CN115391674B - Method, device, equipment and storage medium for efficiently suppressing false information of network community - Google Patents
Method, device, equipment and storage medium for efficiently suppressing false information of network community Download PDFInfo
- Publication number
- CN115391674B CN115391674B CN202211049109.5A CN202211049109A CN115391674B CN 115391674 B CN115391674 B CN 115391674B CN 202211049109 A CN202211049109 A CN 202211049109A CN 115391674 B CN115391674 B CN 115391674B
- Authority
- CN
- China
- Prior art keywords
- false information
- user node
- value
- false
- user
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 230000000750 progressive effect Effects 0.000 claims abstract description 21
- 238000012544 monitoring process Methods 0.000 claims abstract description 18
- 230000005540 biological transmission Effects 0.000 claims abstract description 8
- 230000001629 suppression Effects 0.000 claims description 41
- 230000011218 segmentation Effects 0.000 claims description 13
- 239000013598 vector Substances 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 abstract description 7
- 230000000694 effects Effects 0.000 abstract description 7
- 230000008569 process Effects 0.000 abstract description 6
- 230000001902 propagating effect Effects 0.000 abstract description 5
- 230000010365 information processing Effects 0.000 abstract description 3
- 230000002829 reductive effect Effects 0.000 abstract description 2
- 238000012216 screening Methods 0.000 abstract description 2
- 230000005764 inhibitory process Effects 0.000 abstract 3
- 238000011156 evaluation Methods 0.000 description 5
- 238000002372 labelling Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000012217 deletion Methods 0.000 description 3
- 230000037430 deletion Effects 0.000 description 3
- 230000002265 prevention Effects 0.000 description 3
- 238000013145 classification model Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000036961 partial effect Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000007789 sealing Methods 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Computing Systems (AREA)
- General Business, Economics & Management (AREA)
- Bioethics (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Software Systems (AREA)
- Information Transfer Between Computers (AREA)
Abstract
The invention belongs to the technical field of text information processing, and discloses a method for efficiently suppressing false information of a network community, which comprises the steps of screening false information with larger influence values, determining the grade according to the participation degree of each user node to a classification subject to which the user node belongs, prompting the user node with larger participation degree, and periodically monitoring whether each user node participates in propagating the false information; if the user node participates in the transmission, judging whether the level of the user node is the final candidate level, if not, adjusting the level, and controlling the user node by adopting a control mode corresponding to the adjusted level; if yes, deleting the user node; therefore, the false information with influence can be screened out for inhibition, the efficiency is improved, and meanwhile, the control mode that the inhibition force is sequentially increased is adopted for the user nodes through sequential progressive level adjustment, so that the misjudgment rate of abnormal user nodes in the inhibition process can be reduced, and the user activity of the network community platform is maintained.
Description
Technical Field
The invention belongs to the technical field of text information processing, and particularly relates to a method, a device, equipment and a storage medium for efficiently suppressing false information of a network community.
Background
With the rapid development of the internet, social networks using network communities, online social platforms and the like as carriers are rapidly developed, and a large number of people begin to use the social networks to exchange information, share experience and the like. The social network becomes an important channel for people to obtain mass information; wherein the web community is one of the typical representatives. While providing convenience to people, web communities also generate a large amount of false information (such as rumors, lies, etc.). Serious false information even results in loss of public property and life. However, the network community false information has the characteristics of large propagation strength, wide propagation range, various false information and the like, and provides a new challenge for inhibiting the network community false information. Therefore, how to effectively inhibit the propagation of false information in the network community has great significance.
The traditional method for effectively suppressing the false information of the network community is to process related account numbers and related false contents when the false information is found. Such as "seal", "banning", "cut" etc. The method for deleting the user or the related false content by adopting a simple and rough way of 'one-time cutting' has the defects of simple processing way, poor effect and the like. Moreover, often in this way, false information is easily released or believed but the user is unaware; the method is easy to cause secondary rebound of the false information, and partial public believes that the back has an inner screen, and rather, the confidence degree of the public on the false information is deepened.
In the prior art, some access control-based false information suppression methods are provided, and a specific user is screened out through an algorithm for authorization; the authorized user may access the associated content. Although the propagation of false information can be well prevented and controlled, the method is only suitable for a specific range, a theme and a group, or a special small public forum platform and the like; the method is not suitable for Internet network communities or forum platforms with the characteristics of wide themes, various users, various contents and the like; the application of the method in a typical internet network community has the defects of high difficulty and high cost, and the reason is that proper users are difficult to screen out to carry out access control on related contents.
In addition, some false information suppression methods based on propagation node control are also proposed in the prior art, namely, abnormal user nodes which are propagating or will propagate false information in an internet network community are found through a network information propagation model or a complex network and other theories, and operations such as deleting or sealing numbers are carried out on the abnormal user nodes. Although the method adopts a search correlation algorithm to screen out the specific user for control, the classification of the user is rough, and the condition that the node of the abnormal user is misjudged easily occurs, so that the user loss is caused. Moreover, there is also a risk of injury to the unknown users, which may result in a loss to the network community or forum users. Therefore, in the existing false information suppression process based on propagation node control, the misjudgment rate of abnormal user nodes is high, and the user activity of the network community platform is reduced.
