CN105678317B

CN105678317B - Information processing method and server

Info

Publication number: CN105678317B
Application number: CN201511019197.4A
Authority: CN
Inventors: 陈琼; 周星; 张波
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2015-12-30
Filing date: 2015-12-30
Publication date: 2020-12-04
Anticipated expiration: 2035-12-30
Also published as: CN105678317A

Abstract

The invention discloses an information processing method and a server, wherein the information processing method comprises the following steps: obtaining first data from a data source; extracting at least two types of characteristic parameters which accord with a preset rule from the first data, and establishing correlation between the at least two types of characteristic parameters to obtain a correlation processing result; detecting whether the first data sent by the first terminal generates effective feedback at the second terminal or not to obtain a detection result; generating target data according to the detection result and the correlation processing result; and generating a data analysis model according to the target data, generating an information directional recommendation strategy containing at least two directional condition sequencing results according to the data analysis model, and feeding back the information directional recommendation strategy to the first terminal.

Description

Information processing method and server

Technical Field

The present invention relates to communications technologies, and in particular, to an information processing method and a server.

Background

In the process of implementing the technical solution of the embodiment of the present application, the inventor of the present application finds at least the following technical problems in the related art:

information sending and receiving feedback are common means for information interaction among users. In order to effectively carry out information interaction, an information chain for data analysis is formed, so that an information sending party and an information receiving party can obtain required effective information, the information sending party is required to pertinently select information and pertinently send the information, and therefore the information obtained by the information receiving party is the effective information related to self requirements, and after the information receiving party sends feedback information formed by response feedback of the information to the information sending party, the information sending party analyzes the feedback information to evaluate and analyze whether the sent information is effective or not, so that the fact that how to pertinently select and send the effective information is determined.

It can be seen that: if the information sender has a problem in selecting an information sending directional strategy, the information receiver cannot receive effective information meeting the self requirement, the problem that the information directional pushing is not accurate enough occurs, and the information sender cannot obtain effective feedback information for data analysis aiming at the response feedback of the information receiver to the information. However, in the related art, there is no effective solution to this problem.

Disclosure of Invention

In view of the above, embodiments of the present invention provide an information processing method and a server, which at least solve the problems in the prior art.

The technical scheme of the embodiment of the invention is realized as follows:

an information processing method according to an embodiment of the present invention includes:

obtaining first data from a data source;

extracting at least two types of characteristic parameters which accord with a preset rule from the first data, and establishing correlation between the at least two types of characteristic parameters to obtain a correlation processing result;

detecting whether the first data sent by the first terminal generates effective feedback at the second terminal or not to obtain a detection result;

generating target data according to the detection result and the correlation processing result;

and generating a data analysis model according to the target data, generating an information directional recommendation strategy containing at least two directional condition sequencing results according to the data analysis model, and feeding back the information directional recommendation strategy to the first terminal.

In the above scheme, the method further comprises: after first data are acquired from a data source, the first data are converted into second data in a data format supported by the first data.

In the foregoing scheme, the extracting at least two types of feature parameters that meet a preset rule from the first data, and establishing a correlation between the at least two types of feature parameters to obtain a correlation processing result includes:

extracting a data decomposition condition, and extracting a first class characteristic parameter and a second class characteristic parameter which accord with a preset rule from the first data according to the data decomposition condition;

the first class of characteristic parameters are used for characterizing the orientation condition identified by X split data segments, wherein the X split data segments are derived from the content of each piece of data in the first data, and X is a positive integer greater than 2;

the second type characteristic parameter is used for representing the data classification identified by the attribute of each piece of data of the first data;

and respectively associating the orientation conditions identified by the X split data segments with the data classification to obtain X association results, and determining the obtained X association results as the correlation processing results.

In the foregoing solution, the detecting whether the first data sent by the first terminal generates effective feedback at the second terminal to obtain a detection result includes:

when detecting that the first data sent by the first terminal generates effective feedback at the second terminal, recording the detection result as the conversion rate obtained through feedback;

and when detecting that the first data sent by the first terminal does not generate effective feedback at the second terminal, recording the detection result as that the conversion rate is not obtained through feedback.

In the foregoing solution, generating target data with a classification attribute according to the detection result and the correlation processing result includes:

when the detection result is that the conversion rate is obtained through feedback, matching corresponding first-class training data from the correlation processing result according to the detection result;

when the detection result is that the conversion rate is not obtained through feedback, matching corresponding second type training data from the correlation processing result according to the detection result;

and generating the target data with the classification attribute according to the first class of training data and the second class of training data.

In the foregoing solution, the generating an information-oriented recommendation policy including at least two oriented condition ordering results according to the data analysis model, and feeding back the information-oriented recommendation policy to the first terminal includes:

generating a conversion rate data model for analyzing a conversion rate as the data analysis model according to the target data and the weight value related to the target data;

and estimating the sequencing results of at least two directional conditions according to the conversion rate data model, packaging the sequencing results of the at least two directional conditions into the information directional recommendation strategy, and feeding back the information directional recommendation strategy to the first terminal.

A server according to an embodiment of the present invention includes:

an acquisition unit configured to acquire first data from a data source;

the first processing unit is used for extracting at least two types of characteristic parameters which accord with a preset rule from the first data, and establishing correlation between the at least two types of characteristic parameters to obtain a correlation processing result;

the detection unit is used for detecting whether the first data sent by the first terminal generates effective feedback at the second terminal or not to obtain a detection result;

the second processing unit is used for generating target data according to the detection result and the correlation processing result;

and the directional recommendation unit is used for generating a data analysis model according to the target data, generating an information directional recommendation strategy containing at least two directional condition sequencing results according to the data analysis model, and feeding back the information directional recommendation strategy to the first terminal.

In the above solution, the server further includes:

the data conversion unit is used for converting the first data into second data in a data format supported by the data conversion unit after the first data is acquired from a data source.

In the foregoing solution, the first processing unit is further configured to:

In the foregoing solution, the detecting unit is further configured to:

In the foregoing solution, the second processing unit is further configured to:

In the foregoing solution, the directional recommendation unit is further configured to:

The information processing method of the embodiment of the invention comprises the following steps: obtaining first data from a data source; extracting at least two types of characteristic parameters which accord with a preset rule from the first data, and establishing correlation between the at least two types of characteristic parameters to obtain a correlation processing result; detecting whether the first data sent by the first terminal generates effective feedback at the second terminal or not to obtain a detection result; generating target data according to the detection result and the correlation processing result; and generating a data analysis model according to the target data, generating an information directional recommendation strategy containing at least two directional condition sequencing results according to the data analysis model, and feeding back the information directional recommendation strategy to the first terminal.

By adopting the embodiment of the invention, the correlation is established between the at least two types of characteristic parameters to obtain the correlation processing result; detecting whether the first data sent by the first terminal generates effective feedback at the second terminal or not to obtain a detection result; generating target data according to the detection result and the correlation processing result; and generating a data analysis model according to the target data, and generating an information directional recommendation strategy containing at least two directional condition sequencing results according to the data analysis model, so that the information sender is very accurate in selecting the directional strategy for sending the information, on one hand, the information receiver is ensured to receive the effective information meeting the self requirement, and the effect of directionally pushing the information is accurate, on the other hand, the information receiver is capable of receiving the effective information meeting the self requirement, and then the feedback aiming at the effective information is also an effective feedback response, so that the information sender obtains the effective feedback information for data analysis aiming at the effective feedback response of the information receiver to the information, and the final data analysis result is also accurate enough.

