CN117743581A - Intervention method for agricultural product quality safety network rumors - Google Patents
Intervention method for agricultural product quality safety network rumors Download PDFInfo
- Publication number
- CN117743581A CN117743581A CN202311807450.7A CN202311807450A CN117743581A CN 117743581 A CN117743581 A CN 117743581A CN 202311807450 A CN202311807450 A CN 202311807450A CN 117743581 A CN117743581 A CN 117743581A
- Authority
- CN
- China
- Prior art keywords
- information
- rumor
- agricultural product
- intervention
- rumors
- 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
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 20
- 230000007246 mechanism Effects 0.000 claims abstract description 14
- 238000005516 engineering process Methods 0.000 claims abstract description 13
- 238000004458 analytical method Methods 0.000 claims abstract description 10
- 238000003058 natural language processing Methods 0.000 claims abstract description 10
- 238000012216 screening Methods 0.000 claims abstract description 8
- 238000010801 machine learning Methods 0.000 claims abstract description 5
- 239000013598 vector Substances 0.000 claims description 18
- 238000001514 detection method Methods 0.000 claims description 17
- 230000000694 effects Effects 0.000 claims description 12
- 230000008451 emotion Effects 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 12
- 230000007935 neutral effect Effects 0.000 claims description 10
- 238000007477 logistic regression Methods 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 6
- 238000012544 monitoring process Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000007476 Maximum Likelihood Methods 0.000 claims description 3
- 230000009193 crawling Effects 0.000 claims description 3
- 238000004393 prognosis Methods 0.000 claims description 3
- 235000015277 pork Nutrition 0.000 description 8
- 238000000354 decomposition reaction Methods 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 5
- 238000012545 processing Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 206010037660 Pyrexia Diseases 0.000 description 1
- 241000282898 Sus scrofa Species 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000012271 agricultural production Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000011109 contamination Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000009849 deactivation Effects 0.000 description 1
- 230000006397 emotional response Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000002407 reforming Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to an intervention method for a rumor of an agricultural product quality safety network, which can effectively detect and intervene the rumor of the agricultural product quality safety in the network. The method mainly comprises five steps. Firstly, acquiring information about agricultural product quality safety in social media, news media and other public information sources through a big data technology; then screening and classifying the acquired information by using a natural language processing technology; then, carrying out deep analysis on the resources classified as negative information by adopting a machine learning algorithm to determine whether the resources are rumors or not; then, an intervention mechanism is started aiming at the detected rumors, real and accurate agricultural product quality information is pushed to affected consumers, and rumor information is reported to related departments; by the method, the accuracy of the quality safety information of the agricultural products can be improved, and the fairness and normal operation of the agricultural product market are ensured.
Description
Technical Field
The invention belongs to the field of big data, and particularly relates to an intervention method of a network rumor for agricultural product quality safety.
Background
In recent years, with the rapid development of the internet and information communication technology, networks have become an important platform for people to acquire and spread information. The information on networks is very diverse and includes a lot of information about the quality safety of agricultural products. However, there is a lot of false information, i.e. rumors, interspersed with, which has a great negative impact on scientific knowledge of the quality of the agricultural product and on the correct decisions. Particularly, counterfeit and misleading rumors related to the quality safety of agricultural products often cause panic of consumers, prevent the normal circulation and transaction of the agricultural products, and further have profound effects on agricultural production and the agricultural product market.
Therefore, how to effectively identify and intervene in these rumors is an urgent issue. Traditional rumor intervention mainly relies on manual approaches, such as setting a special reporting platform, and collecting and processing various reporting information. However, this method is time-consuming and labor-consuming, and is difficult to cope with a network environment with a large amount of information and a high update speed.
At present, the network rumor intervention research in China is concentrated in a qualitative stage, takes case analysis as a main method, and lacks a scientific, systematic and quantitative intervention method.