Disclosure of Invention
The invention aims to provide a method, a device, equipment and a storage medium for efficiently suppressing false information of a network community, which can reduce the misjudgment rate of abnormal user nodes in the suppression process of the false information, thereby avoiding the reduction of the user activity of a network community platform.
The invention discloses a method for efficiently suppressing false information of a network community, which comprises the following steps:
determining current false information that appears on the network community;
when the influence value of the current false information reaches a preset influence threshold value, determining the current false information as target false information;
calculating the participation degree of each user node of the network community on a target classification theme to which the target false information belongs;
determining the current level of each user node from at least four candidate levels according to the participation degree; the method comprises the following steps that at least four candidate levels are sequentially advanced, control modes corresponding to the candidate levels are different, and the suppression strength of the control modes is increased along with the advancing direction of the candidate levels;
carrying out false information prompt on the user node with the participation degree larger than a preset participation threshold value;
periodically monitoring whether each user node participates in the propagation of the target false information;
if any user node is monitored to participate in the transmission of the target false information, whether the current level of the user node is the last candidate level in the progressive direction or not is judged;
if the current level of the user node is not the last candidate level in the progressive direction, adjusting the current level of the user node according to the progressive direction of the candidate level, and controlling the user node by adopting a control mode corresponding to the adjusted current level;
and if the current level of the user node is the last candidate level in the progressive direction, deleting the user node.
The second aspect of the present invention discloses an efficient suppression device for false information in a network community, comprising:
a first determining unit, configured to determine current false information occurring on the network community;
a second determining unit, configured to determine the current false information as target false information when an influence value of the current false information reaches a preset influence threshold;
the participation calculation unit is used for calculating the participation degree of each user node of the network community on the target classification theme to which the target false information belongs;
the classification unit is used for determining the current level of each user node from at least four candidate levels according to the participation degree; the method comprises the following steps that at least four candidate levels are sequentially advanced, control modes corresponding to the candidate levels are different, and the suppression strength of the control modes is increased along with the advancing direction of the candidate levels;
the prompting unit is used for carrying out false information prompting on the user nodes with the participation degrees larger than a preset participation threshold;
the monitoring unit is used for periodically monitoring whether each user node participates in the transmission of the target false information;
the judging unit is used for judging whether the current level of any user node is the last candidate level in the progressive direction when the monitoring unit monitors that the user node participates in the transmission of the target false information;
a control unit, configured to, when the determination unit determines that the current level of the user node is not the last candidate level in the forwarding direction, adjust the current level of the user node according to the forwarding direction of the candidate level, and control the user node in a control manner corresponding to the adjusted current level;
and the deleting unit is used for deleting the user node when the judging unit judges that the current level of the user node is the last candidate level in the progressive direction.
A third aspect of the invention discloses an electronic device comprising a memory storing executable program code and a processor coupled to the memory; the processor calls the executable program code stored in the memory for executing the network community false information efficient suppression method disclosed in the first aspect.
In a fourth aspect of the present invention, a computer-readable storage medium is disclosed, and the computer-readable storage medium stores a computer program, where the computer program makes a computer execute the method for efficiently suppressing false information in a network community disclosed in the first aspect.
The method, the device and the storage medium have the advantages that at least four candidate levels which are sequentially progressive are preset, the control modes corresponding to the candidate levels are different, the suppression strength of the control modes is increased along with the progressive direction of the candidate levels, then target false information with the influence value reaching a preset influence threshold value is screened, the participation degree of each user node of the network community on a target classification theme to which the target false information belongs is calculated, the current level of each user node is determined from the at least four candidate levels according to the participation degree, meanwhile, false information prompt is carried out on the user nodes with the participation degree larger than the preset participation threshold value, and then whether the user nodes participate in transmitting the target false information or not is periodically monitored; if any user node participates in the propagation, judging whether the current level of the user node is the last candidate level in the progressive direction or not, if not, adjusting the current level of the user node according to the progressive direction, and controlling the user node by adopting a control mode corresponding to the adjusted current level; if the candidate level is the last level, deleting the user node;
therefore, the method can screen the influential false information for suppression, thereby saving the cost for preventing, controlling and suppressing the false information, improving the suppression efficiency, realizing high-efficiency suppression, and simultaneously carrying out fine-grained classification of at least four levels on the user nodes.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate an embodiment of the invention and, together with the description, serve to explain the principles and effects of the invention.
Unless otherwise specified or defined, the same reference numerals in different figures refer to the same or similar features, and different reference numerals may be used for the same or similar features.
FIG. 1 is a flow chart of a method for efficiently suppressing false information in a network community, which is disclosed by the invention;
FIG. 2 is a schematic structural diagram of an efficient suppression device for false information in a network community, which is disclosed by the present invention;
fig. 3 is a schematic structural diagram of an electronic device disclosed in the present invention.
