CN111191141B

CN111191141B - Text recommendation method and device

Info

Publication number: CN111191141B
Application number: CN202010010667.5A
Authority: CN
Inventors: 胡丁相; 操颖平
Original assignee: Ant Shengxin Shanghai Information Technology Co ltd
Current assignee: Ant Shengxin Shanghai Information Technology Co ltd
Priority date: 2020-01-06
Filing date: 2020-01-06
Publication date: 2023-05-23
Anticipated expiration: 2040-01-06
Also published as: CN111191141A

Abstract

The specification provides a document recommendation method and a device, wherein the document recommendation method comprises the following steps: determining a user tag of a user and an item tag of a business item according to attribute data of the user browsing the business item; inputting the user tag and the item tag into a screening model to obtain multimedia data and a subject format which are output by the screening model and matched with the user; extracting material data matched with the user from a material database pre-established in the business project, and processing the material data according to the material format to obtain material data; and generating a document based on the subject data and the multimedia data, and recommending the document to the user through the display page of the business item.

Description

Text recommendation method and device

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to a document recommendation method. The present description is also directed to a document recommendation apparatus, a computing device, and a computer-readable storage medium.

Background

With the development of internet technology, various internet services have been developed, and in order to widely apply the internet services, it is necessary to continuously popularize the internet services, and the popularization efficiency is improved by means of advertisement and the like, so as to further realize the wide application of the internet services.

However, in the process of recommending the promotion information such as the advertisement to the user, the promotion information such as the advertisement is usually selected according to the attribute information of the user and the current service scene to recommend the promotion information to the user, and the promotion information is usually generated by adopting a piecewise manner, so that the promotion information finally shown to the user may not be in agreement with the intention of the user, further the effect of improving the promotion efficiency cannot be realized, and the operation flow of the process is complex, and the efficiency and the effect are both low, so that a simpler method is needed to realize the generation of the promotion information so as to achieve the effect of high-efficiency promotion.

Disclosure of Invention

In view of this, the present embodiments provide a document recommendation method. The present disclosure also relates to a document recommendation apparatus, a computing device, and a computer-readable storage medium, which solve the technical drawbacks of the prior art.

According to a first aspect of embodiments of the present disclosure, there is provided a document recommendation method, including:

determining a user tag of a user and an item tag of a business item according to attribute data of the user browsing the business item;

inputting the user tag and the item tag into a screening model to obtain multimedia data and a subject format which are output by the screening model and matched with the user;

Extracting material data matched with the user from a material database pre-established in the business project, and processing the material data according to the material format to obtain material data;

and generating a document based on the subject data and the multimedia data, and recommending the document to the user through the display page of the business item.

Optionally, the determining the user tag of the user and the item tag of the service item according to the attribute data of the user browsing the service item includes:

determining a user tag set formed by a first tag corresponding to the user according to the attribute data, and determining an item tag set formed by a second tag corresponding to the service item according to the item data of the service item;

detecting historical user conversion rate corresponding to a first label contained in the user label set and historical item conversion rate corresponding to a second label contained in the item label set;

and selecting the first label with the highest historical user conversion rate as the user label of the user, and selecting the second label with the highest historical item conversion rate as the item label of the business item.

Optionally, the processing the material data according to the material format to obtain material data includes:

determining a text size format, a text color format and a sentence pattern according to the subject format;

performing word size processing on word data in the material data according to the word size format to obtain first material data;

performing text color processing on the first material data according to the text color format to obtain second material data;

and carrying out sentence processing on the second material data according to the sentence pattern to obtain target subject data, and taking the target subject data as the subject data.

Optionally, the generating a document based on the subject data and the multimedia data includes:

selecting a distribution template from a distribution template library pre-established in the business project according to the multimedia data;

determining a subject distribution area corresponding to the subject data based on the distribution template, and determining a multimedia distribution area corresponding to the multimedia data;

adjusting the subject data according to the subject distribution area, and adjusting the multimedia data according to the multimedia distribution area;

And carrying out assembling processing on the adjusted subject data and the adjusted multimedia data based on the distribution template to generate the document.

Optionally, the screening model is trained by:

collecting sample user labels of sample users participating in the business project and sample project labels with corresponding relations with the sample user labels;

constructing a sample tag group based on the sample user tag and a sample item tag corresponding to the sample user tag, and determining sample multimedia data and a sample topic format corresponding to the sample tag group;

and inputting the sample tag group and the sample multimedia data and the sample topic format corresponding to the sample tag group into a screening model constructed based on the correlation relationship between the sample tag group and the sample multimedia data and the sample topic format corresponding to the sample tag group for training, so as to obtain the screening model.

Optionally, after the step of generating the document based on the subject data and the multimedia data and recommending the document to the user through the display page of the service item is performed, the method further includes:

Obtaining the conversion rate corresponding to the text, and judging whether the conversion rate is greater than a conversion rate threshold value;

if yes, storing the text to a text library of the service item, and establishing an association relationship between the text, the subject format and the multimedia data;

if not, the multimedia data, the topic format, the user tag and the item tag are used as negative training samples to optimize the screening model.

Optionally, before the step of extracting the material data matched with the user from the material database pre-established in the service project and processing the material data according to the material format to obtain the material data is performed, the method further includes:

determining project users which successfully participate in the business project according to the historical browsing data of the business project;

determining an item user tag of the item user based on the attribute data of the item user, and sending the item user tag to a material data generation module;

and receiving the material data generated aiming at the project user label and returned by the material data generating module, and storing the material data into the material database.

