CN113781180A - Article recommendation method and device, electronic equipment and storage medium - Google Patents

Article recommendation method and device, electronic equipment and storage medium Download PDF

Info

Publication number
CN113781180A
CN113781180A CN202111087804.6A CN202111087804A CN113781180A CN 113781180 A CN113781180 A CN 113781180A CN 202111087804 A CN202111087804 A CN 202111087804A CN 113781180 A CN113781180 A CN 113781180A
Authority
CN
China
Prior art keywords
matrix
user
similarity
article
target
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202111087804.6A
Other languages
Chinese (zh)
Inventor
李文勇
时宝旭
莫海江
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hubei Tiantian Digital Chain Technology Co ltd
Original Assignee
Hubei Tiantian Digital Chain Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hubei Tiantian Digital Chain Technology Co ltd filed Critical Hubei Tiantian Digital Chain Technology Co ltd
Priority to CN202111087804.6A priority Critical patent/CN113781180A/en
Publication of CN113781180A publication Critical patent/CN113781180A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0631—Item recommendations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Business, Economics & Management (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Marketing (AREA)
  • Algebra (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The application provides an article recommendation method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a user scoring matrix; acquiring a user emotion matrix; generating a user evaluation matrix based on the user scoring matrix and the user emotion matrix; acquiring an object similarity target matrix; recommending the object to the target user based on the object similarity target matrix and the user evaluation matrix. In the embodiment of the application, the comment information of the user on the articles is fully dug, the user emotion matrix is generated, the user emotion matrix is combined with the user rating matrix, the preference of the user is judged from two dimensions of rating and comment, and then the similarity between the articles in the article similarity target matrix is combined, so that the articles favored by the user are accurately recommended.

