CN111444424A

CN111444424A - Information recommendation method and information recommendation system

Info

Publication number: CN111444424A
Application number: CN202010218283.2A
Authority: CN
Inventors: 吴伟兴
Original assignee: Shenzhen Fenqile Network Technology Co ltd
Current assignee: Shenzhen Fenqile Network Technology Co ltd
Priority date: 2020-03-25
Filing date: 2020-03-25
Publication date: 2020-07-24

Abstract

The invention provides an information recommendation method, which comprises the following steps: acquiring source data input by a user; extracting first characteristic information from the source data; inquiring a recommendation model corresponding to the first characteristic information based on a preset comparison table; and substituting the first characteristic information into the recommendation model to generate a recommendation result. The invention also provides an information recommendation system, which realizes automatic search recommendation by configuring a recommendation model based on the search source data input by the user through a universal information recommendation component, so that system codes do not need to be repeatedly developed, and the maintenance is more convenient.

Description

Information recommendation method and information recommendation system

Technical Field

The embodiment of the invention relates to a search recommendation system technology, in particular to an information recommendation method and an information recommendation system.

Background

The current commodity recommendation system and the current search system are based on models to carry out intelligent recommendation, and the recommendation accuracy and the matching degree are improved through deep learning and training of the models. And the models are basically calculated based on the matching degree of the characteristics of the users and the commodity. In current implementations of recommendation and search systems, each recommendation system requires a separate code.

This approach results in a need to maintain a separate set of codes for each recommendation system, and reduced maintainability in the face of multiple different recommendation needs.

Disclosure of Invention

The invention provides an information recommendation method and an information recommendation system, which are used for realizing automatic search recommendation based on a source data configuration recommendation model input by a user and enabling system codes not to need repeated development.

In a first aspect, the present embodiment provides an information recommendation method, including:

Acquiring source data input by a user;

Extracting first characteristic information from the source data;

Inquiring a recommendation model corresponding to the first characteristic information based on a preset comparison table;

And substituting the first characteristic information into the recommendation model to generate a recommendation result.

Further, the first feature information includes one or more of a feature list, a data type, and a data key value.

Further, the comparison table is used for storing the relevance of one or more of the feature list, the data type and the data key value to the recommendation model.

Further, after querying the recommendation model corresponding to the first feature information based on the preset comparison table, the method further includes:

Querying second feature information in a history record, wherein the second feature information corresponds to the feature list of the first feature information;

And substituting the first characteristic information and the second characteristic information into the recommendation model to generate a second recommendation result.

Further, before the obtaining the source data input by the user, the method further includes:

Determining the first characteristic information required by each recommendation model;

Generating the comparison table based on the corresponding relation between the recommendation model and the first characteristic information;

And storing the comparison table into a database of the information recommendation system.

Judging whether the recommendation model needs third feature information or not;

If so, performing derivation processing on the first characteristic information to generate third characteristic information;

And substituting the third characteristic information into the recommendation model to generate a third recommendation result.

Further, the derivation process includes addition, subtraction, multiplication, encoding, and/or neural network training of the first feature information.

In a second aspect, the present invention further provides an information recommendation system, including:

The acquisition module is used for acquiring source data input by a user;

The first characteristic module is used for extracting first characteristic information from the source data;

The query module is used for querying the recommendation model corresponding to the first characteristic information based on a preset comparison table;

And the first calculation module is used for substituting the first characteristic information into the recommendation model to generate a recommendation result.

In a third aspect, the present invention further provides a server, including a memory, a processor, and a program stored in the memory and executable on the processor, where the processor executes the program to implement an information recommendation method as described in any one of the above

In a fourth aspect, a terminal-readable storage medium stores a program that, when executed by a processor, is capable of implementing an information recommendation method as in any one of the above.

According to the method and the device, the recommendation model is determined based on the search source data input by the user and the comparison table, so that unified configuration of various recommendation systems is realized, automatic search recommendation can be performed, and system codes do not need to be repeatedly developed.

Drawings

Fig. 1 is a flowchart of an information recommendation method according to the first embodiment.

Fig. 2 is a flowchart of an information recommendation method according to an alternative embodiment of the second embodiment.

Fig. 3 is a flowchart of an information recommendation method according to an alternative embodiment of the second embodiment.

Fig. 4 is a flowchart of an information recommendation method according to an alternative embodiment of the second embodiment.

Fig. 5 is a flowchart of an information recommendation method according to an alternative embodiment of the second embodiment.

Fig. 6 is a block diagram of an information recommendation system according to a third embodiment.

Fig. 7 is a block diagram of an information recommendation system according to three alternative embodiments of this embodiment.

Fig. 8 is a block diagram of a server in the fourth embodiment.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.

Before discussing exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although a flowchart may describe the steps as a sequential process, many of the steps can be performed in parallel, concurrently or simultaneously. In addition, the order of the steps may be rearranged. A process may be terminated when its operations are completed, but may have additional steps not included in the figure. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc.

Furthermore, the terms "first," "second," and the like may be used herein to describe various orientations, actions, steps, elements, or the like, but the orientations, actions, steps, or elements are not limited by these terms. These terms are only used to distinguish one direction, action, step or element from another direction, action, step or element. For example, the first feature information may be the second feature information or the third feature information, and similarly, the second feature information and the third feature information may be the first feature information without departing from the scope of the present application. The first characteristic information, the second characteristic information and the third characteristic information are characteristic information of the distributed file system, but are not the same characteristic information. The terms "first", "second", etc. are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present invention, "plurality", "batch" means at least two, e.g., two, three, etc., unless specifically limited otherwise.

The terms and abbreviations used in the following examples have the following meanings:

TF-IDF model: a statistical model is used to evaluate the importance of a word to one of a set of documents or a corpus. The importance of a word increases in proportion to the number of times it appears in a document, but at the same time decreases in inverse proportion to the frequency with which it appears in the corpus. Various forms of TF-IDF weighting are often applied by search engines as a measure or rating of the degree of relevance between a document and a user query. In addition to TF-IDF, search engines on the internet use a ranking method based on link analysis to determine the order in which documents appear in search results.

And (3) blocking and boxing: card direction binning is a supervised, bottom-up data discretization method.

EDA (electronic design automation): exploratory Data Analysis, which is a Data Analysis method that explores existing Data (especially original Data obtained by investigation or observation) under the least prior assumption, explores Data structures and rules by means of drawing, tabulation, equation fitting, calculating characteristic quantities and the like, and analyzes Data to summarize main characteristics of the Data.

GDBT decision tree: the Decision Tree of Gradient Boosting Decision Tree is an integrated learning method and can be used for solving the problems of classification, regression and the like.

Example one

The embodiment provides an information recommendation method, which takes an information recommendation system as an execution main body, receives search keywords input by a user and outputs a recommendation result.

The recommendation system mentioned in this embodiment and the following embodiments generally refers to a type of application for determining the articles/services currently needed or interested by the user, and is generally applied to the fields of e-commerce, movies, videos, social contact, personalized reading, extended services, personalized advertisements, and the like, and is not limited to the above fields.

Illustratively, the recommendation system is used for recommending related documents at a document query website, recommending music which a user may like by a music player, recommending similar commodities which the user may need by a shopping website, and the like, and optionally, may also be used for realizing functions of recommending users in the same city, users in the same interest group, and/or people nearby by a social website, and the like, or pushing similar short videos for the user according to user browsing records in a short video app. As shown in fig. 1, the present embodiment includes the following specific steps:

S101, acquiring source data input by a user.

In this step, the information recommendation system obtains source data input by the user, in this embodiment and the following embodiments, the source data includes, but is not limited to, user social relationship data, interest points, a located web environment, search content input in a search bar, a user name, and the like, and also includes text information, picture information, audio information, and the like, and the source data is generally a set of a plurality of data.

And S102, extracting first characteristic information from the source data.

In this step, the first feature information is valid data for generating a recommendation result, and includes a feature list, and optionally, one or more of a data type and a data key value.