Disclosure of Invention
The invention mainly solves the technical problems of effectively collecting the information of the agricultural product quality safety rumors, accurately distinguishing the content of the rumors, effectively turning over the rumors by using a specific intervention mechanism, reforming the brand image of the front agricultural product and resisting the adverse effects of the rumors. In addition, the invention aims at evaluating the intervention effect and providing feedback so as to purposefully adjust and optimize the intervention strategy and provide long-term and continuous guarantee for the quality safety of agricultural products. The series of solutions can resist the propagation of the agricultural product quality safety rumors, help to improve the confidence of the public on the agricultural product quality, stabilize the agricultural product market and promote the sustainable and healthy development of agriculture.
In order to achieve the above purpose, the present invention is realized by adopting the following technical scheme: the intervention method comprises the following steps:
step one, data acquisition, namely performing data crawling on social media, news media and other public real life information through a big data technology to acquire all information resources about agricultural product quality safety;
screening and classifying the acquired information by using a natural language processing technology, and determining whether the information is positive information, negative information or neutral information;
and thirdly, rumor detection, namely performing deep analysis on the resources classified as negative information by adopting a machine learning algorithm to determine whether false information exists, namely rumors.
Step four: an intervention mechanism that is activated upon detection of a rumor; real and accurate agricultural product quality information is quickly pushed to affected consumers, misunderstanding of the agricultural product quality is corrected, and rumor information is reported to related departments;
step five: effect feedback and continuous monitoring, capturing and analyzing information flow change of dry prognosis, evaluating rumor intervention effect and optimizing; continuous data acquisition and rumor detection are performed, so that intervention can be performed in time when a new rumor is found.
Further, the step one data acquisition comprises
S101, setting a crawler program: the specific targets are various large social media, network news platforms and public resources containing agricultural product information, such as forums, blogs and question-answering websites;
s102, keyword setting: setting keywords related to the quality and safety of agricultural products, such as 'quality of agricultural products', 'safety of agricultural products', 'pollution of agricultural products', so as to collect related information in a directed manner;
s103, information acquisition: and information acquisition is carried out on a set website through a crawler program, wherein the information acquisition comprises an original text, a text time and a text place so as to maximally retain the original information.
Further, the step two information screening includes:
s201, text preprocessing: performing text preprocessing on the crawled original information, wherein the text preprocessing comprises word segmentation, stop word removal, punctuation mark removal and irrelevant character removal;
s202 conversion format: converting the preprocessed data into a format suitable for modeling; converting the segmented sentences into vector forms by using a Word2Vec method;
s203, establishing a model: in the step, the emotion analysis model in NLP is used for identifying the semantics and emotion tendencies;
s204, classification information: and (3) classifying the information related to the agricultural products into three types of positive information, negative information and neutral information through the constructed model.
Further, the step of detecting the rumor includes:
s301, extracting features: extracting features of the resources classified as negative information; the characteristics comprise text length, number of negative words in the text, number of emotion color words in the text and reputation degree of an author; these features are denoted as feature vectors X;
s302, logistic regression model setting: the logistic regression model was set as:
P(Y＝1|X)＝1/(1+exp(-(ω·X+b)))
where P (y= 1|X) represents the probability that the text is a rumor, y=1 represents a rumor, y=0 represents a non-rumor, ω is a weight coefficient, b is a constant term, ω·x is the result of multiplying the feature vector and the weight;
s303, training a model: optimizing parameters w and b by using algorithms such as maximum likelihood estimation, gradient descent and the like through the marked data set, including the texts of rumors and non-rumors and the characteristics thereof;
s304 rumor detection: extracting new negative information, extracting characteristic vector X, putting into a trained model, calculating probability P (Y= 1|X) of rumor, and if the probability is greater than a set threshold (such as 0.5), predicting that the text is rumor; otherwise, the non-rumors are used.
Further, the fourth intervention mechanism is as follows: once the rumor detection step 3 determines that rumors are present, an intervention mechanism is initiated;
s401, matching rumor propagation modes by using a batch mode matching algorithm;
s402 using the candidate identification to find users in the matched node paths that may be affected by rumors;
after determining a rumor message, S403 pushes the rumor-forming message to the corresponding candidate user, so as to reduce the rumor propagation.