Optionally, in the case that the multimedia data is sound data and picture data, the generating a document based on the subject data and the multimedia data includes:

generating image-text data based on the picture data and the subject data, fusing the sound data and the image-text data to obtain a target document, and taking the target document as the document.

Optionally, in the case that the multimedia data is picture data, the generating a document based on the subject data and the multimedia data includes:

and generating image-text resources based on the image data and the subject data, and taking the image-text resources as the document.

Optionally, in the case that the multimedia data is sound data, the generating a document based on the subject data and the multimedia data includes:

an audio resource is generated based on the sound data and the subject data, and the audio resource is used as the document.

Optionally, in the case that the multimedia data is video data, the generating a document based on the subject data and the multimedia data includes:

and generating video resources based on the video data and the subject data, and taking the video resources as the document.

Optionally, the recommending the document to the user through the display page of the business item includes:

acquiring browsing time of the user for browsing the display page of the service item;

and recommending the text to the user in a mode of displaying the text on a display window in the display page under the condition that the browsing time is larger than a preset time threshold.

According to a second aspect of embodiments of the present specification, there is provided a document recommendation apparatus, comprising:

a determining tag module configured to determine a user tag of a user browsing a business item and an item tag of the business item according to attribute data of the user;

the model screening module is configured to input the user tag and the item tag into a screening model to obtain multimedia data and a topic format which are output by the screening model and matched with the user;

the processing data module is configured to extract material data matched with the user from a material database pre-established in the business project, and process the material data according to the material format to obtain material data;

and a recommending file module configured to generate a file based on the subject data and the multimedia data and recommend the file to the user through the display page of the business item.

According to a third aspect of embodiments of the present specification, there is provided a computing device comprising:

a memory and a processor;

the memory is for storing computer-executable instructions, and the processor is for executing the computer-executable instructions:

According to a fourth aspect of embodiments of the present specification, there is provided a computer readable storage medium storing computer executable instructions which, when executed by a processor, implement the steps of the document recommendation method.

According to the embodiment of the description, in the process of browsing the business project by the user, the user tag of the user and the project tag of the business project are input into the screening model, multimedia data and the material format matched with the user are screened, the material data are extracted from the material database, the material data are processed according to the material format to obtain the material data, the text recommended to the user is generated based on the material data and the multimedia data, the text capable of effectively triggering the interest intention of the user is generated, the conversion rate of the user browsing the business project is further improved, the interpretability of the text recommendation process is higher, the process of generating the text for the user is conveniently analyzed, the generation efficiency of the text is effectively improved, and the experience effect of the user is further improved.

Drawings

FIG. 1 is a flowchart of a document recommendation method according to an embodiment of the present disclosure;

FIG. 2 is a schematic diagram of a document recommendation method according to an embodiment of the present disclosure;

FIG. 3 is a schematic diagram showing a page display document in a document recommendation method according to an embodiment of the present disclosure;

FIG. 4 is a process flow diagram of a document recommendation method applied to insurance items according to one embodiment of the present disclosure;

FIG. 5 is a schematic diagram of a document recommendation apparatus according to an embodiment of the present disclosure;

FIG. 6 is a block diagram of a computing device according to one embodiment of the present disclosure.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present description. This description may be embodied in many other forms than described herein and similarly generalized by those skilled in the art to whom this disclosure pertains without departing from the spirit of the disclosure and, therefore, this disclosure is not limited by the specific implementations disclosed below.

The terminology used in the one or more embodiments of the specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of the specification. As used in this specification, one or more embodiments and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used in one or more embodiments of the present specification refers to and encompasses any or all possible combinations of one or more of the associated listed items.

It should be understood that, although the terms first, second, etc. may be used in one or more embodiments of this specification to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first may also be referred to as a second, and similarly, a second may also be referred to as a first, without departing from the scope of one or more embodiments of the present description. The word "if" as used herein may be interpreted as "at … …" or "at … …" or "responsive to a determination", depending on the context.

In the present specification, a document recommendation method is provided, and the present specification relates to a document recommendation apparatus, a computing device, and a computer-readable storage medium, which are described in detail in the following embodiments one by one.

Fig. 1 shows a flowchart of a document recommending method according to an embodiment of the present disclosure, fig. 2 shows a schematic diagram of a document in a document recommending method according to an embodiment of the present disclosure, and fig. 3 shows a schematic diagram of a presentation page presentation document in a document recommending method according to an embodiment of the present disclosure, wherein fig. 1 specifically includes the following steps:

Step 102: and determining a user tag of the user and an item tag of the business item according to attribute data of the user browsing the business item.

In an embodiment of the present disclosure, the service item may be an item to which a content browsed by a user through a terminal device belongs, for example, an insurance item or an internet item, where the internet item specifically refers to an item for selling a product or popularizing a product based on an internet background; the attribute data of the user may be age, sex, hobbies, work type, marital situation, etc. of the user, the user tag is determined based on the attribute data and is used for reflecting the intention of the user to browse the service item, for example, the tag of the user is a father, the service item is an insurance item, and the intention of the user to browse the insurance item may be to purchase child insurance for a child; the item label is determined based on the item information of the business item, and is used for representing the field related to the business item, for example, the item label of the business item is insurance, and the business item is represented by the insurance label to relate to the insurance field.

