CN112685656B - Label recommending method and electronic equipment - Google Patents
Label recommending method and electronic equipment Download PDFInfo
- Publication number
- CN112685656B CN112685656B CN202011529096.2A CN202011529096A CN112685656B CN 112685656 B CN112685656 B CN 112685656B CN 202011529096 A CN202011529096 A CN 202011529096A CN 112685656 B CN112685656 B CN 112685656B
- Authority
- CN
- China
- Prior art keywords
- user
- tag
- labels
- scoring
- users
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 239000011159 matrix material Substances 0.000 claims abstract description 63
- 230000006399 behavior Effects 0.000 claims abstract description 22
- 238000012545 processing Methods 0.000 claims description 20
- 238000013527 convolutional neural network Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 16
- 238000000354 decomposition reaction Methods 0.000 claims description 15
- 238000004422 calculation algorithm Methods 0.000 claims description 12
- 230000011218 segmentation Effects 0.000 claims description 7
- 230000000694 effects Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 238000002372 labelling Methods 0.000 description 3
- 238000011176 pooling Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the disclosure discloses a tag recommendation method and electronic equipment. The label recommending method comprises the following steps: scoring the social relationship of the user based on the extracted data to obtain a social relationship score of the user; constructing a user tag scoring matrix based on the extracted data, and selecting social relationship tags of the users based on the user tag scoring matrix and the social relationship scores of the users; and obtaining the score of the user on the social relation label, and recommending corresponding content according to the score of the user on the social relation label. Selecting social relation labels based on scoring and a constructed user label scoring matrix by scoring the social relation of the user, predicting the scoring of the social relation labels by the user, and recommending corresponding contents based on the scoring of the social relation labels by the user, so as to achieve the aim of personalized recommendation of labels according to different user behaviors.
Description
Technical Field
The disclosure belongs to the technical field of internet, and in particular relates to a tag recommendation method and electronic equipment.
Background
In the era of high popularization of the internet, the network scale is continuously expanded, various network application layers are endless, and the network information growing in geometric series is faced, wherein useful information is easily buried and cannot obtain the opportunity of display, so that how to obtain interesting information more quickly becomes a problem that each application needs to consider. In order to solve the problem, personalized recommendation is generated, and unlike popular recommendation, the method can predict the interest points of the user according to the information such as the historical behaviors of the user and provide personalized service for the user, so that the loyalty of the user can be greatly improved, and the loss of the client is prevented.
Tags have found widespread use in many services to assist users in locating classification information. In this type of application of forum, the tag label is selected when the user asks a question so that users familiar with the art can find and answer in time. In addition, the user can label himself to mark the favorites, and the related problems and friends with the same favorites can be found. With the increase of website contents, tags are gradually diversified, and some websites provide hot tag recommendation services for users, so that some users can find useful information.
In the process of implementing the present invention, the inventor finds that at least the following problems exist in the prior art:
the indiscriminate label recommending mode for all users cannot meet the personalized demands of the users, and the exposure of cold labels can be reduced to generate the Martai effect.
Disclosure of Invention
In view of this, the embodiments of the present disclosure provide a tag recommendation method and an electronic device, which at least solve the problem that in the prior art, indiscriminate tag recommendation is not performed for all users, and thus the personalized requirements of the users cannot be met.
In a first aspect, an embodiment of the present disclosure provides a tag recommendation method, including:
scoring the social relationship of the user based on the extracted data to obtain a social relationship score of the user;
Constructing a user tag scoring matrix based on the extracted data;
selecting social relationship labels of users based on the user label scoring matrix and the user social relationship scores;
And obtaining the score of the user on the social relation label, and recommending corresponding content according to the score of the user on the social relation label.
Optionally, the extracted data includes,
User basic information, social relationships and/or historical behavior data are extracted from the historical data.
Optionally, before the step of constructing the user tag scoring matrix based on the extracted data, the method includes:
And (3) carrying out standardization processing on the labels, layering the labels subjected to standardization processing, and marking the membership of the labels among different layers.
Optionally, before the step of normalizing the label, the method includes:
Word segmentation processing is carried out on the extracted data;
And selecting words meeting a set threshold value from words obtained by word segmentation processing by using a TF-IDF algorithm, thereby obtaining the label.