Further, the threshold is 0.5.
The invention has the beneficial effects that:
through big data technology and natural language processing technology, can collect and filter the relevant information of agricultural product quality safety effectively, in time discover and discern the rumor, improved detection efficiency and accuracy to the rumor. The rumor intervention mechanism can timely correct misunderstanding of consumers on the quality of agricultural products, counteract bad influences of rumors, improve the confidence of consumers on the quality of the agricultural products, and is beneficial to stabilizing the agricultural product market.
The mechanism of effective feedback and continuous monitoring ensures that intervention can be timely performed when new agricultural product quality safety rumors are found, can resist continuous transmission of rumors, and slows down the influence of rumors on market and consumer mind states.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a flow chart of data acquisition according to the present invention;
FIG. 3 is a flow chart of the information screening of the present invention;
FIG. 4 is a flow chart of rumor detection according to the present invention;
fig. 5 is a flow chart of rumor intervention in accordance with the present invention.
Detailed Description
In order that the invention may be readily understood, a more complete description of the invention will be rendered by reference to the appended drawings. Exemplary embodiments of the present invention are illustrated in the accompanying drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
As shown in fig. 1, the intervention method of the agricultural product quality safety network rumors comprises the following steps:
step one, data acquisition, namely performing data crawling on social media, news media and other public real life information through a big data technology to acquire all information resources about agricultural product quality safety.
As shown in FIG. 2, the step one data acquisition includes
S101, setting a crawler program: specific crawler programs are written, with the specific goal of large social media, web news platforms, and other published resources that may contain agricultural product information, such as forums, blogs, question-answering websites, etc.
S102, keyword setting: keywords related to the quality and safety of agricultural products, such as 'quality of agricultural products', 'safety of agricultural products', 'pollution of agricultural products', etc., are set so as to collect related information in a directed manner.
S103, information acquisition: and collecting information at the set website through a crawler program. Including original text, time of posting, place of posting, etc., to maximize retention of the original information.
Screening and classifying the acquired information by using a natural language processing technology, and determining whether the information is positive information, negative information or neutral information;
as shown in fig. 3, the step two information filtering includes:
s201, text preprocessing: text preprocessing is carried out on the crawled original information, and the text preprocessing comprises word segmentation, word deactivation (such as' and other words with insubstantial meaning), punctuation marks and removal of irrelevant characters.
S202 conversion format: converting the preprocessed data into a format suitable for modeling; the Word2Vec method is used to convert the segmented sentences into vector form.
For the preprocessed text data, the Word2Vec method specifically comprises the following steps:
first, a sliding window (e.g., a window of size 5) is used to slide over the text, each sliding resulting in a center word and context words within the window (words within the window other than the center word).
For the Skip-gram model, the input is a center word and the output is a context word.
The Skip-gram model is:
L(θ)＝∑ x ∈corpus∑ y ∈context(x)log p(y|x)(1)
where θ represents a parameter of the model, i.e. a vector in the matrix, and p represents a conditional probability.
Two matrices are used: an input matrix and an output matrix, each word having a corresponding vector representation in both matrices. During the training process, the vector parameters in both matrices are updated according to the current task (e.g., predictive center word or predictive context word).
After training, word2Vec vectors of each Word are corresponding vectors in the input matrix or the output matrix.
S203, establishing a model: this step uses the emotion analysis model in NLP to emphasize recognition of semantics and emotion tendencies.
With a naive bayes classifier, under this assumption, the probability of emotion classification of the model predictive text can be calculated as:
P(C|F1,...,Fn)＝P(C)*P(F1|C)...P(Fn|C)/P(F1,...,Fn)(2)
where C is a particular classification, either positive, negative or neutral, F1 to Fn are features of the input text (commonly referred to as words), and P (c|f1, term, fn) represents the probability that the text is classified as C given these features; p (C) represents the probability that any text is classified as C; p (fi|c) represents the probability of containing the feature Fi in the case where the text is classified as C.
S204, classification information: and (3) classifying the information related to the agricultural products into three types of positive information, negative information and neutral information through the constructed model.
After the model is built, the text after preprocessing and format conversion is input into the model, the probability of positive, negative and neutral classification of the text is calculated, and finally the text is classified into the class with the highest probability.
For example, for the text "A", the model computes
P (Positive i a) =0.4, P (Negative i a) =0.3, and P (Neutral i a) =0.3, since the probability of the front is the largest, the text "a" is determined as the front information.
And thirdly, rumor detection, namely performing deep analysis on the resources classified as negative information by adopting a machine learning algorithm to determine whether false information exists, namely rumors.
As shown in fig. 4, the step of the third rumor detection includes:
s301, extracting features: extracting features of the resources classified as negative information; the characteristics comprise text length, number of negative words in the text, number of emotion color words in the text and reputation degree of an author; these features are denoted as feature vectors X;
s302, logistic regression model setting: the logistic regression model was set as:
P(Y＝1|X)＝1/(1+exp(-(ω·X+b))) (3)
where P (y= 1|X) represents the probability that the text is a rumor, y=1 represents a rumor, y=0 represents a non-rumor, ω is a weight coefficient, b is a constant term, ω·x is the result of multiplying the feature vector and the weight;
s303, training a model: optimizing parameters w and b by using algorithms such as maximum likelihood estimation, gradient descent and the like through the marked data set, including the texts of rumors and non-rumors and the characteristics thereof;
s304 rumor detection: extracting new negative information, extracting characteristic vector X, putting into a trained model, calculating probability P (Y= 1|X) of rumor, and if the probability is greater than a set threshold (such as 0.5), predicting that the text is rumor; otherwise, the non-rumors are used.
Step four: an intervention mechanism that is activated upon detection of a rumor; real and accurate agricultural product quality information is quickly pushed to affected consumers, misunderstanding of the agricultural product quality is corrected, and rumor information is reported to related departments.
As shown in fig. 5, S401 matches rumor propagation patterns using a batch pattern matching algorithm.
The algorithm creates an index for the nodes in the regular tree, then traverses the regular tree and expands the partial decomposition step by step based on the index, and finally obtains all the matching node paths of the pattern.
Formalized definition of batch pattern matching
Given a graph G and a regular tree RT that may represent rumor propagation patterns PC, a pattern PC is matched on the graph G to a node path that matches at least one coincidence P C Is provided. The algorithm firstly creates an index for each node in the regular tree, candidate matching of the current node label stored in the index in the large graph is carried out, then hierarchical traversal is adopted on the regular tree to gradually expand partial decomposition, finally all matching node paths of the whole mode are obtained, and the set of the matching node paths is P N And (3) representing. The matching algorithm does not need to traverse the whole large graph, but reduces the traversal on the regular tree, and the size of the matching algorithm is about one thousandth of the large graph, so that the algorithm efficiency is improved. The formal definition is as follows:
question input: an information dissemination network G and a regular tree RT,
problem output: node path set P N 。
The algorithm will then traverse the regular tree through the hierarchy with the help of the index to get each node n i Is a partial solution of S (n) i )，S(n i ) Is formed by all matching canonical paths L (n 0 )…L(n i ) Is defined by the node paths. When the calculation of the leaf node layer is completed, the paths stored in the partial solutions in the leaf nodes are the matching pattern P C Is a node path of (a). Matching pattern P C Is P N 。
Partial decomposition S (n) i ) The definition is as follows: the matching algorithm will traverse the regular tree RT, from top to bottom, to progressively expand the partial decomposition,
the method comprises the following specific steps:
first for each node n in RT i Initialization of partial decomposition
Then traverse RT from top to bottom, access each node ni and call matlab subroutine SubMatch (n) i ,n j ) The method comprises the steps of carrying out a first treatment on the surface of the The program uses I (n) j ) To expand the partial decomposition S (n) i ) And gets S (n) j ) Wherein n is j Is n i Until the regular tree RT access is complete.