In practical application, whether insurance or application program is adopted, more users are expected to join or use, and the better popularization capability is required to be realized when more users are attracted to join in insurance projects or use application programs, and the better popularization mode can improve the conversion rate of the users so as to realize the increase of the retention rate of the users, thereby achieving the wide application or use of business projects; based on this, in the process of recommending the document corresponding to the business item to the user, the document with novelty and attraction to the user needs to be created according to the business item, and the user needs to be enabled to know the type of the business item and the service provided by the user fastest under the condition of receiving the document, so that the user browsing the business item can be converted into the member in the business item, and therefore, recommending the document of interest to the user is very important for the business item.

According to the text recommendation method provided by the specification, in order to recommend the text which is interesting to the user enough to the user, so that the conversion rate of the user relative to the business project is improved, the user tag of the user and the project tag of the business project are input into the screening model, the multimedia data and the subject format matched with the user are screened, the material data are extracted from the material database, the material data are processed according to the subject format to obtain the subject data, the text recommended to the user is generated based on the subject data and the multimedia data, the text which can effectively trigger the interest intention of the user is generated, the conversion rate of the user browsing the business project is further improved, the interpretability of the text recommendation process is higher, the process of generating the text aiming at the user is conveniently analyzed, the generation efficiency of the text is effectively improved, and the experience effect of the user is further improved.

In the specific implementation, in the process of recommending the text related to the service item to the user, the service item can be an insurance item, so that the participation rate of the user in joining the insurance item is increased, or the service item can be an application program recommending item, so that the use rate of the application program used by the user is increased; in this embodiment, the service item is taken as an insurance item to describe recommending an insurance document (insurance advertisement) with higher user interest to the user, so as to improve the participation rate of the user in participating in the insurance item and improve the experience effect of the user.

Further, in the process of determining the user tag and the item tag, since the service item types in which the user participates may be more and different user tags, the service item types provided by the service item may be more and different item tags, and at this time, the user tag capable of representing the user and the item tag capable of representing the service item need to be selected from the plurality of item tags and the user tag, so that the best document generation can be achieved to recommend to the user, in one or more embodiments of this embodiment, specific implementation manners of determining the user tag and the item tag are as follows:

Specifically, the first label specifically refers to all user labels corresponding to the user, the second label specifically refers to all item labels corresponding to the service item, a first label corresponding to the user is determined according to the attribute data, the first label is formed into the user label set, a second label corresponding to the service item is determined according to the item data of the service item, and the second label is formed into the item label set; for example, the attribute data of the user includes age 25 years old, sex men, marital status married, child 1 year old, like dog raising and traveling, participate in public welfare tree planting and public welfare donation, and the determining the first tag according to the attribute data of the user includes: father tags, travel tags, loving animal tags, and loving heart tags; or the item data of the business item includes providing pet insurance, providing child insurance, providing adult insurance and providing elderly insurance, the determining the second tag from the item data of the business item includes: pet insurance labels, child insurance labels, adult insurance labels, and senior insurance labels;

based on the above, under the condition that the user tag set and the item tag set are determined, detecting the historical user conversion rate corresponding to a first tag contained in the user tag set and the historical item conversion rate corresponding to a second tag contained in the item tag set, wherein the historical user conversion rate corresponding to the first tag specifically refers to the probability that a user corresponding to the first tag successfully participates in the business item, the historical item conversion rate corresponding to the second tag specifically refers to the probability that the user successfully invites the business item to participate in according to a document generated by the second tag, and based on the detected historical user conversion rate corresponding to each first tag and the detected historical item conversion rate corresponding to each second tag, in order to be able to recommend a document which best accords with the user intention to the user, selecting the first tag with the highest historical user conversion rate as the user tag, and selecting the second tag with the highest historical item conversion rate as the item tag for the subsequent processing.

For example, in the process of browsing the medical insurance item by the user, in order to be able to recommend the insurance advertisement of sufficient interest to the user, it is necessary to determine the insurance advertisement according to the user tag of the user and the item tag of the item, determine that the user tag of the user includes a father tag, a driver tag, and a basketball tag by collecting the attribute data of the user, determine that the item tag of the insurance item includes a child insurance tag and an adult insurance tag by the item data of the medical insurance item, determine that the conversion rate of the user carrying the father tag into the member of the medical insurance item is 50% by detecting the conversion rate corresponding to each tag, determine that the conversion rate of the user corresponding to the driver tag is 34%, the conversion rate of the user corresponding to the basketball tag is 5%, and select the father tag and the child insurance tag having the highest conversion rate as the user tag and the item tag for the purpose of enhancing the interest of the user to join the insurance item based on the child tag's conversion rate of 77%, and select the father tag and the child insurance tag having the highest conversion rate as the user tag and the item tag for the subsequent input screening model to perform the text generation processing.

In the process of generating the document based on the user tag and the item tag, in order to improve the document capable of generating the user interest intention, a first tag with the highest conversion rate of the historical user is collected to serve as the user tag, a second tag with the highest conversion rate of the historical item is collected to serve as the item tag, and subsequent document generation processing is conducted, so that the finally generated document can effectively trigger the user interest intention, and further the conversion rate of the user and the experience effect of the user are improved.

Step 104: and inputting the user tag and the item tag into a screening model to obtain multimedia data and a subject format which are output by the screening model and matched with the user.