Drawings

FIG. 1 is a diagram of hardware entities performing information interaction in an embodiment of the present invention;

FIG. 2 is a flow chart illustrating an implementation of a first embodiment of the present invention;

FIG. 3 is a flowchart illustrating a second implementation of the second embodiment of the present invention;

FIG. 4 is a schematic diagram of a system configuration according to a fourth embodiment of the present invention;

FIGS. 5-8 are diagrams of a plurality of user interface scenarios in which embodiments of the present invention may be employed;

fig. 9 is a flowchart of information interaction between a terminal and a server according to a sixth embodiment of the present invention;

fig. 10 is a schematic diagram of a hardware configuration according to a seventh embodiment of the present invention;

FIG. 11 is a diagram illustrating an example of training data for an application scenario in which embodiments of the present invention are applied;

FIG. 12 is a schematic diagram of modeling an application scenario to which an embodiment of the present invention is applied.

Detailed Description

The following describes the embodiments in further detail with reference to the accompanying drawings.

Fig. 1 is a schematic diagram of hardware entities performing information interaction in an embodiment of the present invention, where fig. 1 includes: the server 11 … … 1n and the terminal devices 21-24, where the terminal devices 21-24 perform information interaction with the server through a wired network or a wireless network, and the terminal devices include mobile phones, desktop computers, PCs, all-in-one machines, and the like, in an example, the server 11 … … 1n may further perform interaction with a first terminal (e.g., a terminal where an advertiser is located, or referred to as an object providing an advertisement material and a content promotion) through a network, and after the first terminal (e.g., a terminal where an advertiser is located, or referred to as an object providing an advertisement material and a content promotion) submits an advertisement that is desired to be delivered, the advertisement is stored in a server cluster, and a manager may be configured to perform a series of processes such as checking and the like on the advertisement delivered by the first terminal (e.g., a terminal where an advertiser is located, or referred to an object providing an advertisement material and a. In which, compared to a first terminal (e.g. a terminal where an advertiser is located, or referred to as an object providing advertisement material and content promotion), the terminal devices 21-24 may be referred to as a second terminal (e.g. a terminal where an ordinary user is located, or referred to as an object of advertisement display or exposure), may be a user watching a video through a video application, a user playing a game through a game application, a user browsing a page through internet, and so on. Wherein, all applications installed in the terminal equipment or designated applications (such as game application, video application, browser application and the like) can add advertisements to show more recommendation information to the user. By adopting the embodiment of the present invention, based on the system shown in fig. 1, the server obtains the first data (such as historical data and real-time updated data) sent by the first terminal (such as the terminal where the advertiser is located, or called as the object providing the advertisement material and content promotion) and/or the second terminal (such as the terminal where the ordinary user is located, or called as the object providing the advertisement material and content promotion), and detects whether the first data (such as historical data and real-time updated data) sent by the first terminal (such as the terminal where the advertiser is located, or called as the object providing the advertisement material and content promotion) generates effective feedback at the second terminal (such as the terminal where the ordinary user is located, or called as the object providing the advertisement material and content promotion) at the server, so as to obtain the detection result; generating target data with classification attributes according to the detection result and the correlation processing result, such as positive example data (obtaining effective feedback response, obtaining effective feedback response may be referred to as "positive feedback" for short) or negative example data (not obtaining effective feedback response — not obtaining effective feedback response may be referred to as "negative feedback" for short) obtained according to whether effective feedback response exists, wherein training data composed of the positive example data and the negative example data can be used as the training data; generating a data analysis model (such as a conversion rate data model for analyzing conversion rate) according to the target data (such as training data), generating an information targeting recommendation strategy containing at least two targeting condition sequencing results according to the data analysis model (such as a conversion rate data model for analyzing conversion rate), so that when a first terminal (such as a terminal where an advertiser is located or an object for providing advertisement materials and content promotion) is used as an information sending party, the targeting strategy for sending information can ensure accurate positioning, and therefore, when a second terminal (such as a terminal where an ordinary user is located or an object for advertisement display or exposure) is used as an information receiving party, the information receiving party can receive effective information meeting the requirements of the information receiving party, and the information sending party responds to the effective feedback information of the information by the information receiving party to obtain effective feedback information for data analysis, the final data analysis results will be accurate enough to further optimize the data analysis model (e.g., the conversion data model for analyzing conversion) and the information delivery targeting strategy.

The above example of fig. 1 is only an example of a system architecture for implementing the embodiment of the present invention, and the embodiment of the present invention is not limited to the system architecture described in the above fig. 1, and various embodiments of the present invention are proposed based on the system architecture.

The first embodiment is as follows:

as shown in fig. 2, an information processing method according to an embodiment of the present invention includes:

step 101, a server acquires first data from a data source.

And 102, extracting at least two types of characteristic parameters which accord with a preset rule from the first data by the server, and establishing correlation between the at least two types of characteristic parameters to obtain a correlation processing result.

The first data acquired in step 101 at least includes history data, and is not limited to history data, and may be a data set composed of history data and real-time update data. The data source is at least from a first terminal (such as a terminal where an advertiser is located or an object providing advertisement materials and content promotion) and/or a second terminal (such as a terminal where an ordinary user is located or an object providing advertisement display or exposure).

The at least two types of characteristic parameters in step 102 include, but are not limited to, a targeting condition and a data classification, taking advertisement information as an example, the first type of characteristic parameter is a targeting pushing condition for selecting advertisement information to be pushed to a target information receiving group, and the second type of characteristic parameter is an information classification (or referred to as an information field or category) to which the advertisement information belongs. Establishing correlation between the at least two types of characteristic parameters to obtain a correlation processing result, wherein the technical effect to be achieved is as follows: taking the advertisement information as an example, information cross processing is performed between the targeted push condition for pushing the advertisement information to the target information receiving group and the information classification (or referred to as information field or category) to which the advertisement information belongs, so that the two have an intersection, and thus, the result after the information cross combination is more beneficial to accurately describing the targeted recommendation strategy in the step 106. Another example is: not only the intersection of information generation, but also the union of information generation is possible, and in short, the comprehensive information composed of the two types of information is required.

Step 103, the server detects whether the first data sent by the first terminal generates effective feedback at the second terminal, and obtains a detection result.

Here, the first terminal (e.g. the terminal where the advertiser is located, or called as the object providing advertisement material and content promotion) is used as the information sender, it needs to ensure that the targeting strategy for selecting information sending can ensure accurate positioning, the second terminal (e.g. the terminal where the ordinary user is located, or called as the object of advertisement display or exposure) is used as the information receiver, it needs to ensure that it generates effective feedback, and of course, ineffective feedback is also collected, because the data analysis model (e.g. the conversion rate data model for analyzing the conversion rate) is subsequently generated to obtain the targeting recommendation strategy through final calculation, in the data analysis, it needs to collect both positive case data (obtaining effective feedback response, e.g. generating conversion rate) and negative case data (not obtaining effective feedback response, e.g. not generating conversion rate), and for the positive case data (effective feedback), negative case data is directed to reverse feedback (ineffective feedback), so that the final analysis result is accurate only by comprehensively considering two factors, and the directional recommendation strategy obtained in step 106 is ensured to be accurate enough.

And 104, the server generates target data according to the detection result and the correlation processing result, generates a data analysis model according to the target data, generates an information directional recommendation strategy containing at least two directional condition sorting results according to the data analysis model, and feeds back the information directional recommendation strategy to the first terminal.

