CN112418525A - Method and device for predicting social topic group behaviors and computer storage medium - Google Patents
Method and device for predicting social topic group behaviors and computer storage medium Download PDFInfo
- Publication number
- CN112418525A CN112418525A CN202011325005.3A CN202011325005A CN112418525A CN 112418525 A CN112418525 A CN 112418525A CN 202011325005 A CN202011325005 A CN 202011325005A CN 112418525 A CN112418525 A CN 112418525A
- Authority
- CN
- China
- Prior art keywords
- topic
- data
- node
- sequence
- 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.)
- Granted
Links
- 230000006399 behavior Effects 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 66
- 238000003860 storage Methods 0.000 title claims abstract description 17
- 239000013598 vector Substances 0.000 claims abstract description 78
- 230000005012 migration Effects 0.000 claims abstract description 26
- 238000013508 migration Methods 0.000 claims abstract description 26
- 230000007246 mechanism Effects 0.000 claims abstract description 14
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 12
- 230000004927 fusion Effects 0.000 claims abstract description 12
- 238000013507 mapping Methods 0.000 claims abstract description 12
- 230000006870 function Effects 0.000 claims description 24
- 238000012545 processing Methods 0.000 claims description 9
- 238000005457 optimization Methods 0.000 claims description 7
- 230000003993 interaction Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims 1
- 238000003012 network analysis Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 19
- 238000010586 diagram Methods 0.000 description 14
- 239000011159 matrix material Substances 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 230000008878 coupling Effects 0.000 description 4
- 238000010168 coupling process Methods 0.000 description 4
- 238000005859 coupling reaction Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000011176 pooling Methods 0.000 description 3
- 108010001267 Protein Subunits Proteins 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 239000007795 chemical reaction product Substances 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 238000005295 random walk Methods 0.000 description 2
- 230000007480 spreading Effects 0.000 description 2
- 238000003892 spreading Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 229920001296 polysiloxane Polymers 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Computing Systems (AREA)
- Marketing (AREA)
- Computational Linguistics (AREA)
- Primary Health Care (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to the field of social network analysis, and particularly relates to a method and a device for predicting social topic group behaviors and a computer storage medium; the prediction method comprises the steps of constructing a confrontation generation network to carry out data enhancement on topic data; forming a topic sequence by adopting a node walk strategy; extracting a low-dimensional vector of a topic sequence finished by migration by taking the maximum probability entropy as a target; mapping text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism, and extracting text characteristic factors influencing group behaviors; inputting low-dimensional vectors and text characteristic factors of the topic sequence, and predicting whether potential topic node group users in the next time period can participate in the propagation of the hot topic by adopting a convolutional neural network; the method effectively solves the problems caused by effective data sparsity, topic propagation feature space complexity and topic timeliness, and improves the accuracy of social topic group behavior prediction.
Description
Technical Field
The invention belongs to the field of social network analysis, relates to user behavior analysis, particularly relates to analysis of group behaviors under propagation of hot topics in a social network, and particularly relates to a prediction method and device of group behaviors of social topics and a computer storage medium.
Background
In recent years, with the increasing maturity of internet technology, the internet has become inseparable from our lives. Meanwhile, with the rise of a series of social platforms represented by Twitter, Facebook, weibo and the like, an online social network has become an important channel and a carrier for information communication of people in the modern society. The appearance of social networks provides great convenience for the life of people. Users can publish content, comment and forward information shared by friends and spread news hotspots anytime and anywhere through the Twitter platform, the weibo platform and the like. However, the social network is a double-edged sword, so that information transmission becomes convenient, and due to the characteristics of interactivity, openness, burstiness, timeliness and the like, some information can be evolved into hot topics in the transmission process, and if the development rule of the hot topics cannot be accurately mastered, and the topics can be timely found and correctly guided, an emergency which is unfavorable for life of people can be generated, and the social network is influenced.
The generation and the propagation of the hot topics mainly depend on the participation of group users, the propagation trend of the hot information is grasped, and the prediction of group behaviors is mainly based on. By analyzing the group behaviors of the hot topics, the behavior characteristics of group users can be mastered, the situation of the network hot topics can be further mastered, the propagation rule of information can be mastered, the development of hot information can be dynamically mastered, effective supervision can be facilitated, and social public opinion, emergency early warning and the like can be correctly guided. Therefore, the research on the group behaviors of the hot topics has great significance.
In recent years, the research of many expert scholars on behavior prediction of hot topic groups mainly focuses on two aspects: in the aspect of macroscopic view, from the information propagation angle, many expert scholars construct an information propagation model by using a complex network theory, predict the propagation path of a hot topic, further deduce the behavior of group users, and explore the macroscopic propagation trend of information among groups; in a microscopic aspect, from the perspective of a user, a scholars predicts whether a group participates in the propagation of hot topics by researching internal and external factors influencing the user behavior, and grasps the rule of the group behavior.
As deep learning techniques mature, representation learning and neural networks have found widespread use in many areas. Chenyu senn (prediction of propagation based on nettext rumors representing learning [ D ]. university of wuhan, 2018). The paper predicts rumor propagation by using representation learning and neural networks, but due to the particularity of topic propagation, the problem of sparse effective topic data in the social network in the actual rumor topic propagation is not considered.
Although numerous scholars have made extensive research in the field of prediction of hot topic propagation and achieved considerable success, there are still some challenges:
1. the sparsity of effective topic data leads to inaccurate prediction results. Although the social network is full of massive data, in the process of spreading hot topics, when effective topic data are extracted, due to the influence of factors such as the characteristics of the topics and spreading users, data sparsity still exists. The problem of data sparsity can affect the accuracy of the prediction model.
2. Potential features in the topic data are ignored. The traditional part of research mainly predicts the group behaviors by manually extracting the characteristic factors which influence the user behaviors, such as basic attributes, interest preferences and the like of the user. However, due to the complexity of topic propagation feature space, the potential relationship between nodes also affects the behavior of the user, and the manually extracted features often ignore part of features hidden between nodes, resulting in inaccurate prediction results. Similarly, the traditional information text analysis method often ignores deep-level relevance among texts, so that deviation exists in modeling of user interest, and accurate characteristics cannot be provided for prediction.
3. The spread of the hot topics among groups is limited, and the popularity of the topics in each time phase is different. How to dynamically predict the behavior of the user becomes a difficulty in group behavior prediction research.
Disclosure of Invention
Based on the problems in the prior art, the invention provides a method and a device for predicting social topic group behaviors and a computer storage medium, which are mainly used for compensating topic homomorphic data by using a confrontation generation network, extracting a topic sequence, namely a user node sequence, by adopting an improved node walking strategy, carrying out uniform vector expression by adopting a characteristic space representing a learning dialogue topic, and finally dynamically predicting the group behaviors in the hot topic propagation process by combining a convolutional neural network.
The technical scheme adopted by the invention for solving the technical problems comprises the following steps:
in a first aspect of the present invention, the present invention provides a method for predicting social topic group behavior, the method comprising the steps of:
s1, constructing a confrontation generation network by the topic data generator and the topic data discriminator, and performing data enhancement on the topic data;
s2, calculating the probability of the topic node migrating to the next topic node according to the relationship between the topic attribute and the topic data tendency of the data-enhanced topic data by adopting a node migration strategy, and forming a topic sequence after migration is completed;
s3, taking the maximum probability entropy as a target, and extracting a low-dimensional vector of the topic sequence finished by the migration;
s4, mapping the text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism, and extracting text characteristic factors influencing group behaviors;
s5, inputting low-dimensional vectors and text characteristic factors of the topic sequence, and predicting whether potential group users in the next time period can participate in the propagation of the hot topics by adopting a convolutional neural network.
In a second aspect of the present invention, the present invention provides a prediction apparatus of social topic group behavior, the prediction apparatus comprising:
a topic data generator for generating topic data;
the topic data discriminator is used for discriminating the probability that the generated topic data is real data or false data and forming a countermeasure generation network with the topic data generator;
the topic sequence generator is used for calculating the probability of the topic node migrating to the next topic node by adopting a node migration strategy, and a topic sequence is formed after migration is completed;
the topic sequence dimensionality reduction module is used for extracting a low-dimensional vector of the walk-completed topic sequence by taking the maximum probability entropy as a target;
the text feature extraction module is used for mapping the text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism and extracting text feature factors influencing group behaviors;
and the convolutional neural network module is used for inputting the low-dimensional vector and the text characteristic factor of the topic sequence and predicting whether the potential topic node group user outputting the next time period participates in the propagation of the hot topic.
In a third aspect of the present invention, the present invention also provides a computer readable storage medium having stored thereon computer instructions which, when executed, implement the steps of:
constructing a countermeasure generation network by the topic data generator and the topic data discriminator, and performing data enhancement on the topic data;
calculating the probability of the topic node migrating to the next topic node by adopting a node migration strategy according to the relationship between the topic attribute and the topic data tendency of the data-enhanced topic data, and forming a topic sequence after migration is completed;
extracting a low-dimensional vector of a topic sequence finished by migration by taking the maximum probability entropy as a target;
mapping text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism, and extracting text characteristic factors influencing group behaviors;
and inputting low-dimensional vectors and text characteristic factors of the topic sequence, and predicting whether potential topic node group users in the next time period can participate in the propagation of the hot topic by adopting a convolutional neural network.
The invention has the beneficial effects that:
according to the method, the topic data generator and the topic data discriminator are adopted to construct the confrontation generation network, so that the topic data can be effectively enhanced, parameters of the confrontation network in each iteration process are optimized according to the rules of local optimization and global optimization, and meanwhile, the parameters in the whole iteration process reach Nash balance, the precision of the confrontation network is effectively improved, and the enhanced topic data are effectively provided; the invention also improves the node migration strategy, and enables the topic nodes to migrate a more optimal topic sequence on the premise of the relevance among the topic node attributes; and by taking the maximum probability entropy as a target, the dimensionality of the topic sequence is effectively reduced, so that the problems caused by effective data sparsity, topic propagation feature space complexity and topic time limitation are effectively relieved, and the prediction precision of social topic group behaviors is improved.
Drawings
FIG. 1 is a diagram of a model architecture for predicting group behavior based on a confrontation-generating network and representing learning of hot topics employed in an embodiment of the present invention;
FIG. 2 is a flowchart of a method for predicting social topic group behavior in an embodiment of the invention;
FIG. 3 is a schematic diagram of hot topic homomorphic data enhancement employed in an embodiment of the present invention;
fig. 4 is a hot topic forwarding prediction graph adopted in the embodiment of the present invention;
FIG. 5 is a hot topic propagation spatial feature representation construction diagram employed in an embodiment of the present invention;
fig. 6 is a structural diagram of a device for predicting social topic group behavior according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a diagram of a hot topic group behavior prediction model architecture based on a countermeasure generation network and presentation learning, which is adopted in the embodiment of the present invention, and as shown in fig. 1, in the present invention, a user relationship network, a user history behavior, a user history text, and a user basic attribute are mainly collected to generate a network for countermeasure training so as to enhance topic data, and in a topic propagation feature space, text information features in topic data and user structure information are extracted, and the probability of whether a user will participate in a topic and the propagation trend of the topic can be output by inputting the information into a group behavior prediction model.
Fig. 2 is a flowchart of a method for predicting social topic group behavior according to the present invention, where the method shown in fig. 2 includes:
s1, constructing a confrontation generation network by the topic data generator and the topic data discriminator, and performing data enhancement on the topic data;
before entering data enhancement, a data source needs to be acquired, and data can be acquired by directly downloading an existing public data set, a web crawler or an API opened by each social network site. The data acquired in the invention is information in a hot topic transmission cycle, and comprises time for transmitting and commenting topics, basic information of users participating in topic transmission, friend relationship information (including concerned and concerned information) of the users, and historical behavior information (information for transmitting, commenting and publishing user history) of participants. And pre-processes the data.
The process of acquiring the data source is mainly divided into the following processes:
raw data is acquired. The raw data may be obtained by directly downloading an existing data source, through a social networking public API, or by directly downloading an existing data source.
Simple data cleaning. And carrying out structured processing on the data, deleting repeated data and cleaning invalid nodes.
After a data source is obtained, relevant attributes in the data source need to be extracted, and therefore topic data are formed;
whether a user participates in the propagation process of the hot topic can be influenced by various factors, such as: the user's personal interests, whether friends are involved in the dissemination, etc. Based on this, the extracted features of the present invention mainly include:
user relationship network for social networks
A social network is defined as a directed network G ═ V, E composed of a set of nodes N and a set of directed edges E. Wherein V ═ { V ═ V1,v2,…,vnDenotes a set of users of the group, E ═ E1,e2,…,enRepresents a set of relationship edges between users. The invention defines the user relationship network in the social network as follows:
wherein, the whole network user Wt＝(U∪V)tUser U for representing hot topic participating in topic propagation in t time periodtWith potential users VtThe set of (a) and (b),
representing the relationship edges between the network-wide user groups. The users in the whole network and the relationship network form the user relationship network in the whole network
User historical behavior information B
The historical behavior of the user is specifically defined as:
B＝{(b,wi,h)|wi∈(U∪V)} (2)
wherein, (b, w)iH) represents a user wiAnd releasing, forwarding, commenting or approving the information b at the h time.
User text information S
In the social network, whether the user participates in the hot topic is influenced by the interest of the user, and the text information published by the user contains the topic content in which the user is interested. The present invention defines text information as:
S＝{(T,wi)|wi∈(U∪V)} (3)
wherein, (T, w)i) Representing a full network user wiPublished history text T ═ T1,…TkThe historical text comprises original text information and forwarding text information;
user basic Attribute A
The basic attributes of the user comprise the number of fans of the user, the number of concerns, the gender, the location and the like, and the attributes have certain relevance to the propagation of whether the user can be a hot topic. The invention defines the basic attributes of the user as follows:
A＝{(a,wi)|wi∈(U∪V)} (4)
wherein (a, w)i) Represents user wiBasic information of, e.g. user wiThe number of vermicelli, attention number, activity, etc.
The topic-related data set adopted by the invention is expressed as data ═ x1,x2,...,xn]Let us give PdataRepresenting the distribution of the real topic sequence. The process of topic homomorphic data iterative enhancement is shown in fig. 3, and mainly includes the following steps:
z obtained by random sampling from an original topic sequence is input into a topic data generator G, and the random sequence z is converted into topic data G (z) ═ x by the generator G. PG(x, theta) represents generationDistribution of the topic data G (z), and the parameter theta is determined by the maximum likelihood function as shown in formula (5):
assuming that the topic data discriminator is D, x represents arbitrary sequence data input to the topic data discriminator D, and may be [ x ]1,x2,...,xn]Any of the data, D (x) is output as a value of [0,1 ]]The real numbers in the range represent probability values that the sequence data is real topic data. The arbiter D maximizes the discrimination ability, i.e., maximizes the objective function, as shown in equation (6):
wherein,
represents the input arbitrary sequence data x and the real topic sequence Pdata(iii) a desire;
representing the input arbitrary sequence data x and the topic sequence P output by the generatorG(iii) a desire; if the judgment result is the generated data, the confidence of the data generated by G is not high; the confidence of the generator needs to be continuously improved; and D, simultaneously feeding back the judgment result to G, and after G and D update the relevant parameters according to the feedback, G generates new data and inputs the new data into D again to perform a new round of judgment. G continuously improves the authenticity of G (z) to confuse the decision of the decision device D, and minimizes the probability that G (z) is discriminated as the generated data, i.e. minimizes the objective function as shown in equation (7):
in conjunction with equations (2) and (3), the objective function of the entire countermeasure generation network can be expressed by equation (8) as follows:
due to PdataAnd PGRepresenting the distribution of real topic data and generated topic data, respectively, the optimization process of the whole countermeasure generation network can be represented as iterative countermeasures of G and D until PGInfinite proximity Pdata。
In some preferred embodiments, in order to obtain more real topic data, after nash balance is achieved, if the accuracy of the discriminator D reaches a set threshold epsilon, G (z) is reversely transferred to the generator G, and iterative optimization is performed by combining the input of the random sequence z to update G.
Through the iterative optimization process, the topic data is subjected to data enhancement, more real topic data can be expanded, and the attribute characteristics of the topic data are highlighted.
S2, calculating the probability of the topic node migrating to the next topic node according to the relationship between the topic attribute and the topic data tendency of the data-enhanced topic data by adopting a node migration strategy, and forming a topic sequence after migration is completed;
it can be understood that the topic node in the present invention is a user node, the topic node refers to a user node participating in topic propagation and a user node not participating in propagation, and the nodes in the following text also correspond to the user node or topic node, taking fig. 4 as an example, at time T, a user relationship network formed by the user nodes participating in topic propagation and not participating in propagation can be predicted through the prediction model, and a user relationship network formed by the user nodes participating in topic propagation and not participating in propagation at time T +1 can be predicted.
In the topic propagation space, characteristic factors affecting group behaviors are various, such as a network structure of a topic, text characteristics of the topic, and the like. The invention can represent the characteristics into a low-rank dense vector form through representation learning, and the overall details are shown in FIG. 5.
And (5) carrying out structural feature representation on the topic network. In the topic propagation space, the self attribute of the user node and the relationship between the node attributes influence the behavior of the user group. Most of the existing network representations only consider structural property and homogeneity, and often ignore the relevance among node attributes. Therefore, the invention considers the attribute information of the nodes and defines a topic structure feature representation method.
In consideration of the influence factors of the node attributes, the invention improves the node walk strategy.
Specifically, the improved node of the present invention has a transition probability defined as formula (9), and the starting node is defined as c0The ith node in the random walk is defined as ci：
Wherein, P (r | c)i,ci-1) Represents the previous topic node ci-1Based on the current topic node ciProbability of walking to the next topic node r; a isp,q(ci-1R) a weight adjustment parameter for the topic-representing node ci-1A weight adjustment parameter with the topic node r; as shown in equation (10).
β(ciAnd r) represents the current node ciThe similarity of the attribute with the next node r can be expressed by selecting attribute information such as gender, position information, fan number and the like of the user, mapping the attribute information to a characteristic vector space and utilizing the Euclidean distance between vectors.
γ(ciAnd r) represents the edge weight of the topic network. Since the interaction strength between the potential user and the hotspot user can influence the tendency of the potential user to participate in the hotspot topic, the invention applies the interaction degree to the network
And distributing edge weight values. Side weight value gamma (c)iAnd r) is defined as shown in formulas (11) and (12).
In the above embodiment, the invention performs random walk through equations (9) - (12) to determine the route traveled by the user, and combines the route to the topic sequence { c) of the nodes0,…,ci-1,ci,ci+1…, but since the sequence at this time is a high-dimensional vector, it needs to be processed.
S3, taking the maximum probability entropy as a target, and extracting a low-dimensional vector of the topic sequence finished by the migration;
to process this high-dimensional vector, the present invention effectively encodes one-hot from one bit for each node in the sequence by maximizing probability entropy, training as Embedding vector Embedding, and the objective function is defined as:
wherein ns (w) represents the neighborhood of the topic node w, and Pr (ns (w) | f (w)) represents the probability of the topic node vector f (w) appearing in the neighborhood node, which can be obtained by the formula (14):
by controlling the objective function, the node sequence can be expressed as a low-dimensional vector
In a form of (1), wherein dsThe dimension of the feature vector, N, represents the number of nodes.
S4, mapping the text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism, and extracting text characteristic factors influencing group behaviors;
whether a user group participates in a hot topic or not is influenced by the interests of the user, and historical texts published by the user contain topic contents which are interested in different historical stages. The method disclosed by the invention integrates an attention mechanism to map the text information of the topic to a low-dimensional vector space, and extracts the text characteristic factors influencing the group behaviors.
The invention adopts a hierarchical attention mechanism to endow different weights for the text vector and the word vector in the text, and selectively selects the user interest characteristic vector.
For convenience of description, the social network platform in the present invention takes a microblog platform as an example, and first, considering the influence of a context sentence and a topic of a whole paragraph on a text feature, a doc2vec algorithm is adopted to perform vector representation on each microblog of a user, which is specifically as follows:
l'j＝Ds+Uh(lt-d,…,lt+d；L) (15)
y's＝[l'1,l'2,…] (16)
wherein D issRepresents a paragraph identification ID, and U represents a parameter of the softmax classifier; h represents cascading or averaging word vectors; lt-dRepresenting the t-d word vectors, wherein a word vector matrix contains 2d word vectors in total; l represents a word vector matrix; y'sRepresenting the feature vector of the s-th microblog issued by the user; w represents each word in the text,/'jAnd the j word vector in the s microblog released by the user is represented.
Secondly, setting different weights for each word vector in the microblog from the word level, as shown in formulas (16) and (17):
wherein,
is the factor of the regulation of the frequency,
representing band weights
M denotes the total number of word vectors. Screening top-k word vectors according to the weight, and representing the s-th microblog issued by the user as
Then, different weights are set for different microblogs of the user at the text level, as shown in formulas (18) and (19):
wherein,
a feature vector matrix representing the composition of feature vectors of a single microblog, daDimension representing vector, N representing number of users, yiRepresenting band weights
N represents the number of microblogs issued by the user.
S5, inputting low-dimensional vectors and text characteristic factors of the topic sequence, and predicting whether potential group users in the next time period can participate in the propagation of the hot topics by adopting a convolutional neural network.
The method establishes a hot topic group behavior prediction model by combining CNN. The prediction of group behaviors is defined as a two-classification problem, namely whether potential group users in a t +1 time period can participate in the propagation of hot topics is predicted through hot topic whole-network feature vectors in the t time period.
Firstly, feature fusion is carried out on the multi-feature spatial information extracted in the upper section. For each user node uaBy extracting and expressing vectors according to the network structure of the topic, the higher the semantic similarity between the node vectors is, the higher the correlation between the nodes is. Arbitrary node u1And u2The similarity of the node vectors can be measured by the cosine value of the included angle of the node vectors, which is shown in formula (21):
wherein v (u)1) And v (u)2) Respectively represent users u1And user u2The feature vector of (2). Selecting and user node uaTop-k nodes u with high correlation1,u2…, splicing the feature vectors of the k nodes to form a node uaFeature vector matrix of
Wherein Vs(ua)∈Es，Va(ua)∈EsRespectively represent users uaStructural feature vectors and textual feature vectors. Finally, fusing topic multi-feature space information into a feature vector matrix:
the input of the group behavior prediction model is a topic feature vector matrix E of a t time period. The model is composed of a convolution layer, a pooling layer and a full-connection layer. Firstly, the first layer is used as a convolution layer, a convolution kernel with a single channel is selected to perform convolution operation on a matrix E:
wherein Relu (·) represents a nonlinear activation function, wiRepresenting a weight matrix, EiIs a topic feature vector corresponding to the ith channel of the convolutional layer, biIs an offset value, xconvRepresenting the output of the convolution operation.
Performing max pooling maxPool (x) on the convolved topic feature informationconv) In operation, the convolution layer extracted local features are aggregated. Finally, the output of the pooling layer is carried out through the full connection layer, and the group users v at the t +1 moment are obtained by using the formulas (24) and (25)jThe propagation behavior of (c):
the formula (24) represents the probability that the group users participate in hot topic propagation at the time t +1, i takes a value of 0 or 1, theta is a weight matrix, and h represents the output of the full connection layer.
Represents the final predicted result of the invention; the behavior of a user group on a certain topic can be predicted.
Fig. 6 is a structural diagram of a prediction apparatus for group behavior of social topics in an embodiment of the present application, and as shown in fig. 6, the prediction apparatus includes:
a topic data generator for generating topic data;
the topic data discriminator is used for discriminating the probability that the generated topic data is real data or false data and forming a countermeasure generation network with the topic data generator;
the topic sequence generator is used for calculating the probability of the topic node migrating to the next topic node by adopting a node migration strategy, and a topic sequence is formed after migration is completed;
the topic sequence dimensionality reduction module is used for extracting a low-dimensional vector of the walk-completed topic sequence by taking the maximum probability entropy as a target;
the text feature extraction module is used for mapping the text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism and extracting text feature factors influencing group behaviors;
and the convolutional neural network module is used for inputting the low-dimensional vector and the text characteristic factor of the topic sequence and predicting whether the potential topic node group user outputting the next time period participates in the propagation of the hot topic.
A computer readable storage medium in an embodiment of the present application having stored thereon computer instructions that, when executed, perform the steps of:
constructing a countermeasure generation network by the topic data generator and the topic data discriminator, and performing data enhancement on the topic data;
calculating the probability of the topic node migrating to the next topic node by adopting a node migration strategy according to the relationship between the topic attribute and the topic data tendency of the data-enhanced topic data, and forming a topic sequence after migration is completed;
extracting a low-dimensional vector of a topic sequence finished by migration by taking the maximum probability entropy as a target;
mapping text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism, and extracting text characteristic factors influencing group behaviors;
and inputting low-dimensional vectors and text characteristic factors of the topic sequence, and predicting whether potential topic node group users in the next time period can participate in the propagation of the hot topic by adopting a convolutional neural network.
Although the present application provides method steps as described in an embodiment or flowchart, more or fewer steps may be included based on conventional or non-inventive means. The order of steps recited in the embodiments is merely one manner of performing the steps in a multitude of orders and does not represent the only order of execution. When an actual apparatus or end product executes, it may execute sequentially or in parallel (e.g., parallel processors or multi-threaded environments, or even distributed data processing environments) according to the method shown in the embodiment or the figures. The terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, the presence of additional identical or equivalent elements in a process, method, article, or apparatus that comprises the recited elements is not excluded.
The units, devices, modules, etc. set forth in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. For convenience of description, the above devices are described as being divided into various modules by functions, and are described separately. Of course, in implementing the present application, the functions of each module may be implemented in one or more software and/or hardware, or a module implementing the same function may be implemented by a combination of a plurality of sub-modules or sub-units, and the like. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the communication connection of the method or apparatus or electronic device according to the embodiments may be an indirect coupling or communication connection through some interfaces, apparatuses or units, and may be electrical, mechanical or other forms.
It will also be apparent to those skilled in the art that hardware circuitry that implements the logical method flows can be readily obtained by merely slightly programming the method flows with several hardware description languages into an integrated circuit.
The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic for the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller as pure computer readable program code, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may thus be considered a hardware component, and the means included therein for performing the various functions may also be considered as a structure within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.
The apparatuses, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or implemented by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a vehicle-mounted human-computer interaction device, a cellular telephone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
Although the present application provides method steps as described in an embodiment or flowchart, more or fewer steps may be included based on conventional or non-inventive means. The order of steps recited in the embodiments is merely one manner of performing the steps in a multitude of orders and does not represent the only order of execution. When an actual apparatus or end product executes, it may execute sequentially or in parallel (e.g., parallel processors or multi-threaded environments, or even distributed data processing environments) according to the method shown in the embodiment or the figures. The terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, the presence of additional identical or equivalent elements in a process, method, article, or apparatus that comprises the recited elements is not excluded.
For convenience of description, the above devices are described as being divided into various modules by functions, and are described separately. Of course, in implementing the present application, the functions of each module may be implemented in one or more software and/or hardware, or a module implementing the same function may be implemented by a combination of a plurality of sub-modules or sub-units, and the like. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
Those skilled in the art will also appreciate that, in addition to implementing the controller as pure computer readable program code, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may therefore be considered as a hardware component, and the means included therein for performing the various functions may also be considered as a structure within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (10)
1. A prediction method for social topic group behaviors is characterized by comprising the following steps:
s1, constructing a confrontation generation network by the topic data generator and the topic data discriminator, and performing data enhancement on the topic data;
s2, calculating the probability of the topic node migrating to the next topic node according to the relationship between the topic attribute and the topic data tendency of the data-enhanced topic data by adopting a node migration strategy, and forming a topic sequence after migration is completed;
s3, taking the maximum probability entropy as a target, and extracting a low-dimensional vector of the topic sequence finished by the migration;
s4, mapping the text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism, and extracting text characteristic factors influencing group behaviors;
s5, inputting low-dimensional vectors and text characteristic factors of the topic sequence, and predicting whether potential topic node group users in the next time period can participate in the propagation of the hot topic by adopting a convolutional neural network.
2. The method for predicting social topic group behavior as claimed in claim 1, wherein the S1 includes randomly sampling an original topic sequence, and inputting the result of the random sampling into a topic data generator to generate topic data; verifying the probability that the topic data is real topic data by using a topic data discriminator; and the judgment capability of the topic data discriminator is maximized, the probability of the topic data generator being judged as the generated data is minimized, the topic data generator and the topic discriminator are iterated repeatedly until the topic data output by the topic data generator is similar to the real topic data according to the rules of local optimization and global optimization.
3. The method for predicting the social topic group behavior as claimed in claim 2, wherein the original topic sequence comprises a user relationship network in a social network, user historical behavior information, user text information and user basic attributes.
4. The method for predicting social topic group behavior as claimed in claim 1, wherein the formula for calculating the probability of the topic node walking to the next topic node by using the node walking strategy is represented as:
wherein, P (r | c)i,ci-1) Represents the previous topic node ci-1Based on the current topic node ciProbability of walking to the next topic node r; a isp,q(ci-1And r) represents topic node ci-1A weight adjustment parameter with the topic node r; beta (c)iAnd r) represents the current topic node ciSimilarity of attributes with the next topic node r; gamma (c)iAnd r) represents the current topic node ciThe edge weight of the topic network transmitted to the next topic node r.
5. The method for predicting social topic group behavior according to claim 4, wherein the formula for calculating the weight adjustment parameter is represented as:
wherein,
represents the previous topic node ci-1The shortest path between the topic node r and the topic node r has a value range of {0,1,2 }; p denotes a return parameter, i.e., the probability of repeatedly walking to the previous topic node, and q denotes an access parameter, i.e., the walking characteristics of the topic node.
6. The method for predicting social topic group behavior as claimed in claim 4, wherein the calculation formula of the edge weight of the topic network is represented as:
wherein, intract (c)iAnd r) represents topic node ciThe degree of interaction with the topic node r,
indicates whether the topic node r focuses on the topic node ci，ActkbRepresenting topic node r processing topic node c based on behavior biThe kth microblog, t represents the time of the current hot topic, tkRepresents user ciThe time of issuing the kth microblog, K represents the user ciAnd (4) total number of issued microblogs.
7. The method for predicting social topic group behavior as claimed in claim 1, wherein the extracting the low-dimensional vector of the walk-completed topic sequence with the maximized probability entropy as an objective function comprises expressing with the maximized probability entropy as:
wherein, Pr (n)jL f (w)) the appearance of a neighborhood node n in a low-dimensional vector f (w) representing a topic node wjNs (w) represents the set of neighborhood nodes of topic node w.
8. The method for predicting social topic group behavior according to claim 1, wherein the mapping text information of the data-enhanced topic data to a low-dimensional vector space by using a fusion attention mechanism, and extracting text characteristic factors affecting group behavior comprises performing vector representation on each piece of text of a user by using a doc2vec algorithm; a hierarchical attention mechanism is adopted to endow different weights to the text vector and the word vector in the text; and selectively selecting the user interest feature vector.
9. A prediction apparatus of social topic group behavior, characterized in that the prediction apparatus comprises:
a topic data generator for generating topic data;
the topic data discriminator is used for discriminating the probability that the generated topic data is real data or false data and forming a countermeasure generation network with the topic data generator;
the topic sequence generator is used for calculating the probability of the topic node migrating to the next topic node by adopting a node migration strategy, and a topic sequence is formed after migration is completed;
the topic sequence dimensionality reduction module is used for extracting a low-dimensional vector of the walk-completed topic sequence by taking the maximum probability entropy as a target;
the text feature extraction module is used for mapping the text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism and extracting text feature factors influencing group behaviors;
and the convolutional neural network module is used for inputting the low-dimensional vector and the text characteristic factor of the topic sequence and predicting whether the potential topic node group user outputting the next time period participates in the propagation of the hot topic.
10. A computer readable storage medium having computer instructions stored thereon which when executed perform the steps of:
constructing a countermeasure generation network by the topic data generator and the topic data discriminator, and performing data enhancement on the topic data;
calculating the probability of the topic node migrating to the next topic node by adopting a node migration strategy according to the relationship between the topic attribute and the topic data tendency of the data-enhanced topic data, and forming a topic sequence after migration is completed;
extracting a low-dimensional vector of a topic sequence finished by migration by taking the maximum probability entropy as a target;
mapping text information of the topic data after data enhancement to a low-dimensional vector space by adopting a fusion attention mechanism, and extracting text characteristic factors influencing group behaviors;
and inputting low-dimensional vectors and text characteristic factors of the topic sequence, and predicting whether potential topic node group users in the next time period can participate in the propagation of the hot topic by adopting a convolutional neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011325005.3A CN112418525B (en) | 2020-11-24 | 2020-11-24 | Method and device for predicting social topic group behaviors and computer storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011325005.3A CN112418525B (en) | 2020-11-24 | 2020-11-24 | Method and device for predicting social topic group behaviors and computer storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112418525A true CN112418525A (en) | 2021-02-26 |
| CN112418525B CN112418525B (en) | 2022-06-10 |
Family
ID=74777971
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011325005.3A Active CN112418525B (en) | 2020-11-24 | 2020-11-24 | Method and device for predicting social topic group behaviors and computer storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112418525B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113051928A (en) * | 2021-03-17 | 2021-06-29 | 卓尔智联(武汉)研究院有限公司 | Detection comment method and device based on block chain and electronic equipment |
| CN113256024A (en) * | 2021-06-17 | 2021-08-13 | 上海大学 | User behavior prediction method fusing group behaviors |
| CN113592058A (en) * | 2021-07-05 | 2021-11-02 | 西安邮电大学 | Method for quantitatively predicting microblog forwarding breadth and depth |
| CN113946708A (en) * | 2021-10-29 | 2022-01-18 | 重庆邮电大学 | Topic propagation prediction method based on image restoration technology and rumor splitting information |
| CN115878907A (en) * | 2022-12-28 | 2023-03-31 | 人民网股份有限公司 | Social network forwarding behavior prediction method and device based on user dependency relationship |
| CN117454143A (en) * | 2023-09-13 | 2024-01-26 | 广州大学 | Group behavior prediction method based on topic representation and multidimensional feature embedding |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106651030A (en) * | 2016-12-21 | 2017-05-10 | 重庆邮电大学 | Method for predicting user participation behavior of hot topic by improved RBF neural network |
| CN108460019A (en) * | 2018-02-28 | 2018-08-28 | 福州大学 | A kind of emerging much-talked-about topic detecting system based on attention mechanism |
| CN109783798A (en) * | 2018-12-12 | 2019-05-21 | 平安科技（深圳）有限公司 | Method, apparatus, terminal and the storage medium of text information addition picture |
| CN110807556A (en) * | 2019-11-05 | 2020-02-18 | 重庆邮电大学 | Method and device for predicting propagation trend of microblog rumors or/and dagger rumors |
| CN110825980A (en) * | 2019-11-05 | 2020-02-21 | 重庆邮电大学 | Microblog topic pushing method based on countermeasure generation network |
| US20200110623A1 (en) * | 2018-10-04 | 2020-04-09 | Microsoft Technology Licensing, Llc | User-centric contextual information for browser |
| US20200118315A1 (en) * | 2018-10-15 | 2020-04-16 | Shutterstock, Inc. | Image editor for merging images with generative adversarial networks |
-
2020
- 2020-11-24 CN CN202011325005.3A patent/CN112418525B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106651030A (en) * | 2016-12-21 | 2017-05-10 | 重庆邮电大学 | Method for predicting user participation behavior of hot topic by improved RBF neural network |
| CN108460019A (en) * | 2018-02-28 | 2018-08-28 | 福州大学 | A kind of emerging much-talked-about topic detecting system based on attention mechanism |
| US20200110623A1 (en) * | 2018-10-04 | 2020-04-09 | Microsoft Technology Licensing, Llc | User-centric contextual information for browser |
| US20200118315A1 (en) * | 2018-10-15 | 2020-04-16 | Shutterstock, Inc. | Image editor for merging images with generative adversarial networks |
| CN109783798A (en) * | 2018-12-12 | 2019-05-21 | 平安科技（深圳）有限公司 | Method, apparatus, terminal and the storage medium of text information addition picture |
| CN110807556A (en) * | 2019-11-05 | 2020-02-18 | 重庆邮电大学 | Method and device for predicting propagation trend of microblog rumors or/and dagger rumors |
| CN110825980A (en) * | 2019-11-05 | 2020-02-21 | 重庆邮电大学 | Microblog topic pushing method based on countermeasure generation network |
Non-Patent Citations (5)
| Title |
|---|
| ADITYA GROVER: "node2vec: Scalable Feature Learning for Networks", 《PROCEEDINGS OF THE 22ND ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MININGAUGUST 2016》 * |
| ADITYA GROVER: "node2vec: Scalable Feature Learning for Networks", 《PROCEEDINGS OF THE 22ND ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MININGAUGUST 2016》, 13 August 2016 (2016-08-13), pages 855 - 863 * |
| YUNPENG XIAO: "A Rumor & Anti-rumor Propagation Model Based on Data Enhancement and Evolutionary Game", 《IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING》 * |
| YUNPENG XIAO: "A Rumor & Anti-rumor Propagation Model Based on Data Enhancement and Evolutionary Game", 《IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING》, 27 October 2020 (2020-10-27), pages 1 - 13 * |
| 于海等: "基于在线社交网络事件库多因素耦合的流行度预测方法", 《天津大学学报(自然科学与工程技术版)》, no. 12, 22 October 2020 (2020-10-22), pages 1272 - 1280 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113051928A (en) * | 2021-03-17 | 2021-06-29 | 卓尔智联(武汉)研究院有限公司 | Detection comment method and device based on block chain and electronic equipment |
| CN113051928B (en) * | 2021-03-17 | 2023-08-01 | 卓尔智联(武汉)研究院有限公司 | Block chain-based comment detection method and device and electronic equipment |
| CN113256024A (en) * | 2021-06-17 | 2021-08-13 | 上海大学 | User behavior prediction method fusing group behaviors |
| CN113592058A (en) * | 2021-07-05 | 2021-11-02 | 西安邮电大学 | Method for quantitatively predicting microblog forwarding breadth and depth |
| CN113592058B (en) * | 2021-07-05 | 2024-03-12 | 西安邮电大学 | Method for quantitatively predicting microblog forwarding breadth and depth |
| CN113946708A (en) * | 2021-10-29 | 2022-01-18 | 重庆邮电大学 | Topic propagation prediction method based on image restoration technology and rumor splitting information |
| CN115878907A (en) * | 2022-12-28 | 2023-03-31 | 人民网股份有限公司 | Social network forwarding behavior prediction method and device based on user dependency relationship |
| CN115878907B (en) * | 2022-12-28 | 2023-08-11 | 人民网股份有限公司 | Social network forwarding behavior prediction method and device based on user dependency relationship |
| CN117454143A (en) * | 2023-09-13 | 2024-01-26 | 广州大学 | Group behavior prediction method based on topic representation and multidimensional feature embedding |
| CN117454143B (en) * | 2023-09-13 | 2024-08-02 | 广州大学 | Group behavior prediction method based on topic representation and multidimensional feature embedding |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112418525B (en) | 2022-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112418525B (en) | Method and device for predicting social topic group behaviors and computer storage medium | |
| CN111931062B (en) | Training method and related device of information recommendation model | |
| Lin et al. | A survey on reinforcement learning for recommender systems | |
| Lu et al. | GLR: A graph-based latent representation model for successive POI recommendation | |
| CN113536144B (en) | Social network information propagation scale prediction method and device | |
| CN113590976A (en) | Recommendation method of space self-adaptive graph convolution network | |
| Huang et al. | Information fusion oriented heterogeneous social network for friend recommendation via community detection | |
| Wu et al. | Temporal interaction and causal influence in community-based question answering | |
| CN112699310A (en) | Cold start cross-domain hybrid recommendation method and system based on deep neural network | |
| CN118013135A (en) | Graph comparison learning recommendation method based on relationship graph convolution neural network | |
| Zhang et al. | DeepBlue: Bi-layered LSTM for tweet popUlarity Estimation | |
| Tilahun et al. | Vector optimisation using fuzzy preference in evolutionary strategy based firefly algorithm | |
| Abolghasemi et al. | A graph neural approach for group recommendation system based on pairwise preferences | |
| CN116628345B (en) | Content recommendation method and device, electronic equipment and storage medium | |
| Liao et al. | An integrated model based on deep multimodal and rank learning for point-of-interest recommendation | |
| Qi et al. | Graph neural network social recommendation algorithm integrating static and dynamic features | |
| Sangeetha et al. | Predicting personalized recommendations using GNN | |
| Sang | User-centric social multimedia computing | |
| Liu et al. | ICE: Information credibility evaluation on social media via representation learning | |
| Li et al. | ANN: a heuristic search algorithm based on artificial neural networks | |
| Yan et al. | Tackling the achilles heel of social networks: Influence propagation based language model smoothing | |
| Lazzez et al. | DeepVisInterests: CNN-ontology prediction of users interests from social images | |
| CN114358364A (en) | Attention mechanism-based short video frequency click rate big data estimation method | |
| Das et al. | Soft-Computing based recommendation system: a comparative study | |
| Miyake et al. | NETEVOLVE: Social Network Forecasting using Multi-Agent Reinforcement Learning with Interpretable Features |
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 |