Description of the reference numerals:
201. a first determination unit; 202. a second determination unit; 203. a participating computing unit; 204. a classification unit; 205. a presentation unit; 206. a monitoring unit; 207. a judgment unit; 208. a control unit; 209. a deletion unit; 301. a memory; 302. a processor.
Detailed Description
In order that the invention may be readily understood, specific embodiments thereof will be described in more detail below with reference to the accompanying drawings.
Unless specifically stated or otherwise defined, 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. In the case of combining the technical solutions of the present invention in a realistic scenario, all technical and scientific terms used herein may also have meanings corresponding to the purpose of achieving the technical solutions of the present invention. As used herein, "first and second" \ 8230, "are used merely to distinguish between names and do not denote a particular quantity or order. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
As used herein, unless otherwise specified or defined, the terms "comprises," "comprising," and "comprising" are used interchangeably to refer to the term "comprising," and are used interchangeably herein.
It is needless to say that technical contents or technical features which are contrary to the object of the present invention or clearly contradicted by the object of the present invention should be excluded.
As shown in fig. 1, an embodiment of the present invention discloses a method for efficiently suppressing false information in a network community, where the method may be implemented by computer programming (such as Python language programming). The execution subject of the method may be a background server of the network community or other network servers, or a device for efficiently suppressing false information of the network community embedded in the server, which is not limited in the present invention. The server serves as a server and serves the webpage client on a computer, a mobile phone or a tablet and the like. The network community refers to an online communication space including BBS/forum, post bar, bulletin board, personal knowledge publishing, group discussion, personal space, wireless value-added service and the like. The method for efficiently suppressing the false information of the network community comprises the following steps S1-S9:
s1, determining current false information appearing on a network community.
In the step, text information (namely comment sentences) appearing on the network community is mainly monitored in real time, and whether the text information is false information or not is identified. Firstly, dynamically collecting and counting text information appearing in the network community/forum; and merging and fusing the same or similar contents in the text information. Then, a false information recognition algorithm (such as text comparison, SVM or Bi-LSTM) is adopted to recognize the collected text information, and a false information set is established. Then, the suppression can be carried out according to each false information, and a suppression method based on user node control is mainly adopted in the invention.
As a preferred implementation mode, the false information identification can be carried out by adopting a text comparison mode based on a value theme. Particularly, a topic which has a large user concern value and rises to a large extent in a certain time and appears on a network community is determined as a value topic, and the value topic is regarded as a value topic which can cause influence and is valuable for carrying out false information identification, so that a non-value topic which can not cause influence or has small influence (for example, a black hole suddenly appears in a galaxy center) can be eliminated, namely a newly published topic on the network community can be dynamically obtained to screen out a high-value topic, and a typical false text for comparison is updated in real time, so that the identification method has better adaptability, and the identification accuracy is improved; in addition, modeling is not needed, the operation is simple, the speed is high, false information can be rapidly identified, and therefore the identification accuracy is greatly improved while high-efficiency identification is achieved. Based on this, step S1 may include the following steps S101 to S105:
s101, acquiring text information appearing in the network community.
S102, a target value theme matched with the text information is called from a plurality of prestored value themes, and a plurality of typical false texts related to the target value theme are called.
The pre-stored multiple value topics can be screened from candidate topics published historically in the network community, and the specific implementation mode comprises the following steps: the method comprises the steps of obtaining a plurality of candidate topics appearing in a network community at a first moment, calculating a first user attention value of each candidate topic at the first moment, determining the candidate topics of which the first user attention value is larger than a preset attention value (for example, 60% of the total number of users in the whole network community) as hot topics, calculating a second user attention value of each hot topic at a second moment, and finally determining the hot topics of which the second user attention value is higher than the first user attention value and the difference value between the second user attention value and the first user attention value is larger than a preset difference value (for example, 20% of the total number of users in the whole network community) as value topics. The first user attention value and the second user attention value can be calculated through the number of users participating in comment and reading at the first time and the second time respectively, and the larger the number of users is, the higher the user attention is, and therefore the larger the user attention value is. The first time and the second time are continuous or intermittent.
Based on each value topic screened out, a plurality of typical false texts (i.e. typical sentences) D = { D } corresponding to the value topic can be established 1 ,d 2 ,…,d |D| And performing association storage. And performing word segmentation and d-recording on each typical false text k The comparative participle vector of the E D is C (D) k ). The contrast participle vector of each typical false text is also stored in association with the value topic.
In practical application, after one or more text messages (namely comment sentences) appearing in a network community are monitored in real time, the semantics of each text message can be extracted, then a value topic with the highest semantic matching degree between the semantics and the text messages is inquired as a target value topic, and the target value topic and a plurality of typical false texts which are stored in association with the target value topic are called.
And S103, calculating the similarity between the text information and each typical false text.
