CN110427568A - A kind of collaboration attention recommendation system, method and apparatus based on information - Google Patents
A kind of collaboration attention recommendation system, method and apparatus based on information Download PDFInfo
- Publication number
- CN110427568A CN110427568A CN201910675118.7A CN201910675118A CN110427568A CN 110427568 A CN110427568 A CN 110427568A CN 201910675118 A CN201910675118 A CN 201910675118A CN 110427568 A CN110427568 A CN 110427568A
- Authority
- CN
- China
- Prior art keywords
- information
- term vector
- article
- user
- promotion
- 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
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
- G06F16/9536—Search customisation based on social or collaborative filtering
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to technical field of information processing, and in particular to a kind of collaboration attention recommendation system, method and apparatus based on information.Include: preprocessing module, information word to be processed is pre-processed, generates the information term vector of each user and the information term vector of each article;Collaboration pays attention to power module, and under goal behavior, the information term vector of information term vector and each article based on each user generates the weight that each user is directed to the information term vector of each article respectively, constructs fisrt feature matrix;Under goal behavior, the information term vector of information term vector and each article based on each user generates the information term vector weight that each article is directed to each user, constructs second characteristic matrix;Promotion generation module calculates the user for the probability for generating goal behavior to promotion article according to fisrt feature matrix and second characteristic matrix, is ranked up to the probability for generating goal behavior, generates promotion list.The accuracy of promotion can be obviously improved.
Description
Technical field
The invention belongs to technical field of information processing, and in particular to a kind of collaboration attention recommendation system based on information,
Method and apparatus.
Background technique
With the development of internet technology, there is different degrees of information overload phenomenon in all kinds of websites and APP how
Comforming and selecting suitable content promotion in multi information to user is each website and APP developer's problem encountered.And it solves
It is that user carries out personalized promotion that certainly the method for this problem, which is exactly using recommendation system,.
In the prior art, information promotion are carried out usually using collaborative filtering, main method is to establish a kind of user
With a kind of rating matrix of the article under goal behavior (such as buy, shares), and user is in practical applications, often has greatly
Other behavioral datas of amount, and to promotion article itself also include a large amount of satellite informations.Establishing user-article rating matrix
When, it how to be a problem in the urgent need to address by the data fusion in these heterogeneous spaces a to entirety.
Summary of the invention
In view of this, the collaboration attention recommendation system that the main purpose of the present invention is to provide a kind of based on information, side
Method and device can be obviously improved the accuracy and efficiency of promotion.
In order to achieve the above objectives, the technical scheme of the present invention is realized as follows:
A kind of collaboration attention recommendation system based on information, the system comprises:
Preprocessing module pre-processes information word to be processed, generates the information term vector of each user and each
The information term vector of article;
Collaboration pays attention to power module, under goal behavior, the information of information term vector and each article based on each user
Term vector generates the weight that each user is directed to the information term vector of each article respectively, constructs fisrt feature matrix;In target
Under behavior, the information term vector of information term vector and each article based on each user generates each article and is directed to each user
Information term vector weight, construct second characteristic matrix;
Promotion generation module calculates the user according to fisrt feature matrix and second characteristic matrix and is directed to the production of promotion article
The probability of raw goal behavior, is ranked up the probability for generating goal behavior, generates promotion list.
Further, the preprocessing module includes: word processing module, is divided according to Information Granularity information to be processed
Word and/or stop words is gone to handle;Dimension abstraction module extracts word from user's dimension and article dimension respectively;Term vector generates mould
Block generates the information term vector of each user and the information term vector of each article.
Further, the collaboration notices that power module includes: a granularity collaboration attention processing module, works as message complexity
Gao Shi, first distich seed degree carry out collaboration attention processing, then carry out collaboration attention processing to word granularity, in goal behavior
Under, the information term vector of information term vector and each article based on each user generates each user and is directed to each article respectively
Information term vector weight, construct fisrt feature matrix;Under goal behavior, information term vector based on each user and every
The information term vector of a article generates the information term vector weight that each article is directed to each user, constructs second characteristic matrix;
Word granularity cooperates with attention processing module, and when message complexity is low, distich seed degree carries out collaboration attention processing first, then
Collaboration attention processing, under goal behavior, information term vector and each article based on each user are carried out to word granularity
Information term vector generates the weight that each user is directed to the information term vector of each article respectively, constructs fisrt feature matrix;In
Under goal behavior, the information term vector of information term vector and each article based on each user generates each article for each
The information term vector weight of user constructs second characteristic matrix.
Further, the promotion generation module includes: probability evaluation entity, under goal behavior, according to fisrt feature
Matrix and second characteristic matrix calculate the user and are directed to the probability for generating goal behavior to promotion article.
A kind of collaboration attention promotion method based on information, the method execute following steps:
Information word to be processed is pre-processed, the information term vector of each user and the information word of each article are generated
Vector;
Under goal behavior, the information term vector of information term vector and each article based on each user generates each use
Family is directed to the weight of the information term vector of each article respectively, constructs fisrt feature matrix;Under goal behavior, it is based on each use
The information term vector at family and the information term vector of each article generate the information term vector weight that each article is directed to each user,
Construct second characteristic matrix;
The user is calculated according to fisrt feature matrix and second characteristic matrix to be directed to promotion article generation goal behavior
Probability is ranked up the probability for generating goal behavior, generates promotion list.
Further, described to pre-process information word to be processed, generate the information term vector of each user and every
The method of the information term vector of a article executes following steps: information to be processed being segmented and/or gone according to Information Granularity
Stop words processing;Word is extracted respectively from user's dimension and article dimension;Generate the information term vector and each article of each user
Information term vector.
Further, the promotion generation module calculates user's needle according to fisrt feature matrix and second characteristic matrix
The probability that promotion article generates goal behavior is treated, the probability for generating goal behavior is ranked up, generates the side of promotion list
Method executes following steps: probability evaluation entity, and under goal behavior, being calculated according to fisrt feature matrix and second characteristic matrix should
User is directed to the probability that goal behavior is generated to promotion article.
A kind of collaboration attention promotion unit based on information, described device includes: a kind of computer of non-transitory can
Storage medium is read, which stores computations comprising: information word to be processed is pre-processed, is generated every
The code segment of the information term vector of the information term vector and each article of a user;Under goal behavior, based on each user's
The information term vector of information term vector and each article generates the power that each user is directed to the information term vector of each article respectively
Weight constructs fisrt feature matrix;Under goal behavior, the information word of information term vector and each article based on each user to
Amount generates the information term vector weight that each article is directed to each user, constructs the code segment of second characteristic matrix;According to first
Eigenmatrix and second characteristic matrix calculate the user and are directed to the probability for generating goal behavior to promotion article, to generation target line
For probability be ranked up, generate promotion list code segment.
A kind of collaboration attention recommendation system, method and apparatus based on information of the invention, traditional collaborative filtering side
Method be used only information of the user under a certain goal behavior, in this way during promotion often because data it is excessively sparse from
And cause promotion effect not as expected, and user is using in the website app/, often Behavior preference is relatively consistent, I
Can be extended from user terminal using the non-targeted behavioral data information of user's portrait, user, and can be adopted at article end
It is extended with information such as goods attributes, to effectively raise the effect that user-article matrix information improves promotion.
Compared to traditional collaborative filtering method, more users of involvement and the information of article improve promotion effect.
Detailed description of the invention
Fig. 1 is the system structure diagram of the collaboration attention recommendation system of the invention based on information.
Fig. 2 is the method flow schematic diagram of the collaboration attention promotion method of the invention based on information.
Fig. 3 is the structural schematic diagram of the preprocessing module of the collaboration attention recommendation system of the invention based on information.
Fig. 4 is the structural schematic diagram of the promotion generation module of the collaboration attention recommendation system of the invention based on information.
Specific embodiment
With reference to the accompanying drawing and the embodiment of the present invention is described in further detail method of the invention.
A kind of collaboration attention recommendation system based on information, the system comprises:
Preprocessing module pre-processes information word to be processed, generates the information term vector of each user and each
The information term vector of article;
Collaboration pays attention to power module, under goal behavior, the information of information term vector and each article based on each user
Term vector generates the weight that each user is directed to the information term vector of each article respectively, constructs fisrt feature matrix;In target
Under behavior, the information term vector of information term vector and each article based on each user generates each article and is directed to each user
Information term vector weight, construct second characteristic matrix;
Promotion generation module calculates the user according to fisrt feature matrix and second characteristic matrix and is directed to the production of promotion article
The probability of raw goal behavior, is ranked up the probability for generating goal behavior, generates promotion list.
For example, it is assumed that user list includes user UserA, UserB, UserC, UserD, and relative article list includes
Article ProductA, ProductB, ProductC, ProductD.In a kind of possible embodiment, each user point is obtained
Safety pin is expressed as follows shown in table 1 weight of the information term vector of each article:
Table 1: each user is directed to the weight table of the information term vector of each article respectively
ProductA | ProductB | ProductC | ProductD | |
UserA | 0.3 | 0.5 | 0.8 | 1.0 |
UserB | 0.7 | 0.3 | 0.2 | 0.1 |
UserC | 0.7 | 0.4 | 0.5 | 0.6 |
UserD | 1.0 | 0.3 | 0.2 | 0.9 |
Above table correlated digital is solely for illustrating, and represents user to the weight of the information term vector of article, practical behaviour
Respective weights may be a number or a vector or a tensor in work.
Similar, also available each article is as shown in table 2 below for the information term vector weight of each user:
Table 2: each article is directed to the information term vector weight of each user
Similarly, the number in above table is also solely for illustrating, and represents article to the power of the information term vector of user
Weight, can be a number or a vector or a tensor.
The data source of two above eigenmatrix can be generated based on technological means such as similar Word2Vector.It needs
Although bright is that 2 forms above are similar, calculation be using the alternating attention in collaboration attention mechanism,
The attention that product is first generated based on user, obtains new user-product attention weight;Again based on new user-production
Product attention feature weight goes to generate the attention of user.
Further, the preprocessing module includes: word processing module, is divided according to Information Granularity information to be processed
Word and/or stop words is gone to handle;Dimension abstraction module extracts word from user's dimension and article dimension respectively;Term vector generates mould
Block generates the information term vector of each user and the information term vector of each article.
For given information system, the size of Information Granularity reflects the concentration class that knowledge classifies to domain, i.e. knowledge
Classify to domain more coarse, Information Granularity is bigger;Knowledge is thinner to domain classification, then Information Granularity is smaller.
Further, the collaboration notices that power module includes: a granularity collaboration attention processing module, works as message complexity
Gao Shi, first distich seed degree carry out collaboration attention processing, then carry out collaboration attention processing to word granularity, in goal behavior
Under, the information term vector of information term vector and each article based on each user generates each user and is directed to each article respectively
Information term vector weight, construct fisrt feature matrix;Under goal behavior, information term vector based on each user and every
The information term vector of a article generates the information term vector weight that each article is directed to each user, constructs second characteristic matrix;
Word granularity cooperates with attention processing module, and when message complexity is low, distich seed degree carries out collaboration attention processing first, then
Collaboration attention processing, under goal behavior, information term vector and each article based on each user are carried out to word granularity
Information term vector generates the weight that each user is directed to the information term vector of each article respectively, constructs fisrt feature matrix;In
Under goal behavior, the information term vector of information term vector and each article based on each user generates each article for each
The information term vector weight of user constructs second characteristic matrix.
In the present invention, segment and/or go stop words processing can use following technology: as soon as provide a document, document
It is a word sequence such as " ABACBFG ", it is desirable to which it is (past to obtain a corresponding vector to different word each in document
Past is low-dimensional vector) it indicates.For example, perhaps we can finally obtain for a sequence of such " A B A C B F G "
Arrive: the corresponding vector of A is [0.1 0.6-0.5], and the corresponding vector of B is that [- 0.2 0.9 0.7] (numerical value herein is served only for showing
Meaning).Why wish each word to be become a vector, purpose or calculate for convenience, for example " asks that word A's is synonymous
Word ", so that it may be accomplished by " seeking the vector most like under cos distance with word A ".
Further, the promotion generation module includes: probability evaluation entity, under goal behavior, according to fisrt feature
Matrix and second characteristic matrix calculate the user and are directed to the probability for generating goal behavior to promotion article.
A kind of collaboration attention promotion method based on information, the method execute following steps:
Information word to be processed is pre-processed, the information term vector of each user and the information word of each article are generated
Vector;
Under goal behavior, the information term vector of information term vector and each article based on each user generates each use
Family is directed to the weight of the information term vector of each article respectively, constructs fisrt feature matrix;Under goal behavior, it is based on each use
The information term vector at family and the information term vector of each article generate the information term vector weight that each article is directed to each user,
Construct second characteristic matrix;
The user is calculated according to fisrt feature matrix and second characteristic matrix to be directed to promotion article generation goal behavior
Probability is ranked up the probability for generating goal behavior, generates promotion list.
Further, described to pre-process information word to be processed, generate the information term vector of each user and every
The method of the information term vector of a article executes following steps: information to be processed being segmented and/or gone according to Information Granularity
Stop words processing;Word is extracted respectively from user's dimension and article dimension;Generate the information term vector and each article of each user
Information term vector.
Further, the promotion generation module calculates user's needle according to fisrt feature matrix and second characteristic matrix
The probability that promotion article generates goal behavior is treated, the probability for generating goal behavior is ranked up, generates the side of promotion list
Method executes following steps: probability evaluation entity, and under goal behavior, being calculated according to fisrt feature matrix and second characteristic matrix should
User is directed to the probability that goal behavior is generated to promotion article.
A kind of collaboration attention promotion unit based on information, described device includes: a kind of computer of non-transitory can
Storage medium is read, which stores computations comprising: information word to be processed is pre-processed, is generated every
The code segment of the information term vector of the information term vector and each article of a user;Under goal behavior, based on each user's
The information term vector of information term vector and each article generates the power that each user is directed to the information term vector of each article respectively
Weight constructs fisrt feature matrix;Under goal behavior, the information word of information term vector and each article based on each user to
Amount generates the information term vector weight that each article is directed to each user, constructs the code segment of second characteristic matrix;According to first
Eigenmatrix and second characteristic matrix calculate the user and are directed to the probability for generating goal behavior to promotion article, to generation target line
For probability be ranked up, generate promotion list code segment.
It should be pointed out that herein referred information term vector and attention term vector are identical concept.With internet
Fast development, network be people obtain information, transmitting information and exchange emotion provide more by all kinds of means.We are divided with microblogging
Enjoy oneself daily what is seen and heard;It is done shopping using network, then leaves comment;In online booking wine before travelling on official business
User experience is shared after moving in shop.These data enumerate the clothing, food, lodging and transportion -- basic necessities of life of people, to consumer, establishment, even political affairs
There is huge value in mansion department.How emotion information is efficiently excavated from mass data, become currently urgently to be solved and ask
Topic.Machine learning techniques provide many methods for emotional semantic classification problem.The especially development of deep learning in recent years is emotion
The solution of classification problem brings new power.But there are still many deficiencies, have to be solved.Natural language is handled using deep learning
Processing problem first has to text is converted into the manageable form of computer, and currently more commonly used is the mode of term vector.
Although term vector all achieves outstanding representation in many tasks.But current most of term vector training methods are all based on word
The contextual information of language calculates term vector.In Chinese field, the meaning of word is also included among the word for forming it.Set forth herein
The information of word vector is added in term vector training method based on attention mechanism in term vector.Meanwhile in the process of addition
The middle significance level for considering different words.Finally in the tasks such as similarity calculation, reasoning from logic, emotional semantic classification, this method is verified
Obtained term vector has more outstanding expression ability.It is accumulated by many years, existing many different sentiment classification models, example
Such as support vector machines, convolutional neural networks, Recognition with Recurrent Neural Network.These models are based on different hypothesis, from different angles
Degree extracts the knowledge in data.In order to obtain better effect, different models may be integrated.Traditional integrated approach
Ballot method, the method for average or learning method are commonly used in the selection for combining strategy.These approaches increases the calculating of test phase
Amount.This paper presents be based on teachers and students' system integrating learning method.Multiple component classifiers are individually first trained, then initialize one
New neural network is as object classifiers.The training process of object classifiers is additionally added a other than referring to correct category
The judgement information of body classifier.In this way, multiple classifiers are compressed into a classifier, is keeping original property
While energy, less calculating cost is spent.
Person of ordinary skill in the field can be understood that, for convenience and simplicity of description, foregoing description
The specific work process of system and related explanation, can refer to corresponding processes in the foregoing method embodiment, details are not described herein.
It should be noted that system provided by the above embodiment, only illustrate with the division of above-mentioned each functional module
It is bright, in practical applications, it can according to need and complete above-mentioned function distribution by different functional modules, i.e., it will be of the invention
Module or step in embodiment are decomposed or are combined again, for example, the module of above-described embodiment can be merged into a module,
It can also be further split into multiple submodule, to complete all or part of the functions described above.The present invention is implemented
Module, the title of step involved in example, it is only for distinguish modules or step, be not intended as to of the invention improper
It limits.
Person of ordinary skill in the field can be understood that, for convenience and simplicity of description, foregoing description
The specific work process and related explanation of storage device, processing unit, can refer to corresponding processes in the foregoing method embodiment,
Details are not described herein.
Those skilled in the art should be able to recognize that, mould described in conjunction with the examples disclosed in the embodiments of the present disclosure
Block, method and step, can be realized with electronic hardware, computer software, or a combination of the two, software module, method and step pair
The program answered can be placed in random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electric erasable and can compile
Any other form of storage well known in journey ROM, register, hard disk, moveable magnetic disc, CD-ROM or technical field is situated between
In matter.In order to clearly demonstrate the interchangeability of electronic hardware and software, in the above description according to function generally
Describe each exemplary composition and step.These functions are executed actually with electronic hardware or software mode, depend on technology
The specific application and design constraint of scheme.Those skilled in the art can carry out using distinct methods each specific application
Realize described function, but such implementation should not be considered as beyond the scope of the present invention.
Term " first ", " second " etc. are to be used to distinguish similar objects, rather than be used to describe or indicate specific suitable
Sequence or precedence.
Term " includes " or any other like term are intended to cover non-exclusive inclusion, so that including a system
Process, method, article or equipment/device of column element not only includes those elements, but also including being not explicitly listed
Other elements, or further include the intrinsic element of these process, method, article or equipment/devices.
So far, it has been combined preferred embodiment shown in the drawings and describes technical solution of the present invention, still, this field
Technical staff is it is easily understood that protection scope of the present invention is expressly not limited to these specific embodiments.Without departing from this
Under the premise of the principle of invention, those skilled in the art can make equivalent change or replacement to the relevant technologies feature, these
Technical solution after change or replacement will fall within the scope of protection of the present invention.
The foregoing is only a preferred embodiment of the present invention, is not intended to limit the scope of the present invention.
Claims (8)
1. a kind of collaboration attention recommendation system based on information, which is characterized in that the system comprises:
Preprocessing module pre-processes information word to be processed, generates the information term vector and each article of each user
Information term vector;
Collaboration pays attention to power module, under goal behavior, the information word of information term vector and each article based on each user to
Amount generates the weight that each user is directed to the information term vector of each article respectively, constructs fisrt feature matrix;In goal behavior
Under, the information term vector of information term vector and each article based on each user generates the letter that each article is directed to each user
Term vector weight is ceased, second characteristic matrix is constructed;
Promotion generation module calculates the user according to fisrt feature matrix and second characteristic matrix and is directed to promotion article generation mesh
The probability of mark behavior is ranked up the probability for generating goal behavior, generates promotion list.
2. the collaboration attention recommendation system based on information as described in claim 1, which is characterized in that the preprocessing module
Include: word processing module, information to be processed is segmented according to Information Granularity and/or stop words is gone to handle;Dimension extracts mould
Block extracts word from user's dimension and article dimension respectively;Term vector generation module generates the information term vector of each user and every
The information term vector of a article.
3. the collaboration attention recommendation system based on information as claimed in claim 2, which is characterized in that the collaboration attention
Module includes: a granularity collaboration attention processing module, and when message complexity is high, distich seed degree carries out collaboration attention first
Power processing, then word granularity is carried out at collaboration attention
Reason, under goal behavior, the information term vector of information term vector and each article based on each user generates each use
Family is directed to the weight of the information term vector of each article respectively, constructs fisrt feature matrix;Under goal behavior, it is based on each use
The information term vector at family and the information term vector of each article generate the information term vector weight that each article is directed to each user,
Construct second characteristic matrix;Word granularity cooperates with attention processing module, and when message complexity is low, distich seed degree is carried out first
Attention processing is cooperateed with, then collaboration attention processing, under goal behavior, the information word based on each user are carried out to word granularity
The information term vector of each article of vector sum generates the weight that each user is directed to the information term vector of each article respectively, building
Fisrt feature matrix;Under goal behavior, the information term vector of information term vector and each article based on each user is generated
Each article is directed to the information term vector weight of each user, constructs second characteristic matrix.
4. the collaboration attention recommendation system based on information as claimed in claim 3, which is characterized in that the promotion generate mould
Block includes: probability evaluation entity, under goal behavior, calculates the user according to fisrt feature matrix and second characteristic matrix and is directed to
The probability of goal behavior is generated to promotion article.
5. a kind of collaboration attention promotion method based on information, which is characterized in that the method executes following steps:
Information word to be processed is pre-processed, generate each user information term vector and each article information word to
Amount;
Under goal behavior, the information term vector of information term vector and each article based on each user generates each user point
Safety pin constructs fisrt feature matrix to the weight of the information term vector of each article;Under goal behavior, based on each user's
The information term vector of information term vector and each article generates the information term vector weight that each article is directed to each user, building
Second characteristic matrix;
According to fisrt feature matrix and second characteristic matrix calculate the user be directed to promotion article generate goal behavior probability,
The probability for generating goal behavior is ranked up, promotion list is generated.
6. the collaboration attention promotion method based on information as claimed in claim 5, which is characterized in that it is described will be to be processed
Information word is pre-processed, and the method execution for generating the information term vector of each user and the information term vector of each article is following
Step: information to be processed is segmented according to Information Granularity and/or stop words is gone to handle;From user's dimension and article dimension point
It Chou Qu not word;Generate the information term vector of each user and the information term vector of each article.
7. the collaboration attention promotion method based on information as claimed in claim 6, which is characterized in that the promotion generate mould
Block, according to fisrt feature matrix and second characteristic matrix calculate the user be directed to promotion article generate goal behavior probability,
The probability for generating goal behavior is ranked up, the method for generating promotion list executes following steps: probability evaluation entity, in mesh
Under mark behavior, which is calculated according to fisrt feature matrix and second characteristic matrix and is directed to promotion article generation goal behavior
Probability.
8. a kind of collaboration attention promotion unit based on information, which is characterized in that described device includes: a kind of non-transitory
Computer readable storage medium, the storage medium store computations comprising: information word to be processed is located in advance
Reason, generates the code segment of the information term vector of each user and the information term vector of each article;Under goal behavior, based on every
The information term vector of a user and the information term vector of each article generate the information word that each user is directed to each article respectively
The weight of vector constructs fisrt feature matrix;Under goal behavior, information term vector and each article based on each user
Information term vector generates the information term vector weight that each article is directed to each user, constructs the code segment of second characteristic matrix;
The user is calculated for the probability for generating goal behavior to promotion article, to production according to fisrt feature matrix and second characteristic matrix
The probability of raw goal behavior is ranked up, and generates the code segment of promotion list.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910675118.7A CN110427568A (en) | 2019-07-25 | 2019-07-25 | A kind of collaboration attention recommendation system, method and apparatus based on information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910675118.7A CN110427568A (en) | 2019-07-25 | 2019-07-25 | A kind of collaboration attention recommendation system, method and apparatus based on information |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110427568A true CN110427568A (en) | 2019-11-08 |
Family
ID=68412407
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910675118.7A Pending CN110427568A (en) | 2019-07-25 | 2019-07-25 | A kind of collaboration attention recommendation system, method and apparatus based on information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110427568A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112800172A (en) * | 2021-02-07 | 2021-05-14 | 重庆大学 | Code searching method based on two-stage attention mechanism |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150262069A1 (en) * | 2014-03-11 | 2015-09-17 | Delvv, Inc. | Automatic topic and interest based content recommendation system for mobile devices |
CN107145518A (en) * | 2017-04-10 | 2017-09-08 | 同济大学 | Personalized recommendation system based on deep learning under a kind of social networks |
CN108520450A (en) * | 2018-03-21 | 2018-09-11 | 电子科技大学 | Local low-rank matrix based on implicit feedback information approximately recommends method and system |
CN109213868A (en) * | 2018-11-21 | 2019-01-15 | 中国科学院自动化研究所 | Entity level sensibility classification method based on convolution attention mechanism network |
CN109800853A (en) * | 2018-11-30 | 2019-05-24 | 华南师范大学 | Merge matrix disassembling method, device and the electronic equipment of convolutional neural networks and explicit feedback |
-
2019
- 2019-07-25 CN CN201910675118.7A patent/CN110427568A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150262069A1 (en) * | 2014-03-11 | 2015-09-17 | Delvv, Inc. | Automatic topic and interest based content recommendation system for mobile devices |
CN107145518A (en) * | 2017-04-10 | 2017-09-08 | 同济大学 | Personalized recommendation system based on deep learning under a kind of social networks |
CN108520450A (en) * | 2018-03-21 | 2018-09-11 | 电子科技大学 | Local low-rank matrix based on implicit feedback information approximately recommends method and system |
CN109213868A (en) * | 2018-11-21 | 2019-01-15 | 中国科学院自动化研究所 | Entity level sensibility classification method based on convolution attention mechanism network |
CN109800853A (en) * | 2018-11-30 | 2019-05-24 | 华南师范大学 | Merge matrix disassembling method, device and the electronic equipment of convolutional neural networks and explicit feedback |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112800172A (en) * | 2021-02-07 | 2021-05-14 | 重庆大学 | Code searching method based on two-stage attention mechanism |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Yousaf et al. | Emotion recognition by textual tweets classification using voting classifier (LR-SGD) | |
Ahuja et al. | Clustering and sentiment analysis on Twitter data | |
Jain et al. | Sentiments analysis of Twitter data using data mining | |
CN107908715A (en) | Microblog emotional polarity discriminating method based on Adaboost and grader Weighted Fusion | |
Moghaddam et al. | On the design of LDA models for aspect-based opinion mining | |
CN105005594B (en) | Abnormal microblog users recognition methods | |
CN111767725B (en) | Data processing method and device based on emotion polarity analysis model | |
CN108804512A (en) | Generating means, method and the computer readable storage medium of textual classification model | |
CN110909164A (en) | Text enhancement semantic classification method and system based on convolutional neural network | |
CN103164454A (en) | Keyword grouping method and keyword grouping system | |
Ilina et al. | Social event detection on twitter | |
CN105740382A (en) | Aspect classification method for short comment texts | |
CN107885883A (en) | A kind of macroeconomy field sentiment analysis method and system based on Social Media | |
Gui et al. | Intersubjectivity and sentiment: from language to knowledge | |
CN109446423B (en) | System and method for judging sentiment of news and texts | |
Dikiyanti et al. | Sentiment analysis and topic modeling of BPJS Kesehatan based on twitter crawling data using Indonesian Sentiment Lexicon and Latent Dirichlet Allocation algorithm | |
CN109960791A (en) | Judge the method and storage medium, terminal of text emotion | |
Suresh et al. | Mining of customer review feedback using sentiment analysis for smart phone product | |
Hao et al. | How do Mainland Chinese tourists perceive Hong Kong in turbulence? A deep learning approach to sentiment analytics | |
Bharathi et al. | Machine Learning Based Approach for Sentiment Analysis on Multilingual Code Mixing Text. | |
CN113065069B (en) | Bidirectional employment recommendation method and device based on data portrait | |
Choi et al. | Fake review identification and utility evaluation model using machine learning | |
CN109241993A (en) | Merge the evaluation object sensibility classification method and device of user and overall evaluation information | |
CN110427568A (en) | A kind of collaboration attention recommendation system, method and apparatus based on information | |
Mahajan et al. | E3: effective emoticon extractor for behavior analysis from social media |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191108 |
|
RJ01 | Rejection of invention patent application after publication |