CN114663246B - Representation modeling method of information product in propagation simulation and multi-agent simulation method - Google Patents
Representation modeling method of information product in propagation simulation and multi-agent simulation method Download PDFInfo
- Publication number
- CN114663246B CN114663246B CN202210566729.XA CN202210566729A CN114663246B CN 114663246 B CN114663246 B CN 114663246B CN 202210566729 A CN202210566729 A CN 202210566729A CN 114663246 B CN114663246 B CN 114663246B
- Authority
- CN
- China
- Prior art keywords
- information
- information product
- propagation
- product
- simulation
- 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
Images
Classifications
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/31—Indexing; Data structures therefor; Storage structures
- G06F16/313—Selection or weighting of terms for indexing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/34—Browsing; Visualisation therefor
- G06F16/345—Summarisation for human users
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/353—Clustering; Classification into predefined classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
Abstract
The invention discloses a representation modeling method of information products in propagation simulation and a multi-agent simulation method, belonging to the field of cognitive domain propagation simulation and comprising the following steps: explicit and implicit characteristics of the information product are extracted for multi-dimensional modeling, and the problem of information characterization in propagation simulation is solved; designing an interaction mechanism of the modeled information product and the propagation subject; defining a propagation judgment mechanism, taking an interaction result of the information product and a propagation main body as a characteristic, and judging whether to propagate by using a logistic regression function; an information product change judgment mechanism is provided, and a probability model is designed to simulate the change of a transmission subject to an information product in the transmission process; the information product modeling method, the interaction mechanism, the propagation judgment mechanism and the information product change judgment mechanism are applied to multi-agent simulation, and public opinion propagation and evolution in real life are simulated. The invention solves the problems of missing simulation elements of multi-agent transmission and large difference between a simulation result and a real transmission result.
Description
Technical Field
The invention relates to the technical field of cognitive domain propagation simulation, in particular to a representation modeling method of an information product in propagation simulation and a multi-agent simulation method.
Background
With the continuous progress of information technology, the global interconnection of networks such as the Internet, 5G mobile communication and the like is realized, the means of information transmission are greatly enriched, the speed and the breadth of information transmission are improved, and the method has profound influence on the production and the life of the human society. Information dissemination is the process of communicating information between individuals, organizations and groups through symbols and media to convey opinions, attitudes or emotions to other individuals or groups in anticipation of a correspondingly changing activity. The method has the advantages that information transmission is simulated, influence caused by transmission can be scientifically evaluated, decision support is provided for positive energy propaganda, network space management, enterprise brand popularization and the like, and the method has great significance.
Information dissemination simulation involves two important factors, the information artifact and the target audience or media (and the dissemination network that it constitutes), as shown in fig. 1.
At present, the research on information transmission simulation is mainly focused on the latter, for example, chinese patent with publication number CN106682991A discloses an information transmission model based on an online social network and a transmission method thereof, which abstracts transmission into a complex network diagram and interaction behaviors among network nodes based on graph theory and transmission dynamics, and simulates a transmission process with an SIR disease transmission model. Another kind of element contained in the research attention information, for example, chinese patent with publication number CN107122416A discloses a chinese event extraction method, which extracts entity elements such as people, time, and places related to an event based on NLP technology. When the above two types of research results are applied to the information dissemination field, the following problems exist:
(1) the propagation behavior mainly depends on social relationships. The nature of information propagation is driven by information products, the content and the form of the information products are prerequisites for triggering audience propagation determination, social relations are postconditions and secondary factors in a propagation decision chain, and the driving force of neglecting information in propagation simulation causes causal inversion.
(2) The information product is invariable in the process of transmission. Information products interact with target audiences or media, information can be processed through one-time transmission, namely, input and output serving as transmission main bodies are changed according to certain probability, and large deviation is easily caused by product change accumulation after multi-stage transmission.
(3) Audience cognitive abilities solidify. Different target audiences or media have different abilities of recognizing information products, are influenced by various factors such as education level, living habits, hobbies and the like, and if the factors are ignored, the propagation driving force discrimination is poor.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides a representation modeling method of an information product in propagation simulation and a multi-agent simulation method, and solves the technical problems of missing of propagation simulation factors of the existing multi-agent and large difference between a simulation result and a real propagation result.
The purpose of the invention is realized by the following scheme:
a characterization modeling method for information products in propagation simulation comprises the following steps: classifying factors influencing the propagation and diffusion of the information product into an explicit factor and an implicit factor, respectively obtaining an explicit characteristic and an implicit characteristic after corresponding processing, and packaging the obtained explicit characteristic and implicit characteristic into an information product characterization model.
Further, the explicit factors include: language, and/or format, and/or space, and/or named entity, and processing the explicit factor includes the sub-steps of:
and (3) language type processing: the language of the information product is used for judging whether the information can be understood by the transmission main body, and extracting the languages in the text, the picture, the audio and the video to be uniformly expressed by the standard to form an information product language set
；
And/or the presence of a gas in the gas,
and (4) processing the format: dividing the format of the information product into plain text, plain pictures, text and pictures, plain audio, text and audio, plain video, text and video, traversing all format categories and carrying out discretization coding representation on the format categories; for an information product containing pictures and video categories, describing color characteristic information of an image by adopting a color histogram, and representing the characteristic distribution of the image; extracting tone features of an information product containing audio and video, and training a tone classifier to classify tones in the information product;
and/or the presence of a gas in the gas,
processing the space: recording the character length of the text of the information product and the duration of the audio and video, setting a grading threshold value according to historical public opinion propagation data, grading the sections of the information product, and expressing the sections of the multi-mode information product as the highest grade of each type of modal section covered by the sections;
and/or the presence of a gas in the gas,
and (3) processing a named entity: the named entity characteristics of the information product are captured from two aspects: in the first aspect, the mentioning people in the information product are extracted and divided into mentioning people lists according to the symbol @; extracting the release time of the information product as a standard for measuring freshness; in the second aspect, the name of a person, the name of a place and the time contained in the text of the information product are extracted by using a named entity recognition algorithm.