In other possible embodiments, the text information and each typical false text may be aligned, and the alignment rate may be calculated as the similarity between the two. It is preferable in the embodiment to obtain the contrast word segmentation vector C (d) of each typical dummy text k ) And performing word segmentation processing on the text information (namely extracting words in each comment sentence) to obtain a target word segmentation vector, and recording the target word segmentation vector of the jth text information as C (l) j ) And calculating the similarity between the target word segmentation vector and the contrast word segmentation vector of each target typical false text. Based on the method, the similarity between the text information and each typical false text can be obtained through the word segmentation vector, and the calculation efficiency is higher.
And S104, determining the maximum value of the similarity as the false degree of the text information.
And taking the maximum value of the similarity between the text information and each typical false text as the false degree of the text information.
And S105, determining the text information as the current false information when the false degree reaches a preset false threshold value.
Preferably, after the current false information is identified in step S105, an influence value of the current false information may be further calculated, and then the influence value is determined, so that the influence false information may be screened out for suppression, thereby saving the false information prevention and control and suppression cost, improving the suppression efficiency, and implementing an efficient suppression method for the false information of the network community.
Further optionally, the specific implementation of calculating the influence value of the current false information may include the following steps S106 to S107:
s106, determining a target classification subject to which the current false information belongs.
In the step (a), the step (b), the specific method can be word segmentation processing (i.e. extraction) on the current false information related words in the current spurious information s) to obtain a participle vector C(s) = { C 1 ,c 2 \8230 }; and then, carrying out theme classification on the segmentation vectors by adopting a machine learning method (a random forest, an ensemble learning or a deep learning method) to obtain a theme classification result of the current false information, namely obtaining a target classification theme l(s) to which the current false information belongs.
In the embodiment of the invention, a plurality of classification subjects can be preset, a subject classification model is obtained by pre-training, and the subject classification can be carried out on the current false information based on the subject classification model. Each classification subject corresponds to a false information set, and the false information set comprises all historical false information under the corresponding classification subject. Therefore, after determining that the current false information belongs to a target classification subject in the multiple classification subjects, the current false information can be merged into the false information set corresponding to the target classification subject, and all false information included in the false information set corresponding to the target classification subject is updated to include the multiple historical false information and the current false information.
And S107, calculating the influence value of the target classification subject in the first specified time period as the influence value of the current false information.
The first designated time period refers to a recent history time period, that is, a history time period obtained by advancing the current time by a certain time period, for example, last three months or last half a year. And measuring the influence value of the target classification subject according to the user attention degree of the target classification subject in the historical period, and then taking the influence value F (l (s)) of the target classification subject as the influence value of the current false information.
The specific implementation mode can comprise the following steps: calculating the total number of comment users and the total number of reading users when the target classification subject l(s) appears in a first specified time period; calculating to obtain the number of people concerned according to the total number of comment users and the total number of reading users; and determining the ratio of the attention number to the number N of all the topics appearing in the first specified time period on the network community as the influence value of the target classification topic in the first specified time period.
Preferably, when the attention number is calculated, the total number of the comment users and the total number of the reading users can be weighted according to a preset weight coefficient, so that the attention number is obtained. Therefore, the recent influence of the target classification subject can be calculated more accurately.
In practical applications, the influence value F (l (s)) of the target classification subject l(s) within the first specified period can be calculated by the following formula (1):
wherein l(s) represents the target classification topic, F (l (s)) represents the influence value of the target classification topic, N represents the number of all topics appearing in a first specified time period (e.g., last three months or last half year) on the web community, N represents the total number of times the target classification topic appears in the first specified time period, N is equal to N, u is equal to N i Represents the number v of users who participate in comments (including reading and comments) when the ith occurrence in the first specified time period of the target classification subject i Representing the number of users participating in reading (only including reading) at the ith occurrence time in the first specified time period of the target classification subject, wherein alpha and beta are specified influence coefficients, and in practical application, alpha =1 and beta =0.75 can be taken.
And S2, when the influence value of the current false information reaches a preset influence threshold value, determining the current false information as target false information.
When the influence value of the current false information reaches a preset influence threshold value (F (l (s)) > F ≧ F 0 ) Then, the current false information is determined as the target false information, so that a target false information set with large influence can be screened out
And then marking the target false information S on the network community, wherein the marking time is the current time t.
Most false information in the actual network forum is rarely concerned or participated in spreading by the public; the public often cares about the subject or content closely related to the public, which causes the false content related to the public to be easily spread. However, the existing false information suppression method often ignores the relevance of the false information content itself to the public, namely, the adverse effect value or the actual influence on the public caused by the false information.
In contrast, in the embodiment of the invention, the false information with influence can be screened out for suppression, so that the prevention and control cost and the suppression cost of the false information are saved, the suppression efficiency is improved, and the high-efficiency suppression is realized.
And S3, calculating the participation degree of each user node of the network community on the target classification theme to which the target false information belongs.
The participation degree of all user nodes of the network community to the target classification theme to which the target false information belongs can be calculated, and the target false information on the network community can be determined by screening
All user nodes related to the target classification subject to which the node belongs, for example, a user node set U = { U = is counted 1 ,u 2 8230, and then calculating the participation degree of each user node in the user node set to the target classification subject.
