CN113821727A - Item recommendation method, computer device and computer-readable storage medium - Google Patents
Item recommendation method, computer device and computer-readable storage medium Download PDFInfo
- Publication number
- CN113821727A CN113821727A CN202111121782.0A CN202111121782A CN113821727A CN 113821727 A CN113821727 A CN 113821727A CN 202111121782 A CN202111121782 A CN 202111121782A CN 113821727 A CN113821727 A CN 113821727A
- Authority
- CN
- China
- Prior art keywords
- item
- target user
- evaluation information
- score
- user
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000011156 evaluation Methods 0.000 claims abstract description 97
- 230000008859 change Effects 0.000 claims abstract description 18
- 238000012545 processing Methods 0.000 claims description 39
- 238000006243 chemical reaction Methods 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 7
- 238000012896 Statistical algorithm Methods 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 5
- 230000000694 effects Effects 0.000 abstract description 8
- 230000008569 process Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 230000009471 action Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000006866 deterioration Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000002238 attenuated effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- 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
Abstract
An item recommendation method, a computer device and a computer-readable storage medium, the method comprising: obtaining evaluation information of a plurality of articles to be selected; extracting the theme of the evaluation information of a plurality of articles to be selected by using a preset theme model to obtain the comment theme distribution of each evaluation information; acquiring an incidence relation between each piece of evaluation information and a target user, wherein the incidence relation comprises at least one of an evaluation score index, an interest change index and a heat attenuation index; obtaining preference theme distribution corresponding to the target user according to the incidence relation and the comment theme distribution of each piece of evaluation information; calculating to obtain a prediction score of each to-be-selected item in a plurality of to-be-selected items based on the preference theme distribution of the target user; and selecting at least one item from the multiple items to be selected according to the prediction score of each item to be selected and recommending the item to a target user. According to the method and the device, the article recommendation is carried out based on the incidence relation between the evaluation information and the target user, the article recommendation is more accurate, and the recommendation effect is better.
Description
Technical Field
The present application relates to the field of data processing technologies, and in particular, to an item recommendation method, a computer device, and a computer-readable storage medium.
Background
The recommendation system is used as an information screening tool, exists on the basis of mass data, and can effectively solve the problem of information overload. The recommendation system can dig out items (such as information, services, articles and the like) which are interested by the user from the mass data through a recommendation algorithm, and recommend the result to the user so as to meet the requirements of the user.
Although the conventional collaborative filtering algorithm can reduce the influence of data sparsity and cold start on the recommendation result, the factors that the interest of a user may change or the heat degree of an article is attenuated and the like are not considered, so that the recommendation effect is poor and the use experience of the user is influenced.
Disclosure of Invention
In view of the above, it is desirable to provide an item recommendation method, a computer device and a computer-readable storage medium for solving the technical problem that the accuracy of item recommendation is not high.
An embodiment of the present application provides an article recommendation method, including: obtaining evaluation information of a plurality of articles to be selected; extracting the theme of the evaluation information of a plurality of articles to be selected by using a preset theme model to obtain the comment theme distribution of each evaluation information; acquiring an incidence relation between each piece of evaluation information and a target user, wherein the incidence relation comprises at least one of an evaluation score index, an interest change index and a heat attenuation index; obtaining preference theme distribution corresponding to the target user according to the incidence relation and the comment theme distribution of each piece of evaluation information; calculating to obtain a prediction score of each to-be-selected item in a plurality of to-be-selected items based on the preference theme distribution of the target user; and selecting at least one item from the multiple items to be selected according to the prediction score of each item to be selected and recommending the item to a target user.
In some embodiments, the item recommendation method further comprises: preprocessing the evaluation information, wherein the preprocessing comprises at least one of word segmentation processing, stop word removing processing, part of speech tagging processing, case and case conversion processing, part of speech restoring processing and stem extraction processing; and performing text vectorization processing on the preprocessed evaluation information to obtain a vector corresponding to the evaluation information.
In some embodiments, the text vectorization processing is performed on the preprocessed evaluation information, and includes: constructing a text set based on the preprocessed multiple evaluation information; extracting non-repeated words from a text set by using a preset word frequency statistical algorithm, and constructing a word bank table based on the extracted words, wherein the word bank table comprises V words, and V is a positive integer; comparing the preprocessed evaluation information with the word stock table based on a preset comparison rule to obtain a V-dimensional vector corresponding to the evaluation information; wherein, the preset comparison rule comprises: if the evaluation information comprises a certain word in the word bank table, adding n to the word count in the word bank table, wherein n is the number of times of the word appearing in the evaluation information and is a positive integer, and if the evaluation information does not comprise the certain word in the word bank table, adding 0 to the word count in the word bank table.
In some embodiments, the preset topic model is an LDA topic model, and the distribution of the preferred topics corresponding to the target user is calculated according to the following equation:
wherein, PuFor a preferred topic distribution corresponding to the target user u, AuiEvaluation score index of item i to be selected for target user u, BuiInterest change index of object i to be selected for target user u, CuiHeat attenuation index, K, for object user u to select item iuiK-dimensional comment subject distribution extracted from evaluation information of item I to be selected for LDA subject model, IuFor the purpose ofUser u evaluated the collection of items.
In some embodiments, the evaluation score index of the object i to be selected by the target user u is obtained by conversion based on the score of the object i to be selected by the target user u, the interest change index of the object i to be selected by the target user u is obtained by conversion based on the score time of the object i to be selected by the target user u, and the heat attenuation index of the object i to be selected by the target user u is obtained by conversion based on the scored time span of the object i to be selected, the number of clicks of the user u, the number of comments of the user u and the browsing duration of the user u.
In some embodiments, calculating a prediction score for each of a plurality of candidate items based on the distribution of the target user's preferred topics includes: constructing a similar user set which is close to the target user, wherein the similar user set comprises at least one similar user; and calculating to obtain the prediction score of the selected item based on the historical average score evaluated by the target user, the historical average score evaluated by the similar users, the score of the similar users on the selected item, the heat degree of the selected item, the preference theme distribution of the target user and the preference theme distribution of the similar users.
In some embodiments, the predicted score of the item to be selected is calculated by the following equation:
wherein Pre is the prediction score of the item i to be selected, u1A similar user to the target user u,
historical average score evaluated for target user u，
For similar users u1Historical mean score, Hot, evaluatediuN is the similar user set adjacent to the target user u,
for similar users u1Scoring of items to be selected i, QunFor the distribution probability of the target user u on the nth topic,
for similar users u1The probability of the distribution on the nth topic,
for the average preferred topic distribution of the target user u,
for similar users u1Average preference topic distribution of DiuFor the number of times of clicking on the item i to be selected by the target user u, EiuFor the duration that the item i to be selected is browsed by the target user u, FiuFor the number of times the item i to be selected is reviewed by the target user u,
epsilon and eta are preset constants, and k epsilon is (1,2, …, n).
In some embodiments, selecting at least one item from the plurality of items to be selected for recommendation to the target user according to the prediction score of each item to be selected includes: acquiring historical purchase record information of a target user; and selecting the articles which are preset before the predicted score ranking and are not purchased by the target user from the plurality of articles to be selected according to the predicted score and the historical purchase record information of each article to be selected, and recommending the articles to the target user.
An embodiment of the present application provides a computer device, which includes a processor and a memory, where the memory stores a plurality of computer programs, and the processor is configured to implement the steps of the item recommendation method when executing the computer programs stored in the memory.
An embodiment of the present application further provides a computer-readable storage medium, which stores a plurality of instructions, where the plurality of instructions are executable by one or more processors to implement the steps of the item recommendation method.
Compared with the prior art, the item recommendation method recommends the items which are interested by the target user based on the preferences of acquainted user groups, considers the influence of factors such as interest change conditions of the target user, evaluation scores of the user, heat attenuation of the items and the like, and has more accurate item recommendation and better recommendation effect.
Drawings
Fig. 1 is an application environment diagram of an item recommendation method according to an embodiment of the present application.
Fig. 2 is a flowchart of an item recommendation method according to an embodiment of the present application.
Fig. 3 is a functional block diagram of an article recommendation device according to an embodiment of the present application.
Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Description of the main elements
The following detailed description will further illustrate the present application in conjunction with the above-described figures.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
It is further noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
Fig. 1 is an application environment diagram of an item recommendation method according to an embodiment of the present application. Referring to fig. 1, an item recommendation method is applied to an item recommendation system. The item recommendation system may include a terminal 11 and a server 12, and the terminal 11 may be an electronic device used by a target user. The terminal 11 and the server 12 are connected through a wired network or a wireless network. The terminal 11 may specifically be a desktop terminal or a mobile terminal, and the mobile terminal may specifically be at least one of a mobile phone, a tablet computer, a notebook computer, and the like. The server 12 may be implemented as a stand-alone server or as a server cluster of multiple servers.