S402 using the candidate identification to find users in the matched node paths that may be affected by rumors;
node path set P in obtaining matching mode N The nodes in the node path then need to be identified to determine if the user they represent is likely to be a potential propagator more susceptible to rumors.
First, for P N In (a) each node path np=v 0 …v i …v n (0.ltoreq.i.ltoreq.n). The identification rule of each node is as follows. (1) Taking into account v 0 The node selected for the rumor propagation mode (i.e., the first node in the path) should have some similarity to the positive node, which indicates that the user has forwarded or reviewed the rumor microblog, and therefore is represented by v 0 The user of the presentation is considered likely to forward and comment on such rumor microblogs in the past or in the future. (2) Considering that rumor path np is a node path starting with a positive node and not covered by any negative node, it should be unique to rumor propagation.
Thus, other nodes on the path should also have some correlation with rumor propagation. To enhance such correlation, it is assumed that np is represented by sub-pattern p=a 1 …a j …a m (1.ltoreq.j.ltoreq.m) and the algorithm considers for some node vi other than the first node. If each a in pattern p j (1. Ltoreq.j.ltoreq.m), all with edge e= (v, v) i ) Wherein V is V, let a j E L (e) then considers vi to be likely to be affected by the rumor and should be a potential transmitter of the rumor.
Finally, the algorithm obtains a set C of information propagation pairs according to the record of forwarding and commenting (namely the interest of the user P ) Candidate users are identified, i.e. users who may forward or comment on rumor microblogs in the past or later, and the candidate node set C is used u Representing these nodes.
After determining a rumor message, S403 pushes the rumor-forming message to the corresponding candidate user, so as to reduce the rumor propagation.
Set C according to information propagation pairs P The information category to which the candidate node is attached pushes the ballad information of the corresponding category to the user represented by the candidate node set node. For example, set C of certain information propagation pairs P The method is obtained by processing the agricultural product quality information, and then once the rumors with the agricultural product quality are officially identified, the algorithm pushes corresponding rumor-forming information to candidate users, and timely blocks the rumor propagation, so that negative effects are reduced.
Step five: effect feedback and continuous monitoring, capturing and analyzing information flow change of dry prognosis, evaluating rumor intervention effect and optimizing; continuous data acquisition and rumor detection are performed, so that intervention can be performed in time when a new rumor is found.
Real-time monitoring includes, but is not limited to, changes of public information streams such as social media, a network news platform and the like, and records emotional response, comments, consumption behavior changes and the like of consumers after the quality of agricultural products is influenced by rumors.
And observing and measuring the change condition of the information flow, and carrying out data analysis. Such as calculating the rate of change of negative comments after intervention, the change in consumption behavior, etc. From these data, the effect of rumor intervention was assessed quantitatively.
Based on the evaluation results, the advantages and disadvantages of the intervention are analyzed and based on these feedback, the intervention strategy is adjusted and optimized. Such as enhancing the pushing of information to certain specific groups, adjusting the information content to improve its convincing, etc.
Continuous data acquisition and rumor detection. The technology of big data and natural language processing is continuously used, public information such as social media, news media and the like is concerned at any time, and once a new rumors related to quality and safety of agricultural products are found, an intervention mechanism is started in time to intervene, so that the rumors are prevented from being continuously spread and influence is enlarged.
Example 1
Firstly, a crawler program is set, and a target website mainly comprises main social media and news media such as microblogs, weChat public numbers, headline news and the like, and question-answering websites such as agricultural forums and knowledge containing agricultural product information. The keywords are set as keywords related to pork quality safety, such as pork pollution, swine fever and the like.
After the program was set, we began to crawl the website for information, gathering about 2000 pieces of raw information about pork contamination. The original information includes original text, time of posting, place of posting, etc.
After the original information is obtained, text preprocessing is carried out by using a natural language processing technology, wherein the text preprocessing comprises word segmentation, stop words, punctuation marks and irrelevant characters in the text are removed. We then convert the preprocessed data into a vector format suitable for modeling, and then use emotion analysis models to classify the information into three categories, positive, negative and neutral. The processing results in about 1500 pieces of negative information.