It should be noted that, in step S3, a false information set corresponding to a target classification subject to which the target false information belongs may be retrieved, and all false information in the false information set includes a plurality of historical false information under the target classification subject and the target false information.
Thus, a specific implementation of step S3 may include: acquiring the occurrence times of all false information under a target classification theme to which the target false information belongs in a second designated period, determining the propagation times of all false information under the target classification theme, which is participated in the propagation of each user node of the network community in the second designated period, and then calculating the ratio of the propagation times to the occurrence times to acquire the participation degrees of all false information under the target classification theme of each user node. The second designated period may also refer to a recent history period, i.e., a history period that is advanced by a certain time from the current time, for example, last three months or last half a year. The second designated time period may be the same as or different from the first designated time period, and may be set according to actual requirements.
Each false information in the false information set corresponding to the target classification subject is labeled on the network community when being identified as the false information with larger influence, so that each false information corresponds to one labeling time. The false information set corresponding to the target classification subject comprises a plurality of historical false information and the target false information, the marking time of each historical false information is located in a second designated time period, the marking time of each historical false information divides the second designated time period into two time periods, namely a pre-marking time period and a post-marking time period, respectively, and the marking time of the target false information is the current time and is located at the tail end point of the second designated time period, namely the pre-marking time period of the target false information is the whole second designated time period.
Based on this, the determining of the number of times that each user node of the network community participates in propagation of all false information under the target classification topic in the second designated period may specifically include:
firstly, determining the labeling time for labeling all false information in a false information set corresponding to the target classification subject on a network community;
then, determining a first number of times that each user node of the network community participates in propagation of all the false information in a period before marking and a second number of times that all the false information participates in propagation in a period after marking;
and finally, performing weighted calculation on the first times and the second times according to a preset weight coefficient to obtain the propagation times of all the false information of each user node participating in the propagation of the target classification theme in a second specified period.
Specifically, in step S3, the participation degree of each user node in the target classification topic to which the target false information belongs may be calculated by the following formula (2):
wherein the content of the first and second substances,
representing the participation of the user node u in a target classification subject to which the target false information belongs in a second specified period, and m representing the occurrence frequency of all false information in the second specified period under the target classification subject to which the target false information belongs, phi u Represents the first number of times that user node u participates in all false information propagation (including discussion and forwarding) under the target classification subject during the pre-annotation period, and/or>
Representing a second time of propagation of all false information of the user node u participating in the target classification subject in the labeled time period, wherein eta and mu respectively represent weight coefficients of a first time before labeling and a second time after labeling, and eta is less than mu; in practical applications 0 < η < 1, μ > 1, e.g., η =0.3, μ =2, may be preferred.
And S4, determining the current level of each user node from at least four candidate levels according to the participation degree.
The at least four candidate levels are sequentially advanced, the control modes corresponding to the candidate levels are different, and the suppression strength of the control modes is increased along with the advancing direction of the candidate levels. For example, in the embodiment of the present invention, four candidate levels cu, also called user categories, may be preset, which are a user U (-1) that is not easily infected, a user U (0) that is easily infected, a user U (1) that is not informed, and a potentially malicious user U (2), respectively. The control modes corresponding to the four candidate levels are non-processing, prompt processing, warning processing and isolation processing.
After calculating the participation degree g of each user node with respect to the target classification subject, the user node u with respect to the target classification subject to which the target false information belongs can be determined according to the following formula (3)
User evaluation value cu of (1):
wherein cu represents a subject of the user node u about the target classification
-1 represents a user U (-1) that is not susceptible, 0 represents a user U (0) that is susceptible, 1 represents a user U (1) that is not informed, and 2 represents a potentially malicious user U (2); g 0 、g 1 、g 2 、g 3 Respectively represent preset threshold values, and g can be fetched in practical application 0 ＝1、g 1 ＝2.5、g 2 ＝5、g 3 ＝10。
Based on this, a level determination can be made for each user node:
if the user evaluation value cu of the user node is = -1, the current level of the user node is U (-1);
if the user evaluation value cu =0 of the user node, it is indicated that the current level of the user node is U (0);
if the user evaluation value cu =1 of the user node, it is indicated that the current level of the user node is U (1);
if the user evaluation value cu =2 of the user node, it is indicated that the current level of the user node is U (2).
And S5, carrying out false information prompt on the user node with the participation degree larger than the preset participation threshold value.
In the step, the relevant false information potential user nodes are processed for the first time, wherein the preset participation threshold can be independently set as other thresholds or set as the preset threshold g 0 、g 1 、g 2 、g 3 Any one of them, e.g. the preset participation threshold is set to the preset threshold g 0 Then, for g ≦ g 0 The user node of (1), namely the user U (-1) is not easy to infect, and is not processed; for g > g 0 The user nodes of (1) and (2), namely the user U (0) easy to infect, the user U (1) unwittingly spread and the potentially malicious user U (2), carry out false information prompt. Of course, the first processing may be performed according to the current level of the user node, which is not limited.
It should be noted that, step S4 and step S5 may be performed simultaneously, and in the embodiment of the present invention, the order is merely illustrated by way of example, but the order should not be considered as a limitation. In some other possible embodiments, step S5 may be performed first, and then step S4 may be performed. In addition, when steps S4 and S5 are performed simultaneously, it may be on target dummy information
The current time t for marking is executed.
And S6, periodically monitoring whether each user node participates in the transmission of the target false information.
Wherein, a periodic monitoring mode can be started according to a clock instruction, and false information of a target participating in propagation in each current monitoring period is obtained
E.g., at time t +1 at the end of the first monitoring period, statistics is made on the subscriber set ≧ which still participates in the propagation of the target dummy information after the target dummy information is flagged>
And then individually set ≥ the users participating in the propagation>
Taking intersection with a user set U (-1) which is not easy to infect, a user set U (0) which is easy to infect, a user set U (1) which is unknown to propagate and a potential malicious user set U (2) to obtain first intersection->
Second intersection
Third intersection +>
Fourth intersection +>
And S7, if any user node is monitored to participate in propagating the target false information, judging whether the current level of the user node is the last candidate level in the progressive direction. If not, executing the step S8; if yes, go to step S9.
If any user node is monitored not to participate in the propagation of the target false information, the current level of the user node is kept unchanged.
And S8, adjusting the current level of the user node according to the progressive direction of the candidate level, and controlling the user node by adopting a control mode corresponding to the adjusted current level.
If the current level of the user node is not the last candidate level in the progressive direction, the user node possibly belongs to the first intersection
Second intersection +>
Or the third intersection->
If the user node belongs to the first intersection
Adjusting the current level of the user node from U (-1) to U (0) and prompting the current level;
if the user node belongs to the second intersection
Adjusting the current level of the user node from U (0) to U (1) and warning the user node;
if the user node belongs to the third intersection
The current level of the user node is adjusted from U (1) to U (2) and isolated.
And S9, deleting the user node.
If the current level of the user node is the last candidate level in the progressive direction, that is, the user node belongs to the fourth intersection
The user node can be seen to be spread after the prompt of the false information, and the user node is judged to be a malicious user and can be directly deleted.
And then repeating the steps S6 to S9 when the next monitoring period is started, so that the level of each user node of the network community can be dynamically updated in real time.
In summary, by implementing the embodiment of the present invention, the influential false information can be screened out for suppression, so as to save the false information prevention and control and suppression cost, improve the suppression efficiency, while realizing the high-efficiency suppression, the user nodes can be classified into at least four levels of fine granularity, and the control mode of sequentially increasing the suppression degree is adopted for the user nodes through the sequentially progressive level adjustment, so as to reduce the false judgment rate of the abnormal user nodes in the false information suppression process, and greatly reduce the false operation of the user in the false information suppression process, thereby avoiding the false information rebound caused by the traditional false information processing mode, improving the user service degree, and improving the user friendliness of the network community, so as to maintain the user liveness of the network community platform, and avoid the user liveness reduction of the network community platform, therefore, the high-efficiency suppression method for the false information of the network community has the characteristics of strong target and high algorithm adaptability.
As shown in fig. 2, the embodiment of the present invention discloses an efficient suppression apparatus for network community false information, which includes a first determination unit 201, a second determination unit 202, a participation calculation unit 203, a classification unit 204, a prompt unit 205, a monitoring unit 206, a judgment unit 207, a control unit 208, and a deletion unit 209,
a first determining unit 201, configured to determine current false information occurring on a network community;
a second determining unit 202, configured to determine the current false information as the target false information when the influence value of the current false information reaches a preset influence threshold;
the participation calculation unit 203 is used for calculating the participation degree of each user node in the network community on the target classification subject to which the target false information belongs;
a classification unit 204, configured to determine a current level of each user node from at least four candidate levels according to the participation degree; the control method comprises the following steps that at least four candidate levels are sequentially advanced, the control modes corresponding to the candidate levels are different, and the suppression strength of the control modes is increased along with the advancing direction of the candidate levels;
a prompting unit 205, configured to perform false information prompting on a user node with a participation degree greater than a preset participation threshold;
a monitoring unit 206, configured to periodically monitor whether each user node participates in propagating target false information;
a determining unit 207, configured to determine, when the monitoring unit 206 monitors that any user node participates in propagating the target false information, whether a current level of the user node is a last candidate level in the forward direction;
a control unit 208, configured to, when the determining unit 207 determines that the current level of the user node is not the last candidate level in the forwarding direction, adjust the current level of the user node according to the forwarding direction of the candidate level, and control the user node in a control manner corresponding to the adjusted current level;
a deleting unit 209, configured to perform deletion processing on the user node when the judging unit 207 judges that the current level of the user node is the last candidate level in the advancing direction.
As shown in fig. 3, an embodiment of the present invention discloses an electronic device, which includes a memory 301 storing executable program codes and a processor 302 coupled to the memory 301;
the processor 302 calls the executable program code stored in the memory 301 to execute the method for efficiently suppressing the network community false information described in the above embodiments.
The embodiment of the invention also discloses a computer readable storage medium which stores a computer program, wherein the computer program enables a computer to execute the network community false information efficient suppression method described in the embodiments.