Here, the detection result is obtained by step 103: combining the data obtained in step 102 as the results of both forward feedback (effective feedback) for positive case data and reverse feedback (ineffective feedback) for negative case data: the method includes the steps of obtaining comprehensive data by associating (or called information crossing) at least two types of characteristic parameters or combining (or called orientation condition and data classification) at least two types of characteristic parameters, generating training data formed by positive example data and negative example data as the training data, generating a data analysis model (such as a conversion rate data model for analyzing a conversion rate) according to the target data as the target data, and generating an information orientation recommendation strategy containing at least two orientation condition sequencing results according to the data analysis model to be accurate enough so as to ensure that the orientation recommendation strategy fed back to a first terminal (such as a terminal where an advertiser is located or an object providing advertisement materials and content promotion) is accurate enough, so that the first terminal (such as a terminal where the advertiser is located or an object providing the advertisement materials and content promotion) pushes information (such as advertisement information) to a second terminal (or called advertisement information) according to the orientation recommendation strategy Such as a terminal where an ordinary user is located, or called an advertisement display or exposure object), can also be accurately located, and information is accurately pushed.

For example, a first terminal (e.g., a terminal where an advertiser is located, or referred to as an object providing advertisement material and content promotion) pushes information (e.g., advertisement information) to a second terminal (e.g., a terminal where an ordinary user is located, or referred to as an object displaying or exposing an advertisement) according to the targeting recommendation policy, a ranking result of a plurality of targeting conditions in the information targeting recommendation policy can be analyzed for the first terminal, and a most suitable targeting condition (e.g., a first targeting condition and/or a second targeting condition with a top ranking is selected) is selected from the ranking result of the plurality of targeting conditions, so that the information is pushed accurately.

And 105, triggering selection operation aiming at the directional function of information sending at the first terminal, and analyzing to obtain an information classification result.

And 106, aiming at the sequencing result which is matched with the self-classified directional condition from the directional recommendation strategy according to the information classification result, pushing information (such as advertisement information) according to the sequencing result of the directional condition.

Through the above step 101-. The first terminal selects a target orientation condition corresponding to the self requirement from the sequencing results of the at least two orientation conditions according to the information orientation recommendation strategy, the information pushed to the second terminal (or called information receiving group) by the first terminal is targeted, the requirement of the second terminal (or called information receiving group) is met, the positioning is accurate, and the targeted recommendation strategy can be further optimized through the feedback of the second terminal (or called information receiving group), such as the conversion rate formed by the user behavior of the second terminal (or called information receiving group), the optimization of the targeted recommendation strategy can be performed at the first terminal, or the targeted recommendation strategy can be optimized by a server (such as an advertisement analysis processing platform of a third party) after the targeted recommendation strategy is provided to the server (such as an advertisement analysis processing platform of the third party) by the first terminal.

For example, taking advertisement information as an example, the server may generate training data according to targeting conditions (e.g. 18 years old, beijing area, maiden) and advertisement classification (e.g. games, 3C products, daily chemical products) in each advertisement information in the historical data, and the training data needs to be divided into positive example data and negative example data according to conversion rate; generating a conversion rate data model according to the training data and the weight thereof; and estimating a directional condition sorting result recommended to the user for selection based on the conversion rate data model, packaging the directional condition sorting result into a directional recommendation strategy, and pushing the directional recommendation strategy to a first terminal (such as a terminal where an advertiser is located, or called an object for providing advertisement materials and content promotion) by the server for use.

Example two:

as shown in fig. 3, an information processing method according to an embodiment of the present invention includes:

step 201, a server acquires first data from a data source, and converts the first data into second data in a data format supported by the server.

Step 202, the server extracts at least two types of characteristic parameters which accord with a preset rule from the second data, and establishes a correlation between the at least two types of characteristic parameters to obtain a correlation processing result.

The first data acquired in step 201 includes at least history data, and is not limited to history data, and may be a data set composed of history data and real-time update data. The data source is at least from a first terminal (such as a terminal where an advertiser is located or an object providing advertisement materials and content promotion) and/or a second terminal (such as a terminal where an ordinary user is located or an object providing advertisement display or exposure).

In step 201, the first data is used as original data, and for different advertisement analysis processing platforms, there may be a problem that the data is not compatible or the data is not suitable, and there may also be a problem that the data itself has redundancy, so that the first data needs to be converted into a data format of the second data, and the data format of the second data is a data format that can be supported by the current advertisement analysis processing platform for analysis processing.

The at least two types of characteristic parameters in step 202 include, but are not limited to, a targeting condition and a data classification, taking advertisement information as an example, the first type of characteristic parameter is a targeting pushing condition for selecting advertisement information to be pushed to a target information receiving group, and the second type of characteristic parameter is an information classification (or referred to as an information field or category) to which the advertisement information belongs. Establishing correlation between the at least two types of characteristic parameters to obtain a correlation processing result, wherein the technical effect to be achieved is as follows: taking the advertisement information as an example, information cross processing is performed between the targeted push condition for pushing the advertisement information to the target information receiving group and the information classification (or referred to as information field or category) to which the advertisement information belongs, so that the two have an intersection, and thus, the result after the information cross combination is more beneficial to accurately describing the targeted recommendation strategy in the step 206. Another example is: not only the intersection of information generation, but also the union of information generation is possible, and in short, the comprehensive information composed of the two types of information is required.

Step 203, the server detects whether the first data sent by the first terminal generates effective feedback at the second terminal, and obtains a detection result.

Here, the first terminal (e.g. the terminal where the advertiser is located, or called as the object providing advertisement material and content promotion) is used as the information sender, it needs to ensure that the targeting strategy for selecting information sending can ensure accurate positioning, the second terminal (e.g. the terminal where the ordinary user is located, or called as the object of advertisement display or exposure) is used as the information receiver, it needs to ensure that it generates effective feedback, and of course, ineffective feedback is also collected, because the data analysis model (e.g. the conversion rate data model for analyzing the conversion rate) is subsequently generated to obtain the targeting recommendation strategy through final calculation, in the data analysis, it needs to collect both positive case data (obtaining effective feedback response, e.g. generating conversion rate) and negative case data (not obtaining effective feedback response, e.g. not generating conversion rate), and for the positive case data (effective feedback), negative case data is directed to reverse feedback (invalid feedback), so that the final analysis result is accurate only by comprehensively considering two factors, and it can be ensured that the directional recommendation strategy obtained in step 206 is accurate enough.

And 204, the server generates target data with classification attributes according to the detection result and the correlation processing result, generates a data analysis model according to the target data, generates an information directional recommendation strategy containing at least two directional condition sorting results according to the data analysis model, and feeds back the information directional recommendation strategy to the first terminal.

Here, the detection result is obtained through step 203: as a result of the positive case data for the forward feedback (effective feedback) and the negative case data for the reverse feedback (ineffective feedback), combining the data obtained in step 202: the method includes the steps of obtaining comprehensive data by associating (or called information crossing) at least two types of characteristic parameters or combining (or called orientation condition and data classification) at least two types of characteristic parameters, generating training data formed by positive example data and negative example data as the training data, generating a data analysis model (such as a conversion rate data model for analyzing a conversion rate) according to the target data as the target data, and generating an information orientation recommendation strategy containing at least two orientation condition sequencing results according to the data analysis model to be accurate enough so as to ensure that the orientation recommendation strategy fed back to a first terminal (such as a terminal where an advertiser is located or an object providing advertisement materials and content promotion) is accurate enough, so that the first terminal (such as a terminal where the advertiser is located or an object providing the advertisement materials and content promotion) pushes information (such as advertisement information) to a second terminal (or called advertisement information) according to the orientation recommendation strategy Such as a terminal where an ordinary user is located, or called an advertisement display or exposure object), can also be accurately located, and information is accurately pushed.

