CN113128207A - News speaking right evaluation and prediction method based on big data - Google Patents
News speaking right evaluation and prediction method based on big data Download PDFInfo
- Publication number
- CN113128207A CN113128207A CN202110506791.5A CN202110506791A CN113128207A CN 113128207 A CN113128207 A CN 113128207A CN 202110506791 A CN202110506791 A CN 202110506791A CN 113128207 A CN113128207 A CN 113128207A
- Authority
- CN
- China
- Prior art keywords
- news
- index
- network
- order
- praise
- 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.)
- Granted
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 19
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims description 14
- 230000008451 emotion Effects 0.000 claims description 11
- 230000000875 corresponding effect Effects 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 230000002596 correlated effect Effects 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- ONUFESLQCSAYKA-UHFFFAOYSA-N iprodione Chemical compound O=C1N(C(=O)NC(C)C)CC(=O)N1C1=CC(Cl)=CC(Cl)=C1 ONUFESLQCSAYKA-UHFFFAOYSA-N 0.000 claims description 2
- 230000007935 neutral effect Effects 0.000 claims description 2
- 230000010485 coping Effects 0.000 abstract description 2
- 230000008092 positive effect Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 3
- 238000004880 explosion Methods 0.000 description 3
- 238000012552 review Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001680 brushing effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000005259 measurement 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
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Probability & Statistics with Applications (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a news speaking right evaluation and prediction method based on big data. The invention mainly realizes the evaluation of the speaking right of news and the prediction of the speaking right of news, and has positive effects on finding public opinion guidance in time under a big data environment so as to correctly and timely make a public opinion coping scheme.
Description
Technical Field
The invention relates to a news speaking right evaluation and prediction method based on big data, which mainly realizes evaluation of the speaking right of news and prediction of the speaking right of news, and has positive effect on finding public opinion guidance in time under the environment of big data so as to correctly and timely make a public opinion coping scheme.
Background
The traditional news speaking right mainly integrates liveness of users (the liveness of the users comprises news sent by the users and news forwarded by the users) and quality of the news sent by the users (the quality of the news sent by the users comprises the times that the news of the users is forwarded and commented on) as comprehensive indexes to obtain speaking right weight, and calculates the speaking right of the users in the social network by combining with a network plotter structure; or measuring from three dimensions of perception, participation and social attention. The perception degree can be divided into click amount and reading duration; the participation degree is the forwarding amount and the appraisal amount; social concerns are the amount of mention of social media, the amount of reprinting of other media, and the amount of subsequent reports. And finally, estimating and fitting the structural equation model by using a maximum likelihood estimation method, analyzing the overall adaptation degree of the model by using a fitting degree index provided by R software, determining the weight, and finally solving the speaking weight according to the weight. However, in these methods, the evaluation of the speaking right is performed based on simple forwarding or review of the news, it is difficult to distinguish interference of behaviors of malicious forwarding, and in addition, these methods do not consider that the reprinting and review of the news are in direct proportion to fans and attention of authors themselves, in which case, traditional calculation rules are more inclined to large V accounts, and it is difficult to discover explosion of ordinary accounts in time.
A prediction method of user speaking rights comprises the steps that someone establishes a measurement index related to the user speaking rights by analyzing the user speaking rights, basic attributes of each blog are used as main characteristics for measuring the speaking rights of microblogs, the speaking rights of the user are obtained by calculating the sum of the microblog speaking rights sent by the user, and then, xgboost is separately used for training aiming at four characteristic clusters to obtain a prediction model with a good fitting effect; the speaking right of the information is subjected to logarithmic processing, the early speaking right and the late speaking right present strong linear correlation, and a linear regression model is established by utilizing the strong correlation; still another person uses a machine learning algorithm, and takes user characteristics, forwarding behaviors, popularity and other influence factors as input of a machine learning model, and then predicts by using methods such as linear regression, classification regression tree, Gaussian process regression, support vector regression, neural network regression and the like; in the methods, only static indexes are selected, the time correlation is not considered, and the effect is poor.
Disclosure of Invention
The invention aims to provide a news speaking right evaluation and prediction method based on big data.