Description

Article recommendation method and device, electronic equipment and storage medium
Technical Field
The present application relates to the field of data processing technologies, and in particular, to an article recommendation method and apparatus, an electronic device, and a storage medium.
Background
With the continuous expansion of electronic commerce, the number of commodities is increasing, and customers usually need to spend a lot of time to find the articles they want to buy. Therefore, how to recommend favorite items to the user becomes the direction of current research.
Currently, a coordinated filtering algorithm (item-CF) is often adopted for recommending articles, and the algorithm is to recommend articles to a user according to the behavior of selecting articles according to the user history and the scoring condition of the user and by combining the similarity between the articles. However, this approach is too comprehensive and lacks data mining, which in turn results in less accurate and effective recommended items.
Disclosure of Invention
An object of the embodiments of the present application is to provide an article recommendation method, an article recommendation device, an electronic device, and a storage medium, so as to improve accuracy and effectiveness of article recommendation for a user.
The invention is realized by the following steps:
in a first aspect, an embodiment of the present application provides an item recommendation method, including: acquiring a user scoring matrix; wherein the user scoring matrix comprises the scores of the respective scored items by different users; acquiring a user emotion matrix; wherein the user emotion matrix comprises emotion scores of the different users for respective commented items; the emotion score is generated by performing emotion analysis on the comment of the user; generating a user evaluation matrix based on the user scoring matrix and the user emotion matrix; acquiring an object similarity target matrix; the object similarity target matrix comprises similarity between every two objects related in the user scoring matrix; recommending articles to target users based on the article similarity target matrix and the user evaluation matrix; wherein the target user is a user in the user scoring matrix.
In the embodiment of the application, the comment information of the user on the articles is fully dug, the user emotion matrix is generated, the user emotion matrix is combined with the user rating matrix, the preference of the user is judged from two dimensions of rating and comment, and then the similarity between the articles in the article similarity target matrix is combined, so that the articles favored by the user are accurately recommended.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the obtaining an emotion matrix of a user includes: obtaining comments of the different users on the article; scoring the comments based on an emotion analysis tool to obtain emotion scores of the different users on the respective commented articles; and generating the user emotion matrix based on the emotion scores of the different users on the respective commented articles.
In the embodiment of the application, comments of users are scored through the emotion analysis tool, so that emotion scores of different users on respective commented articles can be obtained conveniently.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the generating a user evaluation matrix based on the user scoring matrix and the user emotion matrix includes: acquiring the time factor weight of the different users to the respectively scored goods; wherein the time factor weight characterizes a relationship between the scoring time of the user and the current time; and generating the user evaluation matrix based on the time factor weight, the user scoring matrix and the user emotion matrix.
In the embodiment of the application, a time weight factor is also combined in the process of generating the user evaluation matrix based on the user scoring matrix and the user emotion matrix. The closer the time of the user score is, the larger the time weighting factor is, and the greater the referential of the user score is. By the method, the reference value of the user score can be further improved, and further the reasonability and the accuracy of recommending articles favored by the user are improved.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the expression of an element in the user evaluation matrix is:
Figure BDA0003266439520000021
wherein, cfijThe evaluation value of the ith user on the jth item is shown; c. CijRepresents the ith user pairA score for the jth item;
Figure BDA0003266439520000031
adding elements representing corresponding positions of the matrix; f. ofijRepresenting the sentiment score of the ith user for the jth item; zijRepresenting a time factor weight of the ith user for the jth item; t isjRepresents the time, T, at which the ith user scored the jth itemfirstRepresents the time, T, of the first rating of the ith userallRepresenting the total time from the first to the last scoring of the ith user.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the obtaining an object similarity target matrix includes: acquiring an article similarity matrix and a time similarity matrix; the article similarity matrix represents the similarity of the scored times between the articles; the time similarity characterizes the similarity of time-to-market between the items; and generating the object similarity matrix based on the object similarity matrix and the time similarity matrix.
In the embodiment of the application, the object similarity target matrix comprises an object similarity matrix and a time similarity matrix, the similarity between objects in the time of the market can be determined by combining the time similarity matrix, and then whether two objects are similar objects in the same time of the market or not is determined.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the expression of an element in the time similarity matrix is:
Figure BDA0003266439520000032
wherein, facTa,bRepresents the similarity of the article a and the article b in time, taRepresents the time of the article a on the market, tbRepresenting the time of the listing of said item b.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the obtaining an object similarity target matrix includes: acquiring an article similarity matrix and an attribute similarity matrix; the attribute similarity matrix represents the similarity of attributes among the articles; and generating the object similarity target matrix based on the object similarity matrix and the attribute similarity matrix.
In the embodiment of the application, the object similarity target matrix comprises an object similarity matrix and an attribute similarity matrix, the similarity of the objects on the attributes can be determined by combining the attribute similarity matrix, and then whether the two objects are the same on the attributes is determined.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the expression of an element in the attribute similarity matrix is:
Figure BDA0003266439520000041
wherein, sima,bThe expression represents the similarity of the attributes of the article a and the article b, [ a ]i]∩[bi]The number of the same attributes of the article a and the article b is represented, n represents the total number of the attributes of the article a, and m represents the total number of the attributes of the article b.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the obtaining an object similarity target matrix includes: acquiring an article similarity matrix, a time similarity matrix and an attribute similarity matrix; the article similarity matrix represents the similarity of the scored times between the articles; the time similarity characterizes the similarity of time-to-market between the items; the attribute similarity matrix represents the similarity of attributes among the articles; and generating the object similarity target matrix based on the object similarity matrix, the time similarity matrix and the attribute similarity matrix.