The purpose of the above embodiments is to make an exemplary reproduction and derivation of the technical solutions of the present invention, and to fully describe the technical solutions, objects and effects of the present invention, so as to make the public more thoroughly and comprehensively understand the disclosure of the present invention, and not to limit the protection scope of the present invention.
The above examples are not intended to be exhaustive of the invention and there may be many other embodiments not listed. Any alterations and modifications without departing from the spirit of the invention are within the scope of the invention.
Claims (9)
1. The method for efficiently suppressing false information of the network community is characterized by comprising the following steps:
acquiring a plurality of candidate topics appearing in a network community at a first moment;
calculating a first user attention value of each candidate topic at a first moment;
determining the candidate theme with the first user attention value larger than the preset attention value as a hot theme;
calculating a second user attention value of each hotspot subject at a second moment;
determining the hotspot theme with the second user attention value higher than the first user attention value and the difference value between the second user attention value and the first user attention value larger than a preset difference value as a value theme;
establishing a plurality of typical false texts corresponding to each value theme for associated storage;
acquiring text information appearing on a network community;
calling a target value theme matched with the text information from a plurality of prestored value themes, and calling a plurality of typical false texts associated with the target value theme;
calculating the similarity between the text information and each typical false text;
determining the maximum value of the similarity as the false degree of the text information;
when the false degree reaches a preset false threshold value, determining the text information as current false information;
when the influence value of the current false information reaches a preset influence threshold value, determining the current false information as target false information;
calculating the participation degree of each user node of the network community on a target classification theme to which the target false information belongs;
determining the current level of each user node from at least four candidate levels according to the participation degree; the at least four candidate levels are sequentially advanced, the control modes corresponding to the candidate levels are different, and the suppression strength of the control modes is increased along with the advancing direction of the candidate levels;
carrying out false information prompt on the user node with the participation degree larger than a preset participation threshold value;
periodically monitoring whether each user node participates in the propagation of the target false information;
if any user node is monitored to participate in the transmission of the target false information, whether the current level of the user node is the last candidate level in the progressive direction or not is judged;
if the current level of the user node is not the last candidate level in the progressive direction, adjusting the current level of the user node according to the progressive direction of the candidate level, and controlling the user node by adopting a control mode corresponding to the adjusted current level;
and if the current level of the user node is the last candidate level in the progressive direction, deleting the user node.
2. The method for efficient suppression of false information in web communities according to claim 1, wherein after determining the current false information occurring in a web community, the method further comprises:
determining a target classification subject to which the current false information belongs;
and calculating the influence value of the target classification subject in a first specified time period as the influence value of the current false information.
3. The method for efficiently suppressing the false information of the network community as claimed in claim 2, wherein the determining the target classification topic to which the current false information belongs comprises:
performing word segmentation processing on the current false information to obtain word segmentation vectors;
and performing theme classification on the word segmentation vectors to obtain a target classification theme to which the current false information belongs.
4. The method for efficiently suppressing the false information of the web community as claimed in claim 2, wherein calculating the influence value of the target classification subject in a first specified time period as the influence value of the current false information comprises:
calculating the total number of comment users and the total number of reading users when the target classification theme appears in a first specified time period; calculating to obtain the number of people concerned according to the total number of the comment users and the total number of the reading users;
and determining the ratio of the attention number to the number of all the topics appearing in a first specified time period on the network community as the influence value of the target classification topic in the first specified time period.
5. The method for efficiently suppressing false information of a network community, according to claim 4, wherein the step of calculating the number of people concerned according to the number of total comment users and the number of total reading users comprises the following steps:
and carrying out weighted calculation on the total number of the comment users and the total number of the reading users according to a preset weight coefficient to obtain the number of people concerned.
6. The method for efficiently suppressing the false information of the network community according to any one of claims 1 to 5, wherein calculating the participation degree of each user node of the network community on a target classification subject to which the target false information belongs comprises:
acquiring the occurrence times of all false information in a second designated period under the target classification subject to which the target false information belongs;
determining the propagation times of all false information under the target classification topic participated in the propagation of each user node of the network community in the second specified period;
and calculating the ratio of the propagation times to the occurrence times to obtain the participation of each user node in all false information under the target classification subject.
