An opinion propagation distortion analysis method, device, equipment and storage medium

Abstract

The present application relates to the technical field of information processing, and particularly relates to a public opinion transmission distortion analysis method, device, equipment and storage medium, the method comprises the following steps: determining the emotional transmission distortion information corresponding to the current public opinion text according to the deviation between the text emotion result corresponding to the previous public opinion text which has transmission correlation with the current public opinion text and the text emotion result corresponding to the current public opinion text; determining the negative emotional contribution value corresponding to each public opinion text according to the text emotion result and the emotional transmission distortion information; performing cumulative calculation and decay correction on each negative emotional contribution value in time sequence to obtain the negative emotional accumulation potential corresponding to the target theme; determining the negative diffusion inflection point corresponding to the target theme according to the change result of the negative emotional accumulation potential in the continuous time period; and generating the target public opinion emotion analysis result according to the negative diffusion inflection point. The present application has the analysis ability of improving the emotional deviation and diffusion trend in the public opinion transmission process.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and in particular to a method, apparatus, device, and storage medium for analyzing distortions in public opinion dissemination. Background Technology

[0002] Most existing public opinion analysis techniques focus on identifying the sentiment polarity of a single text or a single time segment. They typically use sentiment classification models to determine whether the text content is positive, negative, or neutral, and then combine this with keyword statistics, popularity statistics, or topic clustering to analyze the state of public opinion. While this approach can reflect the current sentiment tendency of the text to some extent, it focuses more on the static text itself and lacks in-depth analysis of the sentiment shifts between preceding and subsequent texts in the dissemination chain, making it difficult to identify the sentiment distortion that occurs during the dissemination of public opinion.

[0003] Furthermore, existing solutions typically base their assessment of negative public opinion on single sentiment recognition results or simple time-based summations. They lack effective characterization of the differences in the impact of different texts within the dissemination structure, the cumulative changes in negative sentiment during continuous propagation, and the key nodes where local fluctuations transform into rapid spread. This makes existing technologies prone to problems such as recognition lag, significant judgment bias, and inaccurate early warnings when faced with scenarios where public opinion content is continuously amplified, negative sentiment accumulates, and a turning point occurs. Consequently, they struggle to provide reliable support for public opinion monitoring, risk assessment, and subsequent handling. Summary of the Invention

[0004] To improve the ability to analyze the emotional bias and diffusion trend in the process of public opinion dissemination, this application provides a method, apparatus, equipment and storage medium for analyzing the distortion of public opinion dissemination.

[0005] The above-mentioned objective of this application is achieved through the following technical solution:

[0006] A method for analyzing the distortion of public opinion dissemination, comprising:

[0007] Obtain multiple public opinion texts corresponding to the target topic and the dissemination correlation information between these texts;

[0008] Based on the offset between the sentiment results of the preceding public opinion texts that are related to the current public opinion text and the sentiment results of the current public opinion text, the sentiment dissemination distortion information corresponding to the current public opinion text is determined;

[0009] Based on the text sentiment results and the sentiment propagation distortion information, determine the negative sentiment contribution value corresponding to each of the aforementioned public opinion texts;

[0010] The negative sentiment contribution values ​​are cumulatively calculated and attenuated in chronological order to obtain the negative sentiment accumulation potential energy corresponding to the target topic.

[0011] Based on the changes in the negative emotional accumulation potential energy over a continuous time period, the negative diffusion inflection point corresponding to the target topic is determined;

[0012] Based on the negative diffusion inflection point, the target public opinion sentiment analysis results are generated.

[0013] By adopting the above technical solutions, and by acquiring multiple public opinion texts corresponding to the target topic and the dissemination correlation information between these texts, it is possible to analyze public opinion changes from the perspective of the dissemination chain. By determining the emotional dissemination distortion information based on the text sentiment result deviation between the preceding and current public opinion texts, it is possible to identify the emotional deviation of public opinion during the dissemination process. By determining the negative emotional contribution value based on the text sentiment result and the emotional dissemination distortion information, it is possible to distinguish the degree of influence of different public opinion texts on negative diffusion. By performing cumulative calculation and attenuation correction on the negative emotional contribution value, it is possible to reflect the continuous accumulation state of negative emotions. By determining the negative diffusion inflection point based on the change of negative emotional accumulation potential energy over a continuous period of time, it is possible to identify the key turning point where public opinion risks intensify, thereby improving the accuracy and timeliness of public opinion analysis.