Optionally, scoring the social relationship of the user based on the extracted data to obtain a social relationship score of the user, including:
scoring the social relationship of the user based on the user similarity coefficient and the question similarity coefficient;
The user similarity coefficient is related to the number of friends, the number of fans and the number of concerns of the user, and the question similarity coefficient is related to the number of questions, the number of answers and the number of praise.
Optionally, the user similarity coefficient and the question similarity coefficient are Jaccard similarity coefficients, and the scoring of the social relationship of the user by the user similarity coefficient and the question similarity coefficient is as follows:
Where Co (u i,uj) represents the sum of the same friends, fans and interests of users u i and u j, co (u i)、Co(uj) represents the sum of the friends, fans and interests of user u i、uj, co '(u i,uj) represents the sum of questions, answers and praise of the same labels of users u i and u j, co' (u i)、Co′(uj) represents the sum of the questions, answers and praise of user u i、uj, and sim (u i,uj) is the social relationship score of the user.
Optionally, the constructing a user tag scoring matrix based on the extracted data includes:
the user's behavior of the tag is scored,
Searching all the corresponding superior labels of the labels, adding the scores of the labels and the scores of the searched corresponding superior labels, and multiplying the sum by a constant alpha to obtain layering information;
And scoring and layering information are carried out on the behaviors of the labels based on the users to obtain a scoring matrix of the labels of the users.
Optionally, the selecting the social relationship label of the user based on the user label scoring matrix and the user social relationship score includes:
Calculating similarity scores sim (u i,u′i) of the users and friends and concerned users;
searching the scores r i′,j of friends and concerned users of the users in a scoring matrix of the user labels;
The non-empty r i′,j weight sum for each tag is averaged with a weight sim (u i,u′i);
Obtaining a social relationship score s i,j based on sim (u i,u′i) and r i′,j;
The method comprises the following steps:
Wherein K represents a non-zero scoring number corresponding to each tag in the scoring of friends and concerned users, and i' represents a user;
a social relationship label for the user is selected based on the social relationship score s i,j.
Optionally, the obtaining the score of the social relationship label by the user includes:
Predicting a score of the user for each social relationship tag based on the objective function;
The objective function consists of a matrix decomposition and a loss function of CNN, and is specifically as follows:
Where I ij is the indicator function, cnn (W, X i) represents the convolutional neural network, where U is the user matrix, U i is the corresponding information of user I in U, For the transpose of U i, λ U is the regularization term coefficient of matrix U, V is the tag matrix, V j is the corresponding information of tag j in V, λ v is the regularization term coefficient of matrix V, W is the weight matrix, W k is a specific parameter in W, λ w is the regularization term coefficient of matrix W, X i is the social relationship tag feature text, r ij is the score of the tag by the user, i.e. when the tag is scored by the user, i.e. r ij is not empty, I ij =1; otherwise, 0.
In a second aspect, embodiments of the present disclosure further provide an electronic device, including:
a memory storing executable instructions;
a processor executing the executable instructions in the memory to implement the tag recommendation method of any one of the first aspects.
According to the method and the device, the social relationship is scored, the social relationship label is selected based on the scoring and the constructed user label scoring matrix, the scoring of the social relationship label by the user is predicted, and then the corresponding content is recommended based on the scoring of the social relationship label by the user, so that the aim of personalized recommendation of the label according to different user behaviors is fulfilled.
Additional features and advantages of the present disclosure will be set forth in the detailed description which follows.
Drawings
The foregoing and other objects, features and advantages of the disclosure will be apparent from the following more particular descriptions of exemplary embodiments of the disclosure as illustrated in the accompanying drawings wherein like reference numbers generally represent like parts throughout exemplary embodiments of the disclosure.
FIG. 1 illustrates a flow chart of a tag recommendation method of one embodiment of the present disclosure;
fig. 2 illustrates a functional block diagram of a tag recommendation method of one embodiment of the present disclosure.
Detailed Description
Preferred embodiments of the present disclosure will be described in more detail below. While the preferred embodiments of the present disclosure are described below, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein.
The personalized tag is mainly based on user information and historical behavior data, wherein the user information comprises initial registration information, social relations and the like of the user, and the historical behavior data comprises questions, answers, praise and the like. The matrix decomposition algorithm is a common algorithm for realizing personalized recommendation, and the convolutional neural network is an effective method for mining labels and text information. Therefore, according to the data representing the user interests, the user information is fully mined by combining the convolutional neural network and the matrix decomposition algorithm, and a personalized tag recommendation method based on the user social relationship is designed. According to the personalized tag recommendation method, tag standardization processing is carried out according to the actual application condition of the multi-layer tags, the social relationship among users is mainly analyzed, the tags of interest of the users are predicted by using a convolution matrix decomposition algorithm, and good tag recommendation effect is achieved.
As shown in fig. 1, a tag recommendation method includes:
step S101: scoring the social relationship of the user based on the extracted data to obtain a social relationship score of the user;
step S102: constructing a user tag scoring matrix based on the extracted data;
Step S103: selecting social relationship labels of users based on the user label scoring matrix and the user social relationship scores;
Step S104: and obtaining the score of the user on the social relation label, and recommending corresponding content according to the score of the user on the social relation label.
Optionally, the extracted data includes,
User basic information, social relationships and/or historical behavior data are extracted from the historical data.
Optionally, before the step of constructing the user tag scoring matrix based on the extracted data, the method includes:
And (3) carrying out standardization processing on the labels, layering the labels subjected to standardization processing, and marking the membership of the labels among different layers.
Optionally, before the step of normalizing the label, the method includes:
Word segmentation processing is carried out on the extracted data;
And selecting words meeting a set threshold value from words obtained by word segmentation processing by using a TF-IDF algorithm, thereby obtaining the label.
Optionally, scoring the social relationship of the user based on the extracted data to obtain a social relationship score of the user, including:
scoring the social relationship of the user based on the user similarity coefficient and the question similarity coefficient;
The user similarity coefficient is related to the number of friends, the number of fans and the number of concerns of the user, and the question similarity coefficient is related to the number of questions, the number of answers and the number of praise.
Optionally, the user similarity coefficient and the question similarity coefficient are Jaccard similarity coefficients, and the scoring of the social relationship of the user by the user similarity coefficient and the question similarity coefficient is as follows:
Where Co (u i,uj) represents the sum of the same friends, fans and interests of users u i and u j, co (u i)、Co(uj) represents the sum of the friends, fans and interests of user u i、uj, co '(u i,uj) represents the sum of questions, answers and praise of the same labels of users u i and u j, co' (u i)、Co′(uj) represents the sum of the questions, answers and praise of user u i、uj, and sim (u i,uj) is the social relationship score of the user.
Optionally, the constructing a user tag scoring matrix based on the extracted data includes:
the user's behavior of the tag is scored,
Searching all the corresponding superior labels of the labels, adding the scores of the labels and the scores of the searched corresponding superior labels, and multiplying the sum by a constant alpha to obtain layering information;
And scoring and layering information are carried out on the behaviors of the labels based on the users to obtain a scoring matrix of the labels of the users.
Optionally, the selecting the social relationship label of the user based on the user label scoring matrix and the user social relationship score includes:
Calculating similarity scores sim (u i,u′i) of the users and friends and concerned users;
searching the scores r i′,j of friends and concerned users of the users in a scoring matrix of the user labels;
The non-empty r i′,j weight sum for each tag is averaged with a weight sim (u i,u′i);
Obtaining a social relationship score s i,j based on sim (u i,u′i) and r i′,j;
The method comprises the following steps:
Wherein K represents a non-zero scoring number corresponding to each tag in the scoring of friends and concerned users, and i' represents a user;
a social relationship label for the user is selected based on the social relationship score s i,j.
Optionally, the obtaining the score of the social relationship label by the user includes:
Predicting a score of the user for each social relationship tag based on the objective function;
The objective function consists of a matrix decomposition and a loss function of CNN, and is specifically as follows:
Where I ij is the indicator function, cnn (W, X i) represents the convolutional neural network, where U is the user matrix, U i is the corresponding information of user I in U, For the transpose of U i, λ U is the regularization term coefficient of matrix U, V is the tag matrix, V j is the corresponding information of tag j in V, λ V is the regularization term coefficient of matrix V, W is the weight matrix, W k is a specific parameter in W, λ w is the regularization term coefficient of matrix W, X i is the social relationship tag feature text, r ij is the score of the tag by the user, i.e. when the tag is scored by the user, i.e. r ij is not empty, I ij =1; otherwise, 0.
In one particular application scenario of the present invention,
The tag recommending method is characterized in that information such as user basic information, social relations, questions, answers, praise and the like is extracted from a website history record, and personalized tags are predicted through a convolution matrix decomposition algorithm after processing, as shown in fig. 1. As shown in fig. 2, the specific steps are as follows:
step1, extracting and processing historical data:
Basic information, social relations and historical behavior data of the user are extracted from the database. The basic information includes the sex, region, job, etc. of the user. Social relationship data includes friend user information, users of interest, which users are focused on, and the like. The historical behavior data includes questioning, answering, praying, and the like. And deleting the repeated data and processing missing and abnormal data.
Step 2, label standardization:
The label is a keyword representing user interests or summarized text content and is widely applied to the personalized recommendation field, but label standardization processing is performed first, redundant labels are combined, and label description is standardized due to the characteristics of free definition, unconstrained, sharable and the like of the label. Many labels have layering properties, such as labels of 'technology', 'java', 'machine learning', 'knowledge graph', and the like, have a relationship, wherein the 'technology' is an upper-level label, and the 'java', 'machine learning', 'knowledge graph' labels belong to a category of the 'technology', so that in a process of standardizing label description, the existing labels are subjected to layering treatment, and most of the labels are divided into 2-3 layers according to business requirements. In the label labeling process, the language description of each person is slightly different, so that a high-quality label is extracted according to the existing label by combining a keyword extraction technology.
And performing word segmentation on all data of the website and the existing labels, selecting partial words by using TF-IDF and setting a threshold value, manually screening, and searching for inclusion relations to perform layering. The tags are as short as possible, then the tags are searched in the questions and answers, corresponding tags are added to the records containing the words of the tags, and the tags remain unchanged if the tags are already marked. In terms of tag language description, synonyms are recorded.
Step 3, scoring social relationship:
Many network applications today have social properties, and user interests can be better mined according to social relationships of users, including friends, users concerned, fans, and the like. In order to evaluate the similarity between two users, behavior information of the users and friends, concerned users and fans of the users is firstly extracted, the similarity of historical behaviors of the two users is evaluated, and the degree of association between the users is deeply analyzed.
The Jaccard similarity coefficient may compare similarities to differences between a limited set of samples. The larger the Jaccard coefficient value, the higher the sample similarity. Comparing the similarity of two social users by using Jaccard coefficients, and calculating the similarity by considering the similarity of social relations and the similarity of behaviors, wherein the formula is as follows:
Where Co (u i,uj) represents the sum of the same number of friends, fan count and attention count of users u i and u j, co (u i)、Co(uj) represents the sum of the number of friends, fan count and attention count of users u i、uj, co '(u i,uj) represents the sum of the number of questions, the number of answers and the number of praise of the same labels of users u i and u j, and Co' (u i)、Co′(uj) represents the sum of the number of questions, the number of answers and the number of praise of users u i、uj.
Step 4, user labels based on social relations:
user behavior data is collected and quantified using a user-tag scoring matrix. The scoring mode is that the user asks questions and answers the content of a certain label 1 time and adds 1 minute, two or more times and adds 2 minutes, praise adds 0.5 minute, two or more times and adds 1 minute, and the rest is empty. In addition, when labeling, the labels with higher-level labels or more specific lower-level labels with wider descriptions can be used, and the layering relationship of the labels is considered, so that on the basis of a user-label grading matrix, all the higher-level labels corresponding to the user labels are searched, the original labels are added with the scores of a constant alpha to strengthen layering information, data sparsity is reduced, alpha values are manually set according to statistical information, the upper-level label grading is set with an upper grading limit according to actual conditions, and grading data of the labels by users form a user-label grading matrix r.
And mining labels of interest of the user according to the social relationship of the user. Calculating similarity scores sim (u i,u′i) of each user and friends and concerned users, finding scores r i′,j of friends and concerned users in the user-tag scoring matrix, weighting and averaging non-empty r i′,j of each tag j, wherein the weight is sim (u i,u′i), and a social relation score s i,j has the following calculation formula:
Wherein r i′,j is the score for user i' to rank labels j, and K represents the number of non-zero scores corresponding to each label j in the friend and attention user scores. And finally, taking the top M labels with the highest score s i,j of each user i as label characteristics of the user i based on the social relationship.
Step 5, a label recommendation scheme based on social relations:
Matrix decomposition algorithms are a classical and widely used algorithm in the field of recommendation systems. Traditional Convolutional Neural Networks (CNNs) can solve classification tasks such as predicting labels of words, phrases or documents, and the CNN architecture has an embedding layer, a convolutional layer, a pooling layer and an output layer. In order to more effectively mine social relationships and user behavior data, a convolutional matrix decomposition model based on matrix decomposition and CNN is utilized to integrate CNN into a probabilistic matrix decomposition model. Converting the first M labels with the highest scores in the step 4 into vectors at an embedding layer of CNN, selecting proper module size by a convolution layer to extract context label characteristics, and extracting representative characteristics from the convolution layer by a pooling layer through maximum pooling; the output layer is provided with a plurality of full connection layers, finally probability matrix decomposition is carried out, and the objective function of the output layer consists of matrix decomposition and a loss function of CNN, as shown in the following
Wherein I ij is an indicator function, cnn (W, X i) represents a convolutional neural network, wherein W is a weight matrix, X i is a social relationship tag feature text, and when the user has a score r ij for a tag that is not empty, I ij = 1; otherwise, if the label is 0, predicting the score of the user to the unknown label through iterative optimization U, V, W to obtain the score r of the user to each label. And providing a recommended link for the user according to the label with the highest score of the N items before scoring, wherein the scored label can be recommended to the user as well as the text of the recommended content for the user. There is a lot of question and answer content text under each tag link, and the used tags are still needed by the user.
The personalized tag recommendation method based on the user social relationship can store information such as the user social relationship, the user-tag matrix and the like offline, and provide tag recommendation results for users in real time. Tag recommendations may reduce the sparsity of data compared to content recommendations. In the scheme, social relationship characteristics are further mined through social relationship similarity, scoring and convolutional neural networks, so that recommendation is more accurate. In the recommendation process, a convolution matrix decomposition algorithm is adopted according to user registration and historical behavior information and in combination with the social relationship of the user, so that the label recommendation effect is improved.
This embodiment has the following advantages:
1. The label standardization scheme supplements label labeling, designs layered labels, increases the grading of the layered labels in a user-label matrix, and reduces the sparsity of the labels.
2. And excavating social relation data, setting scoring reference proportion by using the similarity of social users, calculating the score of the users to the labels according to the scoring reference proportion, and taking the label with the highest score as the label characteristic of the users to effectively excavate social relation information.
3. Social relation label characteristics are mined through social relation score calculation and a convolutional neural network, and a convolutional matrix decomposition algorithm is utilized to improve label recommendation effect and label prediction accuracy.
The disclosed embodiments provide an electronic device comprising a memory and a processor,
A memory storing executable instructions;
And the processor runs executable instructions in the memory to realize the label recommending method.
The memory is for storing non-transitory computer readable instructions. In particular, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like.
The processor may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device to perform the desired functions. In one embodiment of the present disclosure, the processor is configured to execute the computer readable instructions stored in the memory.
It should be understood by those skilled in the art that, in order to solve the technical problem of how to obtain a good user experience effect, the present embodiment may also include well-known structures such as a communication bus, an interface, and the like, and these well-known structures are also included in the protection scope of the present disclosure.
The detailed description of the present embodiment may refer to the corresponding description in the foregoing embodiments, and will not be repeated herein.
The foregoing description of the embodiments of the present disclosure has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described.
Claims (7)
1. A tag recommendation method, comprising:
scoring the social relationship of the user based on the extracted data to obtain a social relationship score of the user;
Constructing a user tag scoring matrix based on the extracted data;
selecting social relationship labels of users based on the user label scoring matrix and the user social relationship scores;
obtaining the score of the user on the social relation label, and recommending corresponding content according to the score of the user on the social relation label;
wherein the constructing a user tag scoring matrix based on the extracted data comprises:
the user's behavior of the tag is scored,
Searching all the corresponding superior labels of the labels, adding the scores of the labels and the scores of the searched corresponding superior labels, and multiplying the sum by a constant alpha to obtain layering information;
scoring and layering information are carried out on the behaviors of the labels based on the users to obtain a scoring matrix of the labels of the users;
the obtaining the score of the user to the social relationship label comprises the following steps:
Predicting a score of the user for each social relationship tag based on the objective function;
The objective function consists of a matrix decomposition and a loss function of CNN, and is specifically as follows:
,
Wherein the method comprises the steps of To indicate a function,/>Represents a convolutional neural network, where U is the user matrix,/>Is the corresponding information of user i in U,/>For/>Transpose of/>Regularized term coefficients for matrix U, V is the tag matrix,/>Is the corresponding information of tag j in V,/>Regularized term coefficients for matrix V, W being a weight matrix,/>Is a specific parameter in W,/>For regularized term coefficients of matrix W,/>Tag feature text for social relationship,/>For scoring tags by users, i.e. when a user scores tagsWhen not empty,/>=1, Otherwise 0.
2. The tag recommendation method of claim 1, wherein extracting data comprises,
User basic information, social relationships and/or historical behavior data are extracted from the historical data.
3. The tag recommendation method of claim 1, wherein prior to the step of constructing a user tag scoring matrix based on the extracted data, comprising:
And (3) carrying out standardization processing on the labels, layering the labels subjected to standardization processing, and marking the membership of the labels among different layers.
4. A tag recommendation method according to claim 3, wherein before the step of normalizing the tag, it comprises:
Word segmentation processing is carried out on the extracted data;
And selecting words meeting a set threshold value from words obtained by word segmentation processing by using a TF-IDF algorithm, thereby obtaining the label.
5. The tag recommendation method of claim 1, wherein scoring the user social relationship based on the extracted data to obtain a user social relationship score comprises:
scoring the social relationship of the user based on the user similarity coefficient and the question similarity coefficient;
The user similarity coefficient is related to the number of friends, the number of fans and the number of concerns of the user, and the question similarity coefficient is related to the number of questions, the number of answers and the number of praise.
6. The tag recommendation method of claim 5, wherein the user similarity coefficient and the question similarity coefficient are Jaccard similarity coefficients, and the user similarity coefficient and the question similarity coefficient score social relationships of users as follows:
,
Wherein the method comprises the steps of Representing the user/>And/>Sum of the same number of friends, number of fans and number of interests,/>、/>Respectively represent the users/>、/>Sum of friends, fans and attention number,/>Representing the user/>And/>The sum of the number of questions, the number of answers and the number of praise of the same tag,/>、/>Respectively represent the users/>、/>Sum of question number, answer number and praise number,/>Social relationship scores for users.
7. An electronic device, the electronic device comprising:
a memory storing executable instructions;
a processor executing the executable instructions in the memory to implement the tag recommendation method of any one of claims 1-6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011529096.2A CN112685656B (en) | 2020-12-22 | 2020-12-22 | Label recommending method and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011529096.2A CN112685656B (en) | 2020-12-22 | 2020-12-22 | Label recommending method and electronic equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112685656A CN112685656A (en) | 2021-04-20 |
CN112685656B true CN112685656B (en) | 2024-06-21 |
Family
ID=75450552
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011529096.2A Active CN112685656B (en) | 2020-12-22 | 2020-12-22 | Label recommending method and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112685656B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113672736B (en) * | 2021-09-09 | 2023-08-22 | 上海德拓信息技术股份有限公司 | Text multi-label classification method and system |
CN116861101B (en) * | 2023-06-30 | 2024-03-29 | 深圳市爱彼利科技有限公司 | Data processing method and device for social matching |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106649657A (en) * | 2016-12-13 | 2017-05-10 | 重庆邮电大学 | Recommended system and method with facing social network for context awareness based on tensor decomposition |
Family Cites Families (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103092911B (en) * | 2012-11-20 | 2016-02-03 | 北京航空航天大学 | A kind of mosaic society label similarity is based on the Collaborative Filtering Recommendation System of k nearest neighbor |
CN103678431B (en) * | 2013-03-26 | 2018-01-02 | 南京邮电大学 | A kind of recommendation method to be scored based on standard label and project |
US20150187024A1 (en) * | 2013-12-27 | 2015-07-02 | Telefonica Digital España, S.L.U. | System and Method for Socially Aware Recommendations Based on Implicit User Feedback |
CN105550211A (en) * | 2015-12-03 | 2016-05-04 | 云南大学 | Social network and item content integrated collaborative recommendation system |
CN107301583B (en) * | 2017-05-26 | 2020-11-17 | 重庆邮电大学 | Cold start recommendation method based on user preference and trust |
CN107688606A (en) * | 2017-07-26 | 2018-02-13 | 北京三快在线科技有限公司 | The acquisition methods and device of a kind of recommendation information, electronic equipment |
CN110110220B (en) * | 2018-06-21 | 2021-06-01 | 北京交通大学 | Recommendation model fusing social network and user evaluation |
CN108985899B (en) * | 2018-07-13 | 2022-04-22 | 合肥工业大学 | Recommendation method, system and storage medium based on CNN-LFM model |
CN109241454B (en) * | 2018-07-18 | 2021-08-24 | 广东工业大学 | Interest point recommendation method fusing social network and image content |
CN109165975B (en) * | 2018-08-09 | 2023-05-16 | 平安科技(深圳)有限公司 | Label recommending method, device, computer equipment and storage medium |
CN109800424B (en) * | 2018-12-21 | 2023-05-02 | 广东工业大学 | Recommendation method based on improved matrix decomposition and cross-channel convolutional neural network |
CN112016003B (en) * | 2020-08-19 | 2022-07-12 | 重庆邮电大学 | Social crowd user tag mining and similar user recommending method based on CNN |
-
2020
- 2020-12-22 CN CN202011529096.2A patent/CN112685656B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106649657A (en) * | 2016-12-13 | 2017-05-10 | 重庆邮电大学 | Recommended system and method with facing social network for context awareness based on tensor decomposition |
Non-Patent Citations (1)
Title |
---|
融合标签关联关系与用户社交关系的微博推荐方法;马慧芳;贾美惠子;张迪;蔺想红;;电子学报(第01期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN112685656A (en) | 2021-04-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Yuan et al. | Expert finding in community question answering: a review | |
CN110717098B (en) | Meta-path-based context-aware user modeling method and sequence recommendation method | |
CN108536800B (en) | Text classification method, system, computer device and storage medium | |
US11232141B2 (en) | Method and device for processing an electronic document | |
US20160170982A1 (en) | Method and System for Joint Representations of Related Concepts | |
CN112487199B (en) | User characteristic prediction method based on user purchasing behavior | |
CN108491511A (en) | Data digging method and device, model training method based on diagram data and device | |
CN107844533A (en) | A kind of intelligent Answer System and analysis method | |
CN112380331A (en) | Information pushing method and device | |
CN112685656B (en) | Label recommending method and electronic equipment | |
CN110880006A (en) | User classification method and device, computer equipment and storage medium | |
CN112784168B (en) | Information push model training method and device, information push method and device | |
Estevez-Velarde et al. | AutoML strategy based on grammatical evolution: A case study about knowledge discovery from text | |
CN118132856B (en) | Intelligent analysis method and system based on big data | |
CN112215629B (en) | Multi-target advertisement generating system and method based on construction countermeasure sample | |
CN111709225A (en) | Event cause and effect relationship judging method and device and computer readable storage medium | |
Bai et al. | Position-aware deep character-level CTR prediction for sponsored search | |
CN112463966B (en) | False comment detection model training method, false comment detection model training method and false comment detection model training device | |
CN111382254A (en) | Electronic business card recommendation method, device, equipment and computer readable storage medium | |
Nia et al. | A framework for a large-scale B2B recommender system | |
JPWO2012077818A1 (en) | Method for determining transformation matrix of hash function, hash type approximate nearest neighbor search method using the hash function, apparatus and computer program thereof | |
Ali et al. | Identifying and Profiling User Interest over time using Social Data | |
CN115169637A (en) | Social relationship prediction method, device, equipment and medium | |
CN113688633A (en) | Outline determination method and device | |
CN114548242A (en) | User tag identification method, device, electronic equipment and computer readable 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |