CN102012929A - Network consensus prediction method and system - Google Patents
Network consensus prediction method and system Download PDFInfo
- Publication number
- CN102012929A CN102012929A CN2010105682992A CN201010568299A CN102012929A CN 102012929 A CN102012929 A CN 102012929A CN 2010105682992 A CN2010105682992 A CN 2010105682992A CN 201010568299 A CN201010568299 A CN 201010568299A CN 102012929 A CN102012929 A CN 102012929A
- Authority
- CN
- China
- Prior art keywords
- time series
- network
- forecast model
- opinion
- network public
- 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
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a network consensus prediction method and a network consensus prediction system. The method comprises the following steps of: preprocessing network consensus information to acquire a time sequence needed for prediction; establishing a corresponding prediction model according to the time sequence acquired by preprocessing; and predicting a development trend of network consensus based on the prediction model. By the method and the system, the development trend of each hot topic on the network during a short period in the future can be predicted within the shortest time, so that topics which promote social stability are continuously kept by a guide method of the consensus; moreover, topics which destroy social harmony are led to extinction gradually.
Description
Technical field
The present invention relates to network information security analysis and forecasting techniques, more specifically, relate in particular to a kind of network public opinion prediction method and system.
Background technology
In recent years, along with information network technique very fast development worldwide, it is the emerging propagation force outside mass media mode such as newspaper, broadcasting, TV and the interpersonal media of continuing that the network media has become, and network becomes one of main carrier of the social will of the people of reflection.On July 15th, 2010, the statistical report of CNNIC (CNNIC) issue shows, ends in by the end of June, 2010, China netizen quantity has broken through 400,000,000 high pointes, total amount has reached 4.2 hundred million, and Internet penetration rises to 31.8%, and height ranks first in the world.As can be seen, increasing at home people is gathered in the internet, obtains and upgrades information in the internet.
Individuality in the internet is free to deliver viewpoint, and carries out alternately with other individual viewpoints.The macro manifestations of viewpoint can be regarded network public-opinion (networkconsensus) as in the network.More strict definition is as follows:
Network public-opinion is the set of the people that propagate the internet of passing through of producing of the stimulation owing to variety of event for all cognitions, attitude, emotion and the behavior disposition of this incident.In network environment, the main carrier of public feelings information comprises: news, BBS, Blog etc.Network public-opinion has characteristics such as expressing quick, information diversification and mode interaction, and this is that traditional media is incomparable.The network public-opinion development of content Network Based presents following characteristics:
1, sudden, can produce a large amount of discussion in the network in the utmost point short time; 2, the swift and violent property of topic velocity of propagation, topic can propagate into each website soon through the netizen; 3, much-talked-about topic persistence, general much-talked-about topic has a large amount of netizens and participates in discussion repeatedly; 4, coverage is wide, and network public-opinion has influenced daily life to a certain extent.
The characteristic of the sudden and quick propagation of network public-opinion makes it become a kind of reaction formation fast of public opinion, network public-opinion has begun reality society is produced certain influence, because network opening and virtual property have brought very big inconvenience for the internet supervision.Individuality in the network can arbitrarily be expressed viewpoint, if this viewpoint be with a certain focus incident for there being the basis, under the changeable in mood effect of subjectivity, this viewpoint is diffusion rapidly.Some individuality is met with setback in actual life, to social concern unilateral cognition or the like, all can utilize network to be led off.Therefore the speech of vulgar, the grey of easier appearance on network.Therefore, the research to network public-opinion is necessary.
From the angle of network security, predict the development trend of network public-opinion in advance, and the development of network public-opinion guided that this has great significance for social harmony is stable.Therefore the continuation of network much-talked-about topic has determined these much-talked-about topics that the regular hour continuity is arranged, and from the angle of time, is that unit predicts the follow-up developments trend of network much-talked-about topic with the quantity of network public-opinion content.
The function that the main public sentiment product in present home market has has:
1, public sentiment analytic function.This is the Core Feature of public sentiment product, and main contents comprise: (1) much-talked-about topic is found, sensitive subjects identification, can identify the hot issue in section preset time according to parameters such as news source technorati authority, number of reviews, time limit of speech dense degree.Utilize key word to deploy to ensure effective monitoring and control of illegal activities and semantic analysis, the identification sensitive subjects; (2) topic based on sentiment classification.For each topic, viewpoint, tendentiousness that each individuality is delivered are analyzed and added up; (3) theme is followed the tracks of.Analysis is newly published an article, whether the topic of model is identical with existing theme; (4) autoabstract.Can form summary automatically to all kinds of themes, present with the form of reporting; (5) incident analysis.Accident is striden the time, striden the spatial synthesis analysis, know incident generation development course, and the predicted events developing tendency in future; (7) warning system.Accident, the sensitive subjects that relates to content safety are in time found and reported to the police; (8) statistical report generates report according to the results repository after the processing of public sentiment analysis engine, and the user can browse by browser, information retrieval function is provided, according to specified requirements much-talked-about topic, tendentiousness are inquired about, and the particular content of browsing information, decision support is provided.