Step 205, triggering selection operation of the directional function aiming at information sending at the first terminal, and analyzing to obtain an information classification result.

Step 206, aiming at the sorting result which matches the self-classified directional condition from the directional recommendation strategy according to the information classification result, pushing information (such as advertisement information) according to the sorting result of the directional condition.

Through the above step 201-. The first terminal selects a target orientation condition corresponding to the self requirement from the sequencing results of the at least two orientation conditions according to the information orientation recommendation strategy, the information pushed to the second terminal (or called information receiving group) by the first terminal is targeted, the requirement of the second terminal (or called information receiving group) is met, the positioning is accurate, and the targeted recommendation strategy can be further optimized through the feedback of the second terminal (or called information receiving group), such as the conversion rate formed by the user behavior of the second terminal (or called information receiving group), the optimization of the targeted recommendation strategy can be performed at the first terminal, or the targeted recommendation strategy can be optimized by a server (such as an advertisement analysis processing platform of a third party) after the targeted recommendation strategy is provided to the server (such as an advertisement analysis processing platform of the third party) by the first terminal.

Example three:

based on the first to second embodiments, in the information processing method according to the embodiments of the present invention, extracting at least two types of feature parameters that meet a preset rule from the first data, and establishing a correlation between the at least two types of feature parameters to obtain a correlation processing result includes: extracting a data decomposition condition, and extracting a first class characteristic parameter and a second class characteristic parameter which accord with a preset rule from the first data according to the data decomposition condition; the first class of characteristic parameters are used for characterizing the orientation condition identified by X split data segments, wherein the X split data segments are derived from the content of each piece of data in the first data, and X is a positive integer greater than 2; the second type characteristic parameter is used for representing the data classification identified by the attribute of each piece of data of the first data; and respectively associating the orientation conditions identified by the X split data segments with the data classification to obtain X association results, and determining the obtained X association results as the correlation processing results.

In a practical application of the information processing method of the embodiment of the present invention, detecting whether the first data sent by the first terminal generates effective feedback at the second terminal to obtain a detection result includes: when detecting that the first data sent by the first terminal generates effective feedback at the second terminal, recording the detection result as the conversion rate obtained through feedback; and when detecting that the first data sent by the first terminal does not generate effective feedback at the second terminal, recording the detection result as that the conversion rate is not obtained through feedback.

In an actual application of the information processing method according to the embodiment of the present invention, the generation of the target data with the classification attribute according to the detection result and the correlation processing result includes two cases, 1): when the detection result is that the conversion rate is obtained through feedback, matching corresponding first-class training data (such as regular data) from the correlation processing result according to the detection result; 2) and matching corresponding second type training data (such as negative example data) from the correlation processing result according to the detection result when the conversion rate is not obtained through feedback. Subsequently, the target data with the classification attribute may be generated according to the first class of training data and the second class of training data.

In a practical application of the information processing method of the embodiment of the present invention, generating an information-oriented recommendation policy including at least two oriented condition ranking results according to the data analysis model, and feeding back the information-oriented recommendation policy to the first terminal includes: generating a conversion rate data model for analyzing a conversion rate as the data analysis model according to the target data and the weight value related to the target data; and estimating the sequencing results of at least two orientation conditions according to the conversion rate data model, packaging the sequencing results of the at least two orientation conditions into the information orientation recommendation strategy, and feeding back to the first terminal, so that the first terminal can select a target orientation condition corresponding to the self requirement from the sequencing results of the at least two orientation conditions according to the information orientation recommendation strategy.

Example four:

as shown in fig. 4, the information interaction system of the embodiment of the present invention includes: a server 41 (e.g., server 11-1n in fig. 1), a first terminal 42 (e.g., 31-3n in fig. 1), and a second terminal 43 (e.g., terminals 21-24 in fig. 1), as shown in fig. 4, wherein the server 41 includes: an acquisition unit 411 configured to acquire first data from a data source; the first processing unit 412 is configured to extract at least two types of feature parameters that meet a preset rule from the first data, and establish a correlation between the at least two types of feature parameters to obtain a correlation processing result; and a detecting unit 413, configured to detect whether the first data sent by the first terminal generates valid feedback at the second terminal, so as to obtain a detection result; and a second processing unit 414, configured to generate target data according to the detection result and the correlation processing result; and a directional recommendation unit 415, configured to generate a data analysis model according to the target data, generate an information directional recommendation policy including at least two directional condition ordering results according to the data analysis model, and feed back the information directional recommendation policy to the first terminal.

The first terminal 42 includes: the analysis unit 421 is configured to trigger, at the first terminal, a selection operation for an orientation function of information transmission, and analyze the selection operation to obtain an information classification result; and an information pushing unit 422, configured to push information (such as advertisement information) according to the ranking result of the targeting conditions, where the ranking result matches the targeting conditions of its own classification from the targeting recommendation policy according to the information classification result.

The second terminal 43 includes: a display unit 431, configured to receive information (e.g., advertisement information) pushed according to the sorting result of the targeting condition from the first terminal, and display the information (e.g., advertisement information); and a feedback response unit 432, configured to, after displaying the information (e.g., advertisement information), perform a user operation according to the information (e.g., advertisement information), for example, purchase a recommended product in the advertisement information, and return a feedback response result after performing the user operation to the first terminal 42 and/or the server 41 (e.g., a current advertisement analysis processing platform) for processing.

In a practical application to which the embodiment of the present invention is applied, the acquired first data at least includes history data, and is not limited to history data, and may be a data set composed of history data and real-time update data. The data source is at least from a first terminal (such as a terminal where an advertiser is located or an object providing advertisement materials and content promotion) and/or a second terminal (such as a terminal where an ordinary user is located or an object providing advertisement display or exposure).

In a practical application applying the embodiment of the present invention, the at least two types of characteristic parameters include, but are not limited to, a targeting condition and a data classification, taking advertisement information as an example, the first type of characteristic parameter is a targeting pushing condition for selecting advertisement information to be pushed to a target information receiving group, and the second type of characteristic parameter is an information classification (or referred to as an information field or category) to which the advertisement information belongs. Establishing correlation between the at least two types of characteristic parameters to obtain a correlation processing result, wherein the technical effect to be achieved is as follows: taking advertisement information as an example, information cross processing is performed between the targeted push condition for pushing the advertisement information to the target information receiving group and the information classification (or referred to as information field or category) to which the advertisement information belongs, so that the two have intersection, and the result after information cross combination is more favorable for accurately describing the targeted recommendation strategy generated in the targeted recommendation unit. Another example is: not only the intersection of information generation, but also the union of information generation is possible, and in short, the comprehensive information composed of the two types of information is required.

In a practical application applying the embodiment of the present invention, a first terminal (e.g. a terminal where an advertiser is located, or called an object providing advertisement material and content promotion) is used as an information sending party, it is required to ensure that a targeting policy for selecting information sending can ensure accurate positioning, a second terminal (e.g. a terminal where an ordinary user is located, or called an object displaying or exposing an advertisement) is used as an information receiving party, it is required to ensure that an effective feedback is generated, and of course, an invalid feedback is also collected, because a data analysis model (e.g. a conversion rate data model for analyzing a conversion rate) is subsequently generated to obtain a targeting recommendation policy through final calculation, in data analysis, data of both positive case data (obtaining an effective feedback response, e.g. generating a conversion rate) and negative case data (obtaining no effective feedback response, e.g. not generating a conversion rate) need to be collected, for forward feedback (effective feedback), the positive data is for reverse feedback (ineffective feedback), so that the final analysis result is accurate only by comprehensively considering the two factors, and it can be ensured that the directional recommendation strategy generated in the directional recommendation unit is accurate enough.