In order to solve the technical problems, the technical scheme adopted by the invention is that the news speaking right evaluation and prediction method based on big data comprises the following steps:
firstly, evaluation of speaking right:
(1) creating an N-order transshipment network: considering the diffusivity of data volume, N is generally 3; first-order transshipment network G for news A1=(A,A1) Wherein A is1To reprint all news collections of A, A1={A1i,i=1,2,…,m1}, second order transshipment network G2=(A1,A2),A2For transferring corresponding A1By analogy, the n-order transshipment network Gn=(An-1,An),AnFor transferring corresponding An-1News collection of (A)n={Ani,i=1,2,…,mn};
(2) And (3) calculating speaking right:
calculating the emotion index: if the emotion index is positively correlated with the speaking right, the n-order transshipment network G of news A is set as { G ═ G }1,G2,G3,…,GnAll comments Com ═ C of the news in1,C2,…,CmTraining a three-classification emotion analysis model Senti _ model (neu _ index, pos _ index, neg _ index) by using a bert model, and performing C comment on each commentiThe neutral, positive and negative probability distributions are neu _ index, pos _ index and neg _ index after analysis, and C is obtainediHas an emotion index of SCiThe sum of the sentiment indexes of all the comments in the comment index is pos _ index + neg _ index, which is the sum of the sentiment indexes of all the comments in the comment index
Calculating a transshipment index: n-order transshipment network for news a G ═ C1,C2,C3,…,GnAll news new ═ a1,A2,A3,…,An}, then the transshipment index
Wherein | A1| is the number of news reprinted a,to carry over A(k-1)iWhere k is 2,3, …, n;is news A(k-1)iThe reprint following index of the author of (1) is solved by the following steps: for news A(k-1)iAll news of the authors of (a) are ranked from high to low in capacity, i.e., { r1, r2, …, rn }, then,where Σ ri is ≦ i2;
Calculating the like index: similar to the solution of the reprint index, an N-order praise network is firstly constructed, the structure of the N-order praise network is consistent with that of the reprint network, wherein, the first-order praise network R of the news A1=(A,B1) N-th order network of praise Rn=(Bn-1,Bn),BnFor praise corresponding Bn-1News collection of, Bn={Bni,i=1,2,…,mnIs, then like index isWherein, | B1| is the news volume of praise a,to like B(k-1)iK 2,3, …, n;is news B(k-1)iThe praise following index of the author, the solution method andsimilarly, for news B(k-1)iDo asAll news of the people are ranked from high to low in praise number, i.e., { c1, c2, …, cn }, then,where Σ ci is equal to or less than i2;
The final utterance right Speech is Com _ index + rep _ index + fav _ index;
II, predicting the speaking right:
(3) selecting training set and test set, and training set D ═ D1,D2,…,DnWhere n is the number of data sets, DiThe ith news data in the D is obtained;
(4) solving the speech weight feature quantity X of the training set D ═ XD1,XD2,…,XDn}: let time t0,t1,……,tn,ti-ti-1>0, for each news item DiSolving D in different time periods by taking deltak as time intervaliThe speech weight feature quantity X ofDi,XDi={X1,X2,…,Xn-Δk+1In which X isjIs that it is in (t)j,tj+Δk]Evaluating characteristics of the speaking right in a time period, wherein j is 0, 1. Xj={a1,a2,…,a8},a1,a2,…,a8Are respectively (t)j,tj+Δk]Forwarding amount, number of comments, sentiment index, number of praise, Follow within a time periodA,mean(∑FollowG),assitA,mean(∑assitG) Wherein G is all news of the transferred A, and mean is the mean value;
(5) solving the speaking weight Y ═ { Y ] of the training set D according to the speaking weight calculation method in the step (2)1,Y2,…,Yn};
(6) Normalizing X and Y and inputting the normalized X and Y into a seq2last model for learning to obtain a speaking right prediction model; the seq2last model is composed of an LSTM neural network, an average pooling layer and a regression layer, LSTM output at each moment is input into the average pooling layer, the regression layer is connected behind a full connection layer to realize prediction, and an improved sigmoid function of the following formula is adopted in the regression layer:
a limiting parameter alpha is added in the formula, and an optimal prediction model is determined by adjusting the activation functions under different alpha values and utilizing a test set;
(7) solving the data a needing to be predicted according to the step (4) to obtain the speaking weight characteristic quantity XaAfter normalization, inputting the normalized data to the learned model and carrying out inverse normalization to obtain the speaking right.
The invention has the beneficial effects that:
1. most of the evaluation of news speaking right in the traditional method is evaluated based on the forwarding amount (or reading amount), but the cases of malicious forwarding or evaluation brushing cannot be well identified and eliminated, so the invention provides a speaking right evaluation method based on a high-order transshipment relationship network, and the cases of malicious forwarding are eliminated based on a high-order dependency relationship.
2. Because the reprinting and the comment of news are in direct proportion to the fans and attention of the authors, the traditional calculation rules are more inclined to the large V account numbers under the condition and the explosion of the common account numbers is difficult to discover in time, so that the author follows the index concept is introduced, the problem can be effectively solved based on the characteristic normalization of the index, and meanwhile, the sentiment index is introduced, and the evaluation reliability is improved.
3. According to actual experience, the speaking right of a certain period of time is related to the speaking right of the previous period of time, most of the traditional speaking right prediction methods do not consider time factors, and the prediction is carried out based on the time correlation in consideration of the time correlation.
4. Meanwhile, the invention realizes the prediction of the speaking right based on an improved seq2last model, the final LSTM layer result does not directly take the value of the last moment as the final result, but adds a maximum pooling layer to take the average value of each moment as the final result, and the activation function of the regression layer uses an improved sigmoid function, so that the prediction effect is better.
Drawings
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
Fig. 1 is a transfer network of an embodiment of the present invention.
FIG. 2 is a seq2last model of an embodiment of the present invention.
Detailed Description
The assessment of the speaking right of a news A mainly comprises the following steps:
(1) firstly, creating an N-order transshipment network: in consideration of the diffusivity of the data amount, N is generally 3. Taking 3-order transshipment network as an example, news A first-order transshipment network G1=(A,A1) Wherein A is1To transship all news collections of A, the second order transships network G2=(A1,A2),A2For transferring corresponding A1In analogy to the news gathering, G3=(A2,A3),A3For transferring corresponding A2A news collection of (a). Fig. 1 is a 1-N order transfer network.
(2) And (3) calculating speaking right:
calculating the emotion index: if the research finds that the comment emotion index is positively correlated with the speaking right, the 3-order transshipment network G of the news A is determined as G ═ G1,G2,G3All comments Com ═ C of the news in1,C2,…,CmWith a three-classification sentiment analysis bert model Senti _ model (neu _ index, pos _ index, neg _ index), for each comment CiAnalyzing to obtain the probability distribution, then CiHas an emotion index of SCiWhen pos _ index + neg _ index is defined, the index of review is
Calculating a transshipment index: 3 rd order reprinting network for news a G ═ G1,G2,G3All news new ═ a1,A2,A3}, then the transshipment index
Wherein | A1| is the number of news reprinted a,to carry over A(k-1)iWhere k is 2,3, …, n;is news A(k-1)iThe reprint following index of the author of (1) is solved by the following steps: for news A(k-1)iAll news of the authors of (a) are ranked from high to low in capacity, i.e., { r1, r2, …, rn }, then,where Σ ri is ≦ i2;
Calculating the like index: similar to the solution of the reprint index, an N-order praise network is firstly constructed, the structure of the N-order praise network is consistent with that of the reprint network, wherein, the first-order praise network R of the news A1=(A,B1) N-th order network of praise Rn=(Bn-1,Bn),BnFor praise corresponding Bn-1News collection of, Bn={Bni,i=1,2,…,mnIs, then like index isWherein, | B1| is the news volume of praise a,to like B(k-1)iK 2,3, …, n;is news B(k-1)iThe praise following index of the author, the solution method andsimilarly, for news B(k-1)iAll news of the authors of (1) are ranked from high to low in terms of their praise numbers, i.e., { c1, c2, …, cn }, then,where Σ ci is equal to or less than i2;
The final utterance right Speech is Com _ index + rep _ index + fav _ index;
secondly, completing the prediction of the speaking right according to the variable characteristics, mainly constructing a prediction model, and specifically comprising the following steps:
(1) selecting training set and test set, and training set D ═ D1,D2,…,DnWhere n is the number of data sets, DiThe ith news data in the D is obtained;
(2) solving the utterance weight feature quantity X of the data set D ═ { X ═ XD1,XD2,…,XDn}: let time t0,t1……tn,ti-ti-1>0, for each news item DiSolving D in different time periods by taking deltak as time intervaliThe speech weight feature quantity X ofDi,XDi={X1,X2,…,Xn-Δk+1In which X isjIs that it is in (t)j,tj+Δk]Evaluating the speech right evaluation characteristics in the time period, j is 0,1j={a1,a2,…,a8},a1,a2,…,a8Are respectively (t)j,tj+Δk]Forwarding amount, number of comments, sentiment index, number of praise, Follow within a time periodA,mean(∑FollowG),assitA,mean(∑assitG) Wherein G is all news of the transferred A, and mean is the mean value;
(3) solving the speaking weight Y ═ Y of the training set D according to the speaking weight calculation method1,Y2,…,Yn};
(4) Normalizing X and Y and inputting the normalized X and Y into a seq2last model for learning to obtain a speaking right prediction model, wherein the structure of the seq2last model is shown as an attached figure 2;
(5) by optimization of model parameters, e.g. regression layer activation functionAdjusting the limiting parameter alpha, and determining an optimal model by using the test set;
(6) solving the data a needing to be predicted according to the step (2) to obtain the speaking weight characteristic quantity XaAfter normalization, inputting the normalized data to the learned model and carrying out inverse normalization to obtain the speaking right.
The embodiment has the following technical characteristics:
1. the embodiment provides a speaking right evaluation method based on a high-order transshipment relationship network, which excludes the malicious forwarding conditions based on a high-order dependency relationship.
2. Because the reprinting and the comment of the news are in direct proportion to the fans and attention of the authors, the traditional calculation rules are more prone to the large V accounts, and the explosion of the common accounts is difficult to discover in time, so that the author follows the index concept is introduced in the embodiment, the problem can be effectively solved based on the characteristic normalization of the index, and meanwhile, the sentiment index is introduced for evaluation.
3. The present embodiment takes into consideration the correlation of time, and performs prediction based on the time correlation.
4. The embodiment realizes the prediction of speaking right based on an improved seq2last model, the final LSTM layer result does not directly take the value of the last moment as the final result, but adds a maximum pooling layer to take the average value of each moment as the final result, and the activation function of the regression layer uses an improved sigmoid function.
The above-described embodiments of the present invention do not limit the scope of the present invention. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims (1)
1. The news speaking right evaluation and prediction method based on big data comprises the following steps:
firstly, evaluation of speaking right:
(1) creating an N-order transshipment network: considering the diffusivity of data volume, N is generally 3; first-order transshipment network G for news A1=(A,A1) Wherein A is1To reprint all news collections of A, A1={A1i,i=1,2,…,m1}, second order transshipment network G2=(A1,A2),A2For transferring corresponding A1By analogy, the n-order transshipment network Gn=(An-1,An),AnFor transferring corresponding An-1News collection of (A)n={Ani,i=1,2,…,mn};
(2) And (3) calculating speaking right:
calculating the emotion index: if the emotion index is positively correlated with the speaking right, the n-order transshipment network G of news A is set as { G ═ G }1,G2,G3,…,GnAll comments Com ═ C of the news in1,C2,…,CmTraining a three-classification emotion analysis model Senti _ model (neu _ index, pos _ index, neg _ index) by using a bert model, and performing C comment on each commentiThe neutral, positive and negative probability distributions are neu _ index, pos _ index and neg _ index after analysis, and C is obtainediHas an emotion index of SCiThe sum of the sentiment indexes of all the comments in the comment index is pos _ index + neg _ index, which is the sum of the sentiment indexes of all the comments in the comment index
Calculating a transshipment index: n-order transshipment network for news a G ═ G1,G2,G3,…,GnAll news new ═ a1,A2,A3,…,An}, then the transshipment index
Wherein | A1| is the number of news reprinted a,to carry over A(k-1)iWhere k is 2,3, …, n;is news A(k-1)iThe reprint following index of the author of (1) is solved by the following steps: for news A(k-1)iAll news of the authors of (a) are ranked from high to low in capacity, i.e., { r1, r2, …, rn }, then,where Σ ri is ≦ i2;
Calculating the like index: similar to the solution of the reprint index, an N-order praise network is firstly constructed, the structure of the N-order praise network is consistent with that of the reprint network, wherein, the first-order praise network R of the news A1=(A,B1) N-th order network of praise Rn=(Bn-1,Bn),BnFor praise corresponding Bn-1News collection of, Bn={Bni,i=1,2,…,mnIs, then like index isWherein, | B1| is the news volume of praise a,to like B(k-1)iK 2,3, …, n;is news B(k-1)iThe praise following index of the author, the solution method andsimilarly, for news B(k-1)iAll news of the authors of (1) are ranked from high to low in terms of their praise numbers, i.e., { c1, c2, …, cn }, then,where Σ ci is equal to or less than i2;
The final utterance right Speech is Com _ index + rep _ index + fav _ index;
II, predicting the speaking right:
(3) selecting training set and test set, and training set D ═ D1,D2,…,DnWhere n is the number of data sets, DiThe ith news data in the D is obtained;
(4) solving the speech weight feature quantity X of the training set D ═ XD1,XD2,…,XDn}: let time t0,t1,……,tn,ti-ti-1> 0, for each news item DiSolving D in different time periods by taking deltak as time intervaliThe speech weight feature quantity X ofDi,XDi={X1,X2,…,Xn-Δk+1In which X isjIs that it is in (t)j,tj+Δk]Evaluating characteristics of the speaking right in a time period, wherein j is 0, 1. Xj={a1,a2,…,a8},a1,a2,…,a8Are respectively (t)j,tj+Δk]Forwarding amount, number of comments, sentiment index, number of praise, Follow within a time periodA,mean(∑FollowG),assitA,mean(∑assitG) Wherein G is all news of the transferred A, and mean is the mean value;
(5) solving the speaking weight Y ═ { Y ] of the training set D according to the speaking weight calculation method in the step (2)1,Y2,…,Yn};
(6) Normalizing and inputting X and Y into a seq21ast model for learning to obtain a speaking right prediction model; the seq21ast model is composed of an LSTM neural network, an average pooling layer and a regression layer, LSTM output at each moment is input into the average pooling layer, the regression layer is connected behind a full connection layer to realize prediction, and an improved sigmoid function of the following formula is adopted in the regression layer:
a limiting parameter alpha is added in the formula, and an optimal prediction model is determined by adjusting the activation functions under different alpha values and utilizing a test set;
(7) solving the data a needing to be predicted according to the step (4) to obtain the speaking weight characteristic quantity XaAfter normalization, inputting the normalized data to the learned model and carrying out inverse normalization to obtain the speaking right.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110506791.5A CN113128207B (en) | 2021-05-10 | 2021-05-10 | News speaking right assessment and prediction method based on big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110506791.5A CN113128207B (en) | 2021-05-10 | 2021-05-10 | News speaking right assessment and prediction method based on big data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113128207A true CN113128207A (en) | 2021-07-16 |
CN113128207B CN113128207B (en) | 2024-03-29 |
Family
ID=76781254
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110506791.5A Active CN113128207B (en) | 2021-05-10 | 2021-05-10 | News speaking right assessment and prediction method based on big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113128207B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107102976A (en) * | 2017-03-23 | 2017-08-29 | 北京大学 | Entertainment newses autocreating technology and system based on microblogging |
CN108509629A (en) * | 2018-04-09 | 2018-09-07 | 南京大学 | Text emotion analysis method based on emotion dictionary and support vector machine |
CN108932291A (en) * | 2018-05-23 | 2018-12-04 | 福建亿榕信息技术有限公司 | Power grid public sentiment evaluation method, storage medium and computer |
CN109101597A (en) * | 2018-07-31 | 2018-12-28 | 中电传媒股份有限公司 | A kind of electric power news data acquisition system |
WO2020000847A1 (en) * | 2018-06-25 | 2020-01-02 | 中译语通科技股份有限公司 | News big data-based method and system for monitoring and analyzing risk perception index |
US20200302540A1 (en) * | 2019-03-21 | 2020-09-24 | The University Of Chicago | Applying a trained model to predict a future value using contextualized sentiment data |
CN111966878A (en) * | 2020-08-04 | 2020-11-20 | 厦门大学 | Public sentiment event reversal detection method based on machine learning |
CN112464082A (en) * | 2020-11-05 | 2021-03-09 | 重庆邮电大学 | Rumor-splitting game propagation control method based on sparse representation and tensor completion |
-
2021
- 2021-05-10 CN CN202110506791.5A patent/CN113128207B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107102976A (en) * | 2017-03-23 | 2017-08-29 | 北京大学 | Entertainment newses autocreating technology and system based on microblogging |
CN108509629A (en) * | 2018-04-09 | 2018-09-07 | 南京大学 | Text emotion analysis method based on emotion dictionary and support vector machine |
CN108932291A (en) * | 2018-05-23 | 2018-12-04 | 福建亿榕信息技术有限公司 | Power grid public sentiment evaluation method, storage medium and computer |
WO2020000847A1 (en) * | 2018-06-25 | 2020-01-02 | 中译语通科技股份有限公司 | News big data-based method and system for monitoring and analyzing risk perception index |
CN109101597A (en) * | 2018-07-31 | 2018-12-28 | 中电传媒股份有限公司 | A kind of electric power news data acquisition system |
US20200302540A1 (en) * | 2019-03-21 | 2020-09-24 | The University Of Chicago | Applying a trained model to predict a future value using contextualized sentiment data |
CN111966878A (en) * | 2020-08-04 | 2020-11-20 | 厦门大学 | Public sentiment event reversal detection method based on machine learning |
CN112464082A (en) * | 2020-11-05 | 2021-03-09 | 重庆邮电大学 | Rumor-splitting game propagation control method based on sparse representation and tensor completion |
Also Published As
Publication number | Publication date |
---|---|
CN113128207B (en) | 2024-03-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107766929B (en) | Model analysis method and device | |
CN110929164A (en) | Interest point recommendation method based on user dynamic preference and attention mechanism | |
US20230385553A1 (en) | Techniques to add smart device information to machine learning for increased context | |
CN110503531A (en) | The dynamic social activity scene recommended method of timing perception | |
CN108038492A (en) | A kind of perceptual term vector and sensibility classification method based on deep learning | |
CN113469470B (en) | Energy consumption data and carbon emission correlation analysis method based on electric brain center | |
CN111753093A (en) | Method and device for evaluating level of network public opinion crisis | |
CN111538827A (en) | Case recommendation method and device based on content and graph neural network and storage medium | |
CN108229731B (en) | System and method for predicting user behavior under hot topics under multi-message interaction | |
CN111178611A (en) | Method for predicting daily electric quantity | |
CN110689523A (en) | Personalized image information evaluation method based on meta-learning and information data processing terminal | |
CN115688024B (en) | Network abnormal user prediction method based on user content characteristics and behavior characteristics | |
CN112215604A (en) | Method and device for identifying information of transaction relationship | |
CN115115389A (en) | Express customer loss prediction method based on value subdivision and integrated prediction | |
Xiao et al. | User behavior prediction of social hotspots based on multimessage interaction and neural network | |
CN114154557A (en) | Cancer tissue classification method, apparatus, electronic device, and storage medium | |
CN109597944B (en) | Single-classification microblog rumor detection model based on deep belief network | |
CN112541530B (en) | Data preprocessing method and device for clustering model | |
CN112464281B (en) | Network information analysis method based on privacy grouping and emotion recognition | |
CN116955650A (en) | Information retrieval optimization method and system based on small sample knowledge graph completion | |
Feng et al. | Web Service QoS Classification Based on Optimized Convolutional Neural Network | |
CN113128207A (en) | News speaking right evaluation and prediction method based on big data | |
CN115829683A (en) | Power integration commodity recommendation method and system based on inverse reward learning optimization | |
CN111325027B (en) | Sparse data-oriented personalized emotion analysis method and device | |
CN114936890A (en) | Counter-fact fairness recommendation method based on inverse tendency weighting method |
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 |