2, the network information is gathered automatically.According to customer information requirement, set the theme target, by the linking relationship between the Webpage, obtain page info from network automatically, and, finally finish the information gathering task of tailored range by linking constantly to whole extension of network.
3, data scrubbing function.The information of collecting is carried out pre-service, as format conversion, data scrubbing, data statistics.For news analysis, but this function filtering irrelevant information is preserved title, source address, issuing time, content, number of clicks, the participation of news and is commented on information such as people, comment content, number of reviews.For the BBS of forum, the title of record model, spokesman, issuing time, content, money order receipt to be signed and returned to the sender content, money order receipt to be signed and returned to the sender quantity etc. form formatted message at last.
To sum up, existing network public-opinion analytical approach is only analyzed at the network public-opinion that has occurred, do not have a kind of algorithm of more efficiently development trend of network public-opinion being predicted, therefore there is a kind of like this technical need, that is, need a kind of rapid and reliable network public opinion prediction method to predict the development trend of network public-opinion.
Summary of the invention
The object of the present invention is to provide a kind of network public opinion prediction method and system,, can more effectively predict the development trend of network public-opinion based on the present invention.
On the one hand, a kind of network public opinion prediction method of the present invention comprises: pre-treatment step, and network public sentiment information carries out pre-service, obtains and carries out the time series that forecasting institute needs; The forecast model establishment step according to the described time series that obtains through pre-service, is set up corresponding forecast model; Prediction steps is based on the development trend of described forecast model prediction network public-opinion.
In the above-mentioned network public opinion prediction method, preferred described pre-treatment step further comprises: data acquisition and cluster step, and the collection network public feelings information also carries out cluster to described network public sentiment information; The focus obtaining step according to cluster result, obtains the hot spot networks public feelings information; The time series obtaining step carries out data aggregate to described hot spot networks public feelings information, obtains to carry out the time series that forecasting institute needs.
In the above-mentioned network public opinion prediction method, in the preferred described forecast model establishment step, according to described time series, the forecast model of foundation is a reversal error disseminator artificial neural networks model.
In the above-mentioned network public opinion prediction method, in the preferred described forecast model establishment step, described reversal error disseminator artificial neural networks model comprises input layer, hidden layer and output layer; This modelling step further comprises: artificial neural network structure's establishment step, and set up reversal error and propagate the artificial neural network structure, determine the neuron number of described input layer, described hidden layer, described output layer; The parameter value determining step is determined the value of parameter in the described time series; Pre-estimation step by training, is estimated the value of calculating learning rate, two parameters of momentum term; Detect step, check the validity of described forecast model.
On the other hand, a kind of network public-opinion prognoses system of the present invention, comprising: pretreatment module, forecast model are set up module and prediction module.Wherein, pretreatment module is used for network public sentiment information and carries out pre-service, obtains and carries out the time series that forecasting institute needs; Forecast model is set up module and is used for setting up corresponding forecast model according to the described time series through the pre-service acquisition; Prediction module is used for the development trend based on described forecast model prediction network public-opinion.
Above-mentioned network public-opinion prognoses system, preferred described pretreatment module further comprises: data acquisition and cluster cell, focus acquiring unit and time series acquiring unit.Wherein, data acquisition and cluster cell are used for the collection network public feelings information and described network public sentiment information are carried out cluster; The focus acquiring unit is used for according to cluster result, obtains the hot spot networks public feelings information; The time series acquiring unit is used for described hot spot networks public feelings information is carried out data aggregate, obtains to carry out the time series that forecasting institute needs.
Above-mentioned network public-opinion prognoses system, preferred described forecast model is set up in the module, and according to described time series, the forecast model of foundation is a reversal error disseminator artificial neural networks model.
Above-mentioned network public-opinion prognoses system, preferred described forecast model is set up in the module, and described reversal error disseminator artificial neural networks model comprises input layer, hidden layer and output layer; This forecast model is set up module and further comprised: the artificial neural network structure sets up unit, parameter value determining unit, estimates unit and detecting unit in advance.Wherein, the artificial neural network structure sets up the unit and is used to set up reversal error propagation artificial neural network structure, determines the neuron number of described input layer, described hidden layer, described output layer; The parameter value determining unit is used for determining the value of described time series parameter; Pre-estimation unit is used for estimating the value of calculating learning rate, two parameters of momentum term by training; Detecting unit is used to check the validity of described forecast model.
In terms of existing technologies, the present invention has the following advantages: for each much-talked-about topic on the network, can go out the development trend that it (will be generally a day) in a short time in future in the shortest time prediction, be convenient to guide means by public sentiment, continue to keep those topics of promoting social stability, and destroy socially harmonious topic for those, and then to be guided, it is withered away gradually.
Description of drawings
Fig. 1 is the flow chart of steps of network public opinion prediction method embodiment of the present invention;
Fig. 2 is the BP neural network structure figure based on artificial neural network and seasonal effect in time series network public opinion prediction method according to an embodiment of the invention;
Fig. 3 is the 6 groups of predicted value examples of network public opinion prediction method according to an embodiment of the invention;
Fig. 4 is the structured flowchart of network public-opinion prognoses system embodiment of the present invention;
Fig. 5 is among the network public-opinion prognoses system embodiment of the present invention, the structured flowchart of pretreatment module;
Fig. 6 is among the network public-opinion prognoses system embodiment of the present invention, and forecast model is set up the structured flowchart of module.
Embodiment
For above-mentioned purpose of the present invention, feature and advantage can be become apparent more, the present invention is further detailed explanation below in conjunction with the drawings and specific embodiments.
With reference to Fig. 1, Fig. 1 comprises the steps: pre-treatment step S110 for the flow chart of steps of network public opinion prediction method embodiment of the present invention, and network public sentiment information carries out pre-service, obtains and carries out the time series that forecasting institute needs; Forecast model establishment step S120 according to the described time series that obtains through pre-service, sets up corresponding forecast model; Prediction steps S130 is based on the development trend of described forecast model prediction network public-opinion.
Fig. 2 is the schematic flow diagram based on artificial neural network and seasonal effect in time series network public opinion prediction method according to an embodiment of the invention.An embodiment of the invention relate to the prediction to network public-opinion, its forecasting object is the quantity of the news about this theme that occurs in the network in following a period of time, forum's model, blog etc. (below be referred to as " model "), wherein include only main model quantity, do not comprise answer quantity.Because therefore various Word message One's name is legions on the network and do not have rule at first will carry out pre-service to all models, are converted into the seasonal effect in time series form that forecasting institute needs.The experiment link of this algorithm adopt data from forum.
Preprocessing process is divided into following three steps:
1, the model on the network is carried out cluster.Because single model has suddenly in time, does not have general rule, therefore need predict all model total amounts that same topic is discussed on the network.Cluster process is exactly that all models of describing same topic are aggregated in the same classification.The method of cluster can adopt prior art, such as list of references 2 (list of references 2: clustering documents summary, Liu Yuanchao, Wang Xiaolong, Xu Zhiming etc. " Journal of Chinese Information Processing " .2005,20 (3): the technology 55-62).
2, obtain much-talked-about topic.Model quantity on the network is a lot, and pairing topic also is not quite similar, and the scope that most of topic relates to is very little, and the duration is very short, does not have necessity of prediction for this topic, therefore need find the much-talked-about topic in the network to predict.The concrete method of obtaining much-talked-about topic can adopt prior art, (list of references 3: the network much-talked-about topic of the flow content word degree of correlation is extracted such as list of references 3, Zhou Yadong, Sun Qindong, Guan Xiaohong etc. " XI AN JIAOTONG UNIVERSITY Subject Index " .2007.41 (10): 1142-1145,1150) in technology.
3, the model on the network is carried out data aggregate.Because an embodiment of the invention series model service time is predicted, it is input as time series, therefore the model on the network need be carried out data aggregate, obtain a time series, each value constantly be till the current time on the network all about the model of certain topic and the total amount of answer thereof.The method of data aggregate can adopt prior art.
The result that above-mentioned preprocessing process obtains just is to use based on artificial neural network and time series models and predicts needed time series.Suppose that the time series that the pre-service work in early stage obtains is [X
t], X
t=X (t), t=0,1,2....Seasonal effect in time series now or future value is considered in time series analysis and m value of its front between certain funtcional relationship is arranged, i.e. X
N+1=F (X
n, X
N-1..., X
N-m+1).Therefore, come fitting function F exactly based on artificial neural network and seasonal effect in time series network public-opinion prediction algorithm, thereby find out X
N+1And X
n, X
N-1..., X
N-m+1Between relation, be used for the prediction of this time series future value then.
According to an embodiment of the invention, preferably adopt three layers of BP neural network that time series is predicted.
On the basis of above-mentioned preprocessing process,, adopt the BP neural network model of above-mentioned three-decker to predict according to the time series that preprocessing process obtains.
Generally speaking, the structure of three layers of BP neural network is respectively input layer, output layer and hidden layer as shown in Figure 1.Neuronic output all is communicated to down one deck in each layer, and this transmission reaches enhancing, weakens or suppress these output ground effects by connecting power.Except the neuron of input layer, the neuronic clean input of hidden layer and output layer is the weighted sum of preceding one deck neuron output.Each neuron decides its activation degree by its input, activation functions and threshold value.