For these negative information we resort to machine learning methods for deep analysis. Through extracting characteristics such as text length, number of negative words, number of emotion color words, reputation of a publisher and the like, training is carried out by using a logistic regression model, and finally about 200 pieces of information are found to be rumors with pork quality safety.
After discovering the rumors, we start an intervention mechanism, and set up corresponding customs strategies, including publishing real and scientific pork quality information, conducting science popularization on incorrect ideas, and reporting the rumors.
After intervention, we evaluate the effect of rumor intervention by continuously monitoring and analyzing network information, and make adjustments and optimizations.
For example, we found that a social platform has higher negative information receiving degree of pork quality, so we increase information pushing force on the platform and further optimize pushing content to make it more in line with receiving habit of consumers.
Through the steps, the transmission of the rumors with the safe pork quality is successfully controlled, misunderstanding of the pork quality by consumers is corrected, and the confidence of the consumers is enhanced.
Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program stored on a computer readable storage medium, which when executed may comprise the steps of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a read-only memory (ReadOnlyMemory, ROM) or a random access memory (RandomABBessMemory, RAM).
It should be understood that the detailed description of the technical solution of the present invention, given by way of preferred embodiments, is illustrative and not restrictive. Modifications of the technical solutions described in the embodiments or equivalent substitutions of some technical features thereof may be performed by those skilled in the art on the basis of the present description; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (6)
1. An intervention method of a network rumor for agricultural product quality safety is characterized by comprising the following steps: the intervention method comprises the following steps:
data acquisition, namely performing data crawling on social media, news media and other public real life information through a big data technology to acquire all information resources about agricultural product quality safety;
information screening, namely screening and classifying the acquired information by using a natural language processing technology, and determining whether the information is positive information, negative information or neutral information;
rumors are detected, and the resources classified as negative information are subjected to deep analysis by adopting a machine learning algorithm to determine whether false information, namely rumors, exists.
An intervention mechanism, detecting rumors, and starting the intervention mechanism; real and accurate agricultural product quality information is quickly pushed to affected consumers, misunderstanding of the agricultural product quality is corrected, and rumor information is reported to related departments;
effect feedback and continuous monitoring, capturing and analyzing information flow change of dry prognosis, evaluating rumor intervention effect and optimizing; continuous data acquisition and rumor detection are carried out, and intervention is carried out timely when new rumors are found.
2. The method for intervention in a quality of agricultural product safety network rumor as claimed in claim 1, wherein: the data acquisition comprises
S101, setting a crawler program: the specific targets are various large social media, network news platforms and public resources containing agricultural product information, such as forums, blogs and question-answering websites;
s102, keyword setting: setting keywords related to the quality and safety of agricultural products, such as 'quality of agricultural products', 'safety of agricultural products', 'pollution of agricultural products', so as to collect related information in a directed manner;
s103, information acquisition: and information acquisition is carried out on a set website through a crawler program, wherein the information acquisition comprises an original text, a text time and a text place so as to maximally retain the original information.
3. The method for intervention in a quality of agricultural product safety network rumor as claimed in claim 1, wherein: the information screening comprises the following steps:
s201, text preprocessing: performing text preprocessing on the crawled original information, wherein the text preprocessing comprises word segmentation, stop word removal, punctuation mark removal and irrelevant character removal;
s202 conversion format: converting the preprocessed data into a format suitable for modeling; converting the segmented sentences into vector forms by using a Word2Vec method;
s203, establishing a model: in the step, the emotion analysis model in NLP is used for identifying the semantics and emotion tendencies;
s204, classification information: and (3) classifying the information related to the agricultural products into three types of positive information, negative information and neutral information through the constructed model.