In a practical application applying the embodiment of the present invention, as two results of positive case data for forward feedback (effective feedback) and negative case data for reverse feedback (ineffective feedback), the data obtained in step 102 is combined: the method includes the steps of obtaining comprehensive data by associating (or called information crossing) at least two types of characteristic parameters or combining (or called orientation condition and data classification) at least two types of characteristic parameters, generating training data formed by positive example data and negative example data as the training data, generating a data analysis model (such as a conversion rate data model for analyzing a conversion rate) according to the target data as the target data, and generating an information orientation recommendation strategy containing at least two orientation condition sequencing results according to the data analysis model to be accurate enough so as to ensure that the orientation recommendation strategy fed back to a first terminal (such as a terminal where an advertiser is located or an object providing advertisement materials and content promotion) is accurate enough, so that the first terminal (such as a terminal where the advertiser is located or an object providing the advertisement materials and content promotion) pushes information (such as advertisement information) to a second terminal (or called advertisement information) according to the orientation recommendation strategy Such as a terminal where an ordinary user is located, or called an advertisement display or exposure object), can also be accurately located, and information is accurately pushed.

By applying the information interaction system provided by the embodiment of the invention, after historical data are collected and training data (training data consisting of positive examples and negative examples) which are most suitable for analysis are generated by analysis from the historical data, a conversion rate data model for analyzing the conversion rate can be generated according to the training data (training data consisting of positive examples and negative examples) and the weight values related to the training data (training data consisting of positive examples and negative examples), the sequencing results of at least two orientation conditions are estimated according to the conversion rate data model, and the sequencing results of the at least two orientation conditions are encapsulated in the information orientation recommendation strategy and fed back to the first terminal. The first terminal selects a target orientation condition corresponding to the self requirement from the sequencing results of the at least two orientation conditions according to the information orientation recommendation strategy, the information pushed to the second terminal (or called information receiving group) by the first terminal is targeted, the requirement of the second terminal (or called information receiving group) is met, the positioning is accurate, and the targeted recommendation strategy can be further optimized through the feedback of the second terminal (or called information receiving group), such as the conversion rate formed by the user behavior of the second terminal (or called information receiving group), the optimization of the targeted recommendation strategy can be performed at the first terminal, or the targeted recommendation strategy can be optimized by a server (such as an advertisement analysis processing platform of a third party) after the targeted recommendation strategy is provided to the server (such as an advertisement analysis processing platform of the third party) by the first terminal.

In a practical application of the embodiment of the present invention, the server further includes: the data conversion unit is used for converting the first data into second data in a data format supported by the data conversion unit after the first data is acquired from a data source. And a first processing unit, further configured to: extracting a data decomposition condition, and extracting a first class characteristic parameter and a second class characteristic parameter which accord with a preset rule from the first data according to the data decomposition condition; the first class of characteristic parameters are used for characterizing the orientation condition identified by X split data segments, wherein the X split data segments are derived from the content of each piece of data in the first data, and X is a positive integer greater than 2; the second type characteristic parameter is used for representing the data classification identified by the attribute of each piece of data of the first data; and respectively associating the orientation conditions identified by the X split data segments with the data classification to obtain X association results, and determining the obtained X association results as the correlation processing results. And, the detection unit is further configured to: when detecting that the first data sent by the first terminal generates effective feedback at the second terminal, recording the detection result as the conversion rate obtained through feedback; and when detecting that the first data sent by the first terminal does not generate effective feedback at the second terminal, recording the detection result as that the conversion rate is not obtained through feedback.

In a practical application to which the embodiment of the present invention is applied, the second processing unit is further configured to: when the detection result is that the conversion rate is obtained through feedback, matching corresponding first-class training data from the correlation processing result according to the detection result; matching corresponding second type training data from the correlation processing result according to the detection result when the conversion rate is not obtained through feedback; the target data with the classification attribute can be generated according to the first class of training data and the second class of training data. And the directional recommendation unit is further used for: generating a conversion rate data model for analyzing a conversion rate as the data analysis model according to the target data and the weight value related to the target data; and estimating the sequencing results of at least two orientation conditions according to the conversion rate data model, packaging the sequencing results of the at least two orientation conditions into the information orientation recommendation strategy, and feeding back to the first terminal, so that the first terminal can select a target orientation condition corresponding to the self requirement from the sequencing results of the at least two orientation conditions according to the information orientation recommendation strategy.

Example five:

an information processing method according to an embodiment of the present invention includes: the first terminal receives a first operation (for example, selecting a directional function for information transmission on a terminal user interface of the first terminal), and in response to the first operation, the first terminal (for example, a terminal where an advertiser is located) can be classified by using a classifier to obtain which field or category the information classification of the first terminal user (for example, the advertiser) belongs to, for example, e-commerce, games or women's clothing, and the like, and further subdivide the types of the fields or categories such as e-commerce, games or women's clothing, e-commerce-3C-phone, and the like. Thereafter, the information classification (e.g., advertisement information classification including dress-dress, e-business-3C-phone) result is outputted to the first terminal user (e.g., advertiser) operating the first terminal, as shown in the upper right legend of fig. 5: selecting a directional function of information transmission for a user at a terminal user interface of the first terminal to trigger a selection operation, wherein A11 is used for identifying the terminal user interface of the first terminal, and a1 is used for identifying a directional function option of information transmission in the terminal user interface of the first terminal. Legend to the bottom right as shown in fig. 5: after the selection operation is triggered, a classifier is used for classifying a first terminal (such as a terminal where an advertiser is located) to obtain a schematic diagram of which field or category the information classification of a first terminal user (such as the advertiser) belongs to, wherein a11 is used for identifying a terminal user interface of the first terminal, and b1-b3 are respectively used for identifying options of the information classification in the terminal user interface of the first terminal, which belong to a specific field or category, such as an e-commerce, a game or a suit-dress, and further, the suit-dress can be further subdivided, such as options of a dress field or category under the suit-dress identified by b31, and the like.

After a first terminal user (such as an advertiser) operating the first terminal sees a displayed information classification (such as an advertisement information classification including dress-dress and E-business-3C-phone) result in a terminal user interface, the first terminal user selects a directional condition for information transmission according to the consideration of the first terminal user for information (such as advertisement information) transmission and/or according to a directional recommendation strategy pushed by a server. The directional condition of the information transmission is contained in an information directional recommendation strategy generated by the server and pushed to the first terminal, and the information directional recommendation strategy contains sequencing results of at least two directional conditions (such as sequencing results corresponding to the dress-dress and the e-business-3C-phone respectively).

The first terminal performs graphical processing on the directional recommendation policy received from the server to generate a graphical interface, as shown in fig. 6, the graphical interface is displayed on the terminal user interface of the first terminal, and as can be seen from fig. 6: the terminal user interface of the first terminal, which is identified by a11, has a plurality of sorted results of the orientation condition, including a sequence of a plurality of sorted results of the orientation conditions 11-16. It is also possible to include a check box before each orientation condition so that the user can check one orientation condition in the sequence of the plurality of ranking results through the check box, and fig. 6 shows that the user checks the ranking result as the "orientation condition 11" at the first priority.