Specifically, the data types are distinguished according to the data file format, and may include, but are not limited to, an audio type, a video type, a picture type, a text type, and the like, and may also be classified based on the data types that can be processed by a common recommendation model, such as distinguishing the data types as a category type, a continuation type, a date type, a word type, and/or a text type. Each data type corresponds to one or more feature lists.

Illustratively, when the data type is a text type, the feature list includes one or more of a trade name, a search text, a browsing history, a score, and/or a date for distinguishing a format of the source data for storage; the data key value varies depending on the purpose of recommendation. Illustratively, if the data type is a picture class, the feature list is one or more of a feature vector and a feature descriptor of the image.

S103, inquiring a recommendation model corresponding to the first characteristic information based on a preset comparison table.

In this step, the look-up table is used to store the relevance of the recommendation model and one or more of the feature list, the data type and the data key value. Each common recommendation model corresponds to one or more feature lists, one or more data types and/or one or more data key values, corresponding IDs are given to the models and the feature lists respectively and are stored in a database in a unified mode, and information such as the required feature lists, the required data types and the required data key values can be matched through the model IDs.

Specifically, the generation process of the comparison table is as follows: determining the first characteristic information required by each recommendation model; generating the comparison table based on the corresponding relation between the recommendation model and the first characteristic information; and storing the comparison table into a database of the information recommendation system.

Illustratively, in the comparison table, a news recommendation model, an image recognition model, a text retrieval model, and a commodity recommendation model are respectively assigned with model IDs of {1,2,3,4}, and the model assigned with the ID is generally a recommendation model that is already used in the market, for example, a news recommendation algorithm model of today's top, an algorithm model of hundredths knowledge and the like. The comparison table is based on a modularized configuration mode, and model calling is realized through the model ID. The look-up table can be continuously supplemented with new algorithm models based on recommendation requirements. Optionally, to implement the multiplexing function of different recommendation systems, the information recommendation function is implemented, so in an alternative embodiment, after step S103, the method further includes: taking the calculation logic of one or more recommendation models as functional components, and calling one or more functional components to be configured as target recommendation models by the system based on different requirements of business functions so as to calculate recommendation results.

Meanwhile, in the comparison table, each model ID corresponds to feature information that needs to be processed by the model, illustratively, the model ID is given to the news recommendation algorithm model as 1, and the corresponding feature information is given to the feature ID as 1 for one or more of the click rate, reading time, praise, comment, forward, and praise of the news. And recording the corresponding relation between the model ID and the characteristic ID as a comparison table.

Illustratively, the source data is a record of a news page browsed by a user, the extracted first feature list is one or more of a click rate, reading time, praise, comment, forwarding and praise number of the user on news, a corresponding feature ID is 1, a model ID corresponding to the feature ID 1 is obtained by querying from a preset comparison table, and then a preset news recommendation algorithm model is referred to based on the model ID for recommendation calculation.

And S104, substituting the first characteristic information into the recommendation model to generate a recommendation result.

According to the method and the device, the recommendation model is determined based on the search source data input by the user and the comparison table, unified configuration of various recommendation systems is achieved, automatic search recommendation can be performed, and system codes do not need to be repeatedly developed.

Example two

On the basis of the above embodiment, the present embodiment performs functions such as data cleansing, data fusion, sorting, and the like on the source data input by the user, so that redundant data can be removed from the input source data, and valid data is retained. The embodiment takes an information recommendation system as an execution main body, receives search keywords input by a user and outputs a recommendation result. As shown in fig. 2, the specific steps are as follows:

And S2011, acquiring source data input by a user.

In this step, the source data includes valid data, and also includes one or more of missing data, abnormal data, and duplicate data. The missing data, the abnormal data and the repeated data are generated in the searching and browsing process of the user behavior data, for example, the repeated data generally refers to the misoperation of the user, such as the operation that the user generates a large number of clicks in a short time when the network is interrupted. The abnormal data refers to data with high dispersion degree with effective data, and is noise in the network information transmission process, and missing data is generally generated when the network is interrupted and delayed.

S202, extracting first characteristic information from the source data.

S203, inquiring a recommendation model corresponding to the first characteristic information based on a preset comparison table.

S2041, querying second feature information in a history record, wherein the second feature information corresponds to the feature list of the first feature information.

