inspection result.

According to the technical scheme, the method has the following beneficial effects:

the embodiment of the application provides an information recommendation method, which is applied to a block chain system and comprises the following steps: determining a target advertisement and acquiring advertisement characteristic information of the target advertisement. And classifying the advertisement characteristic information through an advertisement characteristic classification model to obtain first classification characteristic information and second classification characteristic information. And screening at least one first user according to the first classification characteristic information to obtain a second user. And acquiring an information matching model, inputting each second classification characteristic information and the user characteristic information of the second user into the information matching model, and acquiring an information matching result corresponding to the second user and output by the information matching model. And the information matching result corresponding to the second user is used for indicating the correlation degree of each second classification characteristic information and the user characteristic information of the second user. And weighting the information matching result corresponding to the second user to obtain the information matching score corresponding to the second user, sorting the second user according to the information matching score to obtain a sorting result, and selecting a preset number of second users according to the sorting result to determine the second users as target users. And recommending the advertisement information of the target advertisement to the target user. According to the technical scheme provided by the embodiment of the application, the users are screened according to the first classification characteristic information and the second classification characteristic information of the advertisement information, the target users meeting the target advertisement requirements are finally determined, and accurate recommendation of the advertisement information is achieved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a schematic diagram of an exemplary application scenario of an information recommendation method provided in an embodiment of the present application;

fig. 2 is a flowchart of an information recommendation method according to an embodiment of the present application;

FIG. 3 is a flow chart of a method for targeting advertisements provided by an embodiment of the present application;

FIG. 4 is a flowchart of a method for training an information matching model according to an embodiment of the present disclosure;

fig. 5 is a schematic diagram of an information recommendation device according to an embodiment of the present application.

Detailed Description

In order to make the aforementioned objects, features and advantages of the present application more comprehensible, embodiments accompanying the drawings are described in detail below.

In order to facilitate understanding and explaining the technical solutions provided by the embodiments of the present application, the following first describes the background art of the embodiments of the present application.

Currently, users may download some applications from the application marketplace. The advertiser can recommend the advertisement information to the user in the application program, and a good effect can be achieved. However, the recommendation method of the advertisement information generally aims at all users of the application program, all users of the application program can receive the advertisement information recommended by the advertiser, but some users are insensitive to the advertisement recommended by the advertiser and do not belong to audience groups of the advertisement recommended by the advertiser, and the advertiser does not know which users are interested in the recommended advertisement information in tracking, so that the accuracy and the efficiency of information recommendation are low, and accurate recommendation cannot be realized.

Based on this, an embodiment of the present application provides an information recommendation method, where the method is applied to a blockchain system, and the method includes: determining a target advertisement and acquiring advertisement characteristic information of the target advertisement. And classifying the advertisement characteristic information through an advertisement characteristic classification model to obtain first classification characteristic information and second classification characteristic information. And screening at least one first user according to the first classification characteristic information to obtain a second user. And acquiring an information matching model, inputting each second classification characteristic information and the user characteristic information of the second user into the information matching model, and acquiring an information matching result corresponding to the second user and output by the information matching model. And the information matching result corresponding to the second user is used for indicating the correlation degree of each second classification characteristic information and the user characteristic information of the second user. And weighting the information matching result corresponding to the second user to obtain the information matching score corresponding to the second user, sorting the second user according to the information matching score to obtain a sorting result, and selecting a preset number of second users according to the sorting result to determine the second users as target users. And recommending the advertisement information of the target advertisement to the target user.

To facilitate understanding of the information recommendation method provided in the embodiment of the present application, an application scenario of the embodiment of the present application is described below with reference to fig. 1, where fig. 1 is a schematic diagram of an exemplary application scenario of the information recommendation method provided in the embodiment of the present application. The information recommendation method provided by the embodiment of the application is applied to the block chain system 101.

In practice, to first target advertisements, the advertiser uploads the targeted advertisements to be recommended to the blockchain system 101. The blockchain system 101 obtains advertisement characteristic information for the targeted advertisement.