7. The device for efficiently suppressing false information of a network community is characterized by comprising:
a first determining unit, configured to determine current false information occurring on a network community;
a second determining unit, configured to determine the current false information as target false information when an influence value of the current false information reaches a preset influence threshold;
the participation calculation unit is used for calculating the participation degree of each user node of the network community on a target classification theme to which the target false information belongs;
the classification unit is used for determining the current level of each user node from at least four candidate levels according to the participation degree; the method comprises the following steps that at least four candidate levels are sequentially advanced, control modes corresponding to the candidate levels are different, and the suppression strength of the control modes is increased along with the advancing direction of the candidate levels;
the prompting unit is used for carrying out false information prompting on the user nodes with the participation degrees larger than a preset participation threshold;
the monitoring unit is used for periodically monitoring whether each user node participates in the transmission of the target false information;
the judging unit is used for judging whether the current level of any user node is the last candidate level in the progressive direction when the monitoring unit monitors that the user node participates in the transmission of the target false information;
a control unit, configured to, when the determination unit determines that the current level of the user node is not the last candidate level in the forwarding direction, adjust the current level of the user node according to the forwarding direction of the candidate level, and control the user node in a control manner corresponding to the adjusted current level;
a deleting unit, configured to delete the user node when the judging unit judges that the current level of the user node is the last candidate level in the forwarding direction;
the first determining unit is specifically configured to obtain a plurality of candidate topics appearing in the network community at a first time; calculating a first user attention value of each candidate topic at a first moment; determining the candidate theme with the first user attention value larger than the preset attention value as a hot theme; calculating a second user attention value of each hotspot topic at a second moment; determining the hotspot theme with the second user attention value higher than the first user attention value and the difference value between the second user attention value and the first user attention value larger than a preset difference value as a value theme; establishing a plurality of typical false texts corresponding to each value theme for associated storage; and (c) a second step of,
acquiring text information appearing on a network community; calling a target value theme matched with the text information from a plurality of prestored value themes, and calling a plurality of typical false texts associated with the target value theme; calculating the similarity between the text information and each typical false text; determining the maximum value of the similarity as the false degree of the text information; and when the false degree reaches a preset false threshold value, determining the text information as the current false information.
8. An electronic device comprising a memory storing executable program code and a processor coupled to the memory; the processor calls the executable program code stored in the memory for executing the network community false information efficient suppression method of any one of claims 1 to 6.
9. Computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, wherein the computer program causes a computer to execute the method for efficient suppression of network community false information according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211049109.5A CN115391674B (en) | 2022-08-30 | 2022-08-30 | Method, device, equipment and storage medium for efficiently suppressing false information of network community |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211049109.5A CN115391674B (en) | 2022-08-30 | 2022-08-30 | Method, device, equipment and storage medium for efficiently suppressing false information of network community |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115391674A CN115391674A (en) | 2022-11-25 |
| CN115391674B true CN115391674B (en) | 2023-04-14 |
Family
ID=84125387
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211049109.5A Active CN115391674B (en) | 2022-08-30 | 2022-08-30 | Method, device, equipment and storage medium for efficiently suppressing false information of network community |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115391674B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117574261A (en) * | 2023-10-19 | 2024-02-20 | 重庆理工大学 | Multi-field false news reader cognition detection method |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019063867A1 (en) * | 2017-09-28 | 2019-04-04 | Nokia Technologies Oy | Reduction of false paging |
| CN110059240A (en) * | 2019-03-20 | 2019-07-26 | 重庆邮电大学 | A kind of network user's responsibility index calculation method based on influence grade |
| CN110990716B (en) * | 2019-11-19 | 2022-06-28 | 浙江工业大学 | False message propagation inhibiting method based on influence maximization |
-
2022
- 2022-08-30 CN CN202211049109.5A patent/CN115391674B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115391674A (en) | 2022-11-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108874777B (en) | Text anti-spam method and device | |
| Martinez-Romo et al. | Detecting malicious tweets in trending topics using a statistical analysis of language | |
| US9201953B2 (en) | Filtering information using targeted filtering schemes | |
| US20100174813A1 (en) | Method and apparatus for the monitoring of relationships between two parties | |
| Rasool et al. | Multi-label fake news detection using multi-layered supervised learning | |
| Ojugo et al. | Memetic algorithm for short messaging service spam filter using text normalization and semantic approach | |
| US20160019659A1 (en) | Predicting the business impact of tweet conversations | |
| US11095588B2 (en) | Social network data processing and profiling | |
| CN110598982B (en) | Active wind control method and system based on intelligent interaction | |
| CN115391674B (en) | Method, device, equipment and storage medium for efficiently suppressing false information of network community | |
| CN106202031A (en) | A kind of system and method group members being associated based on online social platform group chat data | |
| CN111353554B (en) | Method and device for predicting missing user service attributes | |
| Abinaya et al. | Spam detection on social media platforms | |
| CN110516066B (en) | Text content safety protection method and device | |
| WO2024055603A1 (en) | Method and apparatus for identifying text from minor | |
| Khan et al. | The presence of Twitter bots and cyborgs in the# FeesMustFall campaign | |
| Main et al. | Twitterati identification system | |
| CN109922444B (en) | Spam message identification method and device | |
| CN112468444B (en) | Internet domain name abuse identification method and device, electronic equipment and storage medium | |
| CN111464687A (en) | Strange call request processing method and device | |
| Ali et al. | ESMD: enhanced suspicious message detection framework in instant messaging applications | |
| Molina-Gil et al. | Harassment detection using machine learning and fuzzy logic techniques | |
| CN113157993A (en) | Network water army behavior early warning model based on time sequence graph polarization analysis | |
| Yamak et al. | Automatic detection of multiple account deception in social media | |
| CN117271723A (en) | Fraud early warning method, fraud early warning device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |