WO2019095570A1 - 预测事件流行度方法、服务器及计算机可读存储介质 - Google Patents

预测事件流行度方法、服务器及计算机可读存储介质 Download PDF

Info

Publication number
WO2019095570A1
WO2019095570A1 PCT/CN2018/076135 CN2018076135W WO2019095570A1 WO 2019095570 A1 WO2019095570 A1 WO 2019095570A1 CN 2018076135 W CN2018076135 W CN 2018076135W WO 2019095570 A1 WO2019095570 A1 WO 2019095570A1
Authority
WO
WIPO (PCT)
Prior art keywords
event
node
sequence
vector
neural network
Prior art date
Application number
PCT/CN2018/076135
Other languages
English (en)
French (fr)
Inventor
王健宗
吴天博
黄章成
肖京
Original Assignee
平安科技(深圳)有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2019095570A1 publication Critical patent/WO2019095570A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION 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
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION 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
    • G06Q50/00Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/01Social networking

Definitions

  • the present application relates to the field of Internet technologies, and in particular, to a method for predicting event popularity, a server, and a computer readable storage medium.
  • social media has gradually become an indispensable part of people's lives, and has become the main channel for information dissemination in this era.
  • social media is also an important way to spread publicity.
  • predicting the future popularity of this event is of great significance. Aiming at the structure of complex networks and the pattern of information diffusion, an end-to-end social media event heat prediction method is proposed. This method predicts the future popularity of events and provides scientific information support for decision makers by learning the information dissemination mode.
  • the present application proposes a method for predicting event popularity, a server, and a computer readable storage medium to solve the problem.
  • the present application provides a method for predicting event popularity, the method comprising the steps of:
  • the degree of influence of the event is obtained by the bicyclic neural network model output prediction based on the gated loop unit.
  • the present application further provides a server, including a memory, a processor, and a predictive event popularity system executable on the processor, where the predicted event popularity system is stored
  • the processor implements the following steps when executed:
  • the degree of influence of the event is obtained by the bicyclic neural network model output prediction based on the gated loop unit.
  • the present application further provides a computer readable storage medium storing a predicted event popularity system, the predicted event popularity system being executable by at least one processor, The step of causing the at least one processor to perform the predictive event popularity method as described above.
  • the predictive event popularity method, the server, and the computer readable storage medium proposed by the present application firstly abstract the user relationship structure of the social networking site into a node graph; and then, acquire the social networking website at a certain moment.
  • An event, and sampling a sequence that the event may propagate on the node map; further, establishing a dual-loop neural network model based on a gated loop unit; and finally, inputting the sequence of samples to the gate-based A two-cycle neural network model of the control loop unit, predictively obtains the influence degree of the event through the output of the double-loop neural network model based on the gated loop unit, thereby predicting the future popularity of the event and providing scientific information support for the decision maker .
  • 1 is a schematic diagram of an optional hardware architecture of the server of the present application.
  • FIG. 2 is a schematic diagram of a program module of a first embodiment of a predictive event popularity system of the present application
  • FIG. 3 is a schematic diagram of a node of a predictive event popularity system of the present application.
  • FIG. 4 is a schematic diagram of a model of a predictive event popularity system of the present application.
  • FIG. 5 is a schematic diagram of a program module of a second embodiment of a predictive event popularity system of the present application.
  • FIG. 6 is a schematic diagram of a program module of a third embodiment of the predictive event popularity system of the present application.
  • FIG. 7 is a schematic flowchart of a first embodiment of a method for predicting event popularity according to the present application.
  • FIG. 8 is a schematic flowchart of a second embodiment of a method for predicting event popularity according to the present application.
  • FIG. 9 is a schematic flowchart diagram of a third embodiment of a method for predicting event popularity according to the present application.
  • FIG. 1 it is a schematic diagram of an optional hardware architecture of the server 2 of the present application.
  • the server 2 may include, but is not limited to, the memory 11, the processor 12, and the network interface 13 being communicably connected to each other through a system bus. It is pointed out that Figure 1 only shows the server 2 with the components 11-13, but it should be understood that not all illustrated components may be implemented, and more or fewer components may be implemented instead.
  • the server 2 may be a computing device such as a rack server, a blade server, a tower server, or a rack server.
  • the server 2 may be an independent server or a server cluster composed of multiple servers.
  • the memory 11 includes at least one type of readable storage medium including a flash memory, a hard disk, a multimedia card, a card type memory (eg, SD or DX memory, etc.), a random access memory (RAM), a static Random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory, magnetic disk, optical disk, and the like.
  • the memory 11 may be an internal storage unit of the server 2, such as a hard disk or memory of the server 2.
  • the memory 11 may also be an external storage device of the server 2, such as a plug-in hard disk equipped on the server 2, a smart memory card (SMC), and a secure digital (Secure) Digital, SD) cards, flash cards, etc.
  • the memory 11 can also include both the internal storage unit of the server 2 and its external storage device.
  • the memory 11 is generally used to store an operating system installed on the server 2 and various types of application software, such as program code for predicting the event popularity system 200. Further, the memory 11 can also be used to temporarily store various types of data that have been output or are to be output.
  • the processor 12 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments.
  • the processor 12 is typically used to control the overall operation of the server 2.
  • the processor 12 is configured to run program code or process data stored in the memory 11, such as running the predicted event popularity system 200 and the like.
  • the network interface 13 may comprise a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the server 2 and other electronic devices.
  • the present application proposes a predictive event popularity system 200.
  • FIG. 2 it is a program module diagram of the first embodiment of the predictive event popularity system 200 of the present application.
  • the predictive event popularity system 200 includes a series of computer program instructions stored in the memory 11, and when the computer program instructions are executed by the processor 12, the predicted events of the embodiments of the present application may be implemented. Degree operation.
  • the predictive event popularity system 200 can be divided into one or more modules based on the particular operations implemented by the various portions of the computer program instructions. For example, in FIG. 2, the predicted event popularity system 200 can be segmented into an abstraction module 201, a sampling module 202, a construction module 203, an input module 204, and a prediction module 205. among them:
  • the abstracting module 201 is configured to abstract a user relationship structure of the social networking site into a node graph
  • FIG. 3 is a schematic diagram of a node of the first embodiment of the predictive event popularity system 200 of the present application.
  • ABCDEF represents the relationship between different users and arrow users, for example, A focuses on B and D, B focuses on E and C, C focuses on D and F, and D focuses on F, so, different users The relationship between the two is clearly listed.
  • the sampling module 202 is configured to acquire an event at a certain moment, and sample the sequence that the time may propagate on the node graph.
  • random walks are used to sample sequences that the event may propagate.
  • the node being accessed is v
  • the probability of moving to the neighbor node N(v) of v is:
  • is the smoothing factor and sc is the evaluation function, which can be defined as the degree of u or the weight of the edge uv.
  • different node transition sequences can be sampled using the above probabilities.
  • the sampling is terminated when the sequence length reaches the preset value T.
  • the start of the arrow to the end of the arrow is a sequence, for example, For a sequence.
  • the building module 203 is configured to establish a dual-loop neural network model based on a gated loop unit.
  • the input module 204 is configured to input the sequence of the samples to the dual-loop neural network model based on the gated loop unit.
  • the input module 204 is based on the sequence sampled by the sampling module 202 as an input of a dual-loop neural network model based on a gated loop unit. If the sampling module 204 samples from the node graph to K sequences, each sequence has a length T, where T is a variable, that is, the length of each sequence may be different, the input module 204 will have K lengths of T. The sequence is used as an input to a two-loop neural network model based on a gated loop unit.
  • the prediction module 205 is configured to output a prediction result according to the input sequence by the dual loop neural network model based on the gated loop unit.
  • the prediction result is a sequence vector.
  • the prediction module 204 inputs the sequence of samples into the dual-loop neural network model based on the gated loop unit, and sets a prediction target or a prediction condition, such as when an event needs to be predicted at any time, such as an event at time t0.
  • the target to be predicted is the set of nodes Vt affected by the event within the time t after the time t0 at which the event begins to spread.
  • the time t indicates the time period in which the event no longer changes (the event propagation fluctuation is less than a preset value, and the event is no longer changed).
  • the server 2 also converts the output of the bi-loop neural network model based on the gating cycle unit into a representation transformed into a vector graph, which may be represented by the following vector:
  • g c is a subgraph affected by the event
  • k is the sequence number
  • i is the node number
  • B is the number of blocks of the mini-batch.
  • MLP stands for Multilayer Perceptron, which is a forward-structured artificial neural network that maps a set of input vectors to a set of output vectors.
  • An MLP can be thought of as a directed graph consisting of multiple node layers, each connected to the next layer.
  • each node is a neuron (or processing unit) with a nonlinear activation function.
  • FIG. 4 is a schematic diagram of a model of the predicted event popularity system 200 of the present application.
  • the social network structure is first abstracted into a node graph; then different node transition sequences are sampled, as shown by K sequences of length T; then the sequence is converted into a vector as an already established gate-based control.
  • An input of a two-cycle neural network model of the cyclic unit further, converting the sequence vector output by the two-loop neural network model based on the gated loop unit into a map; and then, by using an attention mechanism, the sequence of the graph
  • the vector is converted into a vector graph; finally, the final prediction result is output through the fully connected layer.
  • the predicted event popularity system 200 includes a mapping module, a sampling module 202, a building module 203, an input module 204, and a prediction module 204, which are included in the first embodiment, and includes a mapping. Module 206.
  • the mapping module 206 is configured to map each node of the node graph into a vector.
  • the server 2 maps each node of the sequence into a vector by the mapping module 205.
  • the mapping module 205 maps each of the K sequences of length T into a vector.
  • the mapping module 205 maps each node of the sequence to a vector using Node2Vec.
  • the mapping module 206 is further configured to map the text content of the event propagated by each node into a vector.
  • the mapping module 206 maps the text content propagated by each node into a vector using Word2Vec.
  • the input module 204 is further configured to connect the vector into which the each node is mapped and the vector into which the text content of each node is mapped into a sequence vector as an input of the bidirectional cyclic neural network model.
  • the vector mapped to each node and the vector into which the text content of each node is mapped are connected into a sequence vector, so that the node and the text content propagated by the node are associated with each other, so that the event popularity is The forecast is accurate.
  • the predicted event popularity system 200 further includes a setting module 207 on the basis of the second embodiment.
  • the setting module 207 is configured to set a preset time.
  • the prediction module 205 is further configured to output a heat prediction result regarding the event when the preset time is reached.
  • the server 2 can set a preset time by the setting module 207 to predict the popularity (degree of influence) of the event within a certain period of time after the event occurs.
  • the setting module 207 sets a time interval t0-t, t0 indicates the time when the event occurs, and t indicates the cut-off time of the set time interval. When the time t is reached, the final prediction result is output, and the number of people affected by the event at any time t is predicted.
  • the setting module 207 is further configured to set a preset person value.
  • the prediction module 205 is further configured to output a heat prediction result about the event when the number of people affected by the event reaches the preset number of people.
  • the server 2 may further set a preset person value by using the setting module 207 to predict how long the event is affected or the event is concerned after the event occurs. The number of people reaches the preset person value. After the number of people who have received the influence reaches the preset person value, the final prediction result is output, and the prediction is reached to the value of any person, and how long it takes.
  • the present application also proposes a method for predicting event popularity.
  • FIG. 7 it is a schematic flowchart of a first embodiment of a method for predicting event popularity according to the present application.
  • the order of execution of the steps in the flowchart shown in FIG. 7 may be changed according to different requirements, and some steps may be omitted.
  • Step S301 abstracting a user relationship structure of the social networking site into a node graph
  • ABCDEF represents the relationship between different users and arrow users, for example, A focuses on B and D, B focuses on E and C, C focuses on D and F, and D focuses on F, so, different users The relationship between the two is clearly listed.
  • Step S302 Acquire an event at a certain moment, and sample the sequence that the time may propagate on the node graph.
  • random walks are used to sample sequences that the event may propagate.
  • the node being accessed is v
  • the probability of moving to the neighbor node N(v) of v is:
  • is the smoothing factor and sc is the evaluation function, which can be defined as the degree of u or the weight of the edge uv.
  • different node transition sequences can be sampled using the above probabilities.
  • the sampling is terminated when the sequence length reaches the preset value T.
  • the start of the arrow to the end of the arrow is a sequence, for example, For a sequence.
  • Step S303 establishing a dual-loop neural network model based on a gated loop unit
  • Step S504 input the sequence of the samples into the dual-loop neural network model based on the gated loop unit.
  • the server 2 takes the sampled sequence as an input to a two-loop neural network model based on a gated loop unit. If the server 2 samples from the node graph to K sequences, each sequence has a length T, where T is a variable, that is, the length of each sequence may be different, the server 2 will have K sequences of length T. As an input to a two-loop neural network model based on a gated loop unit.
  • Step S305 outputting a prediction result according to the input sequence by the double loop neural network model based on the gated loop unit.
  • the prediction result is a sequence vector.
  • the server 2 inputs the sequence of the samples into the dual-loop neural network model based on the gated loop unit, and sets a prediction target or a prediction condition, such as when it is necessary to predict an arbitrary time, such as the event at time t0 affects the number of people That is, the node set Vt1 affected by the event at the time t1 is predicted.
  • the target to be predicted is the set of nodes Vt affected by the event within the time t after the time t0 at which the event begins to spread.
  • the time t indicates a time period in which the event no longer changes (the event propagation fluctuation is less than a preset value, and the event is no longer changed).
  • the server 2 also converts the output of the bi-loop neural network model based on the gating cycle unit into a representation transformed into a vector graph, which may be represented by the following vector:
  • g c is a subgraph affected by the event
  • k is the sequence number
  • i is the node number
  • B is the number of blocks of the mini-batch.
  • MLP stands for Multilayer Perceptron, which is a forward-structured artificial neural network that maps a set of input vectors to a set of output vectors.
  • An MLP can be thought of as a directed graph consisting of multiple node layers, each connected to the next layer.
  • each node is a neuron (or processing unit) with a nonlinear activation function.
  • FIG. 8 is a schematic flowchart diagram of a second embodiment of a method for predicting event popularity according to the present application.
  • the step of inputting the sequence of sampling to the dual-loop neural network model based on the gating cycle unit specifically includes the following steps:
  • Step S401 mapping each node of the node graph into a vector.
  • the server 2 maps each node of the sequence into a vector. Assuming that K sequences of length T are sampled, the server 2 maps each of the K sequences of length T into a vector. In this embodiment, the server 2 maps each node of the sequence into a vector using Node2Vec.
  • Step S402 mapping the text content of the event propagated by each node into a vector.
  • the server 2 maps the text content propagated by each node into a vector using Word2Vec.
  • Step S403 connecting the vector into which each node is mapped and the vector into which the text content propagated by each node is mapped into a sequence vector as an input of the bidirectional cyclic neural network model.
  • the vector mapped to each node and the vector into which the text content of each node is mapped are connected into a sequence vector, so that the node and the text content propagated by the node are associated with each other, so that the event popularity is The forecast is accurate.
  • FIG. 9 it is a schematic flowchart of a third embodiment of the method for predicting event popularity in the present application.
  • the method for predicting event popularity of the present application further includes the following steps:
  • step S501 a preset time is set.
  • Step S502 when the preset time is reached, output a heat prediction result regarding the event.
  • the server 2 may set a preset time to predict the popularity (degree of influence) of the event within a certain period of time after the event occurs.
  • the setting module 207 sets a time interval t0-t, t0 indicates the time when the event occurs, and t indicates the cut-off time of the set time interval. When the time t is reached, the final prediction result is output, and the number of people affected by the event at any time t is predicted.
  • step S503 a preset person value is set.
  • Step S504 when the number of people affected by the event reaches the preset number of people, output a heat prediction result regarding the event.
  • the server 2 can also set a preset person value to achieve prediction. How long after the event occurs, the number of people affected by the event or the number of people concerned with the event reaches a preset value. After the number of people who have received the influence reaches the preset person value, the final prediction result is output, and the prediction is reached to the value of any person, and how long it takes.
  • the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be through hardware, but in many cases, the former is better.
  • Implementation Based on such understanding, the technical solution of the present application, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk,
  • the optical disc includes a number of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the methods described in various embodiments of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • Tourism & Hospitality (AREA)
  • General Health & Medical Sciences (AREA)
  • General Business, Economics & Management (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Molecular Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Primary Health Care (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请公开了一种预测事件流行度的方法,该方法包括:将社交网站的用户关系结构抽象成节点图;获取所述社交网站某个时刻事件,并在所述节点图上对所述事件可能传播的序列进行采样;建立基于门控循环单元的双循环神经网络模型;将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型;通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度。本申请还提供一种服务器及计算机可读存储介质。本申请提供的预测事件流行度方法、服务器及计算机可读存储介质能够预测事件未来的流行度,为决策者提供科学的信息支持。

Description

预测事件流行度方法、服务器及计算机可读存储介质
本申请要求于2017年11月17日提交中国专利局、申请号为201711141758.7、发明名称为“预测事件流行度方法、服务器及计算机可读存储介质”的中国专利申请的优先权,其全部内容通过引用结合在申请中。
技术领域
本申请涉及互联网技术领域,尤其涉及一种预测事件流行度方法、服务器及计算机可读存储介质。
背景技术
随着互联网以及移动终端的快速发展,社交媒体已经逐渐成为人们生活中不可缺少的一部分,也成为了这个时代信息传播的主要渠道。同时,社交媒体也是舆情传播的重要途径。在一个事件的早期发展阶段,预测这个事件将来的流行度,具有重大的意义。针对复杂网络的结构和信息扩散的模式,提出端到端的社交媒体事件热度预测方法,该方法通过对信息传播模式的学习,从而预测事件未来的流行度,为决策者提供科学的信息支持。
发明内容
有鉴于此,本申请提出一种预测事件流行度方法、服务器及计算机可读存储介质,以解决如何的问题。
首先,为实现上述目的,本申请提出一种预测事件流行度方法,该方法包括步骤:
将社交网站的用户关系结构抽象成节点图;
获取所述社交网站某个时刻事件,并在所述节点图上对所述事件可能传播的序列进行采样;
建立基于门控循环单元的双循环神经网络模型;
将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型;
通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度。
此外,为实现上述目的,本申请还提供一种服务器,包括存储器、处理器,所述存储器上存储有可在所述处理器上运行的预测事件流行度系统,所述预测事件流行度系统被所述处理器执行时实现如下步骤:
将社交网站的用户关系结构抽象成节点图;
获取所述社交网站某个时刻事件,并在所述节点图上对所述事件可能传播的序列进行采样;
建立基于门控循环单元的双循环神经网络模型;
将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型;
通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度。
进一步地,为实现上述目的,本申请还提供一种计算机可读存储介质,所述计算机可读存储介质存储有预测事件流行度系统,所述预测事件流行度系统可被至少一个处理器执行,以使所述至少一个处理器执行如上述的预测事件流行度方法的步骤。
相较于现有技术,本申请所提出的预测事件流行度方法、服务器及计算机可读存储介质,首先,将社交网站的用户关系结构抽象成节点图;接着,获取所述社交网站某个时刻事件,并在所述节点图上对所述事件可能传播的序列进行采样;进一步地,建立基于门控循环单元的双循环神经网络模型;最后,将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型,通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事 件的影响度,从而预测事件未来的流行度,为决策者提供科学的信息支持。
附图说明
图1是本申请服务器一可选的硬件架构的示意图;
图2是本申请预测事件流行度系统第一实施例的程序模块示意图;
图3是本申请预测事件流行度系统的节点示意图;
图4是本申请预测事件流行度系统的模型示意图;
图5是本申请预测事件流行度系统第二实施例的程序模块示意图;
图6是本申请预测事件流行度系统第三实施例的程序模块示意图;
图7是本申请预测事件流行度方法第一实施例的流程示意图;
图8是本申请预测事件流行度方法第二实施例的流程示意图;
图9是本申请预测事件流行度方法第三实施例的流程示意图。
本申请目的的实现、功能特点及优点将结合实施例,参照附图做进一步说明。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处所描述的具体实施例仅用以解释本申请,并不用于限定本申请。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
需要说明的是,在本申请中涉及“第一”、“第二”等的描述仅用于描述目的,而不能理解为指示或暗示其相对重要性或者隐含指明所指示的技术特征的数量。由此,限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。另外,各个实施例之间的技术方案可以相互结合,但是必须是以本领域普通技术人员能够实现为基础,当技术方案的结合出现相互矛 盾或无法实现时应当认为这种技术方案的结合不存在,也不在本申请要求的保护范围之内。
参阅图1所示,是本申请服务器2一可选的硬件架构的示意图。
本实施例中,所述服务器2可包括,但不仅限于,可通过系统总线相互通信连接存储器11、处理器12、网络接口13。需要指出的是,图1仅示出了具有组件11-13的服务器2,但是应理解的是,并不要求实施所有示出的组件,可以替代的实施更多或者更少的组件。
其中,所述服务器2可以是机架式服务器、刀片式服务器、塔式服务器或机柜式服务器等计算设备,该服务器2可以是独立的服务器,也可以是多个服务器所组成的服务器集群。
所述存储器11至少包括一种类型的可读存储介质,所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如,SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器(PROM)、磁性存储器、磁盘、光盘等。在一些实施例中,所述存储器11可以是所述服务器2的内部存储单元,例如该服务器2的硬盘或内存。在另一些实施例中,所述存储器11也可以是所述服务器2的外部存储设备,例如该服务器2上配备的插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)等。当然,所述存储器11还可以既包括所述服务器2的内部存储单元也包括其外部存储设备。本实施例中,所述存储器11通常用于存储安装于所述服务器2的操作系统和各类应用软件,例如预测事件流行度系统200的程序代码等。此外,所述存储器11还可以用于暂时地存储已经输出或者将要输出的各类数据。
所述处理器12在一些实施例中可以是中央处理器(Central Processing Unit,CPU)、控制器、微控制器、微处理器、或其他数据处理芯片。该处理器12通常用于控制所述服务器2的总体操作。本实施例中,所述处理器12用于 运行所述存储器11中存储的程序代码或者处理数据,例如运行所述的预测事件流行度系统200等。
所述网络接口13可包括无线网络接口或有线网络接口,该网络接口13通常用于在所述服务器2与其他电子设备之间建立通信连接。
至此,己经详细介绍了本申请相关设备的硬件结构和功能。下面,将基于上述介绍提出本申请的各个实施例。
首先,本申请提出一种预测事件流行度系统200。
参阅图2所示,是本申请预测事件流行度系统200第一实施例的程序模块图。
本实施例中,所述预测事件流行度系统200包括一系列的存储于存储器11上的计算机程序指令,当该计算机程序指令被处理器12执行时,可以实现本申请各实施例的预测事件流行度操作。在一些实施例中,基于该计算机程序指令各部分所实现的特定的操作,预测事件流行度系统200可以被划分为一个或多个模块。例如,在图2中,所述预测事件流行度系统200可以被分割成抽象模块201、采样模块202、构建模块203、输入模块204及预测模块205。其中:
所述抽象模块201,用于将社交网站的用户关系结构抽象成节点图;
具体地,所述服务器2通过所述抽象模块201将社交网络结构,例如微博,的用户关系结构抽象为节点图G=(V,E),图的节点V代表用户,边E代表用户之间的联系。
请一并参阅附图3,附图3是本申请预测事件流行度系统200第一实施例的节点示意图。如图3所示,ABCDEF分别表示不同的用户,箭头用户之间的关系,比如,A关注B和D,B关注E和C,C关注D和F,D关注F,如此,将不同用户之间的关系清楚的罗列。
所述采样模块202,用于获取某个时刻的事件,在所述节点图上对所述时间可能传播的序列进行采样。
在本实施例中,采用随机游走对事件可能传播的序列进行采样。在随机 游走的过程中,正在访问的节点为v,接下来转移到v的邻居节点N(v)的概率为:
Figure PCTCN2018076135-appb-000001
其中,α为平滑因子,sc为评价函数,可定义为u的出度或者边uv的权重。
根据马尔科夫性质,使用以上概率,可以采样出不同的节点转移序列。采样在序列长度到达预设值T时终止。
参照图3所示,从箭头的开始到箭头的截止为一个序列,例如,
Figure PCTCN2018076135-appb-000002
为一个序列。
所述构建模块203,用于建立基于门控循环单元的双循环神经网络模型。
所述输入模块204,用于将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型。
具体地,所述输入模块204根据所述采样模块202采样到的序列,作为基于门控循环单元的双循环神经网络模型的输入。若所述采样模块204从节点图采样到K个序列,每个序列的长度为T,其中T为变量,即每个序列的长度可能不一样,则所述输入模块204将K个长度为T的序列作为基于门控循环单元的双循环神经网络模型的输入。
所述预测模块205,用于通过所述基于门控循环单元的双循环神经网络模型根据所述输入序列,输出预测结果。
在本实施方式中,所述预测结果为序列向量。所述预测模块204将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型,并设定预测目标或预测条件,如当需要预测任意时刻,如时刻t0的事件影响到人数,即预测时刻t1的事件影响到的节点集合Vt1。当需要预测事件最终的影响到的人数,即要预测的目标是在事件开始蔓延的时间t0后的时间t内,事件所影响到的节点集合Vt。其中,时间t表示事件不再有变化(事件传播波动小 于一预设值即可认为事件不再有变化)的时间段。
具体地,所述服务器2还将所述基于门控循环单元的双循环神经网络模型的输出转换成转化成向量图的表示,所述向量图可以用以下向量表示:
Figure PCTCN2018076135-appb-000003
其中,g c是被事件影响的子图,k为序列序号,i为节点序号,B为mini-batch的分块数,
Figure PCTCN2018076135-appb-000004
是序列的编码,a c和λ i可以在深度学习的过程中学习得到。最后使用全连接层输出最终预测结果。其中,使用全连接层输出的最终预测结果表示为:
f(g c)=MLP(h(g c))
其中,MLP代表多层感知机(Multilayer Perceptron),是一种前向结构的人工神经网络,映射一组输入向量到一组输出向量。MLP可以被看作是一个有向图,由多个的节点层所组成,每一层都全连接到下一层。除了输入节点,每个节点都是一个带有非线性激活函数的神经元(或称处理单元)。
请一并参阅附图4,附图,4为本申请预测事件流行度系统200的模型示意图。如图4所示,首先将社交网络结构抽象成节点图;进而采样出不同的节点转移序列,如图为K个长度为T的序列;接着将序列转换成向量作为已经建立好的基于门控循环单元的双循环神经网络模型的输入;进一步地,将所述基于门控循环单元的双循环神经网络模型输出的序列向量转换成图;然后,通过使用注意力(attention)机制将图的序列向量转化成向量图;最后通过全连接层输出最终预测结果。
参阅图5所示,是本申请预测事件流行度系统200第二实施例的程序模块图。本实施例中,所述的预测事件流行度系统200除了包括第一实施例中的所述的抽象模块201、采样模块202、构建模块203、输入模块204及预测模块204之外,还包括映射模块206。
所述映射模块206,用于将所述节点图的每个节点映射成向量。
具体地,所述服务器2通过所述映射模块205将所述序列的每个节点映射成向量。假设所述采样模块202采样了K个长度为T的序列,则所述映射模块205将所述K个长度为T的序列中的每个节点映射成向量。在本实施例中,所述映射模块205使用Node2Vec将所述序列的每个节点映射成向量。
所述映射模块206,还用于将每个节点传播的关于所述事件的文本内容映射成向量。
具体地,在本实施例中,所述映射模块206使用使用Word2Vec将所述每个节点传播的文本内容映射成向量。
所述输入模块204,还用于将所述每个节点映射成的向量以及所述每个节点传播的文本内容映射成的向量连接成序列向量作为双向循环神经网络模型的输入。
具体地,所述每个节点映射成的向量以及所述每个节点传播的文本内容映射成的向量连接成序列向量,如此,节点与节点传播的文本内容相互关联映射,使得对于事件流行度的预测跟准确。
参阅图6所示,是本申请预测事件流行度系统200第三实施例的程序模块图。本实施例中,所述的预测事件流行度系统200在第二实施例的基础上,还包括设定模块207。
所述设定模块207,用于设定一预设时间。
所述预测模块205,还用于当到达所述预设时间时,输出关于所述事件的热度预测结果。
具体地,所述服务器2可以通过设定模块207设定一预设时间,以实现预测当事件发生后,在一定时间内,该事件的流行度(影响度)。如所述设定模块207设定一时间区间t0-t,t0表示事件发生的时间,t表示设定的时间区间的截止时间。当到达时间t后,输出最终预测结果,实现预测经过任意时间t的事件影响人数。
所述设定模块207,还用于设定一预设人数值。
所述预测模块205,还用于当所述事件的影响人数达到所述预设人数时,输出关于所述事件的热度预测结果。
具体地,所述服务器2还可以通过所述设定模块207设定一预设人数值,以实现预测当事件发生后,在多长时间内,该受到事件的影响人数或是关注该事件的人数达到预设人数值。当收到影响的人数达到预设人数值之后,输出最终预测结果,实现了预测达到任意人数值,需要经过多长时间。
此外,本申请还提出一种预测事件流行度方法。
参阅图7所示,是本申请预测事件流行度方法第一实施例的流程示意图。在本实施例中,根据不同的需求,图7所示的流程图中的步骤的执行顺序可以改变,某些步骤可以省略。
步骤S301,将社交网站的用户关系结构抽象成节点图;
具体地,所述服务器2将社交网络结构,例如微博,的用户关系结构抽象为节点图G=(V,E),图的节点V代表用户,边E代表用户之间的联系。
如图3所示,ABCDEF分别表示不同的用户,箭头用户之间的关系,比如,A关注B和D,B关注E和C,C关注D和F,D关注F,如此,将不同用户之间的关系清楚的罗列。
步骤S302,获取某个时刻的事件,在所述节点图上对所述时间可能传播的序列进行采样。
在本实施例中,采用随机游走对事件可能传播的序列进行采样。在随机游走的过程中,正在访问的节点为v,接下来转移到v的邻居节点N(v)的概率为:
Figure PCTCN2018076135-appb-000005
其中,α为平滑因子,sc为评价函数,可定义为u的出度或者边uv的权重。
根据马尔科夫性质,使用以上概率,可以采样出不同的节点转移序列。 采样在序列长度到达预设值T时终止。
参照图3所示,从箭头的开始到箭头的截止为一个序列,例如,
Figure PCTCN2018076135-appb-000006
为一个序列。
步骤S303,建立基于门控循环单元的双循环神经网络模型;
步骤S504,将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型。
具体地,所述服务器2将采样到的序列,作为基于门控循环单元的双循环神经网络模型的输入。若所述服务器2从节点图采样到K个序列,每个序列的长度为T,其中T为变量,即每个序列的长度可能不一样,则所述服务器2将K个长度为T的序列作为基于门控循环单元的双循环神经网络模型的输入。
步骤S305,通过所述基于门控循环单元的双循环神经网络模型根据所述输入序列,输出预测结果。
在本实施方式中,所述预测结果为序列向量。所述服务器2将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型,并设定预测目标或预测条件,如当需要预测任意时刻,如时刻t0的事件影响到人数,即预测时刻t1的事件影响到的节点集合Vt1。当需要预测事件最终的影响到的人数,即要预测的目标是在事件开始蔓延的时间t0后的时间t内,事件所影响到的节点集合Vt。其中,时间t表示事件不再有变化(事件传播波动小于一预设值即可认为事件不再有变化)的时间段。
具体地,所述服务器2还将所述基于门控循环单元的双循环神经网络模型的输出转换成转化成向量图的表示,所述向量图可以用以下向量表示:
Figure PCTCN2018076135-appb-000007
其中,g c是被事件影响的子图,k为序列序号,i为节点序号,B为mini-batch的分块数,
Figure PCTCN2018076135-appb-000008
是序列的编码,a c和λ i可以在深度学习的过程中学习得到。最后使 用全连接层输出最终预测结果。其中,使用全连接层输出的最终预测结果表示为:
f(g c)=MLP(h(g c))
其中,MLP代表多层感知机(Multilayer Perceptron),是一种前向结构的人工神经网络,映射一组输入向量到一组输出向量。MLP可以被看作是一个有向图,由多个的节点层所组成,每一层都全连接到下一层。除了输入节点,每个节点都是一个带有非线性激活函数的神经元(或称处理单元)。
如图8所示,是本申请预测事件流行度方法的第二实施例的流程示意图。本实施例中,第一实施例中的,所述将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型的步骤,具体包括如下步骤:
步骤S401,将所述节点图的每个节点映射成向量。
具体地,所述服务器2将所述序列的每个节点映射成向量。假设采样了K个长度为T的序列,则所述服务器2将所述K个长度为T的序列中的每个节点映射成向量。在本实施例中,所述服务器2使用Node2Vec将所述序列的每个节点映射成向量。
步骤S402,将每个节点传播的关于所述事件的文本内容映射成向量。
具体地,在本实施例中,所述服务器2使用使用Word2Vec将所述每个节点传播的文本内容映射成向量。
步骤S403,将所述每个节点映射成的向量以及所述每个节点传播的文本内容映射成的向量连接成序列向量作为双向循环神经网络模型的输入。
具体地,所述每个节点映射成的向量以及所述每个节点传播的文本内容映射成的向量连接成序列向量,如此,节点与节点传播的文本内容相互关联映射,使得对于事件流行度的预测跟准确。
如图9所示,是本申请预测事件流行度方法的第三实施例的流程示意图。本实施例中,本申请预测事件流行度方法还包括如下步骤:
步骤S501,设定一预设时间。
步骤S502,当到达所述预设时间时,输出关于所述事件的热度预测结果。
具体地,所述服务器2可以设定一预设时间,以实现预测当事件发生后,在一定时间内,该事件的流行度(影响度)。如所述设定模块207设定一时间区间t0-t,t0表示事件发生的时间,t表示设定的时间区间的截止时间。当到达时间t后,输出最终预测结果,实现预测经过任意时间t的事件影响人数。
步骤S503,设定一预设人数值。
步骤S504,当所述事件的影响人数达到所述预设人数时,输出关于所述事件的热度预测结果。
具体地,所述服务器2还可以设定一预设人数值,以实现预测当事件发生后,在多长时间内,该受到事件的影响人数或是关注该事件的人数达到预设人数值。当收到影响的人数达到预设人数值之后,输出最终预测结果,实现了预测达到任意人数值,需要经过多长时间。
上述本申请实施例序号仅仅为了描述,不代表实施例的优劣。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中,包括若干指令用以使得一台终端设备(可以是手机,计算机,服务器,空调器,或者网络设备等)执行本申请各个实施例所述的方法。
以上仅为本申请的优选实施例,并非因此限制本申请的专利范围,凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换,或直接或间接运用在其他相关的技术领域,均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种预测事件流行度方法,应用于服务器,其特征在于,所述方法包括步骤:
    将社交网站的用户关系结构抽象成节点图;
    获取所述社交网站某个时刻事件,并在所述节点图上对所述事件可能传播的序列进行采样;
    建立基于门控循环单元的双循环神经网络模型;
    将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型;
    通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度。
  2. 如权利要求1所述的预测事件流行度方法,其特征在于,所述将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型的步骤,具体包括如下步骤:
    将所述节点图的每个节点映射成向量;
    将每个节点传播的关于所述事件的文本内容映射成向量;
    将所述每个节点映射成的向量以及所述每个节点传播的文本内容映射成的向量连接成序列向量作为双向循环神经网络模型的输入。
  3. 如权利要求2所述的预测事件流行度方法,其特征在于,所述通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度的步骤,具体包括:
    使用注意力(attention)机制将图的序列向量转化成向量图的表示;
    使用全连接层输出最终预测结果。
  4. 如权利要求1所述的预测事件流行度方法,其特征在于,在所述节点图上对所述事件可能传播的序列进行采样的步骤,具体包括如下步骤:
    采用随机游走对所述事件可能传播的序列进行采样;
    计算在随机游走中正在访问的节点转移到邻居节点的概率;
    根据马尔科夫性质及所述概率,采样出不同的节点转移序列。
  5. 如权利要求2所述的预测事件流行度方法,其特征在于,在所述节点图上对所述事件可能传播的序列进行采样的步骤,具体包括如下步骤:
    采用随机游走对所述事件可能传播的序列进行采样;
    计算在随机游走中正在访问的节点转移到邻居节点的概率;
    根据马尔科夫性质及所述概率,采样出不同的节点转移序列。
  6. 如权利要求3所述的预测事件流行度方法,其特征在于,在所述节点图上对所述事件可能传播的序列进行采样的步骤,具体包括如下步骤:
    采用随机游走对所述事件可能传播的序列进行采样;
    计算在随机游走中正在访问的节点转移到邻居节点的概率;
    根据马尔科夫性质及所述概率,采样出不同的节点转移序列。
  7. 如权利要求1所述的预测事件流行度方法,其特征在于,所述方法还包括:
    设定一预设时间;
    当到达所述预设时间时,输出关于所述事件的热度预测结果;
    设定一预设人数值;
    当所述事件的影响人数达到所述预设人数时,输出关于所述事件的热度预测结果。
  8. 如权利要求2所述的预测事件流行度方法,其特征在于,所述方法还包括:
    设定一预设时间;
    当到达所述预设时间时,输出关于所述事件的热度预测结果;
    设定一预设人数值;
    当所述事件的影响人数达到所述预设人数时,输出关于所述事件的热度预测结果。
  9. 如权利要求3所述的预测事件流行度方法,其特征在于,所述方法还包括:
    设定一预设时间;
    当到达所述预设时间时,输出关于所述事件的热度预测结果;
    设定一预设人数值;
    当所述事件的影响人数达到所述预设人数时,输出关于所述事件的热度预测结果。
  10. 如权利要求4所述的预测事件流行度方法,其特征在于,所述方法还包括:
    设定一预设时间;
    当到达所述预设时间时,输出关于所述事件的热度预测结果;
    设定一预设人数值;
    当所述事件的影响人数达到所述预设人数时,输出关于所述事件的热度预测结果。
  11. 一种服务器,其特征在于,所述服务器包括存储器、处理器,所述存储器上存储有可在所述处理器上运行的预测事件流行度系统,所述预测事件流行度系统被所述处理器执行时实现如下步骤:
    将社交网站的用户关系结构抽象成节点图;
    获取所述社交网站某个时刻事件,并在所述节点图上对所述事件可能传播的序列进行采样;
    建立基于门控循环单元的双循环神经网络模型;
    将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型;
    通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度。
  12. 如权利要求11所述的服务器,其特征在于,将所述采样的序列输入 至所述基于门控循环单元的双循环神经网络模型的步骤,具体包括如下步骤:
    将所述节点图的每个节点映射成向量;
    将每个节点传播的关于所述事件的文本内容映射成向量;
    将所述每个节点映射成的向量以及所述每个节点传播的文本内容映射成的向量连接成序列向量作为双向循环神经网络模型的输入。
  13. 如权利要求12所述的服务器,其特征在于,所述通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度的步骤,具体包括:
    使用注意力(attention)机制将图的序列向量转化成向量图的表示;
    使用全连接层输出最终预测结果。
  14. 如权利要求12所述的服务器,其特征在于,在所述节点图上对所述事件可能传播的序列进行采样的步骤,具体包括如下步骤:
    采用随机游走对所述事件可能传播的序列进行采样;
    计算在随机游走中正在访问的节点转移到邻居节点的概率;
    根据马尔科夫性质及所述概率,采样出不同的节点转移序列。
  15. 如权利要求12所述的服务器,其特征在于,所述预测事件流行度系统被所述处理器执行时还实现如下步骤:
    设定一预设时间;
    当到达所述预设时间时,输出关于所述事件的热度预测结果;
    设定一预设人数值;
    当所述事件的影响人数达到所述预设人数时,输出关于所述事件的热度预测结果。
  16. 一种计算机可读存储介质,所述计算机可读存储介质存储有预测事件流行度系统,所述预测事件流行度系统可被至少一个处理器执行,以使所述至少一个处理器执行时实现如下步骤:
    将社交网站的用户关系结构抽象成节点图;
    获取所述社交网站某个时刻事件,并在所述节点图上对所述事件可能传播的序列进行采样;
    建立基于门控循环单元的双循环神经网络模型;
    将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型;
    通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度。
  17. 如权利要求16所述的计算机可读存储介质,其特征在于,将所述采样的序列输入至所述基于门控循环单元的双循环神经网络模型的步骤,具体包括如下步骤:
    将所述节点图的每个节点映射成向量;
    将每个节点传播的关于所述事件的文本内容映射成向量;
    将所述每个节点映射成的向量以及所述每个节点传播的文本内容映射成的向量连接成序列向量作为双向循环神经网络模型的输入。
  18. 如权利要求17所述的计算机可读存储介质,其特征在于,所述通过所述基于门控循环单元的双循环神经网络模型输出预测得到所述事件的影响度的步骤,具体包括:
    使用注意力(attention)机制将图的序列向量转化成向量图的表示;
    使用全连接层输出最终预测结果。
  19. 如权利要求17所述的计算机可读存储介质,其特征在于,在所述节点图上对所述事件可能传播的序列进行采样的步骤,具体包括如下步骤:
    采用随机游走对所述事件可能传播的序列进行采样;
    计算在随机游走中正在访问的节点转移到邻居节点的概率;
    根据马尔科夫性质及所述概率,采样出不同的节点转移序列。
  20. 如权利要求17所述的计算机可读存储介质,其特征在于,所述预测事件流行度系统被所述处理器执行时还实现如下步骤:
    设定一预设时间;
    当到达所述预设时间时,输出关于所述事件的热度预测结果;
    设定一预设人数值;
    当所述事件的影响人数达到所述预设人数时,输出关于所述事件的热度预测结果。
PCT/CN2018/076135 2017-11-17 2018-02-10 预测事件流行度方法、服务器及计算机可读存储介质 WO2019095570A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201711141758.7 2017-11-17
CN201711141758.7A CN107944610B (zh) 2017-11-17 2017-11-17 预测事件流行度方法、服务器及计算机可读存储介质

Publications (1)

Publication Number Publication Date
WO2019095570A1 true WO2019095570A1 (zh) 2019-05-23

Family

ID=61931584

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/076135 WO2019095570A1 (zh) 2017-11-17 2018-02-10 预测事件流行度方法、服务器及计算机可读存储介质

Country Status (2)

Country Link
CN (1) CN107944610B (zh)
WO (1) WO2019095570A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112396091A (zh) * 2020-10-23 2021-02-23 西安电子科技大学 社交媒体图像流行度预测方法、系统、存储介质及应用
CN112580879A (zh) * 2020-12-23 2021-03-30 河南广播电视台 一种基于图神经网络的信息流行度预测方法
CN112580878A (zh) * 2020-12-23 2021-03-30 河南广播电视台 一种基于图神经网络的信息流行度预测方法
CN113076335A (zh) * 2021-04-02 2021-07-06 西安交通大学 一种网络模因检测方法、系统、设备及存储介质
CN113205426A (zh) * 2021-05-27 2021-08-03 中库(北京)数据系统有限公司 一种预测社交媒体内容的热度等级的方法和装置
CN113505924A (zh) * 2021-07-09 2021-10-15 福州大学 一种基于级联时空特征的信息传播预测方法及系统
CN115660147A (zh) * 2022-09-26 2023-01-31 哈尔滨工业大学 一种基于传播路径间与传播路径内影响力建模的信息传播预测方法及系统

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108900446B (zh) * 2018-05-28 2021-03-30 南京信息工程大学 基于门控循环单元神经网络的坐标变换常模盲均衡方法
CN108923983B (zh) * 2018-07-13 2021-01-12 南昌航空大学 机会网络链路的预测方法、装置及可读存储介质
CN110032682B (zh) * 2019-04-17 2023-08-11 腾讯科技(上海)有限公司 一种信息推荐列表生成方法、装置及设备
CN111401648B (zh) * 2020-03-20 2021-01-19 李惠芳 一种互联网热点互相影响情况下的事件预测方法
CN112150207A (zh) * 2020-09-30 2020-12-29 武汉大学 基于时空上下文注意力网络的网约车订单需求预测方法
CN113283589B (zh) * 2021-06-07 2022-07-19 支付宝(杭州)信息技术有限公司 事件预测系统的更新方法及装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103617279A (zh) * 2013-12-09 2014-03-05 南京邮电大学 基于Pagerank方法的微博信息传播影响力评估模型的实现方法
CN104182457A (zh) * 2014-07-14 2014-12-03 上海交通大学 在社交网络中基于泊松过程模型的事件流行度预测方法
US20150186378A1 (en) * 2013-12-30 2015-07-02 International Business Machines Corporation System for identifying, monitoring and ranking incidents from social media
CN105913159A (zh) * 2016-04-29 2016-08-31 北京邮电大学 基于社交网络事件的用户影响力预测方法
CN105975504A (zh) * 2016-04-28 2016-09-28 中国科学院计算技术研究所 一种基于循环神经网络的社交网络消息爆发检测方法及系统

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9652525B2 (en) * 2012-10-02 2017-05-16 Banjo, Inc. Dynamic event detection system and method
US20150271557A1 (en) * 2014-03-24 2015-09-24 Joseph Akwo Tabe Multimedia television system for interactive social media and social network
US9817893B2 (en) * 2015-02-18 2017-11-14 Adobe Systems Incorporated Tracking changes in user-generated textual content on social media computing platforms
CN106681989A (zh) * 2015-11-09 2017-05-17 郑州大学 一种预测微博转发概率的方法
CN106339463A (zh) * 2016-08-26 2017-01-18 中国传媒大学 基于Logistic模型的网络舆情预警系统及其预警方法
CN106651030B (zh) * 2016-12-21 2020-08-04 重庆邮电大学 一种改进的rbf神经网络热点话题用户参与行为预测方法
CN106599273A (zh) * 2016-12-23 2017-04-26 贾志娟 一种基于社交网络分析的微博诈骗团体挖掘方法
CN106897388A (zh) * 2017-01-24 2017-06-27 北京航空航天大学 预测微博事件热度的方法及装置
CN107145977B (zh) * 2017-04-28 2020-07-31 电子科技大学 一种对在线社交网络用户进行结构化属性推断的方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103617279A (zh) * 2013-12-09 2014-03-05 南京邮电大学 基于Pagerank方法的微博信息传播影响力评估模型的实现方法
US20150186378A1 (en) * 2013-12-30 2015-07-02 International Business Machines Corporation System for identifying, monitoring and ranking incidents from social media
CN104182457A (zh) * 2014-07-14 2014-12-03 上海交通大学 在社交网络中基于泊松过程模型的事件流行度预测方法
CN105975504A (zh) * 2016-04-28 2016-09-28 中国科学院计算技术研究所 一种基于循环神经网络的社交网络消息爆发检测方法及系统
CN105913159A (zh) * 2016-04-29 2016-08-31 北京邮电大学 基于社交网络事件的用户影响力预测方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
NIU, JIANWEI ET AL.: "Inferring Information Propagation over Online Social Networks: Edge Asymmetry and Flow Tendency", 2015 IEEE 17TH INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE COMPUTING AND COMMUNICATIONS, 26 August 2015 (2015-08-26), pages 629 - 634, XP032819315 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112396091A (zh) * 2020-10-23 2021-02-23 西安电子科技大学 社交媒体图像流行度预测方法、系统、存储介质及应用
CN112396091B (zh) * 2020-10-23 2024-02-09 西安电子科技大学 社交媒体图像流行度预测方法、系统、存储介质及应用
CN112580879A (zh) * 2020-12-23 2021-03-30 河南广播电视台 一种基于图神经网络的信息流行度预测方法
CN112580878A (zh) * 2020-12-23 2021-03-30 河南广播电视台 一种基于图神经网络的信息流行度预测方法
CN112580879B (zh) * 2020-12-23 2023-10-10 河南广播电视台 一种基于图神经网络的信息流行度预测方法
CN112580878B (zh) * 2020-12-23 2023-10-10 河南广播电视台 一种基于图神经网络的信息流行度预测方法
CN113076335A (zh) * 2021-04-02 2021-07-06 西安交通大学 一种网络模因检测方法、系统、设备及存储介质
CN113076335B (zh) * 2021-04-02 2024-05-24 西安交通大学 一种网络模因检测方法、系统、设备及存储介质
CN113205426A (zh) * 2021-05-27 2021-08-03 中库(北京)数据系统有限公司 一种预测社交媒体内容的热度等级的方法和装置
CN113505924A (zh) * 2021-07-09 2021-10-15 福州大学 一种基于级联时空特征的信息传播预测方法及系统
CN113505924B (zh) * 2021-07-09 2023-08-01 福州大学 一种基于级联时空特征的信息传播预测方法及系统
CN115660147A (zh) * 2022-09-26 2023-01-31 哈尔滨工业大学 一种基于传播路径间与传播路径内影响力建模的信息传播预测方法及系统

Also Published As

Publication number Publication date
CN107944610A (zh) 2018-04-20
CN107944610B (zh) 2022-08-02

Similar Documents

Publication Publication Date Title
WO2019095570A1 (zh) 预测事件流行度方法、服务器及计算机可读存储介质
US9807473B2 (en) Jointly modeling embedding and translation to bridge video and language
WO2021169115A1 (zh) 风控方法、装置、电子设备及计算机可读存储介质
WO2021120677A1 (zh) 一种仓储模型训练方法、装置、计算机设备及存储介质
US8438189B2 (en) Local computation of rank contributions
US8069167B2 (en) Calculating web page importance
WO2020224106A1 (zh) 基于神经网络的文本分类方法、系统及计算机设备
CN110855648B (zh) 一种网络攻击的预警控制方法及装置
WO2022110640A1 (zh) 一种模型优化方法、装置、计算机设备及存储介质
WO2022083093A1 (zh) 图谱中的概率计算方法、装置、计算机设备及存储介质
CN116032663A (zh) 基于边缘设备的隐私数据处理系统、方法、设备及介质
US20150207691A1 (en) Preloading content based on network connection behavior
JP2022068123A (ja) センサ非依存のジェスチャ検出のためのコンピュータ実装方法、システムおよびコンピュータプログラム製品(センサ非依存のジェスチャ検出)
CN116684330A (zh) 基于人工智能的流量预测方法、装置、设备及存储介质
CN113988225A (zh) 建立表征提取模型、表征提取、类型识别的方法和装置
CN112528040B (zh) 基于知识图谱的引导教唆语料的检测方法及其相关设备
CN111144473A (zh) 训练集构建方法、装置、电子设备及计算机可读存储介质
CN114625971B (zh) 一种基于用户签到的兴趣点推荐方法及设备
Zhang et al. Proactive serving decreases user delay exponentially: The light-tailed service time case
KR20150029719A (ko) 웹 브라우저 동작 방법 및 시스템
US20200242446A1 (en) Convolutional dynamic boltzmann machine for temporal event sequence
CN115828269A (zh) 源代码漏洞检测模型的构建方法、装置、设备及存储介质
US20230041339A1 (en) Method, device, and computer program product for user behavior prediction
CN115099875A (zh) 基于决策树模型的数据分类方法及相关设备
CN114896412A (zh) 嵌入向量的生成方法、基于企业对的同名人员分类方法及装置

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205N DATED 09/10/2020)

122 Ep: pct application non-entry in european phase

Ref document number: 18878165

Country of ref document: EP

Kind code of ref document: A1