In this step, when the user browses the short video by using the app, the first feature information includes one or more information of search content, a user name, a currently opened short video, a located web environment, browsing duration, and the like, which are input by the user in the search bar, and the second feature information refers to one or more information of search content, a user name, a currently opened short video, a located web environment, and browsing duration, which are input by the corresponding user in the search bar, in the historical browsing record and the historical operation log.

S2042, substituting the first characteristic information and the second characteristic information into the recommendation model to generate a second recommendation result.

As shown in fig. 3, the present invention may be applied to the field of big data, and in an alternative embodiment, after step S2011, the method further includes:

S2012, performing EDA data analysis on the source data.

The step is used for acquiring one or more distribution states of missing values, abnormal values, modes, average values, 1 st median, 2 nd median, 3 rd median, standard deviation, maximum values and minimum values of the source data. By performing EDA analysis on a large amount of user data, various messy dirty data can be well processed, the structure and the characteristics of the data can be truly and directly observed through the EDA, the use efficiency of the data is improved, and the subsequent data cleaning is convenient.

S2013, data cleaning is conducted on the source data. The step is used for acquiring the valid data. The cleaning is to discard, fill, replace, remove duplicate and other operations on the data set to achieve the purposes of removing abnormality, correcting errors and complementing deficiency so as to obtain effective data. Illustratively, this step may employ a chi-square binning method.

As shown in fig. 4, in another alternative embodiment, step S2041 is followed by:

S2043, according to the data source type and the data source key corresponding to the first characteristic information and the second characteristic information, fusing the first characteristic information and the second characteristic information to generate fourth characteristic information.

S2044, ranking the fourth feature information according to the weight, and filtering the fourth feature information with lower weight based on business requirements.

S2045, substituting the fourth feature information with lower filtering weight into the recommendation model to generate a fourth recommendation result.

And processing the data sets screened out by the historical data and the current data by using a machine learning model. By fusing historical data and real-time behavior data, the data sources of the data set are wider, and the calculation accuracy of the model is improved. Meanwhile, since the historical data and the real-time behavior data have different weights in different strategies and different models, the fused data needs to be sorted according to a required rule so as to delete unimportant data.

As shown in fig. 5, in another alternative embodiment, after step S203, the method further includes:

S2051, judging whether the recommended model needs third characteristic information;

S2052, if needed, performing derivation processing on the first characteristic information to generate third characteristic information;

The derivation process of this step includes addition, subtraction, multiplication, encoding and/or neural network training of the first feature information. The third feature information, i.e. the derived feature, refers to a new feature obtained by learning features with the original data, and includes a feature that is not originally present in the data based on the change (e.g. conversion into a vector) of the original data or a new feature generated when learning through a neural network. And the third characteristic information is used for reflecting the data association relationship of the deeper layer of the source data.

Illustratively, the first feature information is trained sequentially through a GBDT model and a decision tree model, the GBDT model establishes a new decision tree in the gradient direction of the residual error reduction at each iteration, the latter tree learns the former residual error, and one or more of new features and combinations different from the first feature can be constructed by using the GBDT model. Optionally, the number of the decision trees is multiple, and the decision trees are used for distinguishing and expressing the feature information, and the specific number is determined according to the data type of the first feature information and the accuracy requirement of the model.

And S2053, substituting the third characteristic information into the recommendation model to generate a third recommendation result.

According to the method and the device, the acquired source data are cleaned, and the characteristic values are fused, sorted, derived and processed through the data source type and the data source key corresponding to the characteristics, so that the processing precision of the source data is improved, the deeper data association relation of the source data is reflected, and the recommendation precision of the recommendation system is finally improved.

EXAMPLE III

As shown in fig. 6, the present embodiment provides an information recommendation system 3, which includes the following modules:

An obtaining module 301, configured to obtain source data input by a user.

A first feature module 302, configured to extract first feature information from the source data.

The query module 303 is configured to query the recommendation model corresponding to the first feature information based on a preset comparison table.

And the first calculation module 304 is configured to substitute the first feature information into the recommendation model to generate a recommendation result. The first feature information includes one or more of a feature list, a data type, and a data key value. The comparison table is used for storing the relevance of one or more of the feature list, the data type and the data key value with the recommendation model.

As in fig. 7, in an alternative embodiment, further comprising:

A second feature module 305, configured to query second feature information in a history, where the second feature information corresponds to the feature list of the first feature information.

And the second recommending module 306 is configured to substitute the first characteristic information and the second characteristic information into the recommending model to generate a second recommending result.

Further comprising:

A comparison module 307, configured to determine the first feature information required by each recommendation model.

A generating module 308, configured to generate the comparison table based on a correspondence between the recommendation model and the first feature information.

The storage module 309 is configured to store the comparison table in the database of the information recommendation system.

Further comprising:

A determining module 310, configured to determine whether the recommendation model requires third feature information.

A third characteristic module 311, configured to derive the first characteristic information if needed, so as to generate the third characteristic information. The derivation process includes addition, subtraction, multiplication, encoding, and/or neural network training of the first feature information.

And a third recommending module 312, configured to substitute the third feature information into the recommending model to generate a third recommending result.

The information recommendation system provided by the embodiment of the invention can execute the information recommendation method provided by any embodiment of the invention, and has corresponding execution methods and beneficial effects of the functional modules.

Example four

The present embodiment provides a schematic structural diagram of a server, as shown in fig. 8, the server includes a processor 401, a memory 402, an input device 403, and an output device 404; the number of the processors 401 in the server may be one or more, and one processor 401 is taken as an example in the figure; the processor 401, memory 402, input device 403 and output device 404 in the device/terminal/server may be linked by a bus or other means, as exemplified by the linking via a bus in fig. 8.

The memory 402 is a computer-readable storage medium and can be used for storing software programs, computer-executable programs, and modules, such as program instructions/modules (e.g., the obtaining module 301, the first feature module 302, etc.) corresponding to the gateway-based link generation method in the embodiment of the present invention. The processor 401 executes various functional applications of the device/terminal/server and data processing by running software programs, instructions and modules stored in the memory 402, that is, implements the above-described information recommendation method.

The memory 402 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory 402 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, the memory 402 may further include memory located remotely from the processor 401, which may be linked to the device/terminal/server through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The input means 403 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the device/terminal/server. The output device 404 may include a display device such as a display screen.

The embodiment of the invention also provides a server which can execute the information recommendation method provided by any embodiment of the invention and has corresponding functional modules and beneficial effects of the execution method.

EXAMPLE five

An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements a push recommendation method of an information recommendation system according to any embodiment of the present invention:

Acquiring source data input by a user;

Extracting first characteristic information from the source data;

The computer-readable storage media of embodiments of the invention may take any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical link having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a storage medium may be transmitted over any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including AN object oriented programming language such as Java, Smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language, or similar programming languages.

It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. An information recommendation method, comprising:

Acquiring source data input by a user;

Extracting first characteristic information from the source data;

2. The information recommendation method according to claim 1, wherein the first feature information comprises one or more of a feature list, a data type and a data key value.

3. The information recommendation method according to claim 2, wherein the lookup table is used for storing the relevance of one or more of the feature list, the data type and the data key value to the recommendation model.

4. An information recommendation method according to claim 2,

After querying the recommendation model corresponding to the first feature information based on the preset comparison table, the method further includes:

5. The information recommendation method according to claim 1, further comprising, before said obtaining source data input by a user:

6. The information recommendation method according to claim 1, wherein after the querying a recommendation model corresponding to the first feature information based on the preset lookup table, the method further comprises:

7. The information recommendation method according to claim 6, wherein the derivation process comprises addition, subtraction, multiplication, coding and/or neural network training of the first feature information.

8. An information recommendation system, comprising:

The acquisition module is used for acquiring source data input by a user;

9. A server comprising a memory, a processor and a program stored on the memory and executable on the processor, wherein the processor implements an information recommendation method as claimed in any one of claims 1-7 when executing the program.

10. A terminal-readable storage medium, on which a program is stored, wherein the program, when executed by a processor, is capable of implementing an information recommendation method according to any one of claims 1-7.