[0014] Preferably, determining the sentiment propagation distortion information corresponding to the current public opinion text based on the offset between the sentiment result corresponding to the preceding public opinion text that has a propagation association with the current public opinion text and the sentiment result corresponding to the current public opinion text includes:

[0015] Extract the preceding sentiment polarity information and preceding sentiment intensity information corresponding to the preceding public opinion text, and extract the current sentiment polarity information and current sentiment intensity information corresponding to the current public opinion text;

[0016] Based on the preceding sentiment polarity information and the current sentiment polarity information, determine the sentiment direction offset information corresponding to the current public opinion text;

[0017] Based on the preceding sentiment intensity information and the current sentiment intensity information, determine the sentiment intensity offset information corresponding to the current public opinion text;

[0018] The core stance expression information is extracted from the preceding public opinion text and the current public opinion text respectively. Based on the difference between the core stance expression information in the preceding public opinion text and the core stance expression information in the current public opinion text, the stance deviation information corresponding to the current public opinion text is determined.

[0019] Based on the emotional direction offset information, the emotional intensity offset information, and the stance deviation information, the emotional propagation distortion information corresponding to the current public opinion text is determined.

[0020] By employing the aforementioned technical solution, and by extracting the preceding emotional polarity and intensity information corresponding to the preceding public opinion text, as well as the current emotional polarity and intensity information corresponding to the current public opinion text, the texts before and after dissemination can be compared from two dimensions: the direction of emotional tendency and the degree of emotional tendency. By determining the information on emotional direction shift, emotional intensity shift, and stance deviation, the changes in emotional expression and stance expression during the dissemination process can be identified. By determining the information on emotional dissemination distortion based on the information on emotional direction shift, emotional intensity shift, and stance deviation, the degree of distortion of public opinion texts during the dissemination process can be more comprehensively reflected, thereby improving the accuracy of identifying emotional dissemination distortion.

[0021] Preferably, determining the negative sentiment contribution value corresponding to each of the public opinion texts based on the text sentiment results and the sentiment propagation distortion information includes:

[0022] Based on the text sentiment results, determine the basic negative sentiment value corresponding to each of the aforementioned public opinion texts;

[0023] The distortion information of emotional transmission is evaluated item by item to obtain the direction distortion value, intensity distortion value and stance distortion value corresponding to each of the public opinion texts;

[0024] Based on the directional distortion value, the intensity distortion value, and the position distortion value, determine the distortion correction value corresponding to each of the public opinion texts;

[0025] Based on the propagation level information, propagation response information, and subsequent follow-up information of each of the aforementioned public opinion texts in the propagation association information, determine the propagation correction value corresponding to each of the aforementioned public opinion texts;

[0026] Based on the basic negative sentiment value, the distortion correction value, and the propagation correction value, the negative sentiment contribution value corresponding to each of the aforementioned public opinion texts is determined.

[0027] By adopting the above technical solutions, and by determining the basic negative sentiment value corresponding to each public opinion text based on the text sentiment results, the degree of negative emotion of each public opinion text can be reflected. By sub-judging the information of emotional dissemination distortion and obtaining the directional distortion value, intensity distortion value, and stance distortion value, the emotional deviation in the dissemination process can be identified from multiple dimensions. By determining the distortion correction value and the dissemination correction value, the degree of emotional distortion and the diffusion impact in the dissemination process can be comprehensively reflected. By determining the negative sentiment contribution value based on the basic negative sentiment value, distortion correction value, and dissemination correction value, the degree of influence of different public opinion texts on the spread of negative emotions can be more accurately distinguished, thereby improving the accuracy of negative sentiment contribution analysis.

[0028] Preferably, the step of performing cumulative calculation and attenuation correction on each of the negative sentiment contribution values ​​in chronological order to obtain the negative sentiment accumulation potential energy corresponding to the target topic includes:

[0029] The negative sentiment contribution values ​​corresponding to each of the aforementioned public opinion texts are sorted according to the publication time order to obtain the time-series negative sentiment contribution sequence corresponding to the target topic;

[0030] Based on the time interval between adjacent negative sentiment contribution values ​​in the time-series negative sentiment contribution sequence, determine the time decay value corresponding to each negative sentiment contribution value;

[0031] Based on the growth difference between the current negative sentiment contribution value and the previous negative sentiment contribution value in the time-series negative sentiment contribution sequence, the incremental reinforcement value corresponding to each negative sentiment contribution value is determined;

[0032] Based on the negative emotion contribution value, the time decay value, and the incremental reinforcement value, determine the stage accumulation value corresponding to each time position;

[0033] By continuously accumulating the accumulated values ​​of each stage in chronological order, the negative emotional potential energy corresponding to the target topic is obtained.

[0034] By adopting the above technical solution and sorting the negative sentiment contribution values ​​corresponding to each public opinion text according to the release time, a time-series negative sentiment contribution sequence corresponding to time changes can be formed. By determining the time decay value and the incremental reinforcement value, the decay of historical negative influence and the reinforcement of current negative growth can be reflected simultaneously. By determining the stage accumulation value corresponding to each time position and performing continuous accumulation, the continuous accumulation state of negative sentiment under the target theme in the time dimension can be reflected more accurately, thereby improving the ability to depict the evolution trend of negative sentiment.

[0035] Preferably, determining the negative diffusion inflection point corresponding to the target topic based on the change in the negative emotional accumulation potential energy over a continuous time period includes:

[0036] The accumulated potential energy of negative emotions is divided into intervals according to continuous time periods to obtain multiple continuous potential energy intervals;

[0037] Based on the negative emotional accumulation potential energy change value corresponding to each of the continuous potential energy intervals, the potential energy growth value corresponding to each of the continuous potential energy intervals is determined.

[0038] Based on the difference in the change between the potential energy growth values ​​corresponding to adjacent continuous potential energy intervals, the slope offset value corresponding to each continuous potential energy interval is determined.

[0039] Based on the amplification difference between the potential energy growth value corresponding to the current continuous potential energy interval and the potential energy growth value corresponding to the previous continuous potential energy interval, the sudden shift value corresponding to each continuous potential energy interval is determined.

[0040] Based on the direction of change of potential energy growth value corresponding to multiple consecutive potential energy intervals, determine the continuous growth value corresponding to each consecutive potential energy interval;

[0041] Based on the slope offset value, the sudden increase offset value, and the continuous growth value, the inflection point trigger value corresponding to each continuous potential energy interval is determined.

[0042] If the inflection point trigger value meets the preset inflection point condition, the corresponding continuous potential energy interval is determined as the negative diffusion inflection point corresponding to the target theme.

[0043] By adopting the above technical solution, the accumulated potential energy of negative emotions is divided into intervals according to continuous time periods. The analysis of each continuous potential energy interval is combined with the potential energy growth value, slope offset value, sudden increase offset value, and continuous growth value. This allows for the identification of the changing characteristics of the negative emotion diffusion process from multiple dimensions such as growth rate, sudden increase degree, and continuous trend. By determining the inflection point trigger value corresponding to each continuous potential energy interval, and determining the negative diffusion inflection point when the preset inflection point conditions are met, the key interval for the transformation of negative emotions from normal accumulation to intensified diffusion can be identified more accurately. This improves the accuracy and timeliness of identifying the turning point of public opinion risk.

[0044] Preferably, generating the target public opinion sentiment analysis result based on the negative diffusion inflection point includes:

[0045] Based on the inflection point trigger value, occurrence time location, and continuous potential energy interval location corresponding to the negative diffusion inflection point, the inflection point level information corresponding to the target theme is determined;

[0046] Based on the inflection point level information and the changes in negative emotional accumulation potential energy before and after the negative diffusion inflection point, the diffusion state information corresponding to the target topic is determined.

[0047] Based on the diffusion state information and the negative sentiment accumulation potential energy corresponding to the target topic, the risk priority information corresponding to the target topic is determined;

[0048] Based on the inflection point level information, the diffusion status information, and the risk priority information, the target public opinion sentiment analysis result is generated.

[0049] By adopting the above technical solution, the inflection point level information is determined based on the inflection point trigger value, occurrence time, and continuous potential energy interval corresponding to the negative diffusion inflection point. The diffusion state information is determined by combining the changes in negative sentiment accumulation potential energy before and after the negative diffusion inflection point. This allows for the identification of the intensity and state of negative sentiment diffusion corresponding to the target topic. By determining risk priority information based on diffusion state information and negative sentiment accumulation potential energy, the degree of risk concern corresponding to different target topics can be distinguished. By generating target public opinion sentiment analysis results based on inflection point level information, diffusion state information, and risk priority information, the ability of public opinion sentiment analysis results to support risk assessment and trend judgment can be improved.

[0050] Preferably, generating the target public opinion sentiment analysis result based on the inflection point level information, the diffusion status information, and the risk priority information includes:

[0051] Based on the inflection point level information, determine the diffusion transition trend information corresponding to the target topic;

[0052] Based on the diffusion state information, determine the state continuation information corresponding to the target topic;

[0053] Based on the risk priority information, determine the risk clustering direction information corresponding to the target topic;

[0054] Based on the diffusion transition trend information, the state continuation information, and the risk aggregation direction information, a negative diffusion evolution path corresponding to the target theme is constructed;

[0055] Based on the negative diffusion evolution path, the target public opinion sentiment analysis results are generated.

[0056] By adopting the above technical solution, the diffusion transition trend information corresponding to the target topic is determined based on inflection point level information, the state continuation information corresponding to the target topic is determined based on diffusion state information, and the risk aggregation direction information corresponding to the target topic is determined based on risk priority information. The target topic can be analyzed from three dimensions: changes in diffusion intensity, changes in state, and directions of risk concern. By constructing a negative diffusion evolution path based on diffusion transition trend information, state continuation information, and risk aggregation direction information, the negative sentiment diffusion evolution process corresponding to the target topic can be more completely depicted, thereby improving the ability of the target public opinion sentiment analysis results to support trend judgment and risk assessment.