4. The method for intervention in a quality of agricultural product safety network rumor as claimed in claim 1, wherein: the rumor detection includes:
s301, extracting features: extracting features of the resources classified as negative information; the characteristics comprise text length, number of negative words in the text, number of emotion color words in the text and reputation degree of an author; these features are denoted as feature vectors X;
s302, logistic regression model setting: the logistic regression model was set as:
P(Y＝1|X)＝1/(1+exp(-(ω·X+b)))
where P (y= 1|X) represents the probability that the text is a rumor, y=1 represents a rumor, y=0 represents a non-rumor, ω is a weight coefficient, b is a constant term, ω·x is the result of multiplying the feature vector and the weight;
s303, training a model: optimizing parameters w and b by using algorithms such as maximum likelihood estimation, gradient descent and the like through the marked data set, including the texts of rumors and non-rumors and the characteristics thereof;
s304 rumor detection: extracting the new negative information, extracting the characteristic vector X, putting the negative information into a trained model, and predicting that the text is rumor by calculating the probability P (Y= 1|X) that the text is rumor if the probability is larger than a set threshold value; otherwise, the non-rumors are used.
5. The method for intervention in a quality of agricultural product safety network rumor as claimed in claim 1, wherein: the intervention mechanism is as follows: once the rumor detection step 3 determines that rumors are present, an intervention mechanism is initiated;
s401, matching rumor propagation modes by using a batch mode matching algorithm;
s402 using the candidate identification to find users in the matched node paths that may be affected by rumors; after determining a rumor message, S403 pushes the rumor-forming message to the corresponding candidate user, so as to reduce the rumor propagation.
6. The method for intervention in a quality of agricultural product safety network rumors as claimed in claim 4, wherein: the threshold is 0.5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311807450.7A CN117743581B (en) | 2023-12-26 | 2023-12-26 | Intervention method for agricultural product quality safety network rumors |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311807450.7A CN117743581B (en) | 2023-12-26 | 2023-12-26 | Intervention method for agricultural product quality safety network rumors |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117743581A true CN117743581A (en) | 2024-03-22 |
| CN117743581B CN117743581B (en) | 2024-06-11 |
Family
ID=90260962
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311807450.7A Active CN117743581B (en) | 2023-12-26 | 2023-12-26 | Intervention method for agricultural product quality safety network rumors |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117743581B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010024184A1 (en) * | 2008-08-26 | 2010-03-04 | 日本電気株式会社 | Rumor information detecting system, rumor information detecting method, and program |
| KR101561464B1 (en) * | 2014-08-25 | 2015-10-20 | 성균관대학교산학협력단 | Collected data sentiment analysis method and apparatus |
| CN110795641A (en) * | 2019-11-05 | 2020-02-14 | 重庆邮电大学 | Network rumor propagation control method based on representation learning |
| CN112231562A (en) * | 2020-10-15 | 2021-01-15 | 北京工商大学 | Network rumor identification method and system |
| CN112560495A (en) * | 2020-12-09 | 2021-03-26 | 新疆师范大学 | Microblog rumor detection method based on emotion analysis |
| CN114298853A (en) * | 2021-12-14 | 2022-04-08 | 天津大学 | Graph mode-based personalized rumor-avoiding method |
| DE202022102766U1 (en) * | 2022-05-19 | 2022-05-31 | Mahendra Bhamu | System to detect and avoid misleading messages and false rumors on social media |
| CN114817541A (en) * | 2022-04-24 | 2022-07-29 | 中国人民武装警察部队工程大学 | Rumor detection method and device based on dual-emotion perception |
| CN116306587A (en) * | 2023-03-27 | 2023-06-23 | 中国人民武装警察部队工程大学 | Internet negative public opinion early warning method |
| CN116680358A (en) * | 2023-05-06 | 2023-09-01 | 浙江华巽科技有限公司 | Rumor detection method and device based on bidirectional graph attention network |
| WO2023173537A1 (en) * | 2022-03-17 | 2023-09-21 | 平安科技（深圳）有限公司 | Text sentiment analysis method and apparatus, device and storage medium |