Further, the implicit factors include: theme, and/or heat, and/or credibility, and/or freshness, and/or emotional tendency, and/or explosiveness, and processing the implicit factors includes the sub-steps of:
and (3) processing the theme: describing the theme of the information product by using three-dimensional features, wherein a first dimension extracts tags in texts in the information product, the tags comprise abstracts of the texts of the social platform and are segmented into tag lists; extracting keywords of the information product by using a keyword extraction algorithm in a second dimension to describe the content of the information product; extracting information product mention events by using an event extraction algorithm in the third dimension to generate quadruplets;
and/or the presence of a gas in the atmosphere,
and (3) heat treatment: the like, forward and comment interaction data are weighted and summed to describe the popularity, theiHeat of bar information product
Is represented as follows:
wherein
Is as followsiThe number of praise for the bar product,
in order to set the weights in favor of the numbers,
is a firstiThe number of retransmissions of the strip article,
in order to forward the number weight(s),
is as followsiThe number of reviews of the bar product,
is a comment number weight;
and/or the presence of a gas in the gas,
and (3) carrying out confidence level processing: the credibility of the information product, namely the authority of the sender, is measured by the forwarding amount of original information of the information product, the initial authority is manually configured for each sender in a public opinion propagation simulation scene, and the calculation formula of the authority of the sender is expressed as follows:
sender authority = (initial authority x total original information forwarding amount)/number of original information pieces
And/or the presence of a gas in the atmosphere,
and (3) freshness treatment: freshness decays with increasing days of information product release, and the freshness uses a decay function represented as:
whereindIndicating the number of days the information was distributed,ηthe attenuation coefficient is expressed and can be set according to the actual simulation condition;
and/or the presence of a gas in the gas,
and (3) processing emotional tendency: dividing the emotion expressed by the information product into happiness, anger, worry, thinking, sadness, terror and surprise by adopting seven types of emotion analysis methods;
and/or the presence of a gas in the atmosphere,
and (3) explosive property treatment: the explosiveness of the information product consists of the relevance with the current network hot spot and the similarity with the explosive topic; clustering analysis for public opinion propagation dataTaking the topics with the hot degree attribute ranking set value to form an existing hot topic list of the network, wherein the relevance between the information product and the current network hot spot is the sum of the relevance between each hot topic and the topic of the information product
(ii) a The topics which are easy to attract and draw attention in real life are sorted to form hot topic word groups, and the similarity between the information product and the explosive topic is the sum of the similarity between each hot topic and the current information product topic
(ii) a The explosiveness of the information article is expressed as:
wherein
And
respectively, are weighting coefficients.
Furthermore, the information product representation model can be used for customizing and modifying the explicit characteristics and the implicit characteristics in the model according to different public opinion simulation scenes.
A multi-agent simulation method, comprising the steps of:
modeling information in a simulation environment according to the information product characterization model obtained by any one of the modeling methods, initializing a propagation main body, and configuring the relationship among the propagation main bodies, wherein the propagation main body is an intelligent body with an independent memory unit;
running the simulation environment inT 0 Randomly inputting information products into partial propagation main bodies at any time, calculating whether the current input information products are propagated by the single propagation main body according to a propagation judgment mechanism, and further judging whether the current input information products are propagated according to an information product change judgment mechanism if the current input information products are propagated by the single propagation main bodyChanging the information product, if changing, selecting the characteristic to change according to a certain probability, otherwise, directly transmitting the original information in the relation chain; if not, abandoning the information;
T 1 at the moment, a transmission main body receiving the information product starts to judge whether to transmit the information or not, whether to change the information or not and execute a transmission behavior; and simulating public opinion propagation and evolution in the real environment through N rounds of iteration.
Further, the single propagation agent calculates whether to propagate the current input information artifact according to a propagation decision mechanism, comprising the sub-steps of:
when a single information product acts on a single propagation subject, the propagation decision process is as follows: firstly, judging the comprehension degree of the information product by the propagation subject, and judging the language set of the information product
Language set with propagation subjects
Whether an intersection exists to decide whether to propagate; if the set is an empty set, judging that the transmission is not carried out; if not empty, i.e.
Then the propagation agent's interaction with the information article, including the form preference, is calculated and then the propagation agent's knowledge is updated
Interest degree
Degree of affinity and hydrophobicity
(ii) a Designing a propagation judgment mechanism corresponding model according to the interaction result of the information product and the propagation subject as：
When in usepLess than 0.5, the propagation subject does not propagate the information article downward, whenpWhen the transmission speed is more than or equal to 0.5, the transmission main body transmits the information product downwards; model parameters
Training by taking historical public opinion transmission data as training data;
to propagate the wish.
Further, the decision of whether to change the information artifact according to the information artifact change decision mechanism includes the substeps of: designing a probability model to simulate the phenomenon that a propagation subject changes an information product, and setting the probability that the information product is changed according to the actual propagation simulation environment asq；
In the process of propagation, firstly, whether a propagation subject can propagate the information product is judged, and if so, the propagation subject has probabilityqThe information product is modified, the modification is embodied in one or more of language, space, format, theme, named entity and emotional tendency of the information product, and each kind of characteristic corresponds to the modified probability
The modifications are selected correspondingly according to the following table:
the changed information product is propagated downwards from the propagation main body and continues to circulate in the simulation environment.
Further, calculating the form preference of the propagation subject to the information product
The method comprises the following substeps: and converting the format, space, color and tone quantitative characteristics of the information product into a characteristic vector, and calculating the matching degree of the characteristic vector with the form preference built in the propagation main body.
Further, calculating interest degree of the propagation subject in the information product
The method comprises the following substeps: calculating the cosine similarity between the user interest tag of the propagation subject and the information product theme feature vector, and representing the interest degree of the propagation subject to the information product;
whether or not to propagate, after the information product is received by the propagation subject, when the interest degree of the information product by the propagation subject exceeds a threshold value
And the transmission main body adds the theme of the information product into the interest tag of the transmission main body and updates the cognition of the transmission main body.
Further, calculating affinity and sparseness of the propagation subject to the information product
The method comprises the following substeps: and (4) colliding the named entity list of the information product with the name, the net name, the attribution and the place of residence of the propagation subject, and expressing the degree of affinity and sparseness by using the accumulated overlapping times.
The beneficial effects of the invention include:
the invention provides a representation modeling method of an information product in propagation simulation, which does not adopt a scheme that the existing scheme depends on social relations, deduces factors influencing the propagation and diffusion of the information product according to public opinion propagation science and cognitive psychology knowledge, innovates the content and the form of the information product, realizes the automatic representation of the obvious and implicit characteristics of the information product through an explicit and implicit characteristic abstract processing means, and can fully represent the driving force of the information product in the propagation process compared with a model based on social relations.
According to the invention, through attribute probability variation design, an interaction mechanism of cognitive ability and information product representation model is established, and the technical problems of missing simulation elements and large difference between simulation results and real propagation results of the conventional multi-agent propagation are solved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a diagram illustrating the relationship between two major elements of information dissemination;
FIG. 2 is an information article model;
fig. 3 is a flow chart of a propagation decision based on an information artifact model.
Detailed Description
The invention is further described with reference to the following figures and examples. All features disclosed in all embodiments in this specification, or all methods or process steps implicitly disclosed, may be combined and/or expanded, or substituted, in any way, except for mutually exclusive features and/or steps.
The cognitive domain information propagation evaluation is quantitative analysis on information influence in an information-oriented society, and mainly comprises three methods, namely investigation statistics, propagation dynamics analysis and multi-agent simulation. The investigation statistics and the propagation dynamics analysis are analyzed from a macroscopic view, and the multi-agent simulation simulates the whole propagation condition by the interaction between individuals and groups from a microscopic view, so that the method has stronger deduction capability, and can particularly effectively simulate phenomena of incubation, emergence, aggregation, attenuation and the like in propagation. However, the existing intelligent agent construction model and the representation method have defects, and aiming at the problems, the embodiment of the invention establishes an interaction mechanism of cognitive competence and an information product representation model through explicit and implicit characteristic abstraction and content attribute probability change design so as to solve the technical problems of loss of simulation elements and large difference between a simulation result and a real transmission result of the existing multi-intelligent agent transmission.
In the embodiment of the invention, a representation modeling method of an information product in propagation simulation and a multi-agent simulation method are provided. The method is detailed by taking cognitive domain propagation simulation as an example. In the real-time example, the representation of the information product, the interaction mechanism of the information product and the propagation main body, the propagation judgment mechanism, the information product change judgment mechanism, the multi-agent propagation behavior simulation and the like are mainly discussed.
Characterization of information articles
In order to fully represent the driving force of the information product in the transmission process, and combine public opinion transmission science and cognitive psychology related knowledge, factors influencing the transmission and diffusion of the information product are deduced and classified into explicit and implicit characteristics for extraction, and the information product model is shown as figure 2. And (3) realizing feature extraction operators for each type of features by utilizing related algorithms such as natural language processing, computer vision and the like, packaging the operators into an information product characterization model, and realizing automatic characterization of the apparent and hidden features of the information product.
(1) Explicit characterization of information articles: including the language, format, spread, named entity, etc. of the information product.
1) The language: the language of the information product is used to determine whether the information can be understood by the propagating subjects. Extracting the language unification in the text, the picture, the audio and the video, and representing the same by using an ISO 639-1 standard to form an information product language set
。
2) The format is as follows: the format of the information product is divided into plain text, plain pictures, text + pictures, plain audio, text + audio, plain video, text + video and the like, and all format types are traversed and represented by discretization coding.
For information products including pictures and video categories, picture color and video color characteristics are also one of the factors influencing propagation. Color histogram is used to describe the color characteristic information of the image, and the characteristic distribution of the image is represented.
The information product containing audio and video is extracted with its timbre features, usually the timbre in the information product contains human voice, music and noise, the human voice is divided into male voice, female voice and child voice, and the timbre classifier is trained to classify the timbre in the information product.
3) Space width: recording the character length of the text of the information product and the duration of the audio and video, setting a grading threshold value according to historical public opinion propagation data, and grading the sections of the information product, wherein the sections of the multi-mode information product are expressed as the highest grade of each type of modal section grades covered by the multi-mode information product.
4) Naming an entity: the named entity characteristics of the information artifact are captured from two aspects. Extracting the mentioning people in the information product in the shape of "@ xxx" superficially, and dividing the information product into a mentioning people list according to the symbol @; the time of release of the information product is extracted as a criterion for measuring freshness. In the deep layer, the name of a person, the name of a place, the time and the like contained in the text of the information product are extracted by using a named entity recognition algorithm.
(2) Implicit characteristics of the information article: including the subject, heat, confidence, freshness, emotional propensity, explosiveness of the information article.
1) Subject matter: the method comprises the steps of describing the theme of an information product by using three-dimensional features, wherein a first dimension is to extract tags in the form of "# xxx #" in texts in the information product, and the tags are generally abstracts of texts of a social platform and are divided into tag lists according to a symbol #; the second dimension is to use a keyword extraction algorithm to extract keywords of the information product and describe the main content of the information product; the third dimension is to extract information product mention events by using an event extraction algorithm to generate four-tuples of the shape of < event type, event trigger, event element and element role >.
2) Heat: carrying out weighted summation on interaction data such as praise, forwarding, comment and the like to describe the popularity, andiheat of bar information article
Is represented as follows:
wherein
Is as followsiThe number of praise for the bar product,
in order to set the weight for the number of praise,
is as followsiThe number of retransmissions of the strip article,
in order to forward the number weight(s),
is as followsiThe number of reviews of the bar product,
is the comment number weight.
3) Reliability: the credibility of the information product, namely the authority of the sender, is measured by the forwarding amount of original information of the information product, the initial authority is manually configured for each sender in the public opinion propagation simulation scene, and the calculation formula of the authority of the sender can be expressed as follows:
sender authority = (initial authority x total original information forwarding amount)/number of original information pieces
4) Freshness: freshness decays as the number of days an information product is released increases, and freshness can be expressed using a decay function as:
whereindIndicating the number of days the information was distributed,ηthe attenuation coefficient is expressed and can be set according to the actual simulation situation.
5) Emotional tendency: seven types of emotion analysis methods are adopted to classify the emotions expressed by the information product into happiness, anger, worry, thinking, sadness, terrorism and surprise.
6) Explosiveness: the explosiveness of an information article consists of relevance to the current network hotspot and similarity to an explosive topic. Performing cluster analysis on public opinion propagation data, taking topics with the heat attribute ranked in the top ten to form an existing network hot topic list, wherein the relevance between the information product and the current network hot spot is the sum of the relevance between each hot topic and the information product topic
. The method is characterized in that topics which are easy to attract attention in real life are sorted to form hot topic phrases such as 'dazzling' and 'star', and the similarity between an information product and an explosive topic is the sum of the similarity between each hot topic and the current information product topic
. The explosiveness of an information article can be expressed as:
wherein
And with
Respectively, are weighting coefficients.
The information product representation model listed in the invention is a general version, and can be customized and modified according to different public opinion simulation scenes.
Second, the interaction mechanism of the information product and the propagation subject
The interaction between the information product and the propagation subject is composed of the comprehension degree, form preference degree, interestingness degree and affinity degree of the propagation subject to the information.
1) Comprehension of information artifacts by the propagating subjects:
and judging whether the language set of the propagation main body and the language set extracted from the information product have intersection or not, wherein the information product can be understood and further propagated only if the condition described by the following formula is met.
Wherein, the first and the second end of the pipe are connected with each other,
represents a collection of languages covered in the information article,
representing a set of languages of the propagating subject.
2) Form preference of the propagating subject for the information article:
converting the format, space, color and tone quantization characteristics of information product into characteristic vector, calculating the matching degree with the form preference characteristic vector built in the transmission main body, and using
And (4) showing.
3) Interest of the propagating subject in the information product:
calculating cosine similarity between the user interest label of the propagation subject and the information product theme feature vector, representing the interest degree of the propagation subject to the information product, and using the similarity to calculate the cosine similarity between the user interest label of the propagation subject and the information product theme feature vector
And (4) showing.
4) Affinity and sparseness of the propagating subject to the information article:
the named entity list of the information product is collided with the attributes of the name, net name, attribution, living place and the like of the propagation subject, and the degree of affinity and sparseness is represented by the accumulated overlapping times
。
Third, propagation decision mechanism
The set of propagation policies for the propagating agent is shown in the following table:
| main body | Policy aggregation |
| A | { propagation, non-propagation } |
| B | { propagation, no propagation } |
Willingness to propagate body
Can be represented by the ratio of the forwarding number in the history record of the propagation body to the total information data of the propagation body.
When a single information product acts on a single propagation subject, the propagation decision process is as follows: first, the comprehension of the information product by the propagating agent is judged, i.e.
Whether it is an empty set. If the information is an empty set, judging that the information is not transmitted, and if the information is not an empty set, calculating the form preference, the interest degree, the intimacy and the disambiguation of the transmission subject to the information productAnd (4) degree. According to the interaction result of the information product and the propagation main body, the propagation judgment mechanism corresponding model is as follows:
when the temperature is higher than the set temperaturepWhen less than 0.5, the transmission body does not transmit the information product downward, whenpAnd when the number is greater than or equal to 0.5, the transmission body transmits the information product downwards. Model parameters
The training data is obtained by training through historical public opinion transmission data.
Fourth, information product change judgment mechanism
The set of change policies for a propagating agent on an information artifact is shown in the following table:
| main body | Policy collection |
| A | { modified, unmodified } |
| B | { modified, unmodified } |
In order to enable public opinion propagation simulation to be closer to the practical situation, a probability model is designed to simulate the phenomenon that a propagation main body changes an information product. According to actual propagation simulation environment settingThe probability that the information product is altered isq. In the process of propagation, firstly, whether a propagation subject can propagate the information product is judged, and if so, the propagation subject has probabilityqChanges are made to the informational article. The change is mainly embodied in the following characteristics including language, space, format, theme, named entity, emotional tendency and the like of the information product, and each type of characteristic corresponds to the changed probability
。
The changed information product is propagated downwards from the propagation main body and continuously circulates in the simulation environment, so that the propagation error caused by the fact that the product is unchanged after multi-stage propagation is avoided.
Five, multi-agent propagation behavior simulation
And combining the models and mechanisms, and adopting a multi-agent modeling technology to simulate the public opinion spreading scene in real life. Modeling information in a simulation environment according to an information product representation model, initializing a propagation main body, and configuring the relationship between the propagation main bodies, wherein the propagation main body is an intelligent body with an independent memory unit. The simulation environment is run and the simulation environment is run,T 0 at any moment, information products are randomly input into a part of propagation main bodies, a single propagation main body calculates whether to propagate the currently input information products according to a propagation judgment mechanism, if the currently input information products are propagated, the information products are further judged whether to be changed according to an information product change judgment mechanism, if the currently input information products are changed, characteristics are selected according to a certain probability to be changed, otherwise, original information is directly propagated in a relation chain, and if the currently input information products are not propagated, the information is abandoned.T 1 At the moment, the propagation agent receiving the information product starts to judge whether to propagate the information or not, whether to change the information or not, and executes propagation behavior. And simulating public opinion propagation and evolution in the real environment through N rounds of iteration.
Whether or not to propagate, the propagating subject, after receiving the information article, when the propagating subject is to the information articleInterest level exceeding threshold
The propagation main body adds the theme of the information product to the interest tag of the propagation main body, updates the cognition of the propagation main body, and a propagation judgment process is shown in fig. 3.
The invention extracts the explicit and implicit characteristics of the information product to carry out multi-dimensional modeling, solves the problem of representation of information in propagation simulation, and fully considers the driving force of the information product in the simulation; designing an interaction mechanism of the modeled information product and the propagation subject; furthermore, a propagation judgment mechanism is defined, the interaction result of the information product and the propagation main body is used as a characteristic, and a logistic regression function is used for judging whether the information product is propagated or not; an information product change judgment mechanism is innovatively provided, and a probability model is designed to simulate the change of a transmission main body to an information product in the transmission process; further, the information product modeling method, the interaction mechanism, the transmission judgment mechanism and the information product change judgment mechanism are applied to multi-agent simulation to simulate the transmission and evolution of the public sentiment in real life.
Example 1: a characterization modeling method for information products in propagation simulation comprises the following steps: classifying factors influencing the propagation and diffusion of the information product into an explicit factor and an implicit factor, respectively obtaining an explicit characteristic and an implicit characteristic after corresponding processing, and packaging the obtained explicit characteristic and implicit characteristic into an information product characterization model.
Example 2: on the basis of embodiment 1, the explicit factors include: language, and/or format, and/or spread, and/or named entity, and processing the explicit factors includes the sub-steps of:
and (3) language type processing: the language of the information product is used for judging whether the information can be understood by the transmission main body, and extracting the languages in the text, the picture, the audio and the video to be uniformly expressed by the standard to form an information product language set
；
And/or the presence of a gas in the gas,
and (4) processing the format: dividing the format of the information product into plain text, plain pictures, text and pictures, plain audio, text and audio, plain video, text and video, traversing all format categories and performing discretization coding representation on the format categories; for an information product containing pictures and video categories, describing color characteristic information of an image by adopting a color histogram, and representing the characteristic distribution of the image; extracting tone features of an information product containing audio and video, and training a tone classifier to classify tones in the information product;
and/or the presence of a gas in the gas,
processing the space: recording the character length of the text of the information product and the duration of the audio and video, setting a grading threshold value according to historical public opinion propagation data, grading the sections of the information product, and expressing the sections of the multi-mode information product as the highest grade of each type of modal section covered by the sections;
and/or the presence of a gas in the gas,
and (3) processing a named entity: the named entity characteristics of the information product are captured from two aspects: in the first aspect, the mentioning people in the information product are extracted and divided into mentioning people lists according to the symbol @; extracting the release time of the information product as a standard for measuring freshness; in the second aspect, the name of a person, the name of a place and the time contained in the text of the information product are extracted by using a named entity recognition algorithm.
Example 3: on the basis of embodiment 1, the implicit factors include: theme, and/or heat, and/or credibility, and/or freshness, and/or emotional tendency, and/or explosiveness, and processing the implicit factors includes the sub-steps of:
and (3) processing the theme: describing the theme of the information product by using three-dimensional features, wherein a first dimension extracts tags in texts in the information product, the tags comprise abstracts of the texts of the social platform and are segmented into tag lists; in the second dimension, extracting keywords of the information product by using a keyword extraction algorithm to describe the content of the information product; extracting information product mention events by using an event extraction algorithm in a third dimension to generate a quadruple;
and/or the presence of a gas in the atmosphere,
and (3) heat treatment: the like, forward and comment interaction data are weighted and summed to describe the popularity, the secondiHeat of bar information article
Is represented as follows:
wherein
Is as followsiThe number of praise for the bar product,
in order to set the weights in favor of the numbers,
is a firstiThe number of transfers of the article of manufacture,
in order to forward the number weight(s),
is as followsiThe number of reviews of the bar product,
is a comment number weight;
and/or the presence of a gas in the atmosphere,
and (4) processing the credibility: the credibility of the information product, namely the authority of the sender, is measured by the forwarding amount of original information of the information product, the initial authority is manually configured for each sender in a public opinion propagation simulation scene, and the calculation formula of the authority of the sender is expressed as follows:
sender authority = (initial authority = total original information forwarding amount)/number of original information pieces
And/or the presence of a gas in the atmosphere,
and (3) freshness treatment: freshness decays with increasing days of information product release, and the freshness uses a decay function represented as:
whereindIndicating the number of days the information was distributed,ηthe attenuation coefficient is expressed and can be set according to the actual simulation condition;
and/or the presence of a gas in the atmosphere,
and (3) processing emotional tendency: dividing the emotion expressed by the information product into happiness, anger, worry, thinking, sadness, terror and surprise by adopting seven types of emotion analysis methods;
and/or the presence of a gas in the atmosphere,
and (3) explosive property treatment: the explosiveness of the information product consists of the relevance with the current network hot spot and the similarity with the explosive topic; performing cluster analysis on the public sentiment propagation data, taking the topics with the hot degree attribute ranking set value to form an existing network hot topic list, wherein the relevance between the information product and the current network hot spot is the sum of the relevance between each hot topic and the topic of the information product
(ii) a Arranging the topics which are easy to attract attention in real life to form hot topic phrases, wherein the similarity between the information product and the explosive topic is the sum of the similarity between each hot topic and the topic of the current information product
(ii) a The explosiveness of the information article is expressed as:
wherein
And
respectively, are weighting coefficients.
Example 4: on the basis of the embodiment 1, the information product characterization model can be used for customizing and modifying the explicit characteristics and the implicit characteristics in the model according to different public opinion simulation scenes.
Example 5: a multi-agent simulation method comprising the steps of:
modeling information in a simulation environment according to the information product characterization model obtained by the modeling method in any one of embodiments 1 to 3, initializing a propagation main body, and configuring a relationship between the propagation main bodies, wherein the propagation main body is an intelligent body with an independent memory unit;
running the simulation environment inT 0 Randomly inputting information products into a part of propagation main bodies at any moment, calculating whether the currently input information products are propagated or not by a single propagation main body according to a propagation judgment mechanism, further judging whether the information products are changed or not according to an information product change judgment mechanism if the information products are propagated, selecting characteristics to change according to a certain probability if the information products are changed, and directly propagating original information in a relation chain if the information products are not propagated; if not, abandoning the information;
T 1 at the moment, a transmission main body receiving the information product starts to judge whether to transmit the information or not, whether to change the information or not and execute a transmission behavior; and simulating public opinion propagation and evolution in the real environment through N rounds of iteration.
Example 6: on the basis of embodiment 5, the single propagation agent calculates whether to propagate the current input information artifact according to a propagation decision mechanism, and comprises the following sub-steps:
when a single information product acts on a single propagation subject, the propagation decision process is as follows: firstly, judging the comprehension degree of the information product by the propagation subject, and judging the language set of the information product
Language set with propagation subject
Whether an intersection exists to decide whether to propagate; if the set is an empty set, judging that the transmission is not carried out; if not empty, i.e.
Then the propagation agent's interaction with the information article, including the form preference, is calculated and then the propagation agent's knowledge is updated
Degree of interest
Degree of affinity and hydrophobicity
(ii) a Designing a propagation judgment mechanism corresponding model according to the interaction result of the information product and the propagation main body as follows:
when in usepLess than 0.5, the propagation subject does not propagate the information article downward, whenpWhen the number is more than or equal to 0.5, the transmission main body transmits the information product downwards; model parameters
Training by taking historical public opinion transmission data as training data;
to propagate the wish.
Example 7: on the basis of the embodiment 5, the decision of whether to change the information product according to the information product change decision mechanism comprises the substeps of: design probabilistic model simulationThe transmission subject changes the phenomenon of the information product, and sets the probability that the information product is changed according to the actual transmission simulation environment asq；
In the process of propagation, firstly, whether a propagation subject can propagate the information product is judged, and if so, the propagation subject has probabilityqThe information product is modified, the modification is embodied in one or more of language, space, format, theme, named entity and emotional tendency of the information product, and each kind of characteristic corresponds to the modified probability
The modifications are selected correspondingly according to the following table:
the changed information product is propagated downwards from the propagation main body and continues to circulate in the simulation environment.
Example 8: on the basis of the embodiment 6, the form preference degree of the information product of the propagation subject is calculated
The method comprises the following substeps: and converting the format, space, color and tone quantitative characteristics of the information product into a characteristic vector, and calculating the matching degree of the characteristic vector with the form preference built in the propagation main body.
Example 9: based on the embodiment 6, the interest degree of the information product in the propagation subject is calculated
The method comprises the following substeps: calculating the cosine similarity between the user interest tag of the propagation subject and the information product theme feature vector, and representing the interest degree of the propagation subject to the information product;
whether or not to propagate, after the information product is received by the propagation subject, when the interest degree of the information product by the propagation subject exceeds a threshold value
And the transmission main body adds the theme of the information product into the interest tag of the transmission main body and updates the cognition of the transmission main body.
Example 10: on the basis of the embodiment 6, the degree of affinity and sparseness of the propagation subject to the information product is calculated
The method comprises the following substeps: and (4) colliding the named entity list of the information product with the name, the net name, the attribution and the place of residence of the propagation subject, and expressing the degree of affinity and sparseness by using the accumulated overlapping times.
The units described in the embodiments of the present invention may be implemented by software, or may be implemented by hardware, and the described units may also be disposed in a processor. Wherein the names of the elements do not in some way constitute a limitation on the elements themselves.
According to an aspect of the application, a computer program product or computer program is provided, comprising computer instructions, the computer instructions being stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the method provided in the various alternative implementations described above.
As another aspect, the present application also provides a computer-readable medium, which may be contained in the electronic device described in the above embodiments; or may exist separately without being assembled into the electronic device. The computer readable medium carries one or more programs which, when executed by an electronic device, cause the electronic device to implement the method described in the above embodiments.
The parts not involved in the present invention are the same as or can be implemented using the prior art.
The above-described embodiment is only one embodiment of the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be easily made based on the application and principle of the present invention disclosed in the present application, and the present invention is not limited to the method described in the above-described embodiment of the present invention, so that the above-described embodiment is only preferred, and not restrictive.
Other embodiments than the above examples may be devised by those skilled in the art based on the foregoing disclosure, or by adapting and using knowledge or techniques of the relevant art, and features of various embodiments may be interchanged or substituted and such modifications and variations that may be made by those skilled in the art without departing from the spirit and scope of the present invention are intended to be within the scope of the following claims.
Claims (9)
1. A method for modeling the representation of an information product in propagation simulation is characterized by comprising the following steps: classifying factors influencing the propagation diffusion of the information product into an explicit factor and an implicit factor, respectively obtaining an explicit characteristic and an implicit characteristic after corresponding processing, and packaging the obtained explicit characteristic and implicit characteristic into an information product representation model; the information product characterization model is used for modeling information in a simulation environment, and when the information product characterization model is used for multi-agent simulation, the method comprises the following steps that a single propagation main body calculates whether to propagate a current input information product according to a propagation judgment mechanism:
when a single information product acts on a single propagation subject, the propagation decision process is as follows: firstly, judging the comprehension degree of the information product by the propagation subject, and judging the language set of the information product
Language set with propagation subject
Whether an intersection exists to decide whether to propagate; if the set is an empty set, judging that the transmission is not carried out; if not empty, i.e.
Calculating the interaction of the propagation subject on the information product and then updating the knowledge of the propagation subject, wherein the interaction of the propagation subject on the information product comprises the form preference
Interest degree
Degree of affinity and hydrophobicity
(ii) a Designing a propagation judgment mechanism corresponding model according to the interaction result of the information product and the propagation main body as follows:
when in usepLess than 0.5, the propagation subject does not propagate the information article downward, whenpWhen the transmission speed is more than or equal to 0.5, the transmission main body transmits the information product downwards; model parameters
Training by taking historical public opinion transmission data as training data;
to propagate the desire;
an attenuation function is used for freshness, wherein,dindicating the number of days the information was distributed.
2. The method of claim 1, wherein the explicit factors comprise: languages, formats, spreads, and/or named entities, and processing explicit factors includes the sub-steps of:
and (3) language type processing: the language of the information product is used for judging whether the information can be understood by the transmission main body, and extracting the languages in the text, the picture, the audio and the video to be uniformly expressed by the standard to form an information product language set
；
And/or the presence of a gas in the gas,
and (4) processing the format: dividing the format of the information product into plain text, plain pictures, text and pictures, plain audio, text and audio, plain video, text and video, traversing all format categories and carrying out discretization coding representation on the format categories; for an information product containing pictures and video categories, describing color characteristic information of an image by adopting a color histogram, and representing the characteristic distribution of the image; extracting tone features of an information product containing audio and video, and training a tone classifier to classify tones in the information product;
and/or the presence of a gas in the gas,
processing the space: recording the character length of the text of the information product and the duration of the audio and video, setting a grading threshold value according to historical public opinion propagation data, and grading the sections of the information product, wherein the sections of the multi-mode information product are expressed as the highest grade of each type of modal section grades covered by the multi-mode information product;
and/or the presence of a gas in the gas,
and (3) processing a named entity: the named entity characteristics of the information product are captured from two aspects: in the first aspect, the mentioning people in the information product are extracted and divided into mentioning people lists according to the symbol @; extracting the release time of the information product as a standard for measuring freshness; in the second aspect, the name of a person, the name of a place and the time contained in the text of the information product are extracted by using a named entity recognition algorithm.
3. The method of claim 1, wherein the implicit factors include: theme, heat, credibility, freshness, emotional propensity, and/or explosiveness, and processing implicit factors includes the sub-steps of:
and (3) processing the theme: describing the theme of the information product by using three-dimensional features, wherein a first dimension extracts tags in texts in the information product, the tags comprise abstracts of the texts of the social platform and are segmented into tag lists; in the second dimension, extracting keywords of the information product by using a keyword extraction algorithm to describe the content of the information product; extracting information product mention events by using an event extraction algorithm in the third dimension to generate quadruplets;
and/or the presence of a gas in the gas,
and (3) heat treatment: the like, forward and comment interaction data are weighted and summed to describe the popularity, the secondiHeat of bar information article
Is represented as follows:
wherein
Is as followsiThe number of praise for the bar product,
in order to set the weights in favor of the numbers,
is as followsiThe number of transfers of the article of manufacture,
in order to forward the number weight(s),
is as followsiThe number of reviews of the bar product,
is a comment number weight;
and/or the presence of a gas in the atmosphere,
and (3) carrying out confidence level processing: the credibility of the information product, namely the authority of the sender, is measured by the forwarding amount of original information of the information product, the initial authority is manually configured for each sender in a public opinion propagation simulation scene, and the calculation formula of the authority of the sender is expressed as follows:
sender authority = (initial authority = total original information forwarding amount)/number of original information pieces
And/or the presence of a gas in the gas,
and (3) freshness treatment: freshness decays with increasing days of information product release, and the freshness uses a decay function represented as:
whereindIndicating the number of days the information was distributed,ηthe attenuation coefficient is expressed and can be set according to the actual simulation condition;
and/or the presence of a gas in the gas,
and (3) processing emotional tendency: seven types of emotion analysis methods are adopted to divide the emotions expressed by the information product into happiness, anger, worry, thinking, sadness, fear and surprise;
and/or the presence of a gas in the gas,
and (3) explosive property treatment: the explosiveness of the information product consists of the relevance with the current network hot spot and the similarity with the explosive topic; performing cluster analysis on public opinion propagation data, taking topics with a hot attribute ranking set value to form an existing network hot topic list, wherein the relevance between an information product and a current network hot spot is the sum of the relevance between each hot topic and the topic of the information product
(ii) a The topics which are easy to attract and attract in real life are arranged to form hot topic phrases and informationThe similarity between the product and the explosive topic is the sum of the similarity between each hot topic and the current information product topic
(ii) a The explosiveness of the information article is expressed as:
wherein
And
respectively, are weighting coefficients.
4. The method for modeling the representation of an information product in the propagation simulation of claim 1, wherein the representation model of the information product can be customized and modified for explicit and implicit features in the model according to different public opinion simulation scenarios.
5. A multi-agent simulation method is characterized by comprising the following steps:
modeling information in a simulation environment according to an information product characterization model obtained by the modeling method of any one of claims 1 to 3, initializing a propagation subject, and configuring the relationship between the propagation subjects, wherein the propagation subjects are intelligent agents with independent memory units;
running the simulation environment inT 0 Randomly inputting information products into a part of propagation main bodies at any moment, utilizing the single propagation main body to calculate whether to propagate the currently input information products according to a propagation judgment mechanism, further judging whether to change the information products according to an information product change judgment mechanism if the information products are propagated, selecting characteristics to change according to a certain probability if the information products are changed, and directly propagating the original information in a relation chain if the information products are not propagated(ii) a If not, abandoning the information;
T 1 at the moment, a transmission main body receiving the information product starts to judge whether to transmit the information or not, whether to change the information or not and execute a transmission behavior; and simulating public opinion propagation and evolution in the real environment through N rounds of iteration.
6. The multi-agent emulation method of claim 5, wherein said deciding whether to modify an information artifact based on an information artifact modification decision mechanism comprises the sub-steps of: designing a probability model to simulate the phenomenon that a propagation subject changes an information product, and setting the probability that the information product is changed according to the actual propagation simulation environment asq；
In the process of propagation, firstly, whether the propagation subject can propagate the information product is judged, if so, the propagation subject has probabilityqThe information product is modified, the modification is embodied in one or more of language, space, format, theme, named entity and emotional tendency of the information product, and each kind of feature corresponds to the modified probability
The modifications are selected correspondingly according to the following table:
the changed information product is propagated downwards from the propagation body and continues to circulate in the simulation environment.
7. The multi-agent simulation method of claim 5, wherein the formal preference of the propagating agent for the information article is calculated
The method comprises the following substeps: converting the format, space, color and tone quantization characteristics of information product into characteristic directionAnd (4) calculating the matching degree of the form preference feature vector built in the propagation subject.
8. The multi-agent simulation method of claim 5, wherein the interest level of the information product in the propagating agent is calculated
The method comprises the following substeps: calculating the cosine similarity between the user interest tag of the propagation subject and the information product theme feature vector, and representing the interest degree of the propagation subject to the information product;
whether or not to transmit, when the transmitting subject receives the information product and the interest degree of the transmitting subject in the information product exceeds the threshold value
And the transmission main body adds the theme of the information product into the interest tag of the transmission main body and updates the self cognition.
9. The multi-agent simulation method of claim 5, wherein the affinity of the propagating agent to the information article is calculated
The method comprises the following substeps: and (4) colliding the named entity list of the information product with the name, the net name, the attribution and the living place attribute of the propagation subject, and expressing the affinity and the sparseness by using the accumulated overlapping times.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210566729.XA CN114663246B (en) | 2022-05-24 | 2022-05-24 | Representation modeling method of information product in propagation simulation and multi-agent simulation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210566729.XA CN114663246B (en) | 2022-05-24 | 2022-05-24 | Representation modeling method of information product in propagation simulation and multi-agent simulation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114663246A CN114663246A (en) | 2022-06-24 |
| CN114663246B true CN114663246B (en) | 2022-09-23 |
Family
ID=82037619
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210566729.XA Active CN114663246B (en) | 2022-05-24 | 2022-05-24 | Representation modeling method of information product in propagation simulation and multi-agent simulation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114663246B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102929918A (en) * | 2012-09-20 | 2013-02-13 | 西北工业大学 | False online public opinion identification method |
| CN102945268A (en) * | 2012-10-25 | 2013-02-27 | 北京腾逸科技发展有限公司 | Method and system for excavating comments on characteristics of product |
| CN104834632A (en) * | 2015-05-13 | 2015-08-12 | 北京工业大学 | Microblog topic detection and hotspot evaluation method based on semantic expansion |
| CN105303998A (en) * | 2014-07-24 | 2016-02-03 | 北京三星通信技术研究有限公司 | Method, device and equipment for playing advertisements based on inter-audience relevance information |
| CN109064347A (en) * | 2017-06-11 | 2018-12-21 | 南京理工大学 | Information based on multiple agent is propagated and public sentiment evolution simulation method |
| CN109299808A (en) * | 2018-06-04 | 2019-02-01 | 西安电子科技大学 | Netizen's viewpoint evolution trend prediction technique of force modeling is advised based on group |
| CN109657330A (en) * | 2018-12-14 | 2019-04-19 | 中国人民解放军国防科技大学 | Large-scale network information propagation calculation method and device based on distributed platform |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101976231A (en) * | 2010-08-25 | 2011-02-16 | 孙强国 | Network supervision method for multi-language short messages |
| CN102262624A (en) * | 2011-08-08 | 2011-11-30 | 中国科学院自动化研究所 | System and method for realizing cross-language communication based on multi-mode assistance |
| CN106682991B (en) * | 2016-12-21 | 2021-02-09 | 重庆邮电大学 | Information propagation model based on online social network and propagation method thereof |
| US11494446B2 (en) * | 2019-09-23 | 2022-11-08 | Arizona Board Of Regents On Behalf Of Arizona State University | Method and apparatus for collecting, detecting and visualizing fake news |
| CN112395417A (en) * | 2020-11-18 | 2021-02-23 | 长沙学院 | Network public opinion evolution simulation method and system based on deep learning |
-
2022
- 2022-05-24 CN CN202210566729.XA patent/CN114663246B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102929918A (en) * | 2012-09-20 | 2013-02-13 | 西北工业大学 | False online public opinion identification method |
| CN102945268A (en) * | 2012-10-25 | 2013-02-27 | 北京腾逸科技发展有限公司 | Method and system for excavating comments on characteristics of product |
| CN105303998A (en) * | 2014-07-24 | 2016-02-03 | 北京三星通信技术研究有限公司 | Method, device and equipment for playing advertisements based on inter-audience relevance information |
| CN104834632A (en) * | 2015-05-13 | 2015-08-12 | 北京工业大学 | Microblog topic detection and hotspot evaluation method based on semantic expansion |
| CN109064347A (en) * | 2017-06-11 | 2018-12-21 | 南京理工大学 | Information based on multiple agent is propagated and public sentiment evolution simulation method |
| CN109299808A (en) * | 2018-06-04 | 2019-02-01 | 西安电子科技大学 | Netizen's viewpoint evolution trend prediction technique of force modeling is advised based on group |
| CN109657330A (en) * | 2018-12-14 | 2019-04-19 | 中国人民解放军国防科技大学 | Large-scale network information propagation calculation method and device based on distributed platform |
Non-Patent Citations (7)
| Title |
|---|
| 传播主体实现本体知识建构与思维创新的认知机制;崔宁;《现代传播(中国传媒大学学报)》;20130515(第05期);第25-30页 * |
| 国内新闻传播领域人工智能技术研究综述;刘德寰等;《中国记者》;20200301(第03期);第78-84页 * |
| 基于潜在主题的知识组合分析研究――以传播学为例;周娜等;《农业图书情报学刊》;20180830(第09期);第87-92页 * |
| 微博信息传播预测研究综述;李洋等;《软件学报》;20160215(第02期);第247-263页 * |
| 新传播生态下体育赛事网络舆论特征及治理——基于结构分析与情感分析的双重视角;万晓红等;《体育科学》;20210415;第41卷(第4期);第23-31页 * |
| 网络舆情研究中的仿真方法应用综述;朱毅华等;《情报杂志》;20131018(第10期);第29-35页 * |
| 音乐情感识别研究进展;陈晓鸥等;《复旦学报(自然科学版)》;20170415(第02期);第136-148页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114663246A (en) | 2022-06-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111858960B (en) | Essential computing-oriented virtual community resource processing method and component crossing DIKW (DiKW) map | |
| Preotiuc-Pietro et al. | Studying the Dark Triad of personality through Twitter behavior | |
| CN110674410B (en) | User portrait construction and content recommendation method, device and equipment | |
| CN111814982B (en) | Multi-hop question-answer oriented dynamic reasoning network system and method | |
| US9213946B1 (en) | Comparing models | |
| US20130066823A1 (en) | Knowledge representation systems and methods incorporating customization | |
| CN101346718A (en) | Method for providing user of chosen content item | |
| CN110309114B (en) | Method and device for processing media information, storage medium and electronic device | |
| Satapathy et al. | Sentiment analysis in the bio-medical domain | |
| CN115577316A (en) | User personality prediction method based on multi-mode data fusion and application | |
| Jorge-Botana et al. | Bridging the theoretical gap between semantic representation models without the pressure of a ranking: some lessons learnt from LSA | |
| EP3340073A1 (en) | Systems and methods for processing of user content interaction | |
| CN110991742A (en) | Social network information forwarding probability prediction method and system | |
| CN112633690A (en) | Service personnel information distribution method, service personnel information distribution device, computer equipment and storage medium | |
| CN115659008A (en) | Information pushing system and method for big data information feedback, electronic device and medium | |
| CN115544252A (en) | Text emotion classification method based on attention static routing capsule network | |
| Lee et al. | Discovering underlying sensations of human emotions based on social media | |
| Li et al. | Configurational patterns for COVID-19 related social media rumor refutation effectiveness enhancement based on machine learning and fsQCA | |
| KR20210079264A (en) | Methods and systems for defining emotional machines | |
| CN114663246B (en) | Representation modeling method of information product in propagation simulation and multi-agent simulation method | |
| Tarwani et al. | Survey of Cyberbulling Detection on Social Media Big-Data. | |
| CN111079084A (en) | Information forwarding probability prediction method and system based on long-time and short-time memory network | |
| CN109254993B (en) | Text-based character data analysis method and system | |
| Sonntag et al. | Sentiment analysis: What’s your opinion? | |
| Wang et al. | Application of Improved Machine Learning and Fuzzy Algorithm in Educational Information Technology |
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 |