It should be noted that 1) the server will send the ranking results of all the targeting conditions in the targeting recommendation policy to the first terminal, and for this case, it is also necessary to perform filtering of a plurality of targeting conditions in the targeting recommendation strategy at the first terminal according to the above information classification (e.g. advertisement information classification including dress-dress, e-business-3C-phone), so as to match the ordering result which accords with the self-classified orientation condition from the orientation recommendation strategy according to the information classification result and record the ordering result as an orientation ordering result A, only the directional sequencing result A is graphically processed, the obtained graphical interface is displayed on a terminal user interface of the first terminal, as shown in fig. 7, the orientation sorting result a includes an orientation condition 21, an orientation condition 22, an orientation condition 23, an orientation condition 24, and an orientation condition 25. 2) The server firstly classifies and matches the sequencing results of all the orientation conditions in the orientation recommendation strategy according to different first terminals, obtains a sequencing result meeting the specified orientation condition of the current first terminal, records the sequencing result as an orientation sequencing result B, sends the orientation sequencing result A to the corresponding current first terminal, and the current first terminal carries out graphical processing on the orientation sequencing result B and displays the obtained graphical interface on a terminal user interface of the first terminal, as shown in fig. 8, wherein the orientation sequencing result B comprises an orientation condition 31, an orientation condition 32 and an orientation condition 33. Wherein, the server filters a plurality of targeting conditions in the targeting recommendation strategy by actively collecting or acquiring the information classification (such as advertisement information classification including women's dress-one-piece dress, and e-business-3C-phone) results reported by different first terminals, so as to match the ranking results of the targeting conditions according to the information classification results, thereby realizing the classification and matching of the ranking results of all the targeting conditions in the targeting recommendation strategy according to different first terminals, and then, at the first terminal side, 1) the first terminal user performs selection operation according to the ranking results of the targeting conditions in the interface graphical interface of the terminal user, so as to select a closest targeting condition from the ranking results of the targeting conditions, as shown in fig. 7, a black re-selection frame option is printed on the ranking results, sending information to an information receiving group (such as a game user group, a shopping user group, a video user group and the like) matched with the orientation condition according to the selected orientation condition, wherein the selection operation is a single selection operation; 2) the first end user may select a plurality of closest directional conditions from the directional condition ranking results when selecting according to the directional condition ranking results in the interface graphical interface of the end user, as shown in fig. 8, a plurality of black re-selection box options are printed on the directional condition ranking results, and information is sent to an information receiving group (such as a game user group, a shopping user group, a video user group and the like) matched with the directional conditions according to the selected plurality of directional conditions, wherein the selecting operation is a multi-level selecting operation, and the directional accuracy is higher.

In an embodiment of the present invention, at the first terminal side, a first terminal user performs various single or multi-level selection operations according to a directional condition sorting result in an interface graphical interface of the terminal user, and after selecting a directional condition that best meets the self requirement from the directional condition sorting result, the first terminal user (advertiser) may adjust a recommendation range for an information receiving group on the basis of the selected directional condition, for example, the first terminal user (advertiser) may select to increase accuracy or increase coverage of a group on the basis of the directional condition, and by selecting to increase accuracy, a feature with a higher rank may be selected for accurate information positioning and pushing; by increasing the coverage population, more features can be selected, and the method can also be used for accurate information positioning and pushing.

Example six:

fig. 9 is a schematic diagram of information interaction between the terminal side and the server side (as shown in steps 401 and 409): the method comprises the steps that a server acquires first data from a data source, extracts at least two types of characteristic parameters which accord with a preset rule from the first data, and establishes correlation between the at least two types of characteristic parameters to obtain a correlation processing result; the server detects whether the first data sent by the first terminal generates effective feedback at the second terminal, if the advertisement information is exposed at the second terminal user (such as a shopping user group), and the shopping user group generates a specified user behavior (such as a purchasing behavior) based on the advertisement information, the advertisement information recommended by the first terminal user (advertiser) can be converted through the purchasing behavior, namely: converting into a conversion rate containing advertiser profits so as to obtain a detection result; the server generates target data with classification attributes (for example, data composed of first class training data and second class training data may be referred to as target data) according to the detection result and the correlation processing result; the server generates a data analysis model (such as a conversion rate data model for analyzing the conversion rate) according to the target data, generates an information oriented recommendation strategy containing at least two oriented condition sequencing results according to the data analysis model (such as the conversion rate data model for analyzing the conversion rate), and feeds the information oriented recommendation strategy back to the first terminal. The method comprises the steps that a first terminal receives a directional recommendation strategy pushed by a server, or the first terminal initiates a request to the server to obtain the directional recommendation strategy, then, a selection operation triggered by a directional recommendation function sent by a first terminal user selection information is obtained, after the selection operation is responded, the first terminal firstly uses a classifier to classify information of the first terminal (such as a terminal where an advertiser is located) to obtain which field or category the information classification of the first terminal user (such as the advertiser) belongs to, such as e-commerce, games or women's clothing, and further carries out type subdivision on the fields or categories of e-commerce, games or women's clothing, such as women's clothing-one-piece dress, e-commerce-3C-phone, and the like, so as to obtain an information classification result; matching a sorting result meeting the self-classified orientation condition from the orientation recommendation strategy according to the information classification result, and marking as an orientation sorting result C, obtaining a final target orientation condition (such as the orientation condition which best meets the self-requirement of the user) aiming at one-time clicking selection operation of a first terminal user (such as an advertiser) on one orientation condition in the orientation sorting result C or multiple-time clicking selection operation of the first terminal user (such as the advertiser) on multiple orientation conditions in the orientation sorting result C, and sending information to a designated information receiving group (such as a game user group, a shopping user group, a video user group and the like) by utilizing the target orientation condition (such as the orientation condition which best meets the self-requirement of the user).

Since the first terminal can select the most suitable target orientation condition (the target orientation condition which best meets the self requirement) from the sequencing results of the plurality of orientation conditions according to the information orientation recommendation strategy, for example, the first terminal sends information to an information receiving group (such as a game user group, a shopping user group, a video user group and the like) matched with the orientation condition according to the selected target orientation condition (such as the orientation condition which best meets the self requirement of the user), so that the terminal user operating the first terminal can select the first orientation condition and/or the second orientation condition which is sequenced at the top to send the information, and the information sending precision is improved.

In an embodiment of the present invention, at a server side, according to an extracted data decomposition condition, a first class feature parameter (e.g., a targeting condition) and a second class feature parameter (e.g., an advertisement information classification in an information classification) that meet a preset rule may be extracted from first data (e.g., historical data) according to the data decomposition condition; the first class of characteristic parameters are used for characterizing the orientation condition identified by X split data segments, wherein the X split data segments are derived from the content of each piece of data in the first data, and X is a positive integer greater than 2; the second type characteristic parameters are used for characterizing data classification (or called information classification, such as e-commerce, games or women's clothing, and the like) identified by the attribute of each piece of data of the first data, and further performing type subdivision on fields or categories of e-commerce, games or women's clothing, and the like, such as women's clothing-one-piece dress, e-commerce-3C-phone, and the like); and respectively associating the orientation conditions identified by the X split data segments with the data classification to obtain X association results, and determining the obtained X association results as the correlation processing results. Here, taking as an example that a piece of advertisement information "a mobile phone suitable for a girl of 18 years is newly introduced in beijing area, a user purchases or experiences activities with a mobile phone brand special cabinet of a certain business building in zhongguan province in a certain month in a certain year or a user purchases or experiences activities with a certain e business platform in a promotion period in a certain month in a certain year" as an example, a plurality of correlation results (beijing area +3C commodity, 18 years +3C commodity, girl +3C commodity, certain month in a certain year, zhongguan province, mobile phone brand special cabinet of a certain business platform, purchasing a mobile phone) are obtained based on a plurality of targeting conditions (the mobile phone belongs to the 3C commodity) obtained by splitting the piece of advertisement information and information classification (the mobile phone belongs to the 3C commodity) corresponding to the piece of advertisement information, since the behavior that the girl finally purchases by the piece of advertisement information is determined as successful, then, a conversion rate is generated by the advertisement, so that the above-mentioned associated result in the piece of advertisement information whose purchasing behavior is determined as successful purchase can be used as a positive example in the training data, and if the final purchasing behavior of the girl is determined as unsuccessful purchase by the piece of advertisement information, the above-mentioned associated result in the piece of advertisement information whose purchasing behavior is determined as unsuccessful purchase can be used as a negative example in the training data. It can be seen that: the method comprises the steps of collecting historical data, analyzing and generating training data (training data consisting of positive examples and negative examples) which are most suitable for analysis from the historical data, generating a conversion rate data model for analyzing the conversion rate according to the training data (training data consisting of positive examples and negative examples) and weight values related to the training data (training data consisting of positive examples and negative examples), estimating sequencing results of at least two orientation conditions according to the conversion rate data model, packaging the sequencing results of the at least two orientation conditions into an information orientation recommendation strategy, and feeding back the information orientation recommendation strategy to a first terminal, so that the first terminal can select a target orientation condition corresponding to self requirements from the sequencing results of the at least two orientation conditions according to the information orientation recommendation strategy. On the first terminal side, since the obtained targeted recommendation strategy is based on the first class of characteristic parameters (such as targeting conditions) and the second class of characteristic parameters (such as advertisement information classification in the information classification) in the initial operation analysis, because the difference requirements of a plurality of first terminals (each first terminal is oriented to a different advertiser, and different advertisers have different advertisement putting requirements), then, on the first terminal side, after obtaining the information classification result, the ranking result meeting the targeting conditions of the self classification is matched from the targeted recommendation strategy according to the information classification result, so as to push information (such as advertisement information) according to the ranking result of the targeting conditions, so that the information pushed to the information receiving population is targeted and meets the requirements of the information receiving population, not only the positioning is accurate, but also the feedback of the information receiving population can be realized (such as the conversion rate is formed by the user behaviors of the information receiving population) The targeted recommendation strategy is further optimized, and the optimization of the targeted recommendation strategy can be performed at the first terminal, or the targeted recommendation strategy can be optimized by a server (such as an advertisement analysis processing platform of a third party) after the targeted recommendation strategy is provided to the server (such as an advertisement analysis processing platform of the third party) by the first terminal.

Example seven:

it should be noted that the terminal devices (such as the first terminal and the second terminal) may be electronic devices such as a PC, portable electronic devices such as a PAD, a tablet computer and a laptop, and may also be smart mobile terminals such as a mobile phone, and are not limited to the description herein; the server may be an electronic device formed by a cluster system, and integrated into one or a plurality of unit functions to implement the unit functions, and both the client and the server at least include a database for storing data and a processor for data processing, or include a storage medium arranged in the server or a storage medium arranged independently.

As for the Processor for data Processing, when executing Processing, the Processor can be implemented by a microprocessor, a Central Processing Unit (CPU), a Digital Signal Processor (DSP), or a Programmable logic Array (FPGA); for the storage medium, the storage medium contains operation instructions, which may be computer executable codes, and the operation instructions implement the steps in the flow of the information processing method according to the above-described embodiment of the present invention.

An example of the terminal device (e.g., the first terminal, the second terminal) and the server as hardware entities S11 is shown in fig. 10. The apparatus comprises a processor 61, a storage medium 62 and at least one external communication interface 63; the processor 61, the storage medium 62 and the external communication interface 63 are all connected by a bus 64.

Here, it should be noted that: the above description related to the terminal device (e.g., the first terminal and the second terminal) and the server item is similar to the above description of the method, and the description of the beneficial effects of the method is omitted for brevity. For technical details not disclosed in the embodiments of the client and server of the present invention, refer to the description of the embodiments of the method of the present invention.

The embodiment of the invention is explained by taking a practical application scene as an example as follows:

the application scene is as follows: the information is advertisement information, and is not limited to game advertisements, entertainment advertisements, shopping advertisements, daily chemical advertisements and the like. Description of technical terms referred to herein: 1) and (3) advertisement targeting indication: each advertisement is tagged with multiple tags, such as: "male", "25-30 years" etc., which serve as targeting conditions for advertisements that are retrieved in the advertising system. For example: "male & Beijing area" means the group who advertises to "Beijing area & male". 2) The conversion rate estimation means: and training a conversion rate model through historical data for predicting the conversion rate of the on-line flow. 3) pCVR means: estimated conversion. 4) pCVR pipeline refers to: a channel of training data is generated. 5) The throwing-in end indicates: the advertiser delivers the front end of the advertisement.

For the above application scenario, in the related art, in the advertisement analysis processing system, when an advertiser puts an advertisement, it selects targeting conditions, such as shenzhen, 20-30 years old, women, etc., which are very many in combination, which may cause two problems: new advertisers seeing so many targeting conditions may not have a choice; the advertiser's prior knowledge is not necessarily very accurate and the deep meaning behind the data may be more meaningful. In summary, it is the first terminal side advertiser that has no choice of targeting conditions, or that the targeting conditions chosen are not precise. In response to this problem, analysis has revealed that a large portion of advertisers are very lost in the early stages of advertisement delivery, and they are not aware that their advertisements should be delivered to those groups of people, in which case the advertisers need to make numerous adjustments over a long period of time to achieve stable results, and such adjustments and attempts waste much of the advertiser's budget and are not unreasonable. It is desirable to provide advertisers with a method of referral targeting. Under the condition that the advertisement cost is certain, compared with other indexes of the advertisement, the advertiser is more aware of the self advertisement effect, namely the conversion rate, and the conversion rate is high, so that the fact that the advertisement information which is directionally pushed by the first terminal side advertiser by the second terminal meets the self requirement is shown, and the fact that the directional condition selected by the first terminal side advertiser is accurate can also be shown, therefore, the application scene adopts the embodiment of the invention, and the directional condition of information pushing can be estimated based on the conversion rate model.

The application scene adopts the embodiment of the invention and is divided into the following online part and offline part. The following operation logic is specifically divided:

firstly, an off-line module: an off-line module is adopted for training out pCVR (pre-estimated conversion rate) model for use by a delivery end, and the model can give out weight values of all characteristics.

a) Historical data (logs) on the ad analytics processing platform, such as data that used a history of 14 days, is collected.

b) The training data is generated through a pCVR pipeline, namely a channel for generating training data, and the pCVR pipeline is used for converting the first data into the second data in the embodiment of the invention so as to obtain the second data in a data format supported by the advertisement analysis and processing platform. Wherein, the training data contains all the characteristics of the intersection of the targeting condition and the advertisement classification, the positive case of the training data is that the advertisement generates the conversion behavior (such as the activation behavior of APP), and the negative case is that the advertisement does not generate the conversion behavior; the format of the training data is: "label (1 identifies positive case, 0 identifies negative case) feature 1 feature 2 feature 3 …".

For example: an advertisement scenario is as follows: at 1 am, an 18 year old male user browses an advertisement for selling mobile phones and generates purchasing behavior; the advertising system records this data and generates the following training data by pCVR pipeline:

1gender(male)*EC(3c_phone)age(18)*EC(3c_phone)time(1)*EC(3c_phone)

specifically, the process of generating the training data is as follows:

since the advertisement information of the data is a mobile phone advertisement, and the mobile phone advertisement belongs to a 3C product of an e-commerce type (EC), the advertisement is classified as EC (3C _ phone), and all information that can be used as a target in the scene is extracted after being crossed with the advertisement classification, as shown in fig. 11:

c) the conversion rate model is trained using a trainer (LR, FM algorithm, etc.), which includes feature values and feature weights, as shown in table 1.

Features extracted from training data	Weight value corresponding to feature
		gender(male)*EC(3c_phone)	0.478
age(18)*EC(3c_phone)	0.112
		time(1)*EC(3c_phone)	-0.564
gender(male)*APP(game_casual)	0.221
		……	……

TABLE 1

Secondly, an advertisement delivery end: and adding a function of intelligently recommending the targeting conditions to the advertiser on the advertisement putting end.

a) Advertiser classification: when a new advertiser selects the recommendation targeting function, the advertiser is first classified using a classifier.

Using a tool: existing classifiers in the system use intra-department classification systems, such as e-commerce-suit-dress, e-commerce-3C-phone; the input of the classifier is the advertisement title, advertisement description, etc., and the output is the classification of the advertisement.

b) Selecting orientation conditions: and filtering out all the features with the classification in the model, for example, the classification of the advertisement is EC (3c _ phone), filtering out the features marked as all the targeting conditions EC (3c _ phone) in the model, and sorting according to the weight of the features in the model, wherein the features with the top weight of 50% are selected by default, and the features are the selected targeting conditions.

c) Adjusting the recommended range: the advertiser can choose to increase the accuracy or increase the coverage population on the basis, the features with more advanced ranks can be selected if the accuracy is increased, and more features can be selected if the coverage population is increased.

FIG. 12 shows an example of an ad targeting recommendation model based on a conversion rate estimation model, which includes: the system comprises an advertisement delivery end, pCVR pipeline, a training data generator, a Model generator (Model) and a sequencing result generator (Mixer). The advertisement delivery end is positioned at the first terminal side and can be used as a data source, so that the server side (such as an advertisement analysis processing platform) obtains first data from the data source for analysis; the pCVR pipeline is a channel for generating training data, and can also convert first data obtained from historical data (the historical data can be stored in a log form) into second data in a data format supported by a server side (such as an advertisement analysis processing platform); generating training data according to the second data and the training data generator, inputting the training data into a Model generator (Model), such as a pCVR Model generator, to obtain a directional recommendation strategy for information transmission based on the estimated conversion rate Model, wherein the directional recommendation strategy comprises at least one directional condition, and the at least one directional condition can be displayed by a sequencing result generator (Mixer).

In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described device embodiments are merely illustrative, for example, the division of the unit is only a logical functional division, and there may be other division ways in actual implementation, such as: multiple units or components may be combined, or may be integrated into another system, or some features may be omitted, or not implemented. In addition, the coupling, direct coupling or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or units may be electrical, mechanical or other forms.

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

In addition, all the functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately regarded as one unit, or two or more units may be integrated into one unit; the integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.

Those of ordinary skill in the art will understand that: all or part of the steps for implementing the method embodiments may be implemented by hardware related to program instructions, and the program may be stored in a computer readable storage medium, and when executed, the program performs the steps including the method embodiments; and the aforementioned storage medium includes: a mobile storage device, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

Alternatively, the integrated unit of the present invention may be stored in a computer-readable storage medium if it is implemented in the form of a software functional module and sold or used as a separate product. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or a part contributing to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, a ROM, a RAM, a magnetic or optical disk, or various other media that can store program code.

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

Claims

1. An information processing method, characterized in that the method comprises:

obtaining first data from a data source;

extracting a plurality of first class characteristic parameters which accord with preset rules and correspond to the orientation conditions and second class characteristic parameters which correspond to the data classification of the first data from the first data;

respectively carrying out cross combination processing on the first type characteristic parameters and the second type characteristic parameters, and taking the obtained cross-combined characteristics as correlation processing results;

generating a data analysis model for analyzing the conversion rate according to the target data;

sorting the correlation processing results according to the weight corresponding to the correlation processing results in the data analysis model to generate an information directional recommendation strategy;

the information directional recommendation strategy is packaged with a sequencing result of a plurality of correlation processing results; the correlation processing result comprises the data classification and at least one orientation condition;

and feeding back the information directional recommendation strategy to the first terminal so that the first terminal matches at least one directional condition corresponding to the data classification of the information from the multiple correlation processing results.

2. The method of claim 1, further comprising: after first data are acquired from a data source, the first data are converted into second data in a data format supported by the first data.

3. The method of claim 1,

the extracting, from the first data, a plurality of first class feature parameters that meet a preset rule and correspond to a directional condition, and a second class feature parameter that corresponds to a data classification of the first data includes:

the cross-combining the first-class characteristic parameters and the second-class characteristic parameters respectively, and taking the obtained cross-combined characteristic as a correlation processing result includes:

4. The method of claim 3, wherein the detecting whether the first data sent by the first terminal generates effective feedback at the second terminal, and obtaining a detection result comprises:

5. The method of claim 4, wherein generating target data from the detection results and the correlation processing results comprises:

and generating target data with classification attributes according to the first class of training data and the second class of training data.

6. The method according to any one of claims 1 to 5,

the generating a data analysis model for analyzing conversion rate according to the target data comprises:

and generating a conversion rate data model for analyzing the conversion rate according to the target data and the weight value related to the target data, wherein the conversion rate data model is used as the data analysis model.

7. A server, characterized in that the server comprises:

an acquisition unit configured to acquire first data from a data source;

the first processing unit is used for extracting a plurality of first-class characteristic parameters which accord with preset rules and correspond to the orientation conditions and second-class characteristic parameters which correspond to the data classification of the first data from the first data; respectively carrying out cross combination processing on the first type characteristic parameters and the second type characteristic parameters, and taking the obtained cross-combined characteristics as correlation processing results;

the directional recommendation unit is used for generating a data analysis model for analyzing the conversion rate according to the target data; sorting the correlation processing results according to the weight corresponding to the correlation processing results in the data analysis model to generate an information directional recommendation strategy; the information directional recommendation strategy is packaged with a sequencing result of a plurality of correlation processing results; the correlation processing result comprises the data classification and at least one orientation condition; and feeding back the information directional recommendation strategy to the first terminal so that the first terminal matches at least one directional condition corresponding to the data classification of the information from the multiple correlation processing results.

8. The server of claim 7, further comprising:

9. The server according to claim 7, wherein the first processing unit is further configured to:

10. The server according to claim 9, wherein the detecting unit is further configured to:

11. The server according to claim 10, wherein the second processing unit is further configured to:

12. The server according to any one of claims 7 to 11, wherein the directional recommendation unit is further configured to:

13. A computer-readable storage medium storing executable instructions for implementing the information processing method according to any one of claims 1 to 6 when executed by a processor.

14. An electronic device, comprising:

a memory for storing executable instructions;

a processor for implementing the information processing method of any one of claims 1 to 6 when executing the executable instructions stored in the memory.