-
2023
- 2023-12-26 CN CN202311807450.7A patent/CN117743581B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010024184A1 (en) * | 2008-08-26 | 2010-03-04 | 日本電気株式会社 | Rumor information detecting system, rumor information detecting method, and program |
| KR101561464B1 (en) * | 2014-08-25 | 2015-10-20 | 성균관대학교산학협력단 | Collected data sentiment analysis method and apparatus |
| CN110795641A (en) * | 2019-11-05 | 2020-02-14 | 重庆邮电大学 | Network rumor propagation control method based on representation learning |
| CN112231562A (en) * | 2020-10-15 | 2021-01-15 | 北京工商大学 | Network rumor identification method and system |
| CN112560495A (en) * | 2020-12-09 | 2021-03-26 | 新疆师范大学 | Microblog rumor detection method based on emotion analysis |
| CN114298853A (en) * | 2021-12-14 | 2022-04-08 | 天津大学 | Graph mode-based personalized rumor-avoiding method |
| WO2023173537A1 (en) * | 2022-03-17 | 2023-09-21 | 平安科技（深圳）有限公司 | Text sentiment analysis method and apparatus, device and storage medium |
| CN114817541A (en) * | 2022-04-24 | 2022-07-29 | 中国人民武装警察部队工程大学 | Rumor detection method and device based on dual-emotion perception |
| DE202022102766U1 (en) * | 2022-05-19 | 2022-05-31 | Mahendra Bhamu | System to detect and avoid misleading messages and false rumors on social media |
| CN116306587A (en) * | 2023-03-27 | 2023-06-23 | 中国人民武装警察部队工程大学 | Internet negative public opinion early warning method |
| CN116680358A (en) * | 2023-05-06 | 2023-09-01 | 浙江华巽科技有限公司 | Rumor detection method and device based on bidirectional graph attention network |
Non-Patent Citations (1)
| Title |
|---|
| 邓君;孙绍丹;王阮;宋先智;李贺;: "基于Word2Vec和SVM的微博舆情情感演化分析", 情报理论与实践, no. 08, 13 May 2020 (2020-05-13), pages 112 - 119 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117743581B (en) | 2024-06-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110163647B (en) | Data processing method and device | |
| CN112199608B (en) | Social media rumor detection method based on network information propagation graph modeling | |
| CN111046171B (en) | Emotion discrimination method based on fine-grained labeled data | |
| CN109871434B (en) | Public opinion evolution tracking method based on dynamic incremental probability graph model | |
| CN113343126B (en) | Rumor detection method based on event and propagation structure | |
| CN113434684B (en) | Rumor detection method, system, equipment and storage medium for self-supervision learning | |
| KR102135074B1 (en) | System for identifying fake news using artificial-intelligence-based fact-checking guidelines | |
| CN113378565A (en) | Event analysis method, device and equipment for multi-source data fusion and storage medium | |
| Bello et al. | Reverse engineering the behaviour of twitter bots | |
| CN116244446A (en) | Social media cognitive threat detection method and system | |
| CN111898038A (en) | Social media false news detection method based on man-machine cooperation | |
| CN111428151A (en) | False message identification method and device based on network acceleration | |
| Kumar et al. | Fake news detection of Indian and United States election data using machine learning algorithm | |
| Chua et al. | Problem Understanding of Fake News Detection from a Data Mining Perspective | |
| CN116522013B (en) | Public opinion analysis method and system based on social network platform | |
| CN117216687A (en) | Large language model generation text detection method based on ensemble learning | |
| CN109254993B (en) | Text-based character data analysis method and system | |
| CN112527866A (en) | Stock trend prediction method and system based on text abstract emotion mining | |
| Chai et al. | Automatically measuring the quality of user generated content in forums | |
| CN117743581B (en) | Intervention method for agricultural product quality safety network rumors | |
| CN109408808B (en) | Evaluation method and evaluation system for literature works | |
| CN110941713B (en) | Self-optimizing financial information block classification method based on topic model | |
| Triyono et al. | Fake News Detection in Indonesian Popular News Portal Using Machine Learning For Visual Impairment | |
| CN114662486A (en) | Emergency sensitive word detection method based on machine learning | |
| Kyaw et al. | Traditional and swarm intelligent based text feature selection for document classification |
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 |