The item recommendation system is a tool for associating users and items based on mass data mining, and can help the users to screen information in which the users are interested in the information overload environment, and personalized decision support and information service are provided for the users. The item recommendation system can refer to recommendation of commodities (such as foods, living goods, electronic products, clothes and the like), recommendation of audio and video contents, recommendation of articles and the like for a user. This is not a limitation of the present application.
Fig. 2 is a flowchart of an item recommendation method according to an embodiment of the present application. The present embodiment is exemplified by applying the method to the server 12 in fig. 1, and according to different requirements, the order of the steps in the flowchart may be changed, and some steps may be omitted.
And step S20, obtaining evaluation information of a plurality of articles to be selected.
In some embodiments, the plurality of candidate items may be preset candidate items recommended to the target user, for example, the plurality of candidate items may be a plurality of items sold on a shopping website or a shopping application. When a user purchases an item on a shopping website or a shopping application through an electronic device, the user generally goes through some steps of click-view-buy-comment. The server of the shopping website or the shopping application stores the evaluation information of a plurality of sold articles, and further can acquire the evaluation information of a plurality of articles to be selected from the server.
In some embodiments, when obtaining the evaluation information of the to-be-selected item, the evaluation information may be preprocessed. The preprocessing may include at least one of word segmentation processing, stop word rejection processing, part-of-speech tagging processing, case conversion processing, part-of-speech restoration processing, and stem extraction processing. For example, for evaluation information of a chinese text, the preprocessing may include: word segmentation processing; removing stop words; and (5) processing part-of-speech tagging. For evaluation information of English texts, preprocessing can comprise capital and lower case conversion processing (converting all words into lower case in a unified way), for example, the words can be identified as one word by 'Like' and 'Like'; stop word eliminating processing, such as eliminating "is", "@", ": "," in "," a ", etc.; performing part-of-speech tagging processing by using a tag device provided with corresponding part-of-speech types in various forms, for example, a common auxiliary word with a suffix of "ly" and a common noun with a suffix of "ness"; the part of speech is restored and processed to obtain the basic form of the vocabulary, for example, "catching" is processed to "catch"; and stem extraction processing, such as processing "applets" into "applets".
In some embodiments, after the evaluation information is preprocessed, text vectorization processing may be performed on the preprocessed evaluation information to obtain a vector corresponding to the evaluation information. For example, the text vectorization processing for each evaluation information can be realized by: based on pretreated polypeptidesConstructing a text set by the evaluation information; extracting non-repeated words from the text set by using a preset word frequency statistical algorithm, and constructing a word bank table based on the extracted words, wherein the word bank table comprises V words, and if the word bank table can be expressed as { word bank table }1Word 2, word3…, the word V }, V being a positive integer; and comparing the preprocessed evaluation information with the word bank table based on a preset comparison rule to obtain a V-dimensional vector corresponding to the evaluation information, namely, each piece of evaluation information is represented by a V-dimensional vector. The preset word frequency statistical algorithm can be selected according to actual requirements, such as a word frequency statistical algorithm based on a deep learning network. The preset alignment rule may include: each word in the word bank table corresponds to the same initial value, if the evaluation information includes a certain word in the word bank table, the word count in the word bank table is increased by n, n is the number of times that the word appears in the evaluation information, n is a positive integer, and if the evaluation information does not include a certain word in the word bank table, the word count in the word bank table is increased by 0.
And 21, extracting the topics of the evaluation information of the multiple articles to be selected by using a preset topic model to obtain the comment topic distribution of each evaluation information.
In some embodiments, the preset topic model may be a Latent Dirichlet Allocation (LDA) topic model. The LDA topic model may present the topic of each document in the set of documents in the form of a probability distribution. The evaluation information of a plurality of articles to be selected may be arranged in a form of a document set, for example, each piece of evaluation information may be regarded as or arranged as a document, all documents are put into an LDA topic model, and the LDA may give the topic of each piece of evaluation information in a form of probability distribution, that is, the comment topic distribution of each piece of evaluation information.
And step 22, acquiring the association relation between each piece of evaluation information and the target user.
