CN108710680A - It is a kind of to carry out the recommendation method of the film based on sentiment analysis using deep learning - Google Patents
It is a kind of to carry out the recommendation method of the film based on sentiment analysis using deep learning Download PDFInfo
- Publication number
- CN108710680A CN108710680A CN201810481986.7A CN201810481986A CN108710680A CN 108710680 A CN108710680 A CN 108710680A CN 201810481986 A CN201810481986 A CN 201810481986A CN 108710680 A CN108710680 A CN 108710680A
- Authority
- CN
- China
- Prior art keywords
- film
- user
- comment
- information
- data
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of methods carrying out the recommendation of the film based on sentiment analysis using deep learning, obtain online film comment and the film status information of user, and carry out cleaning operation to film and its comment information and pre-process；The structure and feature information extraction for carrying out data carry out sentiment analysis by deep learning to characteristic information；User interest model is established based on sentiment analysis data later, then calculates user to film interest-degree and carries out film recommendation.The method that the present invention is analyzed with two-way shot and long term memory network as Sentiment orientation using Bagging algorithms, excavate the subjective emotion that user expresses in text message, effective user interest model is constructed, the precision of commending system, the quality that optimization film is recommended are improved.
Description
Technical field
The present invention relates to film recommended technology field more particularly to a kind of carried out based on sentiment analysis using deep learning
Film recommends method.
Background technology
With the development of Information technology, various the Internet, applications emerge one after another, and gradually penetrate into public daily life amusement
Etc. various aspects.The economy of people is gradually increased with living standard, and film has become a kind of important amusement and leisure of modern
Mode, the online viewing film of more and more people's selection simultaneously participate in online comment.It is abundant with movie resource simultaneously, how from
Oneself interested difficult point as people's selection is found in excessive film information.This makes the information sifting of film becomes to work as
A preceding popular research direction, and film commending system also becomes the effective means to solve the above problems.
It can help user to find valuable information to commending system one side, on the other hand can also allow information can
It is presented in before its interested user plane.Mainly have in the proposed algorithm of mainstream at present：Commending system based on collaborative filtering,
Including the collaborative filtering based on user and the collaborative filtering based on article；In addition have content-based recommendation system,
Knowledge based engineering recommends and mixing commending system.
Collaborative filtering is recommended based on user behavior data by building user interest model, passes through
Using user data, to find to possess the higher user of preference similarity and film, but system may recommend it to user not
The film that interested but its similar users are liked, the cold-start condition of collaborative filtering in addition, i.e., to new user or film
It evaluates small numbers of target user and recommends that there is difficulty, popular film can be recommended instead more.Although these question recommendings
The error effect of the accuracy rate of system entirety is limited, but this illustrates that this recommendation method still has defect.
And the recommendation based on content and article, it is the type by movie contents, the features such as plot carry out carrying numerically
It takes and decomposes, by recurrence to characteristic value or sort operation, obtain score information of the user to film, be then based on scoring
Recommend to be directed to target user.Its shortcoming is that having higher requirement, data that need to have more complete cinematic data structure
Content information be easy extraction condition, it is poor simultaneously for the recommendation effect of sparse data.
The modeling cost of mixed method is higher, need to integrate multiple proposed algorithms, and hybrid algorithm is directed in many cases
The effect is unsatisfactory for single specific actual augmentation.
This is because not making full use of user comment for the influence of recommendation results.The comment of user, which typically contains, compares
Valuable feedback can not only attract potential user, and user can be helped to make a decision, and user couple is typically contained in comment
Interest-degree of the user for commodity different characteristic is obtained, so by the sentiment analysis to historical review in the subjective emotion of film
After to build more accurate interest model more effective.
The deep learning method of rising in recent years achieves huge success in image and speech recognition, and gradually applies
To in natural language processing and machine translation, capable of preferably capable of judging the tendency and intensity of emotion from text message, but its
Using less in commending system.
Invention content
For the analysis of the above prior art defect, present invention proposition is a kind of to be carried out being based on sentiment analysis using deep learning
Film recommend method, can improve film recommendation accuracy rate, optimize user experience.
It is a kind of to carry out the film based on sentiment analysis to recommend method, this method including three aspects using deep learning：
In a first aspect, obtaining film comment and film status data, include the following steps：
S1：Obtain online film comment and the film status information of user.
S2：Cleaning operation is carried out to film and its comment information and is pre-processed.
Second aspect, the data obtained using deep learning method processing above-mentioned steps, adopts the technical scheme that one kind
Based on shot and long term memory network（Long-Short Term Memory, LSTM）Learning method, include the following steps：
S3：The structure and characteristic information of the data of acquisition are extracted.
S4：Sentiment analysis is carried out to characteristic information by deep learning.
The third aspect is recommended according to the sentiment analysis data obtained by training, adopts the technical scheme that and be based on
The data after sentiment analysis have been carried out to establish interest model and recommended.Include the following steps：
S5：User interest model is established based on sentiment analysis data.
S6：It calculates user and to film interest-degree and carries out film recommendation.
As a further improvement on the present invention, in step sl, the master data information of acquisition includes movie name ID, electricity
Shadow type, User ID, comment time, film scoring, comment content and approval number；The mode for obtaining data is to utilize online comment
Web crawlers or film information comment database.
As a further improvement on the present invention, in step s 2, include rejecting data type term to the cleaning operation of data
Mesh lacks and evaluates the data item that number of characters is less than preset characters number, and pretreatment operation includes being carried out to film comment information
Participle operation is operated with part-of-speech tagging.
Feature extraction is carried out to the data after S2 step process in step s3 for further improvement of the present invention
Operation include part of speech sequence template matching, be arranged dimensional parameter, training simultaneously build term vector matrix.
As a further improvement on the present invention, in step s 4, the deep learning network model taken is two-way length
Phase memory network（Bi-directional LSTM）, on this basis, the integrated of base learner is carried out using Bagging algorithms,
Final output corresponds to the affective state information of comment.
As a further improvement on the present invention, in step s 5, user interest model is established, emotion in S4 steps need to be based on
The result statistical model data information of analysis, then analyze the correlativity between film information feature and user.
As a further improvement on the present invention, in step s 6, user interest degree is calculated based on model obtained by step S5
Numerical value recommends the film for meeting user interest degree in film information data on this basis.
Compared with prior art, beneficial effects of the present invention are：
Film commending system through the invention, system can efficiently solve problem of information overload, pass through depth learning technology point
Analysis and the scoring of combination most users are handled with comment information, and the emotion information in comment is extracted and divided
Class excavates the feeling polarities that user expresses in text message, constructs effective user interest model, improve commending system
Precision, and then optimize the quality that film is recommended.
Description of the drawings
Fig. 1 is existing shot and long term memory network cellular construction figure.
Fig. 2 is two-way shot and long term memory network sentiment classification model disclosed in an embodiment of the present invention.
Fig. 3 is the Sentiment orientation of Bagging algorithms and two-way shot and long term memory network disclosed in an embodiment of the present invention
Analysis model figure.
Fig. 4 carries out the recommendation of the film based on sentiment analysis to be a kind of disclosed in an embodiment of the present invention using deep learning
The flow chart of method.
Specific implementation mode
Understand for the ease of those of ordinary skill in the art and implement the present invention, with reference to the accompanying drawings and embodiments to this hair
It is bright to be described in further detail, it should be understood that embodiment described herein is merely to illustrate and explain the present invention, and is not had to
It is of the invention in limiting.Based on the embodiments of the present invention, those of ordinary skill in the art are not before making creative work
The every other embodiment obtained is put, shall fall within the protection scope of the present invention.
As shown in figure 4, a kind of disclosed in a kind of embodiment of present invention offer carry out being based on sentiment analysis using deep learning
Film recommend the flow chart of method, which includes the following steps S1 to step S6：
S1：Obtain online film comment and the film status information of user.
In this embodiment, step S1 obtains master data information and includes movie name ID, film types, User ID, comments
By time, film scoring, comment content and approve of number；The mode for obtaining data is to utilize online comment web crawlers or film
Information comment database.
S2：Film and its comment information cleaning and pretreatment.
In this embodiment, as a further improvement on the present invention, the detailed process optimization of step S2 is as follows：
S21：The cleaning operation of data includes rejecting data type project missing and evaluation number of characters less than preset characters number
Data item.
Wherein, data items missing is mainly movie name ID and comment content, and effective preset characters number can be according to need
It is configured, such as 3 Chinese character numbers.
S22：Pretreatment operation includes carrying out participle operation to film comment information to operate with part-of-speech tagging.
Wherein, participle operation can be used based on the cutting of prefix dictionary and the stammerer for building directed acyclic graph according to dicing position
Participle tool carries out the cutting as unit of word, and utilizes point mutual information（PMI, Pointwise Mutual
Information）Algorithm carries out the extraction of related terms movie features, establishes feature set.
And part-of-speech tagging operation may be configured as four lexemes mark (B, M, E, S) or six lexeme marks according to the case where corpus
Note (B, B1 ,B2, M, E, S), to extract part of speech sequence.
S3：Extract the structure and characteristic information of data.
In this embodiment, as a further improvement on the present invention, the detailed process optimization of step S3 is as follows：
S31：Part of speech sequence template matches.
Wherein, the part of speech sequence that candidate attribute word is extracted according to comment data collection, as part of speech obtained by template matches S22
The text message of successful match can be carried out next step operation by sequence.
S32：Setting dimensional parameter simultaneously builds term vector matrix.
Wherein, structure term vector matrix can be used based on vector space model（Vector Space Model, VSM）Thought
With the Word2Vec tools of RNN networks, treated that word switchs to corresponding term vector by Chinese word segmentation for the tool, then structure text
This term vector matrix.The line number of the matrix is the word number of each text, and columns is the dimension specified by each word correspondence vector.
S4：The sentiment analysis of movie features is carried out by deep learning method.
In this embodiment, as a further improvement on the present invention, as follows to the specific optimization of step S4：
The deep learning network model taken is two-way shot and long term memory network（Bi-directional LSTM）, in this base
On plinth, the integrated of base learner is carried out using Bagging algorithms, final output corresponds to the affective state information of comment.
Wherein, Bagging algorithms belong to one kind of integrated study, complete to learn by structure and in conjunction with multiple learners
Task, the stochastical sampling in algorithm are generally adopted by self-service sampling method, the i.e. original training set for m sample, put back to
Ground random acquisition m times finally obtains a sampling set for including m sample.Each base learner is obtained by self-service sampling method
The training set obtained, theoretically has 37% initial data not to be selected, these ignored data are known as wrapping outer data, the calculation
Method helps to train complex model, improves the generalization ability of learning system.
Wherein, two-way shot and long term memory network is one of deep learning neural network structure, it is gathered around, and there are two different directions
Parallel layer, forward direction layer is identical as the method for operation of the method for operation of reversed layer and feedforward neural network.The two layers respectively from
The front end and end that text starts bring into operation, therefore can store the information of the text from both direction so that learning system
The contextual information till now with future can be considered simultaneously, to make it possess better performance in emotional semantic classification.
Fig. 3 is the Sentiment orientation of Bagging algorithms and two-way shot and long term memory network disclosed in an embodiment of the present invention
Analysis model figure illustrates the embodiment of the present disclosure with reference to Fig. 3.
With reference to Fig. 3, the model of the models coupling deep learning and the thought of integrated study, in the frame of Bagging algorithms
Under, the base learner that two-way shot and long term memory network is analyzed as Sentiment orientation, by sampled data come to two-way shot and long term
Memory network is learnt, and tax power is carried out using the base learner that outer data form training is wrapped, finally according to each base learner
Prediction result and temporal voting strategy carry out emotion prediction, specific implementation flow is as follows：
1）Data set to having marked affective tag pre-processes, and is classified as training set and test set two parts.
2）Self-service sampling method is used to the stochastical sampling of training set, training set is divided into n sampling set and the outer number of n packet
According to collection.
3）Two-way shot and long term memory network is passed to using sampling set data to be trained, the training process of n base learner
Independently of each other.
4）Verification and modified weight are carried out using wrapping in the incoming base learner of outer data set.
5）Repeat step 3）To 4）, the prediction result output until completing whole n base learners.
6）Based on Nearest Neighbor with Weighted Voting strategy, Sentiment orientation analysis prediction is carried out to forecast set sample.
S5：User interest model is established based on sentiment analysis data.
In this embodiment, as a further improvement on the present invention, the detailed process optimization of step S5 is as follows：
S51：Statistical model data information.
If affective state is positive, it is considered as favorable comment.Counting user is concentrated in entire comment data, for film information collection
The average positive rating for each feature closed, and the entirely average positive rating of each feature of user's set pair.
S52：Analyze film information feature fiTo the weight accounting of user's x comments, specific formula is as follows：
Wherein Weight (fi , x) and indicate film information feature fiTo the weight accounting of user's x comments, T（fi , x）Indicate feature
fiThe frequency occurred in the comment collection with x, N indicate the comment number of the comment collection of user x, nfiThere is feature f in expressioniComment
By number, F indicates the film information characteristic set of PMI algorithms extraction.
S53：User x is analyzed to film information feature fiPreference, specific formula is as follows：
Wherein Prefer (fi , x) and indicate user x to film information feature fiPreference, Ui(x) indicate user x to film spy
Levy fiAverage positive rating, ViIndicate overall user to movie features fiAverage positive rating.
S6：It calculates user and to film interest-degree and carries out film recommendation.
In this embodiment, as a further improvement on the present invention, the detailed process optimization of step S6 is as follows：
S61：User x is calculated to movie features fiInterest-degree, specific formula are as follows：
Wherein A (fi , x) and indicate user x to movie features fiInterest-degree.
S62:Film recommendation is carried out from movie database to the higher movie features of movie features interest-degree based on user
Just two-way shot and long term memory network emotional semantic classification base learner provided in this embodiment is described in detail below, to scheme
For 2 models provided.
Fig. 2 is two-way shot and long term memory network sentiment classification model disclosed in an embodiment of the present invention, in the present embodiment
In, the emotional semantic classification process implementation process of deep learning is as follows：
1）Extract the word feature vector of positive sequence and inverted sequence：
Term vector sequence corresponding to information before and after each word is input to two-way LSTM in the form of positive sequence and inverted sequence respectively
In network, corresponding positive and negative sequence signature is extracted.The output sequence of two-way LSTM networks is respectively positive characteristic vector sequence y_
F (0) to y_f (n) and opposite feature sequence vector y_r (n) to y_r (0) carries out sequence merging to two sequences, obtains
To corresponding word feature vector.
2）Extract sentence feature vector：
Term vector feature is averaging feature vector in word neighborhood using average pond while sequence merges, is obtained
Sentence feature vector.
3）Emotional semantic classification：
By sentence characteristic vector sequence, neural network is carried out to the sequence information and is connected entirely, information is passed to softmax layers later
Function probability operation is carried out, affective state classification results are finally obtained（Actively, neutral, it is passive）.
Further, which can be used the regularization method based on Dropout to alleviate the over-fitting of model.
Dropout methods act directly on the hiding node layer of neural network structure, and random selection unit is together with the defeated of them
Enter output connection, all temporarily abandon them from network, and the hiding node layer of random drop is not in training process every time
It is identical.Wherein the selected probability temporarily abandoned of unit can be artificially arranged when training, such as 0.2.
Wherein, the LSTM units of the disclosure embodiment model are as shown in Figure 1.
Fig. 1 is existing shot and long term memory network cellular construction figure, and LSTM efficiently solves Recognition with Recurrent Neural Network
The long range information Dependence Problem that (Recurrent Neural Networks, RNN) occurs, the cell state in LSTM units
(Memory Cell) is for preserving historical information.Historical information is controlled by 3 doors respectively:Input gate (Input Gate),
Forget door (Forget Gate), out gate (Output Gate).The information preservation that this structure inputs before being allowed to is in net
In network, and the transmission that goes ahead, input gate, which opens stylish input, can just change the historic state preserved in network, and out gate is beaten
The historic state preserved when holding can be accessed to, and the output after influencing, forget for emptying previously stored history letter
Breath, to be efficiently used to text sequence information.
The present invention recommends method to excavate user's film comment movie features, and emotion point is carried out by deep learning
Analysis, establishes user interest model based on this, improves the precision of commending system, and then optimizes the quality that film is recommended.
It these are only the preferred embodiment of the present invention, be not intended to restrict the invention, for those skilled in the art
For member, the invention may be variously modified and varied.Any modification made by all within the spirits and principles of the present invention,
Equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (8)
1. a kind of carrying out the recommendation method of the film based on sentiment analysis using deep learning, which is characterized in that include the following steps：
Step S1：Obtain online film comment and the film status information of user；
Step S2：Cleaning operation is carried out to film and its comment information and is pre-processed；
Step S3：The structure and characteristic information of the data of acquisition are extracted；
Step S4：Sentiment analysis is carried out to characteristic information by deep learning；
Step S5：User interest model is established based on sentiment analysis data；
Step S6：It calculates user and to film interest-degree and carries out film recommendation.
2. according to the method described in claim 1, it is characterized in that, in the step S2, the cleaning operation to data includes picking
Except data type entry lacks and evaluate the data item that number of characters is less than preset characters number, and pretreatment operation includes to film
Comment information carries out participle operation and is operated with part-of-speech tagging.
3. participle operation according to claim 3 is operated with part-of-speech tagging, which is characterized in that participle operation is using stammerer point
Word tool carries out the cutting as unit of word and extracts film noun feature set using PMI algorithms；Part-of-speech tagging operation setting is
Four lexemes mark or six lexemes mark.
4. according to the method described in claim 1, it is characterized in that, in the step S3, to the number after S2 step process
Include the matching of part of speech sequence template according to the operation of feature extraction is carried out, dimensional parameter is set, training simultaneously builds term vector matrix.
5. part of speech sequence template according to claim 4 and term vector matrix, which is characterized in that part of speech sequence template is to comment
By the part of speech sequence of the candidate attribute word extracted in data set；
The line number of term vector matrix is the word number of each text, and columns is the dimension specified by each word correspondence vector.
6. according to the method described in claim 1, it is characterized in that, in the step S4, the deep learning network mould taken
Type is two-way shot and long term memory network, and using the model as base learner, the integrated of base learner is carried out using Bagging algorithms,
Final output corresponds to the affective state information of comment.
7. Bagging algorithms according to claim 6 carry out the integrated of base learner, which is characterized in that including walking as follows
Suddenly：
1）Data set to having marked affective tag pre-processes, and is classified as training set and test set two parts；
2）Self-service sampling method is used to the stochastical sampling of training set, training set is divided into n sampling set and the n outer data set of packet；
3）Two-way shot and long term memory network is passed to using sampling set data to be trained, the training process of n base learner is mutual
It is independent；
4）Verification and modified weight are carried out using wrapping in the incoming base learner of outer data set；
5）Repeat step 3）To 4）, the prediction result output until completing whole n base learners；
6）Based on Nearest Neighbor with Weighted Voting strategy, Sentiment orientation analysis prediction is carried out to forecast set sample.
8. according to the method described in claim 1, it is characterized in that, in the step S5, user interest model is established, needs to analyze
For film information feature to the weight accounting of user comment, specific formula is as follows：
Wherein Weight (fi , x) and indicate film information feature fiTo the weight accounting of user's x comments, T（fi , x）Indicate feature
fiThe frequency occurred in the comment collection with x, N indicate the comment number of the comment collection of user x, nfiThere is feature f in expressioniComment
By number, F indicates the film information characteristic set of PMI algorithms extraction；
And analysis user, to the preference of film information feature, specific formula is as follows：
Wherein Prefer (fi , x) and indicate user x to film information feature fiPreference, Ui(x) indicate user x to movie features
fiAverage positive rating, ViIndicate overall user to movie features fiAverage positive rating.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810481986.7A CN108710680A (en) | 2018-05-18 | 2018-05-18 | It is a kind of to carry out the recommendation method of the film based on sentiment analysis using deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810481986.7A CN108710680A (en) | 2018-05-18 | 2018-05-18 | It is a kind of to carry out the recommendation method of the film based on sentiment analysis using deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108710680A true CN108710680A (en) | 2018-10-26 |
Family
ID=63869061
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810481986.7A Pending CN108710680A (en) | 2018-05-18 | 2018-05-18 | It is a kind of to carry out the recommendation method of the film based on sentiment analysis using deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108710680A (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109598451A (en) * | 2018-12-27 | 2019-04-09 | 东北大学 | A kind of non-intrusion type load discrimination method based on PCA Yu LSTM neural network |
| CN109635044A (en) * | 2018-12-13 | 2019-04-16 | 平安医疗健康管理股份有限公司 | Hospitalization data method for detecting abnormality, device, equipment and readable storage medium storing program for executing |
| CN109658268A (en) * | 2018-12-13 | 2019-04-19 | 平安医疗健康管理股份有限公司 | Monitoring method, monitoring service end and the storage medium of the medical abnormal behavior of medical treatment |
| CN109902229A (en) * | 2019-02-01 | 2019-06-18 | 中森云链(成都)科技有限责任公司 | A kind of interpretable recommended method based on comment |
| CN110096618A (en) * | 2019-05-10 | 2019-08-06 | 北京友普信息技术有限公司 | A kind of film recommended method based on fractional dimension sentiment analysis |
| CN110532379A (en) * | 2019-07-08 | 2019-12-03 | 广东工业大学 | A kind of electronics information recommended method of the user comment sentiment analysis based on LSTM |
| CN111144103A (en) * | 2019-12-18 | 2020-05-12 | 北京明略软件系统有限公司 | Film review identification method and device |
| CN111428040A (en) * | 2020-04-01 | 2020-07-17 | 中移雄安信息通信科技有限公司 | Method, device and equipment for determining target object and computer storage medium |
| CN111757186A (en) * | 2020-08-14 | 2020-10-09 | 付良辉 | Cinema showing method and system for interactive cinema films |
| CN111787409A (en) * | 2019-04-04 | 2020-10-16 | 杭州晨熹多媒体科技有限公司 | Movie and television comment data processing method and device |
| CN112818248A (en) * | 2021-02-26 | 2021-05-18 | 西北大学 | Emotion-based article recommendation model construction and recommendation method and system |
| CN112949322A (en) * | 2021-04-27 | 2021-06-11 | 李蕊男 | E-commerce opinion mining recommendation system driven by online text comments |
| CN113095087A (en) * | 2021-04-30 | 2021-07-09 | 哈尔滨理工大学 | Chinese word sense disambiguation method based on graph convolution neural network |
| CN113241193A (en) * | 2021-06-01 | 2021-08-10 | 平安科技（深圳）有限公司 | Drug recommendation model training method, recommendation method, device, equipment and medium |
| CN113688281A (en) * | 2021-07-09 | 2021-11-23 | 温州大学 | Video recommendation method and system based on deep learning behavior sequence |
| CN115089206A (en) * | 2022-05-09 | 2022-09-23 | 吴先洪 | Method for predicting heart sound signals and heart auscultation device using same |
| DE202023104110U1 (en) | 2023-07-23 | 2023-07-28 | Upasana Adhikari | Intelligent encryption-based system for movie recommendations |
| CN117851688A (en) * | 2024-03-06 | 2024-04-09 | 成都理工大学 | Personalized recommendation method based on deep learning and user comment content |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102236646A (en) * | 2010-04-20 | 2011-11-09 | 得利在线信息技术(北京)有限公司 | Personalized item-level vertical pagerank algorithm iRank |
| US20120095862A1 (en) * | 2010-10-15 | 2012-04-19 | Ness Computing, Inc. (a Delaware Corportaion) | Computer system and method for analyzing data sets and generating personalized recommendations |
| CN105426514A (en) * | 2015-11-30 | 2016-03-23 | 扬州大学 | Personalized mobile APP recommendation method |
| CN107038609A (en) * | 2017-04-24 | 2017-08-11 | 广州华企联信息科技有限公司 | A kind of Method of Commodity Recommendation and system based on deep learning |
-
2018
- 2018-05-18 CN CN201810481986.7A patent/CN108710680A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102236646A (en) * | 2010-04-20 | 2011-11-09 | 得利在线信息技术(北京)有限公司 | Personalized item-level vertical pagerank algorithm iRank |
| US20120095862A1 (en) * | 2010-10-15 | 2012-04-19 | Ness Computing, Inc. (a Delaware Corportaion) | Computer system and method for analyzing data sets and generating personalized recommendations |
| CN105426514A (en) * | 2015-11-30 | 2016-03-23 | 扬州大学 | Personalized mobile APP recommendation method |
| CN107038609A (en) * | 2017-04-24 | 2017-08-11 | 广州华企联信息科技有限公司 | A kind of Method of Commodity Recommendation and system based on deep learning |
Non-Patent Citations (3)
| Title |
|---|
| MG ARMENTANO 等: "Movies Recommendation Based on Opinion Mining in Twitter", 《MEXICAN INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE》 * |
| 侯银秀 等: "基于用户偏好与商品属性情感匹配的图书个性化推荐研究", 《数据分析与知识发现》 * |
| 金志刚 等: "一种结合深度学习和集成学习的情感分析模型", 《哈尔滨工业大学学报》 * |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109635044A (en) * | 2018-12-13 | 2019-04-16 | 平安医疗健康管理股份有限公司 | Hospitalization data method for detecting abnormality, device, equipment and readable storage medium storing program for executing |
| CN109658268A (en) * | 2018-12-13 | 2019-04-19 | 平安医疗健康管理股份有限公司 | Monitoring method, monitoring service end and the storage medium of the medical abnormal behavior of medical treatment |
| CN109598451A (en) * | 2018-12-27 | 2019-04-09 | 东北大学 | A kind of non-intrusion type load discrimination method based on PCA Yu LSTM neural network |
| CN109598451B (en) * | 2018-12-27 | 2023-01-13 | 东北大学 | Non-invasive load identification method based on PCA (principal component analysis) and LSTM (least Square TM) neural network |
| CN109902229A (en) * | 2019-02-01 | 2019-06-18 | 中森云链(成都)科技有限责任公司 | A kind of interpretable recommended method based on comment |
| CN111787409A (en) * | 2019-04-04 | 2020-10-16 | 杭州晨熹多媒体科技有限公司 | Movie and television comment data processing method and device |
| CN110096618A (en) * | 2019-05-10 | 2019-08-06 | 北京友普信息技术有限公司 | A kind of film recommended method based on fractional dimension sentiment analysis |
| CN110096618B (en) * | 2019-05-10 | 2021-06-15 | 北京友普信息技术有限公司 | Movie recommendation method based on dimension-based emotion analysis |
| CN110532379A (en) * | 2019-07-08 | 2019-12-03 | 广东工业大学 | A kind of electronics information recommended method of the user comment sentiment analysis based on LSTM |
| CN111144103A (en) * | 2019-12-18 | 2020-05-12 | 北京明略软件系统有限公司 | Film review identification method and device |
| CN111428040B (en) * | 2020-04-01 | 2023-08-18 | 中移雄安信息通信科技有限公司 | Target object determination method, device, equipment and computer storage medium |
| CN111428040A (en) * | 2020-04-01 | 2020-07-17 | 中移雄安信息通信科技有限公司 | Method, device and equipment for determining target object and computer storage medium |
| CN111757186A (en) * | 2020-08-14 | 2020-10-09 | 付良辉 | Cinema showing method and system for interactive cinema films |
| CN112818248A (en) * | 2021-02-26 | 2021-05-18 | 西北大学 | Emotion-based article recommendation model construction and recommendation method and system |
| CN112818248B (en) * | 2021-02-26 | 2023-04-18 | 西北大学 | Emotion-based article recommendation model construction and recommendation method and system |
| CN112949322A (en) * | 2021-04-27 | 2021-06-11 | 李蕊男 | E-commerce opinion mining recommendation system driven by online text comments |
| CN113095087A (en) * | 2021-04-30 | 2021-07-09 | 哈尔滨理工大学 | Chinese word sense disambiguation method based on graph convolution neural network |
| CN113241193A (en) * | 2021-06-01 | 2021-08-10 | 平安科技（深圳）有限公司 | Drug recommendation model training method, recommendation method, device, equipment and medium |
| CN113688281A (en) * | 2021-07-09 | 2021-11-23 | 温州大学 | Video recommendation method and system based on deep learning behavior sequence |
| CN113688281B (en) * | 2021-07-09 | 2023-11-17 | 温州大学 | Video recommendation method and system based on deep learning behavior sequence |
| CN115089206A (en) * | 2022-05-09 | 2022-09-23 | 吴先洪 | Method for predicting heart sound signals and heart auscultation device using same |
| DE202023104110U1 (en) | 2023-07-23 | 2023-07-28 | Upasana Adhikari | Intelligent encryption-based system for movie recommendations |
| CN117851688A (en) * | 2024-03-06 | 2024-04-09 | 成都理工大学 | Personalized recommendation method based on deep learning and user comment content |
| CN117851688B (en) * | 2024-03-06 | 2024-05-03 | 成都理工大学 | Personalized recommendation method based on deep learning and user comment content |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108710680A (en) | It is a kind of to carry out the recommendation method of the film based on sentiment analysis using deep learning | |
| US11631147B2 (en) | Social recommendation method based on multi-feature heterogeneous graph neural networks | |
| CN110083705B (en) | Multi-hop attention depth model, method, storage medium and terminal for target emotion classification | |
| AU2020100710A4 (en) | A method for sentiment analysis of film reviews based on deep learning and natural language processing | |
| CN109145112A (en) | A kind of comment on commodity classification method based on global information attention mechanism | |
| CN107133224B (en) | Language generation method based on subject word | |
| CN112199608B (en) | Social media rumor detection method based on network information propagation graph modeling | |
| CN110263257B (en) | Deep learning based recommendation method for processing multi-source heterogeneous data | |
| CN109299396A (en) | Merge the convolutional neural networks collaborative filtering recommending method and system of attention model | |
| CN109241255A (en) | A kind of intension recognizing method based on deep learning | |
| CN108229582A (en) | Entity recognition dual training method is named in a kind of multitask towards medical domain | |
| CN110516245A (en) | Fine granularity sentiment analysis method, apparatus, computer equipment and storage medium | |
| CN108363695B (en) | User comment attribute extraction method based on bidirectional dependency syntax tree representation | |
| CN110222163A (en) | A kind of intelligent answer method and system merging CNN and two-way LSTM | |
| CN109934261A (en) | A kind of Knowledge driving parameter transformation model and its few sample learning method | |
| CN107038480A (en) | A kind of text sentiment classification method based on convolutional neural networks | |
| CN106610955A (en) | Dictionary-based multi-dimensional emotion analysis method | |
| CN108984523A (en) | A kind of comment on commodity sentiment analysis method based on deep learning model | |
| CN107122455A (en) | A kind of network user's enhancing method for expressing based on microblogging | |
| CN108536801A (en) | A kind of civil aviaton's microblogging security public sentiment sentiment analysis method based on deep learning | |
| CN108038205A (en) | For the viewpoint analysis prototype system of Chinese microblogging | |
| CN106991161A (en) | A kind of method for automatically generating open-ended question answer | |
| CN110825850B (en) | Natural language theme classification method and device | |
| CN110532379A (en) | A kind of electronics information recommended method of the user comment sentiment analysis based on LSTM | |
| CN110909529B (en) | User emotion analysis and prejudgment system of company image promotion system |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181026 |