The block chain system 101 classifies the advertisement characteristic information by an advertisement characteristic classification model to obtain first classification characteristic information and second classification characteristic information. Further, the block chain system 101 filters a first user corresponding to at least one first user terminal 102 according to the first classification feature information, and obtains a second user corresponding to a second user terminal 103.

The block chain system 101 obtains the information matching model, inputs each piece of second classification characteristic information and the user characteristic information of the second user into the information matching model, and obtains an information matching result corresponding to the second user output by the information matching model. And the information matching result corresponding to the second user is used for indicating the correlation degree of each second classification characteristic information and the user characteristic information of the second user.

Further, the block chain system 101 performs weighting processing on the information matching result corresponding to the second user to obtain an information matching score corresponding to the second user, sorts the second user according to the information matching score to obtain a sorting result, and selects a preset number of second users according to the sorting result to determine the second users as target users.

Finally, the blockchain system 101 recommends the advertisement information for the targeted advertisement to the targeted user terminal 104 of the targeted user.

Those skilled in the art will appreciate that the schematic diagram shown in fig. 1 is merely one example in which embodiments of the present application may be implemented and that the scope of applicability of embodiments of the present application is not limited in any way by this framework.

In order to facilitate understanding of the technical solutions provided by the embodiments of the present application, the information recommendation method provided by the embodiments of the present application is described below with reference to the accompanying drawings. Referring to fig. 2, fig. 2 is a flowchart of an information recommendation method according to an embodiment of the present application. As shown in fig. 2, the method may include S201-S209:

s201: a targeted advertisement is determined.

The advertisement is mainly recommended in an application program, the recommended advertisement is checked firstly, and the advertisement which passes the checking is determined as the target advertisement.

It will be appreciated that targeted advertisements may be recommended in an application. It should be noted that the process of targeting advertisements is performed in a blockchain system. The detailed process of determining the target advertisement is shown in the following embodiments, and will not be described herein.

S202: and acquiring advertisement characteristic information of the target advertisement.

After determining the target advertisement, the blockchain system obtains advertisement characteristic information of the target advertisement. The advertisement characteristic information includes first classification characteristic information and second classification characteristic information. The first classification characteristic information is determination screening information, and can be understood as quantitative characteristic information, such as age, asset, and the like. The second classification characteristic information is qualitative screening information, and may be understood as qualitative characteristic information, such as character, preference, and the like.

S203: and classifying the advertisement characteristic information through an advertisement characteristic classification model to obtain first classification characteristic information and second classification characteristic information.

And inputting the obtained advertisement characteristic information to be classified into an advertisement characteristic classification model for classification to obtain the output of the advertisement characteristic classification model, namely the first classification characteristic information and the second classification characteristic information corresponding to the input advertisement characteristic information. For example, as a simple example, if the advertisement characteristic information of the target advertisement is over 60 years old and is like to drink milk, the first classification characteristic information can be output as over 60 years old and the second classification characteristic information is like to drink milk through the advertisement characteristic classification model.

In some embodiments, the advertisement feature classification model employs an expert system artificial intelligence model. In some embodiments, the expert system artificial intelligence model is established by a naive Bayesian classification method, and the expert system artificial intelligence model is trained and verified by using the historical advertisement feature information, the corresponding historical first classification feature information and the historical second classification feature information as training data and verification data, so as to finally obtain the available expert system artificial intelligence model.

S204: and screening at least one first user according to the first classification characteristic information and the first user characteristic information of the first user to obtain a second user.

And the block chain system screens at least one first user according to the first classification characteristic information and the first user characteristic information of the first user, and determines a second user from the first user after screening. Wherein the first user characteristic information comprises one or more of the following items: user age, occupation, income level, monthly payment amount, advertisement of interest, gender. It is understood that the second user is the first user whose first user characteristic information satisfies the first classification characteristic information. It should be noted that the blockchain system obtains the first user characteristic information of the first user.

The first user is a good user of the identified application. In some embodiments, the first user is a user who has registered in the application for a time exceeding a certain threshold. In other embodiments, the first user is a user associated with a targeted advertisement in an application. The selection of the first user may be performed according to actual situations, and the first user is not limited herein.

S205: and acquiring an information matching model.

After the blockchain system determines the second user, the second user needs to be further filtered by using the information matching model. It should be noted that the screening by the second user using the information matching model is substantially a further screening of the second user according to the second classification feature information.

In some embodiments, the information matching model may be chosen as a neural network model. The detailed information matching model obtaining process is described in the following embodiments, and is not described herein again.

S206: inputting each second classification characteristic information and the user characteristic information of the second user into an information matching model to obtain an information matching result output by the information matching model and corresponding to the second user; and the information matching result corresponding to the second user is used for indicating the correlation degree of each second classification characteristic information and the user characteristic information of the second user.

And after the block chain system acquires the information matching model, inputting each second classification characteristic information and the user characteristic information of the second user into the information matching model as input to obtain the output of the information matching model, and outputting the output as an information matching result corresponding to the second user.

The second user characteristic information includes information such as the personality of the user and the preference of the user. It should be noted that, if the second classification characteristic information is at least one, the output of the obtained information matching model is at least one information matching result. And the information matching result is used for the degree of correlation between each second classification characteristic information and the user characteristic information of the second user, and the output of the information matching model is the degree of correlation between at least one second classification diagnosis-specific information and the user characteristic information of the second user. For example, the second classification characteristic information is a preference for quiet and a preference for milk. Through the information matching model, the matching result of the 'hobby quiet' and the second user characteristic information of the second user can be obtained, and the matching result of the 'favorite milk' and the second user characteristic information of the second user can also be obtained. In some embodiments, the information matching result is expressed as a percentage, that is, the percentage represents the degree of correlation, for example, the information matching result of "like quiet" and the second user characteristic information is 60%, and the information matching result of "like drink milk" and the second user characteristic information is 80%.

It can be understood that, in the embodiment of the present application, the first user is subjected to the first screening according to the first classification characteristic information, and then the obtained user is subjected to the second screening according to the second classification characteristic information. Because the process of screening the users according to the first classification characteristic information is simple and easy to complete, the screening efficiency can be improved by screening according to the sequence of the first classification characteristic information and the second classification characteristic information, and the efficiency of determining the target users is further improved.

S207: and carrying out weighting processing on the information matching result corresponding to the second user to obtain the information matching score corresponding to the second user.

And after the information matching result corresponding to the second user is obtained, carrying out weighting processing on the information matching result of the second user to obtain the information matching score corresponding to the second user. It should be noted that the information matching result corresponding to the second user is at least one. Therefore, different weights can be assigned to different information matching results, and the weighted information matching results are summed to finally obtain the information matching score corresponding to the second user. For example, the information matching result of "like quiet" and the second user profile information is 60%, the weight assigned to "like quiet" is 0.4, the information matching result of "like drink milk" and the second user profile information is 80%, and the weight assigned to "like drink milk" is 0.6. It can be understood that a more accurate information matching score can be obtained by means of weighted summation.

S208: and sorting the second users according to the information matching scores to obtain a sorting result, and selecting a preset number of second users according to the sorting result to determine the second users as target users.

Since the number of the second users is at least one, the information matching score corresponding to the at least one second user can be obtained. And sequencing at least one information matching score from high to low to obtain a sequencing result. And selecting a preset number of second users to determine the second users as target users according to the obtained sequencing result. The preset number may be set according to the needs of the advertiser of the targeted advertisement, for example, the preset number is 100.

S209: and recommending the advertisement information of the target advertisement to the target user.

After the target user is determined, the advertisement information of the target advertisement is recommended to the target user. The advertisement information of the target advertisement is information such as the nature of the advertisement, the service object, the advertisement content, and the company size.

According to the technical scheme provided by the embodiment, the users are screened according to the first classification characteristic information and the second classification characteristic information of the advertisement information, the target users meeting the target advertisement requirements are finally determined, and accurate recommendation of the advertisement information is achieved.

In order to facilitate understanding of the technical solution provided by the embodiment of the present application, S201 in the information recommendation method provided by the embodiment of the present application is described in detail below, referring to fig. 3, and fig. 3 is a flowchart of a method for determining a target advertisement provided by the embodiment of the present application. As shown in FIG. 3, the targeted advertising method may include S301-S305:

s301: advertisement information and certification information for a preselected advertisement is obtained.

The blockchain system obtains advertising information and certification information for the preselected advertisement. The preselected advertisements are advertisements uploaded to be audited in the block chain system by the advertiser. The certification information includes company name, company license, company product qualification information, and the like.

S302: and acquiring a check standard model, inputting the advertisement information of the preselected advertisements into the check standard model, and acquiring a first check standard corresponding to the preselected advertisements output by the check standard model.

And the block chain system acquires the inspection standard model, inputs the advertisement information of the preselected advertisement into the inspection standard model by taking the advertisement information as input, obtains the output of the inspection standard model, and outputs the output as the first inspection standard corresponding to the preselected advertisement. In some embodiments, the first examination criteria are some criteria by which the advertiser self-reviews its targeted advertisements. It is understood that the first inspection standard is an inspection standard corresponding to the preselected advertisement, for example, if the preselected advertisement is a dairy product, the first inspection standard includes criteria such as a production qualification standard of the dairy product and a production process qualification standard of the dairy product.

In some embodiments, the inspection criteria model is an inspection item artificial intelligence model. It should be noted that the blockchain system obtains the inspection standard model that can be finally used by training and verifying the inspection standard model. The block chain system trains and verifies the check standard model through historical advertisement information in the historical data and historical first check standard, wherein the historical advertisement information serves as input of the check standard model, and the historical first check standard serves as output of the check standard model. The historical advertisement information is divided into training data and verification data, the historical inspection standard is divided into the training data and the verification data, and the inspection standard model is trained and verified.

In some embodiments, to make the inspection criteria model more flexible, the first inspection criteria used to train the inspection criteria model may be criteria uploaded into the blockchain system by some advertising companies and agreed upon by a specified number of first users in the blockchain system.

S303: and judging whether the advertisement information and the certification information of the preselected advertisements meet the first check standard corresponding to the preselected advertisements.

After the first check standard corresponding to the preselected advertisement is obtained, the advertisement information and the certification information of the preselected advertisement are examined according to the first check standard, and whether the advertisement information and the certification information of the preselected advertisement meet the first check standard or not is judged. For example, when the preselected advertisement is a dairy product, it is determined whether the advertisement information and the certification information of the preselected advertisement meet the criteria such as the production qualification criteria of the dairy product and the production process qualification criteria of the dairy product.

In specific implementation, the method for judging whether the advertisement information and the certification information of the preselected advertisements meet the first inspection standard corresponding to the preselected advertisements comprises the following steps:

and when the type of the check standard corresponding to the preselected advertisement is automatic check, the first check standard is a professional detection standard, and whether the advertisement information and the certification information of the preselected advertisement meet the professional detection standard or not is judged.

It should be noted that the type of the check criteria corresponding to the preselected advertisement may be automatically checked, i.e., not manually intervened, and is automatically performed directly by the blockchain. In this case, the first check criterion is a professional detection criterion, i.e. a detection criterion which has been shaped in the field to which the pre-selected advertisement belongs. The block chain system judges whether the advertisement information and the certification information of the preselected advertisements meet the professional detection standard corresponding to the preselected advertisements by using professional detection labels.

When the type of the check standard corresponding to the preselected advertisement is manual check, the first check standard is a manual check standard, and whether the advertisement information and the certification information of the preselected advertisement meet the first check standard corresponding to the preselected advertisement is judged, including:

receiving a manual inspection result sent by a first user, wherein the manual inspection result is obtained by the first user inspecting advertisement information and certification information of a preselected advertisement according to a manual inspection standard;

and judging whether the advertisement information and the certification information of the preselected advertisements meet the manual inspection standard or not through the manual inspection result.

It should be noted that the type of inspection criteria corresponding to the preselected advertisement may be manually inspected. In this case, the first user is required to check the advertisement information and the certification information of the preselected advertisement according to the manual check criteria to obtain a manual check result, and the block chain system determines whether the advertisement information and the certification information of the preselected advertisement meet the manual check criteria according to the obtained manual check result. It is understood that the manual inspection standard can be set according to actual needs, and is not limited herein.

When the type of the check standard corresponding to the preselected advertisement is a certification check, the first check standard is a certification check standard, and whether the advertisement information and the certification information of the preselected advertisement meet the first check standard corresponding to the preselected advertisement is judged, including:

determining a proof checking user from the first users;

receiving a certification inspection result sent by an inspection user, wherein the certification inspection result is obtained by the inspection user through inspecting advertisement information and certification information of a preselected advertisement according to a certification inspection standard;

and judging whether the advertisement information and the certification information of the preselected advertisements meet the certification inspection standard or not according to the certification inspection result.

It should be noted that the type of the check criteria corresponding to the preselected advertisement may be a proof check, that is, the checking user of the first user is required to actually check the advertisement information and the proof information in the preselected advertisement and generate a proof check result. Wherein the actual check is a check that the user performed an offline check on the preselected advertisement. In addition, the checking user may be determined from the first user according to actual needs, and a specific manner of determining the checking user is not limited herein.

The block chain system judges whether the advertisement information and the certification information of the preselected advertisement meet the certification inspection standard or not according to the received certification inspection result.

S304: if the first check criterion is met, whether the advertisement information and the certification information of the preselected advertisement meet the second check criterion is judged.

When the first inspection criterion is met, it is also determined whether the advertisement information and the certification information of the preselected advertisement satisfy the second inspection criterion. It should be noted that the second check criterion is a criterion set in the application program that allows the advertiser to upload the target advertisement to the application program for recommendation.

In specific implementation, the first inspection standard is met, and the following conditions are included:

when the type of the check criteria corresponding to the pre-selected advertisement is automatic check, the meeting the first check criteria comprises: the advertisement information and the certification information of the preselected advertisement meet professional detection standards.

When the type of the inspection standard corresponding to the preselected advertisement is manual inspection, dividing the manual inspection result into inspection passing and inspection failing; compliance with the first inspection criteria includes: the manual inspection result is that the ratio of passing inspection is greater than a first ratio threshold. The first proportional threshold may be set according to actual needs, for example, 90%.

When the type of the check standard corresponding to the preselected advertisement is proof check, the proof check result is divided into proof passing and proof not passing; compliance with the first inspection criteria includes: the ratio of proof checking results as proof of pass is greater than the second ratio threshold. The first proportional threshold may be set according to actual needs, for example, 90%.

When it is determined that the advertisement information and the certification information of the preselected advertisement do not meet the first inspection criteria, the advertisement is primarily certified or otherwise modified for the corresponding location that does not meet the first inspection criteria.

S305: and determining the pre-selected advertisements meeting the second check criterion as the target advertisements.

The preselected advertisement may be determined to be the targeted advertisement after the preselected advertisement meets the second inspection criteria.

By the method for determining the target advertisement, the preselected advertisement uploaded by the advertiser can be audited, and after the advertiser audits whether the preselected advertisement meets the first check standard, the block chain system audits whether the preselected advertisement meets the second check standard. The truthfulness and credibility of the preselected advertisement can be improved by two types of examination criteria.

In order to better understand the technical solution of the embodiment of the present application, a process of training the information matching model method is described with reference to fig. 4. Fig. 4 is a flowchart of a method for training an information matching model according to an embodiment of the present application, where the method includes the following steps S401 to S405:

s401: and determining labels of whether the historical second classification characteristic information, the user characteristic information of the historical second users and the user characteristic information of the historical second users are related as target data, and dividing the target data into training data and verification data.

And determining whether the historical second classification characteristic information, the user characteristic information of the historical second user and the label of whether the historical second classification characteristic information and the user characteristic information of the historical second user are related as target data, wherein the target data are collected and sorted actual data and are obtained through non-calculation. And dividing the target data into training data and verification data, wherein model parameters of the product information recommendation model are continuously adjusted by using the training data and the verification data so as to obtain an optimal product information recommendation model.