In some embodiments, the target object recommended by the target user for the current item may refer to a user using the item recommendation system, for example. The correlation may include at least one of an evaluation score index, an interest change index, and a heat fading index. The following description will take an example in which the correlation includes an evaluation score index, an interest change index, and a heat deterioration index.
In some embodiments, each piece of evaluation information is an evaluation of a user on a certain article, and the evaluation score index may represent a user's preference degree on the article, and may be obtained by performing 0-1 normalization processing on the evaluation score of the user. For example, the evaluation score index of a certain article can be calculated by the following equation (i):
wherein A isuiAn evaluation score index of the object i to be selected for the target user u, auiScoring of item to be selected i for target user u, aminMinimum score for item i to be selected, a, for target user umaxAnd b, the maximum score of the object user u to be selected i is obtained, beta is an action parameter of the adjustment score on the evaluation score index, the value of beta can be 0-1, the larger the value of beta is, the larger the effect of the score on the comment score index is, otherwise, the smaller the value of beta is, the smaller the effect of the score on the comment score index is. w is a1Is a preset weight, w1Is a constant and e is a mathematical constant. From the above equation (i), it can be found that, when the score of the user is higher, the corresponding comment score index is higher, which indicates that the user has a higher degree of favor and activity on the item.
In some embodiments, the interest change index can be used for predicting interest change of a user on an article, and can be obtained by performing 0-1 standardization processing on comment time of the user. For example, the interest change index of an item can be calculated by the following equation (ii):
wherein, BuiInterest change index of object to be selected i for target user u, buiTime of scoring of item to be selected i for target user u, bminFor the purpose ofEarliest scoring time of item i to be selected by user u, bmaxAnd d, the latest scoring time of the object i to be selected is the target user u, x is an action parameter of adjusting the scoring time on the interest change index, and the value of x can be between 0 and 1. w is a2Is a preset weight, w2Is a constant. From the above equation (ii), it can be concluded that the longer the time the user scores the item, the smaller the influence of the score on the item recommendation result, and the closer the time the user scores the item, the larger the influence of the score on the item recommendation result.
In some embodiments, the heat decay index may be used to predict heat decay changes of the item, and may be converted based on the scored time span, the number of clicks by the user, the number of comments by the user, and the browsing duration of the item by the user. For example, the heat deterioration index of an article can be calculated by the following equation (iii):
wherein, CuiHeat attenuation index of item i to be selected for target user u, cuiTime span scored for item to be selected i, cminIs the earliest scored time of the item i to be selected, cmaxThe latest scoring time of the item i to be selected is delta is an action parameter for adjusting the scoring time to the heat attenuation index, and the value of delta can be between 0 and 1. w is a3Is a preset weight, w3Is a constant. HotiuAs heat of the item i to be selected, DiuFor the number of times of clicking on the item i to be selected by the target user u, EiuFor the duration that the item i to be selected is browsed by the target user u, FiuFor the number of times the item i to be selected is reviewed by the target user u,
both ε and η are predetermined constants, e.g.
The values of epsilon and eta can be both 0-1.
And step 23, obtaining preference theme distribution corresponding to the target user according to the incidence relation and the comment theme distribution of each piece of evaluation information.
In some embodiments, the comment subject distribution of the evaluation information may be converted into a preference subject distribution corresponding to the target user based on the association relationship between the evaluation information and the target user, and the preference subject distribution corresponding to the target user may be calculated by the following equation (iv):
wherein, PuFor a preferred topic distribution corresponding to the target user u, AuiEvaluation score index of item i to be selected for target user u, BuiInterest change index of object i to be selected for target user u, CuiHeat attenuation index, K, for object user u to select item iuiA K-dimensional comment subject distribution, I, extracted from evaluation information of an item I to be selected based on an LDA subject modeluThe evaluated item sets for the target user u.
And 24, calculating to obtain the prediction score of each to-be-selected item in the multiple to-be-selected items based on the preference theme distribution of the target user.
In some embodiments, the predicted score of the item to be selected may be used to characterize the degree of purchase intent of the item by the target user, with a higher predicted score indicating a higher degree of matching of the item with the target user, a higher likelihood of reaching a transaction.
In some embodiments, the predicted score of the item to be selected may be calculated by constructing a similar user set in close proximity to the target user, wherein the similar user set includes at least one similar user, and the predicted score of the item to be selected is calculated based on a historical average score evaluated by the target user (e.g., an average of all scores submitted on the shopping website or shopping application), a historical average score evaluated by the similar user, scores of the item to be selected by the similar user, a popularity of the item to be selected, a preference topic distribution of the target user, and a preference topic distribution of the similar user. Similar users may refer to users who have a user profile with a target user, and the information included in the user profile may be: age, gender, occupation, city, historical purchase records, etc., similar users may also refer to users that have an intersection with the target user (e.g., users in the target user's buddy list/address book).
In some embodiments, the prediction score of each candidate item may be calculated by the following equation (v):
wherein Pre is the prediction score of the item i to be selected, u1A similar user to the target user u,
the historical average score evaluated for the target user u,
representing similar users u1Historical mean score, Hot, evaluatediuN is the similar user set adjacent to the target user u,
for similar users u1Scoring of items to be selected i, QunFor the distribution probability of the target user u on the nth topic,
for similar users u1The probability of the distribution on the nth topic,
for the average preferred topic distribution of the target user u,
for similar users u1K ∈ (1,2, …, n). sim (u, u)1) May be used to characterize the similarity between the target user u and the similar user u 1. In some embodiments, the target user and a plurality of similar users may also be ranked according to the similarity, and the items purchased by the acquaintance users with high similarity and not purchased by the target user are recommended to the target user.
And 25, selecting at least one item from the multiple items to be selected according to the prediction score of each item to be selected and recommending the item to the target user.
In some embodiments, when the prediction score of each candidate item is calculated, the candidate items may be ranked according to the prediction scores, and m (e.g., m top ranked) candidate items with higher prediction scores are recommended to the target user.
In some embodiments, historical purchase record information of the target user can be acquired, and according to the prediction score of each article to be selected and the historical purchase record information of the target user, an article which is preset before the prediction score and is not purchased by the target user is selected from a plurality of articles to be selected and recommended to the target user, so that the recommendation effect of the article can be further improved.
According to the item recommendation method, the items which are interested by the target user are recommended based on the preferences of the acquainted user groups, the influence of factors such as interest change conditions of the target user, evaluation scores of the user, heat attenuation of the items and the like is considered, the item recommendation is more accurate, and the recommendation effect is better.
Based on the same idea as the item recommendation method in the above embodiment, the present application also provides an item recommendation apparatus, which may be used to execute the above item recommendation method. For convenience of explanation, only the parts related to the embodiments of the present application are shown in the schematic structural diagram of the embodiments of the article recommendation device, and those skilled in the art will understand that the illustrated structure does not constitute a limitation of the device, and may include more or less components than those illustrated, or combine some components, or arrange different components.
As shown in fig. 3, the item recommendation apparatus 100 includes a first obtaining module 101, a processing module 102, an extracting module 103, a second obtaining module 104, a converting module 105, a calculating module 106, and a recommending module 107. In some embodiments, the modules may be programmable software instructions stored in a memory and invoked for execution by a processor. It will be appreciated that in other embodiments, the modules may also be program instructions or firmware (firmware) that are resident in the processor.
The first obtaining module 101 is configured to obtain evaluation information of a plurality of articles to be selected.
The processing module 102 is configured to perform preprocessing and text vectorization on the evaluation information to obtain a vector corresponding to the evaluation information.
In some embodiments, the preprocessing may include at least one of a word segmentation process, a stop word culling process, a part-of-speech tagging process, a case conversion process, a part-of-speech restoration process, and a stem extraction process. The processing module 102 may construct a text set based on the preprocessed multiple evaluation information, extract non-repeated words from the text set by using a preset word frequency statistical algorithm, construct a word bank table based on the extracted words, and compare the preprocessed evaluation information with the word bank table based on a preset comparison rule to obtain a V-dimensional vector corresponding to the evaluation information.
The extraction module 103 is configured to perform theme extraction on the evaluation information of the multiple articles to be selected by using a preset theme model, so as to obtain comment theme distribution of each evaluation information. For example, the preset topic model may be an LDA topic model.
The second obtaining module 104 is configured to obtain an association relationship between each piece of evaluation information and the target user. The correlation may include at least one of an evaluation score indicator, an interest change indicator, and a heat decay indicator.
The conversion module 105 is configured to obtain a preference topic distribution corresponding to the target user according to the association relationship and the comment topic distribution of each piece of evaluation information.
The calculating module 106 is configured to calculate a prediction score of each of the multiple candidate items based on the preference topic distribution of the target user.
In some embodiments, the calculation module 106 may construct a similar user set that is close to the target user, where the similar user set includes at least one similar user, and calculate a prediction score of the to-be-selected item based on the historical average score evaluated by the target user, the historical average score evaluated by the similar user, the score of the to-be-selected item by the similar user, the popularity of the to-be-selected item, the preference topic distribution of the target user, and the preference topic distribution of the similar user.
The recommending module 107 is configured to select at least one item from the multiple items to be selected according to the prediction score of each item to be selected and recommend the at least one item to the target user.
In some embodiments, the recommending module 107 may obtain historical purchase record information of the target user, and select an item, which is a preset position before the predicted score ranking and has not been purchased by the target user, from the multiple items to be selected according to the predicted score of each item to be selected and the historical purchase record information, and recommend the item to the target user.
Fig. 4 is a schematic diagram of a hardware structure of the electronic device 10 according to an embodiment of the present disclosure. As shown in fig. 4, the electronic device 10 may include a processor 1001, a memory 1002, and a communication bus 1003. The memory 1002 is used to store one or more computer programs 1004. One or more computer programs 1004 are configured to be executed by the processor 1001. The one or more computer programs 1004 include instructions that may be used to implement the method for item recommendation described in FIG. 2 for execution in the electronic device 10.
The Processor 1001 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. A general purpose processor may be a microprocessor or the like.
The memory 1002 may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other non-volatile solid state storage device.
It is to be understood that the illustrated structure of the present embodiment does not constitute a specific limitation to the electronic device 10. In other embodiments, electronic device 10 may include more or fewer components than shown, or combine certain components, or split certain components, or a different arrangement of components.
The present embodiment also provides a computer storage medium, where computer instructions are stored in the computer storage medium, and when the computer instructions are run on an electronic device, the electronic device is caused to execute the above related method steps to implement the item recommendation method in the above embodiment.
The present embodiment also provides a computer program product, which when running on a computer, causes the computer to execute the relevant steps described above, so as to implement the item recommendation method in the above embodiments.
Through the above description of the embodiments, it is clear to those skilled in the art that, for convenience and simplicity of description, the foregoing division of the functional modules is merely used as an example, and in practical applications, the above function distribution may be completed by different functional modules according to needs, that is, the internal structure of the device may be divided into different functional modules to complete all or part of the above described functions.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described embodiments of the apparatus are illustrative, and for example, the division of the module or unit into one logical functional division may be implemented in another way, for example, a plurality of units or components may be combined or integrated into another apparatus, or some features may be omitted, or not implemented. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present application may be essentially or partially contributed to by the prior art, or all or part of the technical solutions may be embodied in the form of a software product, where the software product is stored in a storage medium and includes several instructions to enable a device (which may be a single chip, a chip, or the like) or a processor (processor) to execute all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above description is only an embodiment of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions within the technical scope of the present disclosure should be covered by the scope of the present application.
Claims (10)
1. An item recommendation method, characterized in that the method comprises:
obtaining evaluation information of a plurality of articles to be selected;
extracting the theme of the evaluation information of the multiple articles to be selected by using a preset theme model to obtain the comment theme distribution of each evaluation information;
acquiring an incidence relation between each piece of evaluation information and a target user, wherein the incidence relation comprises at least one of an evaluation score index, an interest change index and a heat attenuation index;
obtaining preference theme distribution corresponding to the target user according to the incidence relation and the comment theme distribution of each piece of evaluation information;
calculating to obtain a prediction score of each to-be-selected item in the multiple to-be-selected items based on the preference theme distribution of the target user;
and selecting at least one item from the multiple items to be selected according to the prediction score of each item to be selected and recommending the at least one item to the target user.
2. The item recommendation method of claim 1, further comprising:
preprocessing the evaluation information, wherein the preprocessing comprises at least one of word segmentation processing, stop word removing processing, part of speech tagging processing, case and case conversion processing, part of speech restoring processing and stem extraction processing;
and performing text vectorization processing on the preprocessed evaluation information to obtain a vector corresponding to the evaluation information.
3. The item recommendation method according to claim 2, wherein the subjecting the preprocessed evaluation information to text vectorization processing includes:
constructing a text set based on the preprocessed multiple evaluation information;
extracting non-repeated words from the text set by using a preset word frequency statistical algorithm, and constructing a word bank table based on the extracted words, wherein the word bank table comprises V words, and V is a positive integer;
comparing the preprocessed evaluation information with the word stock table based on a preset comparison rule to obtain a V-dimensional vector corresponding to the evaluation information;
wherein, the preset comparison rule comprises: if the evaluation information includes a word in the word bank table, adding n to the word count in the word bank table, where n is the number of times that the word appears in the evaluation information and n is a positive integer, and if the evaluation information does not include the word in the word bank table, adding 0 to the word count in the word bank table.
4. The item recommendation method of claim 1, wherein the preset topic model is an LDA topic model, and the distribution of the preferred topics corresponding to the target user is calculated by the following equation:
wherein, PuFor a preferred topic distribution corresponding to the target user u, AuiAn evaluation score index of the object to be selected i for the target user u, BuiAn interest change index of the target user u to the article i to be selected, CuiThe heat attenuation index, K, of the target user u to the object i to be selecteduiA K-dimensional comment theme distribution, I, extracted from the evaluation information of the LDA theme model on the item I to be selecteduAnd (4) evaluating the item set for the target user u.
5. The item recommendation method according to claim 4, wherein the evaluation score index of the target user u for the item i to be selected is obtained through conversion based on the score of the target user u for the item i to be selected, the interest variation index of the target user u for the item i to be selected is obtained through conversion based on the score time of the target user u for the item i to be selected, and the heat attenuation index of the target user u for the item i to be selected is obtained through conversion based on the scored time span of the item i to be selected, the number of clicks of the user u, the number of comments of the user u, and the browsing duration of the user u.
6. The item recommendation method of claim 1, wherein said calculating a prediction score for each of the plurality of candidate items based on the target user's preferred topic distribution comprises:
constructing a similar user set which is close to the target user, wherein the similar user set comprises at least one similar user;
and calculating to obtain the prediction score of the selected item based on the historical average score evaluated by the target user, the historical average score evaluated by the similar users, the score of the similar users on the selected item, the heat degree of the selected item, the preference theme distribution of the target user and the preference theme distribution of the similar users.
7. The item recommendation method of claim 6, wherein the predicted score for the item to be selected is calculated by the following equation:
wherein Pre is the prediction score of the item to be selected i, u1Being a similar user to said target user u,
the historical average score evaluated by the target user u,
for the similar user u1Historical mean score, Hot, evaluatediuIs the popularity of the item i to be selected, N is the similar user set adjacent to the target user u,
for the similar user u1Scoring, Q, of the item to be selected iunThe distribution probability of the target user u on the nth subject is obtained,
for the similar user u1The probability of the distribution on the nth topic,
for the average preferred topic distribution of the target user u,
for the similar user u1Average preference topic distribution of DiuThe number of times that the item i to be selected is clicked by the target user u, EiuThe time length of the item i to be selected browsed by the target user u is FiuFor the number of times that the item i to be selected is commented on by the target user u,
epsilon and eta are preset constants, and k epsilon is (1,2, …, n).
8. The item recommendation method according to any one of claims 1 to 7, wherein said selecting at least one item from the plurality of items to be selected for recommendation to the target user according to the prediction score of each item to be selected comprises:
acquiring historical purchase record information of the target user;
and selecting the articles which are preset before the predicted score ranking and are not purchased by the target user from the plurality of articles to be selected according to the predicted score of each article to be selected and the historical purchase record information, and recommending the articles to the target user.
9. Computer arrangement comprising a processor and a memory, said memory having stored thereon a number of computer programs, characterized in that said processor is adapted to carry out the steps of the item recommendation method according to any one of claims 1-8 when executing the computer programs stored in the memory.
10. A computer-readable storage medium having stored thereon instructions executable by one or more processors to perform the steps of the item recommendation method of any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111121782.0A CN113821727A (en) | 2021-09-24 | 2021-09-24 | Item recommendation method, computer device and computer-readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111121782.0A CN113821727A (en) | 2021-09-24 | 2021-09-24 | Item recommendation method, computer device and computer-readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113821727A true CN113821727A (en) | 2021-12-21 |
Family
ID=78915376
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111121782.0A Pending CN113821727A (en) | 2021-09-24 | 2021-09-24 | Item recommendation method, computer device and computer-readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113821727A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114297489A (en) * | 2021-12-28 | 2022-04-08 | 北京百度网讯科技有限公司 | Content recommendation method and device, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170193106A1 (en) * | 2015-12-30 | 2017-07-06 | Yahoo! Inc. | Method and system for recommending content |
| CN112052401A (en) * | 2020-08-26 | 2020-12-08 | 南京邮电大学 | Recommendation method based on user comments |
| CN112507248A (en) * | 2020-09-18 | 2021-03-16 | 西北大学 | Tourist attraction recommendation method based on user comment data and trust relationship |
| CN112989208A (en) * | 2021-04-30 | 2021-06-18 | 武汉卓尔数字传媒科技有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113254777A (en) * | 2021-06-07 | 2021-08-13 | 武汉卓尔数字传媒科技有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113254780A (en) * | 2021-06-08 | 2021-08-13 | 武汉卓尔数字传媒科技有限公司 | Information processing method and device, electronic equipment and computer storage medium |
-
2021
- 2021-09-24 CN CN202111121782.0A patent/CN113821727A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170193106A1 (en) * | 2015-12-30 | 2017-07-06 | Yahoo! Inc. | Method and system for recommending content |
| CN112052401A (en) * | 2020-08-26 | 2020-12-08 | 南京邮电大学 | Recommendation method based on user comments |
| CN112507248A (en) * | 2020-09-18 | 2021-03-16 | 西北大学 | Tourist attraction recommendation method based on user comment data and trust relationship |
| CN112989208A (en) * | 2021-04-30 | 2021-06-18 | 武汉卓尔数字传媒科技有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113254777A (en) * | 2021-06-07 | 2021-08-13 | 武汉卓尔数字传媒科技有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113254780A (en) * | 2021-06-08 | 2021-08-13 | 武汉卓尔数字传媒科技有限公司 | Information processing method and device, electronic equipment and computer storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 刘庆，姚丽娜，余美华: "Python编程案例教程", 北京：航空工业出版社, pages: 165 - 167 * |
| 赵俊华: "基于改进的用户偏好与物品特征主题的混合协同过滤算法研究", 《中国优秀硕士学位论文全文数据库 社会科学Ⅱ辑》/基于改进的用户偏好与物品特征主题的混合协同过滤算法研究, pages 165 - 167 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114297489A (en) * | 2021-12-28 | 2022-04-08 | 北京百度网讯科技有限公司 | Content recommendation method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fauzi | Random forest approach fo sentiment analysis in indonesian | |
| CN109992646B (en) | Text label extraction method and device | |
| CN105224699B (en) | News recommendation method and device | |
| CN103870973B (en) | Information push, searching method and the device of keyword extraction based on electronic information | |
| US10423648B2 (en) | Method, system, and computer readable medium for interest tag recommendation | |
| US9798797B2 (en) | Cluster method and apparatus based on user interest | |
| CN102193936B (en) | Data classification method and device | |
| US10685181B2 (en) | Linguistic expression of preferences in social media for prediction and recommendation | |
| CN112347778B (en) | Keyword extraction method, keyword extraction device, terminal equipment and storage medium | |
| Suresh | An unsupervised fuzzy clustering method for twitter sentiment analysis | |
| WO2018040068A1 (en) | Knowledge graph-based semantic analysis system and method | |
| US11127063B2 (en) | Product and content association | |
| CN112667899A (en) | Cold start recommendation method and device based on user interest migration and storage equipment | |
| Tewari et al. | Sequencing of items in personalized recommendations using multiple recommendation techniques | |
| US20150066904A1 (en) | Integrating and extracting topics from content of heterogeneous sources | |
| US20150032753A1 (en) | System and method for pushing and distributing promotion content | |
| CN109086375A (en) | A kind of short text subject extraction method based on term vector enhancing | |
| CN105468649B (en) | Method and device for judging matching of objects to be displayed | |
| CN110287409B (en) | Webpage type identification method and device | |
| CN110427453B (en) | Data similarity calculation method, device, computer equipment and storage medium | |
| CN111582932A (en) | Inter-scene information pushing method and device, computer equipment and storage medium | |
| Baishya et al. | SAFER: sentiment analysis-based fake review detection in e-commerce using deep learning | |
| CN110569349A (en) | Big data-based method, system, equipment and storage medium for pushing articles for education | |
| CN113821727A (en) | Item recommendation method, computer device and computer-readable storage medium | |
| JP6870467B2 (en) | Advertising effectiveness estimation device, advertising effectiveness estimation method and advertising effectiveness estimation program |
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 |