Specifically, on the basis of determining the user tag and the item tag of the service item according to the attribute data, the user tag and the item tag are further used as input of the screening model to screen multimedia data and a subject format matched with the user, wherein the multimedia data is data required for generating the text, the multimedia data can comprise picture data, text data, sound data, video data and the like, the subject format is a format for processing material data, and the subject format specifically refers to format adjustment of data such as text or pictures in the material data, for example, processing of the size of the text, processing of the color of the text, typesetting of the text and the like;

based on the above, the screening model screens the subsequent material formats and multimedia data for generating the document according to the user tag and the item tag, wherein the specific screening process of the screening model screens the material formats and multimedia data corresponding to the user tag and the item tag, and in order to improve the prediction accuracy of the output of the screening model, a large number of samples need to be collected to train the screening model.

In a specific implementation, in order to improve the prediction accuracy of the output of the screening model, the historical data of the service item needs to be collected, and then a training sample is determined based on the historical data, so that the prediction accuracy of the output of the screening model can be improved.

Specifically, the sample user tag specifically refers to a user tag corresponding to a user who successfully participates in the service item, the sample item tag specifically refers to a item tag added to the service item through a text corresponding to the sample item tag, based on this, a sample tag group is constructed according to the sample user tag and the sample item tag, each pair of sample user tag and sample item tag included in the sample tag group have a corresponding relationship, and when the sample multimedia data and the sample topic format corresponding to the sample tag group and the sample tag group are input into a screening model constructed based on the sample tag group and the association relationship between the sample multimedia data and the sample topic format corresponding to the sample tag group, training is performed, so that the screening model applied in the text recommendation process can be obtained.

In addition, the screening model can be trained based on the negative-sample label group, the negative-sample multimedia data and the negative-sample subject format, so that the result output by the screening model is more close to the multimedia data and the subject format matched with the user, and the output accuracy of the model is improved.

Step 106: and extracting material data matched with the user from a material database pre-established in the business project, and processing the material data according to the material format to obtain material data.

Specifically, on the basis of inputting the item tag and the user tag into the screening model to obtain the multimedia data and the topic format output by the screening model, further, the topic data can be obtained by extracting the material data matched with the user from a material database pre-established in the business item and processing the material data according to the topic format.

In practical application, the material database is used for storing and generating the material data of the text, the material data can be text data, picture data or ornament data, and the like, the material data specifically refers to the material data after processing, for example, the material data is "i like parachuting", "i like limit sports", "i like bungee", "and" because … is … ", and the material data is in typesetting format, and the material data is typeset according to the typesetting format," i like parachuting "," i like limit sports "," i like bunge "," and "because … is …", and the obtained material data is "because i like limit sports, i like bunge and parachuting", and is used for subsequent text generation and recommendation.

In a specific implementation, as the internet technology is continuously developed and the material data capable of generating the material data is continuously enriched, the material database is dynamically updated, the newly added material data is added to the material database at any time, and in order to improve the quality of the material data in the material database, the material data in the material database needs to be determined according to the item label corresponding to the user who selects to successfully add the service item, and in one or more implementations of this embodiment, the storage process of the material data in the material database is as follows:

Specifically, determining a project user who successfully participates in the service project according to the historical browsing data of the service project, determining a project user tag corresponding to the user who successfully participates in the service project according to attribute data of the project user, and under the condition of determining the project user tag, indicating that the user with the project user tag participates in the service project with higher probability, wherein at the moment, the material data generating module can send the project user tag;

the material data generating module is configured to generate corresponding material data according to the item user tag, in practical application, the process of generating the material data based on the item user tag may be designed and generated by an experience design user for the item user tag, or the material data corresponding to the item user tag is generated by an experience design model, and the specific implementation process is not limited in any way herein;

Based on the above, the material data corresponding to the project user tag returned by the material generation module is received and stored in the material database pre-established by the business project, and when the material data in the material database is needed, the material data matched with the user is directly extracted from the material database.

In addition, the material data in the material database further comprises material data related to the business item, so that the finally generated document is a document related to the business and capable of attracting users to join.

Further, on the basis of extracting material data matched with the user from a material database pre-established in the service project, the material data is processed based on the material format to obtain material data for subsequently generating a document, and the material data is processed differently by the material format to realize that the document with value can be subsequently generated, and in one or more implementations of the embodiment, the specific processing procedure of the material data is as follows:

Specifically, determining the text size format, the text color format and the sentence pattern according to the subject material format, wherein the text size format specifically refers to the size of the text in the material data to be adjusted, the text color format specifically refers to the color of the text in the material data to be adjusted, and the sentence pattern specifically refers to the sentence pattern to be adopted by the text in the material data; the first material data specifically refers to character data with the character size adjusted, and the second material data refers to character data with the character color adjusted;

based on the above, when the topic format is determined, text data in the material data is processed according to the text size format to obtain the first material data, then the first material data is processed according to the text color format to obtain the second material data, finally the second material data is processed based on the sentence pattern to obtain the target topic data, and the target topic data is used as the topic data to perform subsequent text generation processing.

For example, the material data is "invite you to buy a guarantee for family", "invite you to buy a guarantee for own children", and "invite you to buy a guarantee for old people at home", and the material format is a composition omission sentence, the text is thickened, black, the material data is processed based on the material format, the obtained material data is "invite you to buy a guarantee for family, children, old people", and the obtained material data is used for generating the document later.