During specific implementation, optionally, the quantization and vectorization are performed on each historical second classification feature information and the user feature information of the historical second user to obtain a final feature vector, and the feature vector containing each historical second classification feature information and the user feature information of the historical second user is used as the input of the information matching model. And vectorizing whether the labels of the historical second classification characteristic information and the user characteristic information of the historical second user are related or not, and taking the obtained vector as an expected output value of the information matching model.

The obtained characteristic information as input and the label as output are collectively referred to as target data, the target data is real data collected and sorted, and the target data is divided into two groups of training data and verification data to be used for training a product information recommendation model. For example, the target data is 10 groups in total, and the target data is divided into training data and verification data 2 classes, wherein the training data is 5 groups of data, and the verification data is another 5 groups of data.

S402: and adjusting the model parameters of the information matching model according to the training data and the verification result after the last iteration to generate the information matching model after the current iteration, wherein the verification result after the last iteration is zero when the model parameters of the information matching model are adjusted for the first time.

The training data comprises each historical second classification characteristic information, the user characteristic information of the historical second users, and labels of whether each historical second classification characteristic information and the user characteristic information of the historical second users are related. The verification result is a difference between a tag value, which is determined whether each historical second classification characteristic information in the verification data is related to the user characteristic information of the historical second user, and an information matching result corresponding to the second user output by the information matching model, which is described in detail in S404. And adjusting the model parameters of the information matching model by using the historical second classification characteristic information, the user characteristic information of the historical second user, the label of whether the historical second classification characteristic information and the user characteristic information of the historical second user are related or not and the verification result in the training data. And after the model parameters are adjusted, generating the information matching model after the iteration. When the information is adjusted for the first time to match the model parameters of the model, the verification result after the last iteration is set to zero because the last iteration does not exist.

In some embodiments, the information matching model may be selected as a neural network model. Optionally, a three-layer neural network model is established, the historical second classification characteristic information and the historical user characteristic information of the second user are input into the neural network model as model characteristics, and the model characteristics are output as an information matching result corresponding to the second user. According to the kolmogorov principle, one three-layer neural network can complete any mapping from n dimensions to m dimensions, namely, a plurality of items of historical second classification characteristic information and user characteristic information of historical second users can be mapped to information matching results corresponding to the second users.

S403: and inputting each historical second classification characteristic information in the verification data and the user characteristic information of the historical second user into the information matching model after the iteration, and obtaining an information matching result corresponding to the second user output by the information matching model after the iteration.

After the information matching model after the iteration is generated, the historical second classification characteristic information and the user characteristic information of the historical second user in the verification data are used as the input of the information matching model after the iteration. And outputting an information matching result corresponding to the second user, wherein the output is the output of the information matching model after the iteration.

S404: and calculating to obtain a verification result after the iteration according to the label whether the historical second classification characteristic information and the historical user characteristic information of the second user are related and the information matching result corresponding to the second user.

And for the verification data, after an information matching result corresponding to a second user is obtained through the information matching model after the iteration, the difference between the label of whether each historical second classification characteristic information in the verification data is related to the second user characteristic information of the historical second user and the information matching result corresponding to the second user is obtained, and the verification result after the iteration is obtained.

S405: and re-executing the model parameters of the information matching model according to the training data and the verification result after the last iteration, generating the information matching model after the current iteration and subsequent steps until a preset stopping condition is reached, and training to obtain the information matching model.

S401-S404 are steps of adjusting model parameters, generating an information matching model after the iteration, obtaining an iterative verification result by using the verification data and the information matching model after the iteration, executing S401, namely, reusing the training data and the iterative verification result (called as the verification result after the last iteration) to act on the information matching model after the iteration, adjusting the parameters of the model, updating the information matching model again, and generating the information matching model after the iteration. And (3) training by using the training data and the verification data in a circulating manner, continuously adjusting parameters of the model, and stopping training until the accuracy is continuously improved until a preset stopping condition is reached to obtain the finally trained information matching model. Optionally, the preset stop condition may be set to stop training when the deviation is smaller than a set threshold, or stop training when the number of times of training is reached.

According to the technical scheme of the embodiment, the information matching model is obtained by training by using the historical second classification characteristic information, the second user characteristic information of the historical second user and the label of whether the historical second classification characteristic information and the second user characteristic information of the historical second user are related or not as training data and verification data. Therefore, the trained information matching model can be used for determining the target users of the target advertisements from the second users, the accuracy of information recommendation is improved, and the requirements of the users are met.

Referring to fig. 5, fig. 5 is a schematic diagram of an information recommendation device according to an embodiment of the present application. The apparatus may include:

a determining unit 501, configured to determine a target advertisement;

a first obtaining unit 502, configured to obtain advertisement characteristic information of the target advertisement;

a classifying unit 503, configured to classify the advertisement feature information through an advertisement feature classification model to obtain first classification feature information and second classification feature information;

a screening unit 504, configured to screen at least one first user according to the first classification feature information to obtain a second user;

a second obtaining unit 505, configured to obtain an information matching model;

a third obtaining unit 506, configured to input each piece of the second classification feature information and second user feature information of the second user into the information matching model, so as to obtain an information matching result corresponding to the second user output by the information matching model; the information matching result corresponding to the second user is used for indicating the degree of correlation between each second classification characteristic information and the second user characteristic information of the second user;

a weighting processing unit 507, configured to perform weighting processing on the information matching result corresponding to the second user by the user to obtain an information matching score corresponding to the second user;

a sorting unit 508, configured to sort the second users according to the information matching scores to obtain a sorting result, and select a preset number of second users according to the sorting result to determine that the second users are target users;

a recommending unit 509, configured to recommend the advertisement information of the target advertisement to the target user.

Optionally, in some implementations of this embodiment, the second obtaining unit 505 includes:

Optionally, in some implementations of this embodiment, the determining unit 501 includes:

Optionally, in some implementations of this embodiment, when the type of the check criterion corresponding to the preselected advertisement is automatic check, the first check criterion is a professional detection criterion, and the first determining subunit includes:

Optionally, in some implementations of this embodiment, when the type of the check criterion corresponding to the preselected advertisement is automatic check, the meeting the first check criterion includes: the advertisement information and the certification information of the preselected advertisements meet the professional detection standard;

Through the device provided by the embodiment of the application, the users are screened according to the first classification characteristic information and the second classification characteristic information of the advertisement information, the target users meeting the target advertisement requirements are finally determined, the accurate recommendation of the advertisement information is realized, and the information recommendation efficiency is improved.

As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that all or part of the steps in the above embodiment methods can be implemented by software plus a necessary general hardware platform. Based on such understanding, the technical solution of the present application may be essentially or partially implemented in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the method according to the embodiments or some parts of the embodiments of the present application.

It should be noted that, in the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. The method disclosed by the embodiment corresponds to the system disclosed by the embodiment, so that the description is simple, and the relevant points can be referred to the system part for description.

It should also be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An information recommendation method is applied to a block chain system, and comprises the following steps:

determining a target advertisement;

acquiring advertisement characteristic information of the target advertisement;

acquiring an information matching model;

2. The method of claim 1, wherein obtaining the information matching model comprises:

3. The method of claim 1, wherein the targeting advertisements comprises:

4. The method of claim 3,

determining a proof check user from the first user;

5. The method of claim 4,

when the type of the check criteria corresponding to the preselected advertisement is automatic check, the meeting the first check criteria comprises: the advertisement information and the certification information of the preselected advertisements meet the professional detection standard;

6. An information recommendation device, wherein the information recommendation device is applied to a blockchain system, and the device comprises:

a determining unit for determining a target advertisement;

a second obtaining unit configured to obtain an information matching model;

7. The apparatus of claim 6, wherein the second obtaining unit comprises:

8. The apparatus of claim 6, wherein the determining unit comprises:

9. The apparatus of claim 8,

when the type of the check standard corresponding to the preselected advertisement is automatic check, the first check standard is a professional detection standard, and the first judging subunit includes:

10. The apparatus of claim 9,