For convenience of description, input neuron is numbered with i, and the hidden layer neuron is numbered with j, and the output layer neuron is numbered with k.Other symbols that need illustrate are as follows:
o
i: input layer i neuronic output;
o
j: hidden layer j neuronic output;
o
k: k your output of neuron of output layer;
w
Ji: connect weight between i neuron of input layer and j neuron of hidden layer;
w
Kj: connect weight between j neuron of hidden layer and k neuron of output layer;
θ: threshold value, also be bias;
α: momentum term, its span are α ∈ (0,1);
η: learning rate, its span are η ∈ (0,1);
X
Max: time series [X
t] in maximal value;
X
Min: time series [X
t] in minimum value.
Hidden layer j neuronic being input as:
J neuronic being output as:
o
j＝f(net
j) (2)
Output layer k neuronic being input as:
Be output as accordingly:
o
k＝f(net
k) (4)
Wherein, f (x) is an activation functions, can be expressed as
Detailed process is as follows:
Step 1: determine network structure.Choose 3 layers of BP network, the neuron number of input layer, hidden layer, output layer is respectively 3,5,1.
Step 2: with the time series standardization that obtains previously.Can adopt one of following two methods:
Method 1:
Method 2:
N is the number of element in the related time series.
Step 3: give biasing θ and connect weights W in interval (1,1) interior random assignment
Ji, W
Kj
Step 4: be provided with weight variation delta W (i, j)=0, Δ W (j, k)=0, frequency of training n=1;
Step 5: use formula 8 to calculate the value of hidden layer neuron output
Step 6: use formula 9 to calculate the value of output layer neuron output
Step 7: calculate the connection weight and change and average error
Δw
ij＝ηδ
jo
i (11)
Δw
jk＝η(t
k-o
k)o
k(1-o
k)o
j (12)
Wherein, MSE is an average error, t
kBe desired output, and
δ
k=(t
k-o
k) o
k(1-o
k)
Step 8: whether the training process of judgement time sequence stops.
If average error MSE is less than e given in advance, then training process is finished, and forwards step 9 to; Otherwise, upgrade the connection weight according to formula (13), (14), n=n+1, and forward step 4 to;
w
ij(t+1)＝w
ij(t)+ηδ
jo
i+α[w
ij(t+1)-w
ij(t)] (13)
w
jk(t+1)＝w
jk(t)+ηδ
ko
j+α(w
jk(t+1)-w
jk(t)) (14)
Wherein, (t+1) expression (t+1) inferior iteration.
Step 9: preserve connection weight, bias;
Step 10: predict, and return and predict the outcome.
Table 1 is observed reading (original value) sequence,, themes as " room rate " data from the People's Net forum on making the country prosperous, intercepts the day amount of posting on July 25,1 day to 2010 June in 2010.Concrete numerical value is referring to table 1.
Table 1 raw data
The successful implementation of this algorithm need be to the training of experimentizing property of Several Parameters.Detailed process is as follows:
1, the time series degree of association (m chooses)
Discuss X according to the front
N+1Value and m of its front value X
n, X
N-1..., X
N-m+1Relevant, thus m determine to seem particularly important.Based on this, the data of choosing on June 23rd, 8 days 1 June in 2010 are training data, and preceding m days value is as input, and the value of this day is trained successively as output.Final this algorithm predicts amount of posting on June 24th, 2010 of using.Set two end conditions: each maximum cycle is 20000; MSE is 0.001, and promptly as long as satisfy wherein any one condition, algorithm all can stop.
Need to prove that MSE described in the present invention is after the raw data standardization in the training process of neural network the square error (as formula 10) as the Control Training precision.In addition, because the randomness of connection weight and biasing, predict the outcome also can exist certain randomness at every turn, in order to overcome this drawback, adopt the method repeatedly average for predicting the outcome of algorithm, promptly under the identical conditions, the predicted value of averaging repeatedly predicting the outcome as final.In addition, the MSE of 10 predictions is also made average handle, obtain MMSE (average MSE) as the error that finally predicts the outcome.The value of m is respectively 1,2,3,4,5, the concrete outcome of 6, six kind of prediction is as shown in table 2.
Table 2 concrete outcome table
" 6-1 " statement is when m gets 6, promptly to utilize the 7th day the amount of posting of preceding 6 days value prediction in the table 2.The implication of other statements and the like.
Take all factors into consideration predicted value and error, the value of decision m is 4 the most effective.
2, learning rate and momentum term
The data of choosing on June 23rd, 8 days 1 June in 2010 are training data, get m=4, maximum cycle N=10000, and least error is e=0.001.The value of learning rate and momentum term gets 0.1 to 0.9 respectively, then training data and giving a forecast under each combination.Following table is a corresponding results.
Result under the different value combined situation with momentum term of table 3 learning rate shows
Annotate: FD represents the consensus forecast value, and MMSE represents average MSE.
Table in the observation, because actual value is 17, so FD is the closer to 17, experiment effect is good more.In addition, for the Control Training precision, the more little experimental result of MMSE is good more.In order to weigh two parameters, the present invention proposes:
result＝N(FD-17)+N(MMSE) (15)
Wherein, function N (X) expression is carried out standardization to X.Result is more little, and it is good more that the present invention just thinks that it predicts the outcome, and training result is also comparatively accurate.The data of his-and-hers watches 3 are handled, result who obtains such as following table.
Result after table 4 his-and-hers watches 3 data processing shows
As can be seen, learning rate is 0.9, and momentum term is 0.3 o'clock, and value is minimum, and the best predicts the outcome this moment.
On the basis that parameter is selected, further work carries out trend prediction, and experimental data is chosen from table 1.Setting m is 4, and learning rate is 0.9, and momentum term is 0.3, training cycle index 30000 times, and MSE is set at 0.001, and setting training data is 22 groups.The roughly step of experiment: the amount of posting that promptly will predict 2010-07-15, need train preceding 22 days data, after training finished, to system input totally four days the day amount of posting from 2010-07-11 to 2010-07-14, system's return results promptly was the prediction amount of posting of 2010-07-15.It is as shown in table 5 specifically to predict the outcome:
The some signals that predict the outcome of table 5
From The above results as can be seen, be not very accurate though some predicts the outcome, public sentiment overall development trend is comparatively accurately, referring to Fig. 3, and horizontal ordinate express time (715 expression on July 15th, 2010) wherein, ordinate is represented the amount of posting.Explanation has directive significance preferably for some application scenario.
On the other hand, the invention also discloses a kind of network public-opinion prognoses system,, comprising with reference to Fig. 4: pretreatment module 1010, be used for network public sentiment information and carry out pre-service, obtain and carry out the time series that forecasting institute needs; Forecast model is set up module 1020, is used for setting up corresponding forecast model according to the described time series through the pre-service acquisition; Prediction module 1030 is used for the development trend based on described forecast model prediction network public-opinion.
With reference to Fig. 5, pretreatment module 1010 further comprises: data acquisition and cluster cell 10101, focus acquiring unit 10102 and time series acquiring unit 10103.Wherein, data acquisition and cluster cell 10101 are used for the collection network public feelings information and network public sentiment information are carried out cluster; Focus acquiring unit 10102 is used for according to cluster result, obtains the hot spot networks public feelings information; Time series acquiring unit 10103 is used for the hot spot networks public feelings information is carried out data aggregate, obtains to carry out the time series that forecasting institute needs.
In the foregoing description, forecast model is set up in the module 1020, and according to described time series, the forecast model of foundation is that reversal error is propagated (BP) artificial nerve network model.Reversal error disseminator artificial neural networks model comprises input layer, hidden layer and output layer; With reference to Fig. 6, this forecast model is set up module and further comprised: the artificial neural network structure sets up unit 10201, parameter value determining unit 10202, estimates unit 10203 and detecting unit 10204 in advance.Wherein, the artificial neural network structure sets up unit 10201 and is used to set up reversal error propagation artificial neural network structure, determines the neuron number of described input layer, described hidden layer, described output layer; Parameter value determining unit 10202 is used for determining the value of described time series parameter; Pre-estimation unit 10203 is used for estimating the value of calculating learning rate, two parameters of momentum term by training; Detecting unit 10204 is used to check the validity of described forecast model.
System embodiment is identical with the principle of method embodiment, mutually with reference to getting final product, does not give unnecessary details at this each other.
The present invention at first carries out pre-service to network public sentiment information (blog, bbs, news), obtain required time series,, set up forecast model then according to the time series that is obtained, and correlation parameter trained, adopt described forecast model prediction network public-opinion development trend at last.The present invention is for each much-talked-about topic on the network, can go out the development trend that it (will be generally a day) in a short time in future in the shortest time prediction, be convenient to guide means by public sentiment, continue to keep those topics of promoting social stability, and destroy socially harmonious topic for those, then to be guided, it is withered away gradually.
More than a kind of network public opinion prediction method provided by the present invention and system are described in detail, used specific embodiment herein principle of the present invention and embodiment are set forth, the explanation of above embodiment just is used for helping to understand method of the present invention and core concept thereof; Simultaneously, for one of ordinary skill in the art, according to thought of the present invention, part in specific embodiments and applications all can change.In sum, this description should not be construed as limitation of the present invention.
Claims (8)
1. a network public opinion prediction method is characterized in that, comprising:
Pre-treatment step is carried out pre-service to network public sentiment information, obtains and carries out the time series that forecasting institute needs;
The forecast model establishment step according to the described time series that obtains through pre-service, is set up corresponding forecast model;
Prediction steps is based on the development trend of described forecast model prediction network public-opinion.
2. network public opinion prediction method according to claim 1 is characterized in that, described pre-treatment step further comprises:
Data acquisition and cluster step, the collection network public feelings information also carries out cluster to described network public sentiment information;
The focus obtaining step according to cluster result, obtains the hot spot networks public feelings information;
The time series obtaining step carries out data aggregate to described hot spot networks public feelings information, obtains to carry out the time series that forecasting institute needs.
3. network public opinion prediction method according to claim 2 is characterized in that, in the described forecast model establishment step, according to described time series, the forecast model of foundation is a reversal error disseminator artificial neural networks model.
4. network public opinion prediction method according to claim 3 is characterized in that, in the described forecast model establishment step, described reversal error disseminator artificial neural networks model comprises input layer, hidden layer and output layer; This modelling step further comprises:
Artificial neural network structure's establishment step is set up reversal error and is propagated the artificial neural network structure, determines the neuron number of described input layer, described hidden layer, described output layer;
The parameter value determining step is determined the value of parameter in the described time series;
Pre-estimation step by training, is estimated the value of calculating learning rate, two parameters of momentum term;
Detect step, check the validity of described forecast model.
5. a network public-opinion prognoses system is characterized in that, comprising:
Pretreatment module is used for network public sentiment information and carries out pre-service, obtains and carries out the time series that forecasting institute needs;
Forecast model is set up module, is used for setting up corresponding forecast model according to the described time series through the pre-service acquisition;
Prediction module is used for the development trend based on described forecast model prediction network public-opinion.
6. network public-opinion prognoses system according to claim 5 is characterized in that, described pretreatment module further comprises:
Data acquisition and cluster cell are used for the collection network public feelings information and described network public sentiment information are carried out cluster;
The focus acquiring unit is used for according to cluster result, obtains the hot spot networks public feelings information;
The time series acquiring unit is used for described hot spot networks public feelings information is carried out data aggregate, obtains to carry out the time series that forecasting institute needs.
7. network public-opinion prognoses system according to claim 6 is characterized in that described forecast model is set up in the module, and according to described time series, the forecast model of foundation is a reversal error disseminator artificial neural networks model.
8. network public-opinion prognoses system according to claim 7 is characterized in that described forecast model is set up in the module, and described reversal error disseminator artificial neural networks model comprises input layer, hidden layer and output layer; This forecast model is set up module and is further comprised:
The artificial neural network structure sets up the unit, is used to set up reversal error and propagates the artificial neural network structure, determines the neuron number of described input layer, described hidden layer, described output layer;
The parameter value determining unit is used for determining the value of described time series parameter;
Pre-estimation unit is used for estimating the value of calculating learning rate, two parameters of momentum term by training;
Detecting unit is used to check the validity of described forecast model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105682992A CN102012929A (en) | 2010-11-26 | 2010-11-26 | Network consensus prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105682992A CN102012929A (en) | 2010-11-26 | 2010-11-26 | Network consensus prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102012929A true CN102012929A (en) | 2011-04-13 |
Family
ID=43843102
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010105682992A Pending CN102012929A (en) | 2010-11-26 | 2010-11-26 | Network consensus prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102012929A (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708153A (en) * | 2012-04-18 | 2012-10-03 | 中国信息安全测评中心 | Self-adaption finding and predicting method and system for hot topics of online social network |
| CN102999539A (en) * | 2011-09-13 | 2013-03-27 | 富士通株式会社 | Method and device for forecasting future development trend of given topic |
| CN103116651A (en) * | 2013-03-05 | 2013-05-22 | 南京理工大学常熟研究院有限公司 | Public sentiment hot topic dynamic detection method |
| CN103258248A (en) * | 2013-05-21 | 2013-08-21 | 中国科学院计算技术研究所 | Method, device and system for predicting microblog fashion trend |
| CN104008150A (en) * | 2014-05-20 | 2014-08-27 | 中国科学院信息工程研究所 | Method and system for predicting social network information transmission trend |
| CN104408108A (en) * | 2014-11-18 | 2015-03-11 | 重庆邮电大学 | Hot topic group influence analysis system and method based on grey system theory |
| CN105426391A (en) * | 2015-10-27 | 2016-03-23 | 张贝贝 | Method of acquiring diffusion pattern of network hot topic |
| CN105608200A (en) * | 2015-12-28 | 2016-05-25 | 湖南蚁坊软件有限公司 | Network public opinion tendency prediction analysis method |
| CN106156107A (en) * | 2015-04-03 | 2016-11-23 | 刘岩松 | A kind of discovery method of hot news |
| CN107783984A (en) * | 2016-08-25 | 2018-03-09 | 上海硕恩网络科技有限公司 | A kind of method of public sentiment monitoring |
| CN107784010A (en) * | 2016-08-29 | 2018-03-09 | 上海掌门科技有限公司 | A kind of method and apparatus for being used to determine the temperature information of theme of news |
| CN107862081A (en) * | 2017-11-29 | 2018-03-30 | 四川无声信息技术有限公司 | Network Information Sources lookup method, device and server |
| CN107895053A (en) * | 2017-12-13 | 2018-04-10 | 福州大学 | Emerging much-talked-about topic detecting system and method based on topic cluster momentum model |
| CN107992976A (en) * | 2017-12-15 | 2018-05-04 | 中国传媒大学 | Much-talked-about topic early-stage development trend predicting system and Forecasting Methodology |
| CN109214562A (en) * | 2018-08-24 | 2019-01-15 | 国网山东省电力公司电力科学研究院 | A kind of power grid scientific research hotspot prediction and method for pushing based on RNN |
| CN109299374A (en) * | 2018-10-24 | 2019-02-01 | 重庆理工大学 | Online social network information based on artificial neural network is propagated and the positive construction method of public sentiment evolution and system |
| CN109408620A (en) * | 2018-10-11 | 2019-03-01 | 杭州安恒信息技术股份有限公司 | A kind of method, apparatus, equipment and the storage medium of network public opinion trend prediction |
| CN109977393A (en) * | 2017-12-28 | 2019-07-05 | 中国科学院计算技术研究所 | A kind of popular news prediction technique and system based on content controversial |
| CN110083699A (en) * | 2019-03-18 | 2019-08-02 | 中国科学院自动化研究所 | News Popularity prediction model training method based on deep neural network |
| CN110084395A (en) * | 2019-02-18 | 2019-08-02 | 天津科技大学 | Prediction technique, device, server and the storage medium of network public-opinion evolution result |
| CN111178586A (en) * | 2019-12-06 | 2020-05-19 | 浙江工业大学 | Method for tracking, predicting and dredging public opinion events of network patriots |
| CN118152566A (en) * | 2024-03-04 | 2024-06-07 | 南京特尔顿信息科技有限公司 | Deep learning-based public opinion information identification method and system |
-
2010
- 2010-11-26 CN CN2010105682992A patent/CN102012929A/en active Pending
Cited By (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102999539B (en) * | 2011-09-13 | 2015-11-25 | 富士通株式会社 | Predict the method and apparatus of the future developing trend of given topic |
| CN102999539A (en) * | 2011-09-13 | 2013-03-27 | 富士通株式会社 | Method and device for forecasting future development trend of given topic |
| CN102708153A (en) * | 2012-04-18 | 2012-10-03 | 中国信息安全测评中心 | Self-adaption finding and predicting method and system for hot topics of online social network |
| CN102708153B (en) * | 2012-04-18 | 2015-05-06 | 中国信息安全测评中心 | Self-adaption finding and predicting method and system for hot topics of online social network |
| CN103116651A (en) * | 2013-03-05 | 2013-05-22 | 南京理工大学常熟研究院有限公司 | Public sentiment hot topic dynamic detection method |
| CN103258248A (en) * | 2013-05-21 | 2013-08-21 | 中国科学院计算技术研究所 | Method, device and system for predicting microblog fashion trend |
| CN103258248B (en) * | 2013-05-21 | 2016-12-07 | 中国科学院计算技术研究所 | A kind of microblog fashion trend Forecasting Methodology, Apparatus and system |
| CN104008150B (en) * | 2014-05-20 | 2017-07-07 | 中国科学院信息工程研究所 | A kind of method and system for propagating social network information trend prediction |
| CN104008150A (en) * | 2014-05-20 | 2014-08-27 | 中国科学院信息工程研究所 | Method and system for predicting social network information transmission trend |
| CN104408108A (en) * | 2014-11-18 | 2015-03-11 | 重庆邮电大学 | Hot topic group influence analysis system and method based on grey system theory |
| CN104408108B (en) * | 2014-11-18 | 2018-08-14 | 重庆邮电大学 | Much-talked-about topic group influence power analysis system based on gray system theory and method |
| CN106156107A (en) * | 2015-04-03 | 2016-11-23 | 刘岩松 | A kind of discovery method of hot news |
| CN106156107B (en) * | 2015-04-03 | 2019-12-13 | 刘岩松 | Method for discovering news hotspots |
| CN105426391A (en) * | 2015-10-27 | 2016-03-23 | 张贝贝 | Method of acquiring diffusion pattern of network hot topic |
| CN105608200A (en) * | 2015-12-28 | 2016-05-25 | 湖南蚁坊软件有限公司 | Network public opinion tendency prediction analysis method |
| CN107783984A (en) * | 2016-08-25 | 2018-03-09 | 上海硕恩网络科技有限公司 | A kind of method of public sentiment monitoring |
| CN107784010A (en) * | 2016-08-29 | 2018-03-09 | 上海掌门科技有限公司 | A kind of method and apparatus for being used to determine the temperature information of theme of news |
| CN107862081A (en) * | 2017-11-29 | 2018-03-30 | 四川无声信息技术有限公司 | Network Information Sources lookup method, device and server |
| CN107895053A (en) * | 2017-12-13 | 2018-04-10 | 福州大学 | Emerging much-talked-about topic detecting system and method based on topic cluster momentum model |
| CN107895053B (en) * | 2017-12-13 | 2021-09-24 | 福州大学 | Emerging hot topic detection system and method based on topic cluster momentum model |
| CN107992976A (en) * | 2017-12-15 | 2018-05-04 | 中国传媒大学 | Much-talked-about topic early-stage development trend predicting system and Forecasting Methodology |
| CN107992976B (en) * | 2017-12-15 | 2020-09-29 | 中国传媒大学 | Hot topic early development trend prediction system and prediction method |
| CN109977393A (en) * | 2017-12-28 | 2019-07-05 | 中国科学院计算技术研究所 | A kind of popular news prediction technique and system based on content controversial |
| CN109977393B (en) * | 2017-12-28 | 2021-09-03 | 中国科学院计算技术研究所 | Popular news prediction method and system based on content disputeness |
| CN109214562A (en) * | 2018-08-24 | 2019-01-15 | 国网山东省电力公司电力科学研究院 | A kind of power grid scientific research hotspot prediction and method for pushing based on RNN |
| CN109408620A (en) * | 2018-10-11 | 2019-03-01 | 杭州安恒信息技术股份有限公司 | A kind of method, apparatus, equipment and the storage medium of network public opinion trend prediction |
| CN109299374A (en) * | 2018-10-24 | 2019-02-01 | 重庆理工大学 | Online social network information based on artificial neural network is propagated and the positive construction method of public sentiment evolution and system |
| CN110084395A (en) * | 2019-02-18 | 2019-08-02 | 天津科技大学 | Prediction technique, device, server and the storage medium of network public-opinion evolution result |
| CN110083699A (en) * | 2019-03-18 | 2019-08-02 | 中国科学院自动化研究所 | News Popularity prediction model training method based on deep neural network |
| CN110083699B (en) * | 2019-03-18 | 2021-01-12 | 中国科学院自动化研究所 | News popularity prediction model training method based on deep neural network |
| CN111178586A (en) * | 2019-12-06 | 2020-05-19 | 浙江工业大学 | Method for tracking, predicting and dredging public opinion events of network patriots |
| CN111178586B (en) * | 2019-12-06 | 2022-09-23 | 浙江工业大学 | Method for tracking, predicting and dredging network patriotic public opinion events |
| CN118152566A (en) * | 2024-03-04 | 2024-06-07 | 南京特尔顿信息科技有限公司 | Deep learning-based public opinion information identification method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102012929A (en) | Network consensus prediction method and system | |
| Hwong et al. | What makes you tick? The psychology of social media engagement in space science communication | |
| Wang et al. | Understanding the power of opinion leaders’ influence on the diffusion process of popular mobile games: Travel Frog on Sina Weibo | |
| CN106776841B (en) | Method and system for acquiring Internet public opinion event propagation index | |
| Naveed et al. | Bad news travel fast: A content-based analysis of interestingness on twitter | |
| CN101394311A (en) | Network public opinion prediction method based on time sequence | |
| Weninger | An exploration of submissions and discussions in social news: Mining collective intelligence of Reddit | |
| CN102708153B (en) | Self-adaption finding and predicting method and system for hot topics of online social network | |
| CN101763401A (en) | Network public sentiment hotspot prediction and analysis method | |
| CN110336700B (en) | Microblog popularity prediction method based on time and user forwarding sequence | |
| Wiek et al. | Plausibility indications in future scenarios | |
| CN104182457A (en) | Poisson-process-model-based method for predicting event popularity in social network | |
| CN103218412B (en) | Public feelings information processing method and device | |
| CN102394798A (en) | Multi-feature based prediction method of propagation behavior of microblog information and system thereof | |
| CN103177129B (en) | Internet real-time information recommendation prognoses system | |
| CN103745105A (en) | Method and system for predicting user property in social network | |
| Li et al. | Lifecycle research of social media rumor refutation effectiveness based on machine learning and visualization technology | |
| CN106339463A (en) | Network public opinion early-warning system based on logistic model and early-warning method thereof | |
| Guille et al. | Predicting the temporal dynamics of information diffusion in social networks | |
| Dayeen et al. | A text mining analysis of the climate change literature in industrial ecology | |
| Xu et al. | A dynamic dissemination model for recurring online public opinion | |
| Zapico et al. | Climate persuasive services: changing behavior towards low-carbon lifestyles | |
| Basu et al. | Harnessing twitter data for analyzing public reactions to transportation policies: Evidences from the odd-even policy in Delhi, India | |
| CN103092921A (en) | Dynamic prediction method facing microblog hot topic community and system | |
| CN103838806A (en) | Analysis method for subject participation behaviors of user in social network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20110413 |