The material data is processed according to the material format to obtain the material data, so that the experience effect of a user is further improved, and in the process of processing the material data, the material data can be processed by combining any format, so that the richness of a subsequent generated document is improved, and the effect of high enough attraction to the user is achieved.

In addition, the subject format can typeset the material data, part of characters in the character data in the material data can be enlarged, and part of characters are reduced, so that main content related to the business project is embodied, for example, the material data is 'inviting you to join in an A insurance project', and in the process of processing the material data, the characters of the 'A insurance project' can be enlarged, the characters of the 'inviting you to join in' are reduced, and the variety of the outstanding insurance project is realized; in practical application, the subject format may be set according to the practical application scenario, and the present disclosure is not limited in any way.

Step 108: and generating a document based on the subject data and the multimedia data, and recommending the document to the user through the display page of the business item.

Specifically, on the basis of processing the material data according to the material format to obtain the material data, further, the document is generated by combining the material data and the multimedia data, the document is advertisement or promotion information for recommending the service item, under the condition that the document is determined, the document is recommended to the user through the display page of the service item, and the interpretation capability is strong by generating the document based on the multimedia data and the material data, so that the generated document can sufficiently attract the user, and the conversion effect of the user is improved.

Further, in the process of generating the document, the document can be generated by combining the subject data and the multimedia data together, for example, combining text and pictures, or combining video and sound together, and based on this, in order to generate the document of interest to the user, the document needs to be generated according to a template based on the subject data and the multimedia data, so that the user does not feel rejection to the recommended document, and in one or more embodiments of this embodiment, a specific implementation manner of generating the document is as follows:

Specifically, the distribution template is selected from a distribution template library pre-established in the service project based on the multimedia data, wherein the distribution template specifically refers to a template capable of combining the multimedia data and the subject data to generate a document, for example, the multimedia data is picture data, the subject data is text data, the picture data and the text data are combined to generate the document, at the moment, the adopted distribution template can be that pictures corresponding to the picture data are distributed at a bottom layer, characters corresponding to the text data are placed above the pictures, and the pictures with the characters are combined to be the document;

Based on the above, in the case of selecting a distribution template, determining a topic distribution area corresponding to the topic data and a multimedia distribution area corresponding to the multimedia data in the distribution template, where the topic distribution area specifically refers to a position of the topic data in the distribution template, and the multimedia distribution area specifically refers to a position of the multimedia data in the distribution template; adjusting the subject data according to the subject distribution area, and adjusting the multimedia data according to the multimedia distribution area; and carrying out assembly processing on the adjusted subject data and the adjusted multimedia data based on the distribution template to generate the document.

For example, in the process of browsing the insurance types of the insurance items by the user, determining that the subject data is 'one more guarantee and one more safety' by identifying the user label of the user, wherein the multimedia data is picture data, selecting a picture-text distribution template from a distribution template library pre-established in the insurance items, determining a text distribution area of texts in the picture-text distribution template based on the picture-text distribution template, determining a picture distribution area of pictures in the picture-text distribution template, adjusting the 'one more guarantee and one more safety' according to the text distribution area, adjusting the pictures according to the picture distribution area, and assembling according to the adjusted texts and pictures to generate the popularization advertisement of the insurance items as shown in fig. 2.

In the process of generating the document, different distribution templates can be selected based on the multimedia data, and the multimedia data and the subject data are assembled according to the distribution templates to generate the document, so that the optimal document can be generated, and the attraction to a user is improved.

In a specific implementation, since the multimedia data includes sound data, picture data, video data, text data, and the like, different text files are generated based on the subject data and different multimedia data, and the description will explain the different text files from four aspects;

in one or more embodiments of the present embodiment, in the case where the multimedia data is sound data and picture data, the generated text will include text, sound and picture, which may be understood as a process of displaying the text to the user is further configured with a sound explanation, and the specific implementation manner is as follows:

For example, in the process of recommending the application program a to the user, promotion information for the application program a needs to be generated according to the picture data, the subject data and the sound data, and the application program a is recommended to the user, at this time, the application program a is displayed based on the picture data, the advantages of the application program a are described to the user based on the subject data, the downloading mode is reminded to the user a based on the sound data, the promotion information for the application program a is generated by combining the picture data, the subject data and the sound data, and the downloading mode of the application program a can be reminded by describing some points of the application program a through characters in the process of displaying the promotion information and by means of voice.

In one or more implementation manners of this embodiment, in the case where the multimedia data is picture data in the second aspect, the generated document includes a picture and a text, which may be understood as showing the document with the text and the picture to the user, so as to implement recommendation of the service item, and specific implementation manners are as follows:

In one or more embodiments of the present embodiment, in a case where the third aspect of the present invention is that the multimedia data is sound data, the generated document includes sound and text, which may be understood as showing the document carrying the text and the sound to the user, so as to implement recommendation of the service item, and specific implementation manners are as follows:

In one or more implementation manners of this embodiment, in a case where the fourth aspect of the present invention is that the multimedia data is video data, the generated document includes video and text, which may be understood as showing the document carrying the text and the video to the user, so as to implement recommendation of the service item, and specific implementation manners are as follows:

Under the condition that the multimedia data are different kinds of data, different documents are generated by combining the subject data, so that documents with higher richness can be generated, the experience effect of a user is improved, and the conversion rate of the user can be effectively improved.

Further, on the basis of generating the document, the document needs to be recommended to the user, and in order to avoid interference with the user's browsing of business items in the process of recommending the document to the user, the document may be displayed by way of a display window, and in one or more embodiments of the present disclosure, the specific process of recommending the document to the user is as follows:

Specifically, in the process of recommending the document, the browsing time of the user for browsing the service item is obtained, and when the browsing time is greater than a preset time threshold, the user is indicated to have higher interest in the service item, and at this time, the document can be recommended to the user by popping up the display window in the display page.

In practical application, the document can be displayed at the lowest part of the display page or displayed by jumping from the display page to another page under the condition that the content in the display page is slid to the lowest part in the process of displaying the document.

For example, in the process of browsing the insurance item by the user, browsing time is determined to be 5 minutes by acquiring the time of browsing the insurance item display page by the user, the browsing time is compared with a time threshold value, and if the browsing time is determined to be greater than the time threshold value, popularization information of the insurance item is displayed in the display page in a window display mode, as shown in fig. 3.

By determining the recommendation of the document to the user in combination with the time of browsing the business item by the user in the process of displaying the document, the recommendation can be determined under the condition that the user is interested, and the conversion rate of the user is further improved.

In a specific implementation, after recommending the document to the user, if the final conversion effect of the document is good enough, in order to accelerate the document showing efficiency to the user, the document may be stored in the document library, and if the user tag of another user browsing the service item is consistent with the user tag corresponding to the generated document, the document may be directly recommended to another user, so that the document recommending efficiency may be effectively improved, and in one or more embodiments of the present embodiment, specific embodiments are as follows:

Specifically, in the process of recommending the text to the user, the situation that the user tag and the project tag are the same may exist, the finally generated text may be the same, at this time, the conversion rate corresponding to the same text is obtained, that is, the user who clicks the text to add the business project occupies the user who browses the text, then whether the conversion rate is greater than the conversion rate threshold value is judged, if so, the throwing effect of the text is better, the conversion rate of the user can be effectively improved, at this time, the text can be stored in the text library of the business project, and the association relationship between the text, the subject format and the multimedia data is established; under the condition that the multimedia data and the subject formats determined based on the user tags and the item tags of other users are the same as the subject data and the multimedia data for generating the document, the document can be directly recommended to other users; if not, the throwing effect of the description scheme is poor, and at the moment, the accuracy of the output multimedia data and the subject format of the screening model possibly exists is not high, and the screening model can be optimized by taking the multimedia data, the subject format, the user tag and the item tag as negative training samples, so that the accuracy of the screening model is improved.

According to the text recommendation method provided by the specification, the user label of the user and the item label of the business item are input into the screening model, the multimedia data and the material format matched with the user are screened, the material data are extracted from the material database, the material data are processed according to the material format to obtain the material data, the text recommended to the user is generated based on the material data and the multimedia data, the text capable of effectively triggering the interest intention of the user is generated, the conversion rate of the user browsing the business item is further improved, the interpretation capability of the text recommendation process is higher, the process of generating the text for the user is conveniently analyzed, the generation efficiency of the text is effectively improved, and the experience effect of the user is further improved.

The application of the document recommendation method provided in the present specification to insurance projects will be further described with reference to fig. 4. Fig. 4 shows a process flow chart of a document recommendation method applied to an insurance item according to an embodiment of the present disclosure, which specifically includes the following steps:

step 402: and determining a user tag set and an item tag set of the insurance item according to the attribute data of the user browsing the insurance item.

Specifically, in order to improve the conversion rate of the user adding the insurance item, the platform corresponding to the insurance item needs to generate different insurance popularization information according to different user labels of each user to recommend the different insurance popularization information to the user, so that the conversion rate of the user is improved;

based on the information, in the process that the user browses the information related to the insurance item through the mobile phone, the attribute information of the user and the item information of the browsed insurance item are collected to determine the user tag and the item tag.

Step 404: and selecting the label with the highest historical user conversion rate in the user label set as the user label corresponding to the user, and selecting the label with the highest historical item conversion rate in the item label set as the item label corresponding to the insurance item.

Specifically, since the number of user tags is large, and the number of user tags is large, in order to recommend the popularization information of the greatest interest of the user to the user, the user tag with the highest conversion rate according to the historical user and the item tag with the highest conversion rate of the historical item are used for generating the popularization information of the insurance item subsequently.

Step 406: and inputting the user tag and the item tag into the screening model to obtain picture data, sentence format and text color format output by the screening model.

Step 408: and extracting text material data matched with the user from a material database pre-established in the security project.

Specifically, the text material data specifically refers to popularization text of the popularization insurance item.

Step 410: and carrying out sentence processing on the text material data according to the sentence format to obtain intermediate material data.

Specifically, the intermediate material data specifically refers to text material data after sentence pattern processing is performed on the text material data.

Step 412: and performing color format processing on the intermediate material data according to the character color format to obtain character subject data.

Step 414: and generating popularization information of the insurance item based on the text subject data and the picture data.

Step 416: and recommending popularization information to the user through the display page of the safety item.

Specifically, under the condition that text material data and picture data are determined, it can be determined that the promotion information recommended to the user is promotion information of the insurance item with combined pictures and texts, and the promotion information of the insurance item can be generated by assembling the picture data and the text material data, and the promotion information specifically refers to promotion information capable of triggering the intention of the user to join in the insurance item;

based on the information, the insurance item is recommended to the user in a mode of displaying the popularization information in the display page of the insurance item.

According to the text recommendation method provided by the specification, the user label of the user and the item label of the insurance item are input into the screening model, the picture data and the material format matched with the user are screened, the text material data are extracted from the material database, the text material data are processed according to the material format to obtain the text material data, the promotion information recommended to the user is generated based on the text material data and the picture data, the promotion information of the insurance item which can effectively trigger the interest intention of the user is generated, the conversion rate of the user browsing the insurance item is further improved, the interpretation ability of the promotion information recommendation process is higher, the process of generating the promotion information for the user is conveniently analyzed, the generation efficiency of the promotion information is effectively improved, and the experience effect of the user is further improved.

Corresponding to the method embodiment, the present disclosure further provides an embodiment of a document recommendation apparatus, and fig. 5 shows a schematic structural diagram of a document recommendation apparatus according to an embodiment of the present disclosure. As shown in fig. 5, the apparatus includes:

a determining tag module 502 configured to determine a user tag of a user browsing a business item and an item tag of the business item according to attribute data of the user;

A model filtering module 504 configured to input the user tag and the item tag into a filtering model, and obtain multimedia data and a topic format output by the filtering model and matched with the user;

a processing data module 506, configured to extract material data matched with the user from a material database pre-established in the service project, and process the material data according to the material format to obtain material data;

a recommending document module 508 configured to generate a document based on the subject data and the multimedia data, and recommend the document to the user through a presentation page of the business item.

In an alternative embodiment, the determining tag module 502 includes:

a composition set unit configured to determine a user tag set composed of first tags corresponding to the users according to the attribute data, and determine an item tag set composed of second tags corresponding to the service items according to the item data of the service items;

a conversion rate detection unit configured to detect a historical user conversion rate corresponding to a first tag contained in the user tag set and a historical item conversion rate corresponding to a second tag contained in the item tag set;

A selection tag unit configured to select a first tag having a highest conversion rate of the historical user as the user tag of the user, and a second tag having a highest conversion rate of the historical item as the item tag of the business item.

In an alternative embodiment, the processing data module 506 includes:

a format determining unit configured to determine a text size format, a text color format, and a sentence pattern according to the subject format;

the first processing unit is configured to perform word size processing on word data in the material data according to the word size format to obtain first material data;

the second processing unit is configured to perform text color processing on the first material data according to the text color format to obtain second material data;

and the third processing unit is configured to perform sentence processing on the second material data according to the sentence pattern to obtain target material data, and take the target material data as the material data.

In an alternative embodiment, the recommended documents module 508 includes:

a selecting template unit configured to select a distribution template from a distribution template library pre-established in the service item according to the multimedia data;

A determining area unit configured to determine a topic distribution area corresponding to the topic data based on the distribution template, and determine a multimedia distribution area corresponding to the multimedia data;

an adjustment data unit configured to adjust the subject data according to the subject distribution area, and adjust the multimedia data according to the multimedia distribution area;

and the document generation unit is configured to perform assembly processing on the adjusted subject data and the adjusted multimedia data based on the distribution template to generate the document.

In an alternative embodiment, the screening model is trained by:

In an alternative embodiment, the document recommendation apparatus further includes:

the acquisition module is configured to acquire the conversion rate corresponding to the document and judge whether the conversion rate is greater than a conversion rate threshold value;

if yes, operating the storage module;

the storage module is configured to store the text to a text library of the service item and establish an association relationship between the text, the subject format and the multimedia data;

if not, operating the optimization module;

the optimization module is configured to optimize the screening model using the multimedia data and the topic format and the user tag and the project tag as negative training samples.

a determining project user module configured to determine project users who successfully participate in the business project according to the historical browsing data of the business project;

a transmission tag module configured to determine an item user tag of the item user based on attribute data of the item user, and transmit the item user tag to a material data generation module;

and the receiving data module is configured to receive the material data generated aiming at the project user label and returned by the material data generating module, and store the material data into the material database.

In an alternative embodiment, in the case that the multimedia data is sound data and picture data, the recommended text module 508 is further configured to:

In an alternative embodiment, in the case that the multimedia data is picture data, the recommended-text module 508 is further configured to: and generating image-text resources based on the image data and the subject data, and taking the image-text resources as the document.

In an alternative embodiment, in the case that the multimedia data is sound data, the recommended-text module 508 is further configured to: an audio resource is generated based on the sound data and the subject data, and the audio resource is used as the document.

In an alternative embodiment, in the case where the multimedia data is video data, the recommended-text module 508 is further configured to: and generating video resources based on the video data and the subject data, and taking the video resources as the document.

In an alternative embodiment, the recommended documents module 508 includes:

the acquisition time unit is configured to acquire the browsing time of the user for browsing the display page of the service item;

and the document recommending unit is configured to recommend the document to the user in a mode of displaying the document on a display window in the display page under the condition that the browsing time is larger than a preset time threshold.

According to the text recommendation device provided by the specification, the user tag of the user and the item tag of the business item are input into the screening model, the multimedia data and the material format matched with the user are screened, the material data are extracted from the material database, the material data are processed according to the material format to obtain the material data, the text recommended to the user is generated based on the material data and the multimedia data, the text capable of effectively triggering the interest intention of the user is generated, the conversion rate of the user browsing the business item is further improved, the interpretation capability of the text recommendation process is higher, the process of generating the text for the user is conveniently analyzed, the generation efficiency of the text is effectively improved, and the experience effect of the user is further improved.

The above is a schematic solution of a document recommendation apparatus of this embodiment. It should be noted that, the technical solution of the document recommendation device and the technical solution of the document recommendation method belong to the same concept, and details of the technical solution of the document recommendation device, which are not described in detail, can be referred to the description of the technical solution of the document recommendation method.

Fig. 6 illustrates a block diagram of a computing device 600 provided in accordance with an embodiment of the present specification. The components of computing device 600 include, but are not limited to, memory 610 and processor 620. The processor 620 is coupled to the memory 610 via a bus 630 and a database 650 is used to hold data.

Computing device 600 also includes access device 640, access device 640 enabling computing device 600 to communicate via one or more networks 660. Examples of such networks include the Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the internet. The access device 640 may include one or more of any type of network interface (e.g., a Network Interface Card (NIC)) whether wired or wireless, such as an IEEE802.11 Wireless Local Area Network (WLAN) wireless interface, a worldwide interoperability for microwave access (Wi-MAX) interface, an ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a bluetooth interface, a Near Field Communication (NFC) interface, and so forth.

In one embodiment of the present description, the above-described components of computing device 600, as well as other components not shown in FIG. 6, may also be connected to each other, such as by a bus. It should be understood that the block diagram of the computing device shown in FIG. 6 is for exemplary purposes only and is not intended to limit the scope of the present description. Those skilled in the art may add or replace other components as desired.

Computing device 600 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., tablet, personal digital assistant, laptop, notebook, netbook, etc.), mobile phone (e.g., smart phone), wearable computing device (e.g., smart watch, smart glasses, etc.), or other type of mobile device, or a stationary computing device such as a desktop computer or PC. Computing device 600 may also be a mobile or stationary server.

Wherein the processor 620 is configured to execute the following computer-executable instructions:

The foregoing is a schematic illustration of a computing device of this embodiment. It should be noted that, the technical solution of the computing device and the technical solution of the above document recommendation method belong to the same concept, and details of the technical solution of the computing device, which are not described in detail, can be referred to the description of the technical solution of the above document recommendation method.

An embodiment of the present disclosure also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, are configured to:

The above is an exemplary version of a computer-readable storage medium of the present embodiment. It should be noted that, the technical solution of the storage medium and the technical solution of the above document recommendation method belong to the same concept, and details of the technical solution of the storage medium, which are not described in detail, can be referred to the description of the technical solution of the above document recommendation method.

The foregoing describes specific embodiments of the present disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The computer instructions include computer program code that may be in source code form, object code form, executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, a software distribution medium, and so forth. It should be noted that the computer readable medium contains content that can be appropriately scaled according to the requirements of jurisdictions in which such content is subject to legislation and patent practice, such as in certain jurisdictions in which such content is subject to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

It should be noted that, for the sake of simplicity of description, the foregoing method embodiments are all expressed as a series of combinations of actions, but it should be understood by those skilled in the art that the present description is not limited by the order of actions described, as some steps may be performed in other order or simultaneously in accordance with the present description. Further, those skilled in the art will appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily all necessary in the specification.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and for parts of one embodiment that are not described in detail, reference may be made to the related descriptions of other embodiments.

The preferred embodiments of the present specification disclosed above are merely used to help clarify the present specification. Alternative embodiments are not intended to be exhaustive or to limit the invention to the precise form disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the disclosure and the practical application, to thereby enable others skilled in the art to best understand and utilize the disclosure. This specification is to be limited only by the claims and the full scope and equivalents thereof.

Claims

1. A document recommendation method, comprising:

2. The document recommendation method according to claim 1, wherein the determining a user tag of a user browsing a business item and an item tag of the business item according to attribute data of the user, comprises:

3. The document recommendation method according to claim 1, wherein the processing the material data according to the material format to obtain material data includes:

4. The document recommendation method of claim 1, the generating a document based on the subject data and the multimedia data, comprising:

5. The document recommendation method of claim 1, the screening model being trained by:

6. The document recommending method according to claim 1, wherein after the step of generating a document based on the subject data and the multimedia data and recommending the document to the user through the presentation page of the service item is performed, further comprising:

7. The document recommendation method according to claim 1, wherein before the step of extracting the material data matched with the user from the material database pre-established in the business project and processing the material data according to the material format to obtain the material data, the method further comprises:

8. The document recommendation method according to claim 1, wherein in a case where the multimedia data is sound data and picture data, the generating a document based on the subject data and the multimedia data, comprises:

9. The document recommendation method according to claim 1, wherein in a case where the multimedia data is picture data, the generating a document based on the subject data and the multimedia data, comprises:

10. The document recommendation method according to claim 1, wherein in a case where the multimedia data is sound data, the generating a document based on the subject data and the multimedia data, comprises:

11. The document recommendation method according to claim 1, wherein, in a case where the multimedia data is video data, the generating a document based on the subject data and the multimedia data, comprises:

12. The document recommending method according to any one of claims 8 to 11, the recommending the document to the user through the presentation page of the business item, comprising:

13. A document recommendation apparatus comprising:

14. A computing device, comprising:

a memory and a processor;

15. A computer readable storage medium storing computer instructions which, when executed by a processor, implement the steps of the document recommendation method of any one of claims 1 to 12.