In the embodiment of the application, the object similarity target matrix simultaneously comprises an object similarity matrix, a time similarity matrix and an attribute similarity matrix, the similarity between objects in the time of marketing can be determined through the time similarity matrix, and then whether the two objects are similar objects in the time of marketing is determined, the similarity between the objects in the attribute can be determined through the attribute similarity matrix, and then whether the two objects are the same in the attribute is determined. Through the combination of the three, the accuracy and the rationality of the similarity between the objects can be further improved.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the recommending an item to a target user based on the item similarity objective matrix and the user evaluation matrix includes: based on the object similarity target matrix and the user evaluation matrix, acquiring the love degree of the target user to the object related to the user evaluation matrix; recommending the unscored items of the target user to the target user based on the ranking sequence of the high-low likeness.
In the embodiment of the application, the objects which are not scored by the target user are recommended to the target user based on the ranking sequence of the favorite degrees from high to low, so that the favorite objects of the target user can be recommended preferentially, the scored objects are prevented from being recommended to the target user, and the user experience is improved.
With reference to the technical solution provided by the first aspect, in some possible implementation manners, the love degree is obtained through the following expression:
Figure BDA0003266439520000051
wherein u represents the target user, and p (u, j) represents the favorite degree of the target user to the item j; s (j, k) represents a set of k items with the highest similarity to the item j, NuA set of items representing the target user's liking; djiRepresenting the similarity of the article j and the article i, and obtaining the similarity through the article similarity target matrix; cfuiAnd the evaluation value of the target user on the item i is represented and obtained through the user evaluation matrix.
In a second aspect, an embodiment of the present application provides an article recommendation device, including: the first acquisition module is used for acquiring a user scoring matrix; wherein the user scoring matrix comprises the scores of the respective scored items by different users; the second acquisition module is used for acquiring the user emotion matrix; wherein the user emotion matrix comprises emotion scores of the different users for respective commented items; the emotion score is generated by performing emotion analysis on the comment of the user; the generating module is used for generating a user evaluation matrix based on the user scoring matrix and the user emotion matrix; the third acquisition module is used for acquiring an object similarity target matrix; the object similarity target matrix comprises similarity between every two objects related in the user scoring matrix; the recommending module is used for recommending the object to the target user based on the object similarity target matrix and the user evaluation matrix; wherein the target user is a user in the user scoring matrix.
In a third aspect, an embodiment of the present application provides an electronic device, including: a processor and a memory, the processor and the memory connected; the memory is used for storing programs; the processor is configured to invoke a program stored in the memory to perform a method as provided in the above-described first aspect embodiment and/or in combination with some possible implementations of the above-described first aspect embodiment.
In a fourth aspect, embodiments of the present application provide a computer-readable storage medium, on which a computer program is stored, which, when executed by a processor, performs the method as set forth in the above first aspect embodiment and/or in combination with some possible implementations of the above first aspect embodiment.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
Fig. 1 is a block diagram of an electronic device according to an embodiment of the present disclosure.
Fig. 2 is a flowchart illustrating steps of an item recommendation method according to an embodiment of the present application.
Fig. 3 is a block diagram of modules of an article recommendation device according to an embodiment of the present application.
Icon: 100-an electronic device; 110-a processor; 120-a memory; 200-item recommendation means; 210-a first obtaining module; 220-a second acquisition module; 230-a generation module; 240-a third acquisition module; 250-recommendation module.
Detailed Description
The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.
Referring to fig. 1, a block diagram of an electronic device 100 applying an article recommendation method and apparatus according to an embodiment of the present application is provided. In the embodiment of the present application, the electronic device 100 may be a terminal or a server. The terminal may be, but is not limited to, a Personal Computer (PC), a smart phone, a tablet PC, a Personal Digital Assistant (PDA), a Mobile Internet Device (MID), and the like. The server may be, but is not limited to, a web server, a database server, a cloud server, or a server assembly composed of a plurality of sub-servers, etc. Of course, the above-mentioned devices are only used to facilitate understanding of the embodiments of the present application, and should not be taken as limiting the embodiments.
Structurally, electronic device 100 includes a processor 110 and a memory 120. The processor 110 and the memory 120 are electrically connected directly or indirectly to enable data transmission or interaction, for example, the components may be electrically connected to each other via one or more communication buses or signal lines. The item recommendation device includes at least one software module that may be stored in the memory 120 in the form of software or Firmware (Firmware) or solidified in an Operating System (OS) of the electronic device 100. The processor 110 is used for executing executable modules stored in the memory 120, such as software functional modules and computer programs included in the item recommendation device, so as to implement the item recommendation method. The processor 110 may execute the computer program upon receiving the execution instruction.
The processor 110 may be an integrated circuit chip having signal processing capabilities. The Processor 110 may also be a general-purpose Processor, for example, a Central Processing Unit (CPU), a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a discrete gate or transistor logic device, or a discrete hardware component, which may implement or execute the methods, steps, and logic blocks disclosed in the embodiments of the present Application. Further, a general purpose processor may be a microprocessor or any conventional processor or the like.
The Memory 120 may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), and an electrically Erasable Programmable Read-Only Memory (EEPROM). The memory 120 is used for storing a program, and the processor 110 executes the program after receiving the execution instruction.
It should be noted that the structure shown in fig. 1 is only an illustration, and the electronic device 100 provided in the embodiment of the present application may also have fewer or more components than those shown in fig. 1, or have a different configuration than that shown in fig. 1. Further, the components shown in fig. 1 may be implemented by software, hardware, or a combination thereof.
The article recommendation method provided by the embodiment of the application can be applied to various application scenes needing article recommendation, such as: commodity recommendation in shopping software, movie ticket recommendation and other application scenes.
Furthermore, the item recommendation method may be applied to a recommendation platform relating to a corresponding application scenario, for example: a recommendation platform of shopping software, a recommendation platform of movie tickets and the like. These recommendation platforms typically have a local user database based on which recommendations for items can be made.
Referring to fig. 2, fig. 2 is a flowchart illustrating steps of an article recommendation method according to an embodiment of the present application, where the method is applied to the electronic device 100 shown in fig. 1. It should be noted that, the method for recommending an item provided in the embodiment of the present application is not limited by the sequence shown in fig. 2 and the following, and the method includes: step S101-step S105.
Step S101: and acquiring a user scoring matrix.
Wherein the user scoring matrix comprises the scores of the respective scored items by different users. In this embodiment, the user scoring matrix includes scoring conditions of items in the same item set by different users. The scoring may characterize the user's scoring of a certain item in the set of items, as well as the value of the score for the scored item. Specifically, the user scoring matrix C1 is of the form:
Figure BDA0003266439520000081
in the above-mentioned user rating matrix C1, C is usedijRepresenting the elements of the matrix, cijIndicating the rating of the ith user for the jth item, e.g. c11Represents the rating of the 1 st user for the 1 st item, c22Represents the rating of the 2 nd user for the 2 nd item, cwnIndicating the rating of the nth item by the w-th user.
It should be noted that all the articles and users involved may be numbered in advance, so as to form the user scoring matrix. For example, according to the order of user registration, the ID (Identity Document) when the user registers is numbered. And numbering according to the time of market of all items currently on the platform, for example. Illustratively, the number of movies that can be currently shown by the movie platform is 15, and the numbering is performed according to the showing time of the 15 movies.
The score can be any score of 1-5 or any score of 1-10, and certainly, some platforms can also score in a star hitting mode, for example, the platforms can be divided into one to five stars according to the love degree of the object, at this time, the star hitting mode can be converted into a numerical mode, one star corresponds to 1, and five stars corresponds to 5. The present application is not limited thereto. Further, when the user does not rate an item, then the corresponding element is filled with 0.
It should be noted that the ranges of the scores in the same user scoring matrix should be the same, for example, the ranges of the scores in the same user scoring matrix are all 1-5 points, that is, the same user scoring matrix adopts the same scoring rule, and by this way, the problem of inaccurate item recommendation caused by different scoring modes in the following process can be avoided.
Of course, for convenience of subsequent unified calculation, each score may be normalized to a [0, 1] interval or a [0, 5] interval, and the present application is not limited thereto.
Step S102: and acquiring a user emotion matrix.
The user emotion matrix comprises emotion scores of different users for respective articles which are commented on by the users; the sentiment score is generated by sentiment analysis of the comments of the user. In this embodiment, the user range related to the user emotion matrix is the same as the user range related to the user score matrix. And the item set related to the user emotion matrix is the same as the item set related to the user scoring matrix.
Optionally, the obtaining the user emotion matrix specifically includes: obtaining comments of different users on the article; scoring the comments based on an emotion analysis tool to obtain emotion scores of different users for respective articles which are commented; and generating a user emotion matrix based on the emotion scores of different users on the respective commented articles.
The emotion analysis tool may adopt SnowNLP (a word processing tool), which can perform word segmentation on a text and extract sensitive words to analyze the emotion of a user. Illustratively, when the extracted sensitive word is "true" or "comfortable", the user preference is characterized to be higher, the user evaluation is higher, and the emotion score is higher. When the extracted sensitive words are 'bad comment' and 'junk', the representing user preference degree is low, the user evaluation is poor, and at the moment, the emotion score is low. Since the emotion analysis tools are well known in the art, they will not be described in detail herein. Therefore, the comments of the users are scored through the emotion analysis tool, so that emotion scores of different users on the articles which are commented by the users can be obtained conveniently.
In addition, the score can be made according to the word number of the comment of the user, and in general, the more a user likes a certain product, the more time is taken to comment on the product, so the score can be made according to the word number of the comment of the user, and the larger the word number is, the higher the corresponding emotion score is.
The user emotion matrix F1 is of the form:
Figure BDA0003266439520000101
in the user emotion matrix F1, F is usedijRepresenting the elements of the matrix, fijIndicating the sentiment score of the ith user for the jth item, e.g. f11Representing the sentiment score of the 1 st user for the 1 st item, f22Representing the sentiment score of the 2 nd user for the 2 nd item, fwnRepresenting the sentiment score of the w-th user for the n-th item. It should be noted that the user emotion matrix and the user score matrix have the same arrangement order of elements, such as c11And f11Representing the same user and the same item.
When the user does not comment on a certain item, then the corresponding element is filled with 0.
For convenience of subsequent unified calculation, normalization processing may be performed on each emotion score, and the emotion score is normalized to be an interval of [0, 1] or an interval of [0, 5], which is not limited in this application.
Step S103: and generating a user evaluation matrix based on the user scoring matrix and the user emotion matrix.
After the user rating matrix and the user emotion matrix are obtained, the user rating matrix and the user emotion matrix can be directly added to obtain a user evaluation matrix.
Wherein, the expression of the elements in the user evaluation matrix is as follows:
Figure BDA0003266439520000102
in the formula (1), cfijThe evaluation value of the ith user on the jth item is shown; c. CijIndicating the grade of the ith user on the jth item;
Figure BDA0003266439520000103
adding elements representing corresponding positions of the matrix; f. ofijRepresenting the sentiment score of the ith user for the jth item. It should be noted that the condition that the above formula exists may be cijAnd fijNone of which is zero. That is, when cijAnd fijNeither is zero, the two are added and divided by 2. When any one of the two parameters is zero, the value corresponding to the other parameter is taken as an evaluation value, such as cijIs zero, and fij0.5, cfij＝fij，cfijAlso 0.5.
The user evaluation matrix CF1 generated in the above manner has the form:
Figure BDA0003266439520000111
in the user evaluation matrix CF1, CF is usedijRepresenting the elements, cf, in the matrixijRepresenting the value of the ith user's rating for the jth item, e.g. cf11Represents the evaluation value, cf, of the 1 st user for the 1 st item22Represents the evaluation value, cf, of the 2 nd user for the 2 nd itemwnIndicating the evaluation value of the w user to the n item.
Optionally, time factor weights may also be incorporated in the generation of the user rating matrix. Wherein the time factor weight characterizes the relationship between the scoring time of the user and the current time. The steps may specifically include: acquiring time factor weights of different users for respective scored articles; and generating a user evaluation matrix based on the time factor weight, the user scoring matrix and the user emotion matrix.
In this way, the expression of the element in the user evaluation matrix is:
Figure BDA0003266439520000112
in the formula (2), cfijThe evaluation value of the ith user on the jth item is shown; c. CijIndicating the grade of the ith user on the jth item;
Figure BDA0003266439520000113
adding elements representing corresponding positions of the matrix; f. ofijRepresenting the sentiment score of the ith user on the jth item; zijRepresenting the time factor weight of the ith user to the jth item; t isjIndicates the time, T, at which the ith user scored the jth itemfirstIndicates the time of the first rating of the ith user, TallRepresenting the total time from the first to the last scoring of the ith user.
Similarly, the condition for the above formula may be cijAnd fijNone of which is zero. That is, when cijAnd fijNeither is zero, the two are added and divided by 2. When any one of the two parameters is zero, the value corresponding to the other parameter is taken as an evaluation value, such as cijIs zero, and fij0.5, cfij＝fij，cfijAlso 0.5. The user evaluation matrix generated in combination with the time factor weight may refer to the form of the user evaluation matrix CF1, which is not described herein again.
When T isjWhen the latest time of scoring of the ith user is represented, Tall＝Tj-Tfirst；Zij1. It can be seen that the closer the time a user scores,the larger the temporal weighting factor. The larger the temporal weighting factor, the greater the referential of the user score. By the method, the reference value of the user score can be further improved, and further the reasonability and the accuracy of recommending articles favored by the user are improved.
Step S104: and acquiring an object similarity target matrix.
The object similarity target matrix comprises similarity between every two objects related in the user scoring matrix.
As a first embodiment, the item similarity target matrix includes only the item similarity matrix. The item similarity matrix characterizes the similarity of the scored times between items.
The calculation formula of the article similarity in the article similarity matrix is as follows:
Figure BDA0003266439520000121
in the formula (3), WijIndicates the item similarity of item i and item j, n (i) and n (j) indicate the number of times item i and item j are scored, respectively, and n (i) and n (j) indicate the number of times item i and item j are scored simultaneously.
And (4) calculating the article similarity between every two articles through a formula (3) to obtain an article similarity matrix. The form of the item similarity matrix D1 is as follows:
Figure BDA0003266439520000122
in the article similarity matrix D1, D is usedijRepresenting an element in the matrix, dijRepresenting item similarity between item i and item j, e.g. d11Indicates the similarity of the article 1 to itself, d12Indicates the article similarity of article 1 and article 2, dwnIndicating the item similarity between item w and item n.
As a second embodiment, the object similarity target matrix includes an object similarity matrix and a time similarity matrix. Correspondingly, the steps specifically include: acquiring an article similarity matrix and a time similarity matrix; the time similarity characterizes the similarity of time-to-market between the items; and generating an article similarity target matrix based on the article similarity matrix and the time similarity matrix.
Since the article similarity matrix has been described in the first embodiment, it is not described herein any more, and the same parts may be referred to each other.
The expression of the elements in the time similarity matrix, i.e. the calculation formula of the time similarity, is as follows:
Figure BDA0003266439520000131
in the formula (4), facTa,bRepresents the similarity of the article a and the article b in time, taIndicates the time of sale of item a, tbIndicating the time of the sale of item b.
Because the same kind of products are on the market together, for example, a large number of love movies appear in the event of a lover, the similarity between the articles can be judged according to the time. The larger the time difference, the lower the correlation.
T aboveaAnd tbThe time can refer to a specific time on the market, such as a certain day of a certain month of a certain year, and the time can also be determined by the form of numbers. For example, if item a is the first marketed item, item a is numbered 1, and if item b is the second marketed item, item a is numbered 2, and so on, the numbers of all items are obtained. The order of the articles on the market can be judged according to the size of the numbers, and the representation mode is adopted, so that the application is not limited. Of course, for the convenience of subsequent unified calculation, the time may be normalized to [0, 1]]The interval of (2).
And (4) calculating the time similarity between every two articles through a formula (4) to obtain a time similarity matrix. The temporal similarity matrix D2 is of the form:
Figure BDA0003266439520000132
in the time similarity matrix D2, facT is usedijRepresenting an element in the matrix, facTijRepresenting time similarity between item i and item j, e.g. facT11Represents the time similarity, facT, of the item 1 to itself12Represents the time similarity, facT, of item 1 to item 2wnRepresenting the time similarity between item w and item n.
After the article similarity matrix D1 and the time similarity matrix D2 are obtained, the two are combined to obtain an article similarity target matrix D4. Note that the two matrices are combined in such a manner that corresponding elements are multiplied.
It can be seen that, in the embodiment of the application, the object similarity target matrix comprises an object similarity matrix and a time similarity matrix, and the similarity between objects in the time of the market can be determined through the time similarity matrix, so that whether two objects are similar objects in the time of the market or not can be determined.
As a third embodiment, the object similarity degree target matrix includes an object similarity degree matrix and an attribute similarity degree matrix. Correspondingly, the steps specifically include: acquiring an article similarity matrix and an attribute similarity matrix; the attribute similarity matrix represents the similarity of attributes among the articles; and generating an article similarity target matrix based on the article similarity matrix and the attribute similarity matrix.
Since the article similarity matrix has been described in the first embodiment, it is not described herein any more, and the same parts may be referred to each other.
Tag information may exist in the original data of the article, and the tag information represents the property of the article, for example, the tag information of article a includes [ a ]1,a2,a3...an]，a1～anEach corresponding to an attribute of article a; the label information of the article b includes [ b1,b2,b3...bn]，b1～bnAll corresponding to an attribute of the article b, the similarity of the attributes between the articles can be obtained according to the attributes between the articles.
The expression of the elements of the attribute similarity matrix, i.e. the calculation formula of the attribute similarity, is as follows:
Figure BDA0003266439520000141
in the formula (5), sima,bThe expression represents the similarity of the attributes of the article a and the article b, [ a ]i]∩[bi]The number of the same attributes of the articles a and b is shown, n is the total number of the attributes of the articles a, and m is the total number of the attributes of the articles b.
And (4) calculating the attribute similarity between every two articles through a formula (5) to obtain an attribute similarity matrix. The form of the attribute similarity matrix D3 is as follows:
Figure BDA0003266439520000151
in the attribute similarity matrix D3, sim is usedijRepresenting the elements, sim, in the matrixijRepresenting similarity of attributes between item i and item j, e.g. sim11Representing the similarity of the properties of the item 1 to itself, sim12Representing the similarity of the attributes, sim, of article 1 and article 2wnRepresenting the similarity of attributes between item w and item n.
After the article similarity matrix D1 and the attribute similarity matrix D3 are obtained, the two are combined to obtain an article similarity target matrix D4. Note that the two matrices are combined in such a manner that corresponding elements are multiplied.
It can be seen that, in the embodiment of the application, the object similarity target matrix comprises an object similarity matrix and an attribute similarity matrix, and the similarity between the objects in the attribute can be determined by combining the attribute similarity matrix, so that whether the two objects are the same in the attribute can be determined.
As a fourth embodiment, the object similarity degree matrix includes an object similarity degree matrix, a time similarity degree matrix, and an attribute similarity degree matrix. Correspondingly, the steps specifically include: acquiring an article similarity matrix, a time similarity matrix and an attribute similarity matrix; the article similarity matrix represents the similarity of the scored times among the articles; the time similarity characterizes the similarity of time-to-market between the items; the attribute similarity matrix represents the similarity of attributes among the articles; and generating an article similarity target matrix based on the article similarity matrix, the time similarity matrix and the attribute similarity matrix.
Since the article similarity matrix, the time similarity matrix, and the attribute similarity matrix have been described in the foregoing embodiments, they are not described herein again, and the same portions may be referred to each other. Here, it should be noted that the three matrices are combined in a manner that corresponding elements in the three matrices are multiplied.
It can be seen that, in the embodiment of the present application, the object similarity target matrix simultaneously includes an object similarity matrix, a time similarity matrix, and an attribute similarity matrix, and the similarity between the objects in the time of marketing can be determined through the time similarity matrix, so as to determine whether the two objects are similar objects in the time of marketing, and the similarity between the objects in the attribute can be determined through the attribute similarity matrix, so as to determine whether the two objects are the same in the attribute. Through the combination of the three, the accuracy and the rationality of the similarity between the objects can be further improved.
Note that the matrices of the above combinations are arranged in the same order.
Step S105: recommending the object to the target user based on the object similarity target matrix and the user evaluation matrix.
And finally, recommending the object to the target user according to the object similarity target matrix and the user evaluation matrix. The target user is a user in the user scoring matrix.
Optionally, the steps specifically include: acquiring the love degree of a target user to the article related to the user evaluation matrix based on the article similarity target matrix and the user evaluation matrix; and recommending the objects which are not scored by the target user to the target user based on the ranking sequence of the high-preference degrees to the low-preference degrees.
Wherein the like degree is obtained by the following expression:
Figure BDA0003266439520000161
in the formula (6), u represents a target user, and p (u, j) represents the favorite degree of the target user to the item j; s (j, k) represents a set of k articles with the highest similarity to the article j, and the k articles with the highest similarity to the article j can be obtained through an article similarity target matrix; n is a radical ofuRepresenting a collection of items liked by the target user, NuN may be constructed by a user scoring matrix determination, such as determining the three highest scoring itemsu；DjiRepresenting the similarity of the article j and the article i, and obtaining the similarity through an article similarity target matrix; cfuiAnd the evaluation value of the target user on the article i is obtained through the user evaluation matrix.
Therefore, in the embodiment of the application, the objects which are not scored by the target user are recommended to the target user based on the ranking sequence of the favorite degrees from high to low, so that the favorite objects of the target user are recommended preferentially, the scored objects are prevented from being recommended to the target user, and the user experience is improved.
In summary, in the embodiment of the application, the comment information of the user on the articles is fully dug, the user emotion matrix is generated, the user emotion matrix is combined with the user rating matrix, the user preference is judged from two dimensions of rating and comment, and then the similarity between the articles in the article similarity target matrix is combined, so that the articles favored by the user are accurately recommended.
Referring to fig. 3, based on the same inventive concept, an embodiment of the present application further provides an article recommendation device 200, including: a first obtaining module 210, a second obtaining module 220, a generating module 230, a third obtaining module 240 and a recommending module 250.
A first obtaining module 210, configured to obtain a user scoring matrix; wherein the user scoring matrix comprises the scores of the respective scored items by different users.
A second obtaining module 220, configured to obtain a user emotion matrix; wherein the user emotion matrix comprises emotion scores of the different users for respective commented items; the sentiment score is generated by sentiment analysis of the comments of the user.
A generating module 230, configured to generate a user evaluation matrix based on the user scoring matrix and the user emotion matrix.
A third obtaining module 240, configured to obtain an object similarity target matrix; and the item similarity target matrix comprises the similarity between every two items related in the user scoring matrix.
A recommending module 250, configured to recommend an item to a target user based on the item similarity target matrix and the user evaluation matrix; wherein the target user is a user in the user scoring matrix.
Optionally, the second obtaining module 220 is specifically configured to obtain comments of the different users on the item; scoring the comments based on an emotion analysis tool to obtain emotion scores of the different users on the respective commented articles; and generating the user emotion matrix based on the emotion scores of the different users on the respective commented articles.
Optionally, the generating module 230 is specifically configured to obtain time factor weights of the different users for the respective scored items; wherein the time factor weight characterizes a relationship between the scoring time of the user and the current time; and generating the user evaluation matrix based on the time factor weight, the user scoring matrix and the user emotion matrix.
Optionally, the third obtaining module 240 is specifically configured to obtain an article similarity matrix and a time similarity matrix; the article similarity matrix represents the similarity of the scored times between the articles; the time similarity characterizes the similarity of time-to-market between the items; and generating the object similarity matrix based on the object similarity matrix and the time similarity matrix.
Optionally, the third obtaining module 240 is specifically configured to obtain an article similarity matrix and an attribute similarity matrix; the attribute similarity matrix represents the similarity of attributes among the articles; and generating the object similarity target matrix based on the object similarity matrix and the attribute similarity matrix.
Optionally, the third obtaining module 240 is specifically configured to obtain an article similarity matrix, a time similarity matrix, and an attribute similarity matrix; the article similarity matrix represents the similarity of the scored times between the articles; the time similarity characterizes the similarity of time-to-market between the items; the attribute similarity matrix represents the similarity of attributes among the articles; and generating the object similarity target matrix based on the object similarity matrix, the time similarity matrix and the attribute similarity matrix.
Optionally, the recommending module 250 is specifically configured to obtain, based on the object similarity target matrix and the user evaluation matrix, a degree of liking of the target user to the object related to the user evaluation matrix; recommending the unscored items of the target user to the target user based on the ranking sequence of the high-low likeness.
It should be noted that, as those skilled in the art can clearly understand, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
Based on the same inventive concept, embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed, the computer program performs the methods provided in the above embodiments.
The storage medium may be any available medium that can be accessed by a computer or a data storage device including one or more integrated servers, data centers, and the like. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and there may be other divisions when actually implemented, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form.
In addition, units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
Furthermore, the functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
In this document, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims (14)

1. An item recommendation method, comprising:
acquiring a user scoring matrix; wherein the user scoring matrix comprises the scores of the respective scored items by different users;
acquiring a user emotion matrix; wherein the user emotion matrix comprises emotion scores of the different users for respective commented items; the emotion score is generated by performing emotion analysis on the comment of the user;
generating a user evaluation matrix based on the user scoring matrix and the user emotion matrix;
acquiring an object similarity target matrix; the object similarity target matrix comprises similarity between every two objects related in the user scoring matrix;
recommending articles to target users based on the article similarity target matrix and the user evaluation matrix; wherein the target user is a user in the user scoring matrix.
2. The method of claim 1, wherein obtaining the user emotion matrix comprises:
obtaining comments of the different users on the article;
scoring the comments based on an emotion analysis tool to obtain emotion scores of the different users on the respective commented articles;
and generating the user emotion matrix based on the emotion scores of the different users on the respective commented articles.
3. The method of claim 1, wherein generating a user rating matrix based on the user rating matrix and the user emotion matrix comprises:
acquiring the time factor weight of the different users to the respectively scored goods; wherein the time factor weight characterizes a relationship between the scoring time of the user and the current time;
and generating the user evaluation matrix based on the time factor weight, the user scoring matrix and the user emotion matrix.
4. The method of claim 3, wherein the expression of the elements in the user evaluation matrix is:
Figure FDA0003266439510000021
wherein, cfijThe evaluation value of the ith user on the jth item is shown; c. CijRepresenting a rating of the jth item by the ith user;
Figure FDA0003266439510000022
adding elements representing corresponding positions of the matrix; f. ofijRepresenting the sentiment score of the ith user for the jth item; zijRepresenting a time factor weight of the ith user for the jth item; t isjRepresents the time, T, at which the ith user scored the jth itemfirstRepresents the time, T, of the first rating of the ith userallRepresenting the total time from the first to the last scoring of the ith user.
5. The method of claim 1, wherein the obtaining an item similarity target matrix comprises:
acquiring an article similarity matrix and a time similarity matrix; the article similarity matrix represents the similarity of the scored times between the articles; the time similarity characterizes the similarity of time-to-market between the items;
and generating the object similarity matrix based on the object similarity matrix and the time similarity matrix.
6. The method of claim 5, wherein the expression of the elements in the temporal similarity matrix is:
Figure FDA0003266439510000023
wherein, facTa,bRepresents the similarity of the article a and the article b in time, taRepresents the time of the article a on the market, tbRepresenting the time of the listing of said item b.
7. The method of claim 1, wherein the obtaining an item similarity target matrix comprises:
acquiring an article similarity matrix and an attribute similarity matrix; the attribute similarity matrix represents the similarity of attributes among the articles;
and generating the object similarity target matrix based on the object similarity matrix and the attribute similarity matrix.
8. The method of claim 7, wherein the expression of the elements in the attribute similarity matrix is:
Figure FDA0003266439510000031
wherein, sima,bThe expression represents the similarity of the attributes of the article a and the article b, [ a ]i]∩[bi]The number of the same attributes of the article a and the article b is represented, n represents the total number of the attributes of the article a, and m represents the total number of the attributes of the article b.
9. The method of claim 1, wherein the obtaining an item similarity target matrix comprises:
acquiring an article similarity matrix, a time similarity matrix and an attribute similarity matrix; the article similarity matrix represents the similarity of the scored times between the articles; the time similarity characterizes the similarity of time-to-market between the items; the attribute similarity matrix represents the similarity of attributes among the articles;
and generating the object similarity target matrix based on the object similarity matrix, the time similarity matrix and the attribute similarity matrix.
10. The method of claim 1, wherein recommending the item to the target user based on the item similarity objective matrix and the user rating matrix comprises:
based on the object similarity target matrix and the user evaluation matrix, acquiring the love degree of the target user to the object related to the user evaluation matrix;
recommending the unscored items of the target user to the target user based on the ranking sequence of the high-low likeness.
11. The method according to claim 10, wherein the like degree is obtained by the following expression:
Figure FDA0003266439510000041
wherein u represents the target user, and p (u, j) represents the favorite degree of the target user to the item j; s (j, k) represents a set of k items with the highest similarity to the item j, NuA set of items representing the target user's liking; djiRepresenting the similarity of the article j and the article i, and obtaining the similarity through the article similarity target matrix; cfuiAnd the evaluation value of the target user on the item i is represented and obtained through the user evaluation matrix.
12. An item recommendation device, comprising:
the first acquisition module is used for acquiring a user scoring matrix; wherein the user scoring matrix comprises the scores of the respective scored items by different users;
the second acquisition module is used for acquiring the user emotion matrix; wherein the user emotion matrix comprises emotion scores of the different users for respective commented items; the emotion score is generated by performing emotion analysis on the comment of the user;
the generating module is used for generating a user evaluation matrix based on the user scoring matrix and the user emotion matrix;
the third acquisition module is used for acquiring an object similarity target matrix; the object similarity target matrix comprises similarity between every two objects related in the user scoring matrix;
the recommending module is used for recommending the object to the target user based on the object similarity target matrix and the user evaluation matrix; wherein the target user is a user in the user scoring matrix.
13. An electronic device, comprising: a processor and a memory, the processor and the memory connected;
the memory is used for storing programs;
the processor is configured to execute a program stored in the memory to perform the method of any of claims 1-11.
14. A computer-readable storage medium, on which a computer program is stored which, when executed by a computer, performs the method of any one of claims 1-11.
CN202111087804.6A 2021-09-16 2021-09-16 Article recommendation method and device, electronic equipment and storage medium Pending CN113781180A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111087804.6A CN113781180A (en) 2021-09-16 2021-09-16 Article recommendation method and device, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111087804.6A CN113781180A (en) 2021-09-16 2021-09-16 Article recommendation method and device, electronic equipment and storage medium

Publications (1)

Publication Number Publication Date
CN113781180A true CN113781180A (en) 2021-12-10

Family

ID=78851589

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111087804.6A Pending CN113781180A (en) 2021-09-16 2021-09-16 Article recommendation method and device, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN113781180A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114429384A (en) * 2021-12-30 2022-05-03 杭州盟码科技有限公司 Intelligent product recommendation method and system based on e-commerce platform

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104699687A (en) * 2013-12-04 2015-06-10 中国移动通信集团广东有限公司 Item recommendation method and server
CN105069072A (en) * 2015-07-30 2015-11-18 天津大学 Emotional analysis based mixed user scoring information recommendation method and apparatus
CN109635291A (en) * 2018-12-04 2019-04-16 重庆理工大学 A kind of recommended method of fusion score information and item contents based on coorinated training
CN110827118A (en) * 2019-10-18 2020-02-21 天津大学 Method for automatically analyzing user comments in application store and recommending user comments to developer
CN111339439A (en) * 2020-02-18 2020-06-26 广东工业大学 Collaborative filtering recommendation method and device fusing comment text and time sequence effect
CN111523055A (en) * 2020-04-28 2020-08-11 安徽农业大学 Collaborative recommendation method and system based on agricultural product characteristic attribute comment tendency
CN113268667A (en) * 2021-05-28 2021-08-17 汕头大学 Chinese comment emotion guidance-based sequence recommendation method and system

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104699687A (en) * 2013-12-04 2015-06-10 中国移动通信集团广东有限公司 Item recommendation method and server
CN105069072A (en) * 2015-07-30 2015-11-18 天津大学 Emotional analysis based mixed user scoring information recommendation method and apparatus
CN109635291A (en) * 2018-12-04 2019-04-16 重庆理工大学 A kind of recommended method of fusion score information and item contents based on coorinated training
CN110827118A (en) * 2019-10-18 2020-02-21 天津大学 Method for automatically analyzing user comments in application store and recommending user comments to developer
CN111339439A (en) * 2020-02-18 2020-06-26 广东工业大学 Collaborative filtering recommendation method and device fusing comment text and time sequence effect
CN111523055A (en) * 2020-04-28 2020-08-11 安徽农业大学 Collaborative recommendation method and system based on agricultural product characteristic attribute comment tendency
CN113268667A (en) * 2021-05-28 2021-08-17 汕头大学 Chinese comment emotion guidance-based sequence recommendation method and system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114429384A (en) * 2021-12-30 2022-05-03 杭州盟码科技有限公司 Intelligent product recommendation method and system based on e-commerce platform
CN114429384B (en) * 2021-12-30 2022-12-09 杭州盟码科技有限公司 Intelligent product recommendation method and system based on e-commerce platform

Similar Documents

Publication Publication Date Title
CN109559208B (en) Information recommendation method, server and computer readable medium
US9785989B2 (en) Determining a characteristic group
US9569499B2 (en) Method and apparatus for recommending content on the internet by evaluating users having similar preference tendencies
TWI512653B (en) Information providing method and apparatus, method and apparatus for determining the degree of comprehensive relevance
JP6022056B2 (en) Generate search results
CN109064293B (en) Commodity recommendation method and device, computer equipment and storage medium
CN109300003B (en) Enterprise recommendation method, enterprise recommendation device, computer equipment and storage medium
US20130254181A1 (en) Aggregation and Categorization
WO2020192013A1 (en) Directional advertisement delivery method and apparatus, and device and storage medium
CN108509466A (en) A kind of information recommendation method and device
US20160306890A1 (en) Methods and systems for assessing excessive accessory listings in search results
CN112529663B (en) Commodity recommendation method, commodity recommendation device, terminal equipment and storage medium
US20200134689A1 (en) Product and content association
CN109241403A (en) Item recommendation method, device, machinery equipment and computer readable storage medium
CN113032668A (en) Product recommendation method, device and equipment based on user portrait and storage medium
CN110825977A (en) Data recommendation method and related equipment
CN111582932A (en) Inter-scene information pushing method and device, computer equipment and storage medium
US20180182030A1 (en) Determination device, determination method, and non-transitory computer-readable recording medium
CN107305615B (en) Data table identification method and system
CN113781180A (en) Article recommendation method and device, electronic equipment and storage medium
US20220114624A1 (en) Digital Content Text Processing and Review Techniques
Zhang et al. The approaches to contextual transaction trust computation in e‐Commerce environments
CN111353836B (en) Commodity recommendation method, device and equipment
JP6554306B2 (en) Information processing system, information processing method, and computer program
CN111339419A (en) Information recommendation method and device, electronic equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination