CN-121998633-A - Agent opinion prediction method and device in social network and electronic equipment

CN121998633ACN 121998633 ACN121998633 ACN 121998633ACN-121998633-A

Abstract

The invention provides a method, a device and electronic equipment for predicting agent opinions in a social network, wherein the method, the device and the electronic equipment are used for obtaining current information of at least one information source at the current moment, obtaining current opinion leader agent opinions based on the current information and information preference of opinion leader to the information source, determining current common agent opinions based on the current opinion leader agent opinions, iteratively executing the steps of obtaining the current opinion leader agent opinions based on the current information and the information preference of opinion leader to the information source until the obtained current opinion leader agent opinion reaches an opinion state steady state and the current common agent opinion reaches an opinion state steady state, obtaining steady state opinion leader agent opinions based on the current opinion leader agent opinion prediction, and obtaining steady state common agent opinions based on the current common agent opinion prediction. Accurate prediction of opinion leader agent opinion and general agent opinion is achieved.

Inventors

ZHAO HONG
WANG HUISHENG
LIN YIQING

Assignees

清华大学

Dates

Publication Date: 20260508
Application Date: 20251217

Claims (10)

1. A method for proxy opinion prediction in a social network, the method comprising: Acquiring current information of at least one information source at a current time, wherein the information source is an information source related to a preset social event; Obtaining a current opinion leader agent opinion of the opinion leader at a current moment based on the current information and information preference of the opinion leader on the information source, wherein the information preference is used for representing confirmation deviation of the opinion leader; Determining the current common agent opinion of the common agent at the current moment based on the current opinion leader agent opinion; Iteratively executing a step of obtaining a current opinion leader agent opinion based on the current information and information preference of an opinion leader to the information source until a current common agent opinion step of determining a common agent opinion based on the current opinion leader agent opinion at the current time is reached to an opinion state steady state and the obtained current opinion leader agent opinion reaches to an opinion state steady state; And predicting and obtaining steady-state opinion leader agent opinions of the opinion leader on the preset social event based on the current opinion leader agent opinion, and predicting and obtaining steady-state common agent opinions of the common agent on the preset social event based on the current common agent opinion.
2. The method for proxy opinion prediction in a social network according to claim 1, wherein before said obtaining a current opinion leader proxy opinion of said opinion leader at a current time based on said current information and information preference of opinion leader for said information source, said method further comprises: Obtaining an initial opinion leader agent opinion of the opinion leader on the preset social event and opinion leader characteristic parameters of the opinion leader, wherein the opinion leader characteristic parameters are used for describing the degree of fixation of the opinion leader on own opinion; the obtaining the current opinion leader agent opinion of the opinion leader at the current time based on the current information and the information preference of the opinion leader to the information source comprises the following steps: And obtaining the current opinion leader agent opinion of the opinion leader at the current time based on the current information, the initial opinion leader agent opinion, the opinion leader characteristic parameters and the information preference of the opinion leader to the information source.
3. The method of claim 2, wherein the information sources for the predetermined social event include a plurality of information sources, and wherein prior to said obtaining a current opinion leader agent opinion at a current time based on the current information, the initial opinion leader agent opinion, the opinion leader characteristic parameter, and information preference of an opinion leader for the information sources, the method further comprises: Respectively determining the information preference of the opinion leader to each information source; For any one of the information sources, carrying out normalization processing on the information preference of the information source based on each information preference to obtain normalized information preference; Taking the normalized information preference of the information source as the influence weight of the information source on the opinion leader; The obtaining the current opinion leader agent opinion of the opinion leader at the current time based on the current information, the initial opinion leader agent opinion, the opinion leader characteristic parameter, and the information preference of the opinion leader to the information source includes: The influence weight of any information source on the opinion leader is multiplied with the current information respectively to obtain a plurality of multiplied parameters; Adding the multiplied parameters to obtain opinion weighted average information; Based on the opinion leader characteristic parameters, respectively determining a first weight of the initial opinion leader agent opinion and a second weight of the opinion weighted average information; And obtaining the current opinion leader agent opinion of the opinion leader at the current time based on the first weight, the initial opinion leader agent opinion, the second weight and the opinion weighted average information.
4. The method for predicting an opinion in a social network as recited in claim 1, wherein prior to said determining a current opinion of a normal agent at a current time based on said current opinion leader opinion, said method further comprises: Acquiring initial common agent opinion of the common agent on the preset social event, neighbor common agent opinion of the neighbor common agent at the last moment of the preset social event, common agent characteristic parameters of the common agent, neighbor common agent influence proportion and opinion leader influence proportion, wherein the common agent characteristic parameters are used for describing the degree of fixation of the common agent on own opinion; the determining the current common agent opinion under the current time by the common agent based on the current opinion leader agent opinion comprises the following steps: And carrying out weighted summation on the initial common agent opinion, the common agent characteristic parameter, the neighbor common agent opinion, the neighbor common agent influence proportion, the current opinion leader opinion and the opinion leader influence proportion at the previous moment to obtain the current common agent opinion of the common agent at the current moment.
5. The method for predicting the opinion in a social network as recited in claim 4, wherein the neighbor ordinary agent is a plurality of neighbor ordinary agents, and the opinion in the neighbor ordinary agent at the previous time is a plurality of opinion in the neighbor ordinary agent at the previous time, and the opinion in the neighbor ordinary agent at the previous time is determined by: determining, for any one of the neighbor ordinary agents, an influence weight of the neighbor ordinary agent on the ordinary agent, and And carrying out weighted summation on each neighbor ordinary agent opinion at the previous moment and the influence weight of the neighbor ordinary agent on the ordinary agent to obtain the neighbor ordinary agent opinion at the previous moment.
6. The method of claim 4, wherein the opinion leader is a plurality of opinion leaders and the current opinion leader agent opinion is a plurality of current opinion leader agent opinions, and wherein the current opinion leader agent opinion is determined by: Determining, for any one of the opinion leader, an influence weight of the opinion leader on the normal agent, and And carrying out weighted summation on the influence weights of the current opinion leader agent and the opinion leader on the common agent to obtain the opinion of the current opinion leader agent.
7. The method for proxy opinion prediction in a social network according to claim 1, wherein before said obtaining a current opinion leader proxy opinion of said opinion leader at a current time based on said current information and information preference of opinion leader for said information source, said method further comprises: Acquiring the opinion leader agent opinion which is deliberately seen at the last moment of the opinion leader at the last moment; Determining information preference of an opinion leader on the information source based on the distance between the opinion of the opinion leader agent and the current information at the previous moment; The iterative execution is based on the current information and the information preference of the opinion leader to the information source, and the steps from the current opinion leader opinion step to the current common agent opinion step of determining the common agent opinion at the current time based on the current opinion leader agent opinion until the obtained current opinion leader agent opinion reaches an opinion state steady state and the current common agent opinion reaches an opinion state steady state, include: And taking the opinion leader agent opinion at the current moment as the opinion leader agent opinion at the last moment of the next moment, and repeatedly and iteratively executing the opinion leader agent opinion obtaining step at the last moment of the opinion leader till the current common agent opinion step of the common agent at the current moment is determined, until the opinion difference value between the obtained opinion leader agent opinion at the current moment and the opinion leader at the last moment of the opinion leader at the last moment is smaller than a threshold value, and the opinion difference value between the current common agent opinion at the current moment and the common agent opinion at the last moment of the common agent at the current moment is smaller than the threshold value.
8. An agent opinion prediction apparatus in a social network, the apparatus comprising: the system comprises an acquisition module, a storage module and a control module, wherein the acquisition module is used for acquiring current information of at least one information source at the current time, wherein the information source is an information source related to a preset social event; The generating module is used for obtaining the current opinion leader agent opinion of the opinion leader at the current moment based on the current information and the information preference of the opinion leader to the information source, wherein the information preference is used for representing the confirmation deviation of the opinion leader; the determining module is used for determining the current common agent opinion of the common agent at the current time based on the current opinion leader agent opinion; The processing module is used for iteratively executing the step of obtaining the current opinion leader agency opinion based on the current information and the information preference of the opinion leader to the information source until the current common agency opinion based on the current opinion leader agency opinion is determined until the obtained current opinion leader agency opinion reaches an opinion state steady state and the current common agency opinion reaches an opinion state steady state; And the prediction module is used for predicting and obtaining steady-state opinion leader agent opinion of the opinion leader for the preset social event based on the current opinion leader agent opinion and predicting and obtaining steady-state common agent opinion of the common agent for the preset social event based on the current common agent opinion.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the proxy opinion prediction method in a social network of any of claims 1 to 7 when the computer program is executed.
10. A non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the proxy opinion prediction method in a social network according to any of claims 1 to 7.

Description

Agent opinion prediction method and device in social network and electronic equipment Technical Field The present invention relates to the field of agent opinion prediction technologies, and in particular, to a method and an apparatus for agent opinion prediction in a social network, and an electronic device. Background The related art shows that in the information transmission process of the real social network, a two-step transmission mode of 'information source, opinion leader and common agency' is generally followed. However, the existing classical opinion dynamics model has significant limitation in architecture design, most models do not clearly distinguish heterogeneous roles of opinion leaders and common agents, default information can directly act on all agent nodes without distinction, and special functions of the opinion leaders as information screeners and amplifiers in two-step propagation paths are drawn and lost, so that the prediction capability of the model on agent opinions is limited. Disclosure of Invention The invention provides a method, a device and electronic equipment for predicting agent opinions in a social network, which realize accurately describing a hierarchical structure of information transmission and opinion exchange in a real social network and ensure accurate prediction of opinion leader agent opinions and common agent opinions. The invention provides a method for predicting an agent opinion in a social network, which comprises the steps of obtaining current information of at least one information source at the current moment, wherein the information source is an information source related to a preset social event, obtaining a current opinion leader agency opinion of an opinion leader at the current moment based on the current information and the information preference of the opinion leader to the information source, wherein the information preference is used for representing the confirmation deviation of the opinion leader, determining the current common agency opinion of a common agency at the current moment based on the current opinion leader agency opinion, iteratively executing the opinion leader agency opinion step based on the current information and the opinion leader to the information source, and obtaining the current opinion leader agency opinion of the opinion leader at the current moment until the obtained current opinion leader opinion state steady state of a common agency is reached based on the current opinion leader agency opinion, and obtaining the preset steady state opinion leader opinion event of the common agency based on the preset opinion leader, and obtaining the common opinion leader opinion state steady state of the common opinion leader of the common agency based on the preset opinion leader. The method for predicting the agency opinion in the social network further comprises the steps of obtaining initial opinion leader agency opinions of the opinion leader on the preset social event and opinion leader characteristic parameters of the opinion leader before obtaining the current opinion leader agency opinion of the opinion leader on the current moment based on the current information and the information preference of the opinion leader on the information source, wherein the opinion leader characteristic parameters are used for describing the degree of fixation of the opinion leader on own opinion, and obtaining the current opinion leader agency opinion of the opinion leader on the current moment based on the current information and the information preference of the opinion leader on the information source. According to the agent opinion prediction method in the social network, an information source related to a preset social event comprises a plurality of information sources, before the information source is subjected to information preference based on the current information, the initial opinion leader agent opinion, the opinion leader characteristic parameters and the opinion leader, the current opinion leader agent opinion of the opinion leader at the current moment is obtained, the method further comprises the steps of respectively determining each information preference of each information source by the opinion leader, carrying out normalization processing on the information preference of the information source based on each information preference to obtain normalized information preference for any information source, taking the normalized information preference of the information source as influence weight of the information source on the opinion leader, multiplying the opinion leader by the current opinion leader agent opinion, the initial opinion leader characteristic parameters and the opinion leader by the opinion leader average weight, multiplying the opinion leader average weight by the current opinion leader average weight, multiplying the initial opinion leader average weight by the initial opinion leader average weight and the op