[0057] The second objective of this invention is achieved through the following technical solution:

[0058] A public opinion dissemination distortion analysis device, the public opinion dissemination distortion analysis device comprising:

[0059] The public opinion association acquisition module is used to acquire multiple public opinion texts corresponding to the target topic and the dissemination association information between the public opinion texts;

[0060] The emotion propagation distortion determination module is used to determine the emotion propagation distortion information corresponding to the current public opinion text based on the offset between the text emotion result corresponding to the previous public opinion text that has a propagation relationship with the current public opinion text and the text emotion result corresponding to the current public opinion text.

[0061] The negative sentiment contribution determination module is used to determine the negative sentiment contribution value corresponding to each of the aforementioned public opinion texts based on the text sentiment results and the sentiment propagation distortion information.

[0062] The negative emotion accumulation potential energy determination module is used to perform cumulative calculation and attenuation correction on each negative emotion contribution value in chronological order to obtain the negative emotion accumulation potential energy corresponding to the target topic.

[0063] The negative diffusion inflection point determination module is used to determine the negative diffusion inflection point corresponding to the target topic based on the change results of the negative emotion accumulation potential energy over a continuous time period.

[0064] The public opinion sentiment analysis result generation module is used to generate target public opinion sentiment analysis results based on the negative diffusion inflection point.

[0065] By adopting the above technical solutions, and by acquiring multiple public opinion texts corresponding to the target topic and the dissemination correlation information between these texts, it is possible to analyze public opinion changes from the perspective of the dissemination chain. By determining the emotional dissemination distortion information based on the text sentiment result deviation between the preceding and current public opinion texts, it is possible to identify the emotional deviation of public opinion during the dissemination process. By determining the negative emotional contribution value based on the text sentiment result and the emotional dissemination distortion information, it is possible to distinguish the degree of influence of different public opinion texts on negative diffusion. By performing cumulative calculation and attenuation correction on the negative emotional contribution value, it is possible to reflect the continuous accumulation state of negative emotions. By determining the negative diffusion inflection point based on the change of negative emotional accumulation potential energy over a continuous period of time, it is possible to identify the key turning point where public opinion risks intensify, thereby improving the accuracy and timeliness of public opinion analysis.

[0066] The above-mentioned objective three of this application is achieved through the following technical solution:

[0067] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aforementioned method for analyzing distortion in public opinion dissemination.

[0068] The fourth objective of this application is achieved through the following technical solution:

[0069] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned method for analyzing distortion in public opinion dissemination.

[0070] In summary, this application includes at least one of the following beneficial technical effects:

[0071] By acquiring multiple public opinion texts corresponding to the target topic and the dissemination correlation information between these texts, we can analyze public opinion changes from the perspective of the dissemination chain. By determining the emotional dissemination distortion information based on the text sentiment result deviation between previous and current public opinion texts, we can identify the emotional deviation in the dissemination process. By determining the negative emotional contribution value based on the text sentiment result and emotional dissemination distortion information, we can distinguish the degree of influence of different public opinion texts on negative diffusion. By performing cumulative calculation and attenuation correction on the negative emotional contribution value, we can reflect the continuous accumulation state of negative emotions. By determining the inflection point of negative diffusion based on the change of negative emotional accumulation potential energy over a continuous period of time, we can identify the key turning point where public opinion risks intensify, thereby improving the accuracy and timeliness of public opinion analysis. Attached Figure Description

[0072] Figure 1 This is a flowchart of a public opinion dissemination distortion analysis method according to an embodiment of this application.

[0073] Figure 2 This is a schematic diagram of a public opinion dissemination distortion analysis device according to one embodiment of this application; Detailed Implementation

[0074] The present application will be further described in detail below with reference to the accompanying drawings.

[0075] In one embodiment, such as Figure 1 As shown, this application discloses a method for analyzing the distortion of public opinion dissemination, which specifically includes the following steps:

[0076] S10: Obtain multiple public opinion texts corresponding to the target topic and the dissemination correlation information between these texts.

[0077] In this embodiment, the target topic refers to the target event, target topic, or target object pre-determined for the public opinion analysis task; multiple public opinion texts refer to multiple text data formed within the scope of the target topic, used to represent public opinions, attitudes, or evaluations; and dissemination association information refers to the information on the preceding and following relationships formed during the dissemination of each public opinion text, used to represent the reprint relationship, citation relationship, reply relationship, or forwarding relationship between each public opinion text.

[0078] Specifically, when acquiring multiple public opinion texts corresponding to a target theme and the dissemination association information between these texts, the process first involves determining the text filtering range corresponding to the target theme, then collecting and processing text data containing the target theme identifier within the text filtering range to obtain multiple public opinion texts. Next, based on the publication time, source identifier, citation marker, reply marker, and forwarding marker corresponding to each public opinion text, the process matches and determines the dissemination correspondence between these texts, and uses this dissemination correspondence as the dissemination association information between these texts.

[0079] S20: Based on the offset between the sentiment result of the preceding public opinion text that is related to the current public opinion text and the sentiment result of the current public opinion text, determine the sentiment dissemination distortion information corresponding to the current public opinion text.

[0080] In this embodiment, the current public opinion text refers to the public opinion text that is the current object of analysis in the dissemination of related information; the preceding public opinion text refers to the public opinion text that precedes the current public opinion text in the dissemination order and has a dissemination relationship with the current public opinion text; the text sentiment result refers to the result information formed after judging the emotional tendency and emotional degree expressed by the public opinion text; the deviation refers to the change between the text sentiment result corresponding to the preceding public opinion text and the text sentiment result corresponding to the current public opinion text; and the emotional dissemination distortion information refers to the result information formed after the current public opinion text deviates from the emotional expression of the preceding public opinion text during the dissemination process.

[0081] Specifically, to determine the emotional dissemination distortion information corresponding to the current public opinion text based on the offset between the text sentiment results corresponding to the preceding public opinion text that has a dissemination association with the current public opinion text, the preceding public opinion text corresponding to the current public opinion text is first identified in the dissemination association information, and the text sentiment results corresponding to the preceding public opinion text and the current public opinion text are obtained. Then, the text sentiment results corresponding to the preceding public opinion text and the current public opinion text are compared and processed to determine the changes in the direction and degree of emotional tendency between the two, and the offset is obtained. Then, based on the offset, the degree of emotional change of the current public opinion text in the dissemination process is judged to obtain the emotional dissemination distortion information corresponding to the current public opinion text.

[0082] S30: Determine the negative sentiment contribution value corresponding to each public opinion text based on the text sentiment results and sentiment propagation distortion information.

[0083] In this embodiment, the negative sentiment contribution value refers to the quantitative result of the accumulation of negative sentiment corresponding to the target topic under the combined effect of the text sentiment result and the information of sentiment dissemination distortion. It is used to characterize the degree of influence of the corresponding public opinion text on the negative sentiment change of the target topic.

[0084] Specifically, when determining the negative emotional contribution value of each public opinion text based on the text sentiment results and the emotional dissemination distortion information, the text sentiment results of each public opinion text are first judged to determine the negative expression of each public opinion text in terms of emotional tendency direction and emotional degree. Then, combined with the emotional dissemination distortion information of each public opinion text, the degree of deviation of the negative expression in the dissemination process is corrected so that the negative degree represented in the text sentiment results and the degree of emotional deviation represented in the emotional dissemination distortion information form a correspondence. Then, based on the corrected correspondence, the negative emotional contribution value of each public opinion text is determined respectively.

[0085] S40: Perform cumulative calculation and attenuation correction on each negative sentiment contribution value in chronological order to obtain the negative sentiment accumulation potential energy corresponding to the target topic.

[0086] In this embodiment, the negative sentiment accumulation potential energy refers to the comprehensive result formed by the continuous accumulation and decay correction of the negative sentiment contribution value corresponding to each public opinion text over time, which is used to characterize the accumulation state of the negative sentiment corresponding to the target topic in the continuous dissemination process.

[0087] Specifically, to obtain the negative sentiment accumulation potential energy corresponding to the target topic, the negative sentiment contribution values ​​are accumulated and attenuated according to the time sequence of each negative sentiment text. First, the negative sentiment contribution values ​​are sorted according to the release time sequence of each public opinion text to form a time sequence arrangement result corresponding to the time sequence. Then, based on the time sequence arrangement result, the negative sentiment contribution value corresponding to the current time sequence position is accumulated with the negative sentiment contribution value corresponding to the previous time sequence position in chronological order to obtain the stage accumulation result corresponding to each time sequence position. Then, combined with the time interval between adjacent time sequence positions, the stage accumulation result corresponding to the previous time sequence position is attenuated and corrected to reduce the continuous impact of historical negative sentiment contribution values ​​with large time intervals on the current time sequence position. Finally, the attenuated and corrected stage accumulation result is superimposed with the negative sentiment contribution value corresponding to the current time sequence position to determine the negative sentiment accumulation potential energy corresponding to the target topic.

[0088] S50: Based on the changes in the potential energy of negative emotions accumulated over a continuous period of time, determine the inflection point of negative diffusion corresponding to the target theme.

[0089] In this embodiment, a continuous time period refers to multiple adjacent time intervals formed by sequentially dividing the analysis period corresponding to the target topic according to a preset time division rule. The change result refers to the increase or decrease of the negative emotional accumulation potential energy between each continuous time period. The negative diffusion inflection point refers to the key time position corresponding to when the negative emotional accumulation potential energy changes from a relatively slow change to a significantly enhanced change within a continuous time period. It is used to characterize the turning point where the negative emotions corresponding to the target topic enter a state of intensified diffusion.

[0090] Specifically, when determining the negative diffusion inflection point corresponding to the target theme based on the changes in the negative emotional accumulation potential energy over a continuous time period, the analysis process corresponding to the negative emotional accumulation potential energy is first divided into multiple continuous time periods according to a preset time division rule, and the negative emotional accumulation potential energy value corresponding to each continuous time period is determined. Then, the negative emotional accumulation potential energy values ​​corresponding to adjacent continuous time periods are compared to obtain the changes in the negative emotional accumulation potential energy between continuous time periods. Based on the changes between each continuous time period, it is determined whether the negative emotional accumulation potential energy has changed from a continuous accumulation state to a significantly enhanced state. If the determination result indicates that the negative emotional accumulation potential energy has shown an enhancing inflection point, the position of the corresponding continuous time period is determined as the negative diffusion inflection point corresponding to the target theme.

[0091] S60: Generate target public opinion sentiment analysis results based on the negative diffusion inflection point.

[0092] In this embodiment, the target public opinion sentiment analysis result refers to the result information formed after analyzing the negative sentiment diffusion state corresponding to the target topic based on the negative diffusion inflection point, which is used to characterize the public opinion sentiment change state corresponding to the target topic.

[0093] Specifically, when generating target public opinion sentiment analysis results based on the negative diffusion inflection point, the position of the negative diffusion inflection point in the analysis process corresponding to the target topic is first determined, and the negative diffusion inflection point is correlated with the changes in the negative sentiment accumulation potential energy before and after the negative diffusion inflection point. Then, based on the position of the negative diffusion inflection point and the changes in the negative sentiment accumulation potential energy before and after the negative diffusion inflection point, the negative sentiment diffusion state corresponding to the target topic is judged, and the result information corresponding to the public opinion sentiment change state corresponding to the target topic is obtained. The result information is then determined as the target public opinion sentiment analysis result.

[0094] In one embodiment, step S20, namely, determining the sentiment propagation distortion information corresponding to the current public opinion text based on the offset between the sentiment result corresponding to the preceding public opinion text that has a propagation association with the current public opinion text and the sentiment result corresponding to the current public opinion text, includes:

[0095] S201: Extract the preceding sentiment polarity information and preceding sentiment intensity information corresponding to the preceding public opinion text, and extract the current sentiment polarity information and current sentiment intensity information corresponding to the current public opinion text.

[0096] In this embodiment, prior sentiment polarity information refers to the result information in the text sentiment result corresponding to the prior public opinion text used to characterize the direction of sentiment tendency, prior sentiment intensity information refers to the result information in the text sentiment result corresponding to the prior public opinion text used to characterize the degree of sentiment tendency, current sentiment polarity information refers to the result information in the text sentiment result corresponding to the current public opinion text used to characterize the direction of sentiment tendency, and current sentiment intensity information refers to the result information in the text sentiment result corresponding to the current public opinion text used to characterize the degree of sentiment tendency.

[0097] Specifically, when extracting the preceding sentiment polarity information and preceding sentiment intensity information corresponding to the preceding public opinion text, and extracting the current sentiment polarity information and current sentiment intensity information corresponding to the current public opinion text, the text sentiment results corresponding to the preceding public opinion text and the text sentiment results corresponding to the current public opinion text are read separately. Then, the result content used to represent the direction of emotional tendency is separated from the text sentiment results corresponding to the preceding public opinion text as preceding sentiment polarity information, and the result content used to represent the degree of emotional tendency is separated from the text sentiment results corresponding to the preceding public opinion text as preceding sentiment intensity information. Then, the result content used to represent the direction of emotional tendency is separated from the text sentiment results corresponding to the current public opinion text as current sentiment polarity information, and the result content used to represent the degree of emotional tendency is separated from the text sentiment results corresponding to the current public opinion text as current sentiment intensity information.

[0098] S202: Based on the preceding sentiment polarity information and the current sentiment polarity information, determine the sentiment direction offset information corresponding to the current public opinion text.

[0099] In this embodiment, the sentiment direction offset information refers to the result information formed after the current public opinion text changes in the sentiment tendency direction relative to the previous public opinion text. It is used to characterize the change in the sentiment tendency direction corresponding to the current public opinion text relative to the sentiment tendency direction corresponding to the previous public opinion text.

[0100] Specifically, when determining the emotional direction shift information corresponding to the current public opinion text based on the preceding emotional polarity information and the current emotional polarity information, the preceding emotional polarity information is first used as the representation basis for the preceding emotional tendency direction, and the current emotional polarity information is used as the representation basis for the current emotional tendency direction. Then, the preceding emotional polarity information and the current emotional polarity information are compared to determine whether the current emotional tendency direction has changed relative to the preceding emotional tendency direction. Finally, based on the determination result after the comparison, the emotional direction shift information corresponding to the current public opinion text is determined.

[0101] S203: Determine the sentiment intensity offset information corresponding to the current public opinion text based on the preceding sentiment intensity information and the current sentiment intensity information.

[0102] In this embodiment, the emotional intensity shift information refers to the result information formed after the current public opinion text changes in the degree of emotional tendency relative to the previous public opinion text. It is used to characterize the change in the degree of emotional tendency corresponding to the current public opinion text relative to the degree of emotional tendency corresponding to the previous public opinion text.

[0103] Specifically, when determining the emotional intensity shift information corresponding to the current public opinion text based on the preceding emotional intensity information and the current emotional intensity information, the preceding emotional intensity information is first used as the basis for representing the degree of the preceding emotional tendency, and the current emotional intensity information is used as the basis for representing the degree of the current emotional tendency. Then, the preceding emotional intensity information and the current emotional intensity information are compared to determine whether the degree of the current emotional tendency has changed relative to the degree of the preceding emotional tendency. Finally, based on the determination result after the comparison, the emotional intensity shift information corresponding to the current public opinion text is determined.

[0104] S204: Extract the core stance information from the previous public opinion text and the current public opinion text respectively, and determine the stance deviation information corresponding to the current public opinion text based on the differences between the core stance information in the previous public opinion text and the core stance information in the current public opinion text.

[0105] In this embodiment, core stance expression information refers to the expression content in the public opinion text used to characterize the basic attitude, evaluation orientation, or judgment tendency of the corresponding text towards the target topic. Stance deviation information refers to the result information formed after the current public opinion text changes in the core stance expression content relative to the previous public opinion text, and is used to characterize the deviation of the stance orientation corresponding to the current public opinion text from the stance orientation corresponding to the previous public opinion text.

[0106] Specifically, the core stance information is extracted from both the preceding and current public opinion texts. To determine the stance deviation information in the current public opinion text based on the differences between these two texts, the process involves first merging and judging the statements in the preceding text to identify the expressions representing the basic attitude of the preceding text towards the target topic, which are then considered the core stance information in the preceding text. Next, the statements in the current text are merged and judged to identify the expressions representing the basic attitude of the current text towards the target topic, which are also considered the core stance information in the current text. Then, the core stance information in the preceding and current texts is compared to determine the differences in stance orientation, evaluation focus, and attitude tendency. Finally, based on these differences, the stance deviation information in the current text is determined.

[0107] S205: Based on the emotional direction offset information, emotional intensity offset information, and stance deviation information, determine the emotional dissemination distortion information corresponding to the current public opinion text.

[0108] In this embodiment, the emotional dissemination distortion information refers to the result information formed after the current public opinion text deviates from the previous public opinion text in terms of emotional tendency direction, emotional tendency degree and stance orientation during the dissemination process. It is used to characterize the deviation of the current public opinion text in emotional expression during the dissemination process.

[0109] Specifically, when determining the emotional dissemination distortion information corresponding to the current public opinion text based on the emotional direction shift information, emotional intensity shift information, and stance deviation information, the emotional direction shift information is first used as the basis for the change in the emotional tendency direction of the current public opinion text, the emotional intensity shift information is used as the basis for the change in the emotional tendency degree of the current public opinion text, and the stance deviation information is used as the basis for the change in the stance orientation of the current public opinion text. Then, the emotional direction shift information, emotional intensity shift information, and stance deviation information are jointly judged to determine the comprehensive deviation of the current public opinion text in the dissemination process. Finally, based on the comprehensive deviation, the emotional dissemination distortion information corresponding to the current public opinion text is determined.

[0110] In one embodiment, step S30, namely determining the negative sentiment contribution value corresponding to each public opinion text based on the text sentiment result and sentiment propagation distortion information, includes:

[0111] S301: Based on the text sentiment results, determine the basic negative sentiment value corresponding to each public opinion text.

[0112] In this embodiment, the basic negative sentiment value refers to the basic result value determined by each public opinion text based on the corresponding text sentiment result, which is used to characterize the degree of negative sentiment of the corresponding public opinion text itself.

[0113] Specifically, when determining the basic negative sentiment value corresponding to each public opinion text based on the text sentiment results, the text sentiment results corresponding to each public opinion text are first read separately. Based on the content of the results representing the direction and degree of emotional tendency in each text sentiment result, it is determined whether each public opinion text has a negative emotional tendency. If the determination result indicates that the corresponding public opinion text has a negative emotional tendency, the basic negative sentiment value of the corresponding public opinion text is determined based on the degree of negative emotion represented in the text sentiment results. If the determination result indicates that the corresponding public opinion text does not have a negative emotional tendency, the basic negative sentiment value of the corresponding public opinion text is determined as the preset basic value.

[0114] S302: Analyze the distorted information in the emotional communication and obtain the directional distortion value, intensity distortion value, and stance distortion value corresponding to each public opinion text.

[0115] In this embodiment, the directional distortion value refers to the result value in the emotional transmission distortion information used to characterize the degree of deviation of the current public opinion text from the previous public opinion text in the direction of emotional tendency; the intensity distortion value refers to the result value in the emotional transmission distortion information used to characterize the degree of deviation of the current public opinion text from the previous public opinion text in the degree of emotional tendency; and the stance distortion value refers to the result value in the emotional transmission distortion information used to characterize the degree of deviation of the current public opinion text from the previous public opinion text in terms of stance orientation.

[0116] Specifically, when determining the distortion information in emotional communication and obtaining the direction distortion value, intensity distortion value, and stance distortion value corresponding to each public opinion text, the preceding emotional polarity information and the current emotional polarity information are first mapped to polarity parameters. and In this context, positive values ​​are assigned 1, neutral values ​​are assigned 0, and negative values ​​are assigned -1. Furthermore, prior and current sentiment intensity information are mapped to intensity parameters. and ,in, and The values ​​of all values ​​are in the range of [0,1]. Then, the core stance information expressed in the preceding public opinion text and the core stance information expressed in the current public opinion text are mapped to stance parameters respectively. and ,in, and The values ​​of all values ​​are in the range of [0,1], and then a negative inductance coefficient is constructed. The calculation formula is as follows: Among them, the negative transfer coefficient This is used to characterize the degree to which the current public opinion text has shifted in a negative direction relative to the previous public opinion text, and then a stance gating coefficient is constructed. The calculation formula is as follows: ,in, To preset the position coupling coefficient, and then based on the polarity parameter and Directional differences and negative conversion coefficients between them and position gating coefficient Determine the directional distortion value The calculation formula is as follows: ,in, To pre-determine the negative amplification factor, and then based on the strength parameters and Differences in degree between them, negative transfer coefficient and position gating coefficient Determine the intensity distortion value The calculation formula is as follows: ,in, To preset the intensity linkage coefficient, and then based on the position parameters and Differences in positions and negative transfer coefficient between them and strength parameters and The degree of difference between them determines the position distortion value. The calculation formula is as follows: ,in, To preset the negative position linkage coefficient, To preset the strength-position coupling coefficient, finally, the directional distortion value is... Intensity distortion value and position distortion value The directional distortion value, intensity distortion value, and stance distortion value were respectively determined for each public opinion text.

[0117] S303: Determine the distortion correction value corresponding to each public opinion text based on the directional distortion value, intensity distortion value, and position distortion value.

[0118] In this embodiment, the distortion correction value refers to the result value formed after jointly correcting the directional distortion value, intensity distortion value, and position distortion value, and is used to characterize the comprehensive distortion correction degree generated by the corresponding public opinion text during the dissemination process.

[0119] Specifically, when determining the distortion correction value for each public opinion text based on the orientation distortion value, intensity distortion value, and stance distortion value, the orientation distortion value is first recorded as follows: The intensity distortion value is denoted as The position distortion value is recorded as Reconstruct the directional dominant coefficient The calculation formula is as follows: ,in, A preset smoothing coefficient is used to avoid the denominator being zero, and then the position traction coefficient is constructed. Its calculation formula is Then based on the orientation distortion value Intensity distortion value and position distortion value The coupling relationship between them determines the joint distortion. The calculation formula is as follows: ,in, For the preset strength coupling coefficient, To preset the position traction coefficient, The preset directional linkage coefficient is then used, based on the joint distortion amount. With direction distortion value The degree of deviation between them constructs the distortion focusing coefficient. The calculation formula is as follows: ,in, To preset the focusing coefficient, finally, based on the joint distortion... and distortion focus coefficient Determine the distortion correction value The calculation formula is as follows: and the distortion correction value The distortion correction value is determined for each public opinion text.

[0120] S304: Determine the corresponding propagation correction value for each public opinion text based on its propagation hierarchy, propagation response, and follow-up information within the propagation-related information.

[0121] In this embodiment, the propagation hierarchy information refers to the result information of the hierarchical position of each public opinion text in the propagation chain corresponding to the propagation association information; the propagation response information refers to the result information related to the degree of activity of the propagation feedback corresponding to each public opinion text; the subsequent follow-up information refers to the result information corresponding to the follow-up propagation situation formed after each public opinion text continues along the propagation association information; and the propagation correction value refers to the result value formed after correcting the degree of diffusion influence of the corresponding public opinion text in the propagation process according to the propagation hierarchy information, propagation response information, and subsequent follow-up information, which is used to characterize the degree of propagation amplification of the corresponding public opinion text in the propagation chain.

[0122] Specifically, when determining the propagation correction value for each public opinion text based on its propagation hierarchy information, propagation response information, and subsequent follow-up information within the propagation-related information, the propagation hierarchy information corresponding to each public opinion text is first mapped to a hierarchy position parameter, denoted as... The dissemination response information corresponding to each public opinion text is mapped to a response activity parameter, denoted as... And map the subsequent follow-up information corresponding to each public opinion text to follow-up extension parameters, denoted as Among them, the hierarchical position parameter Response active parameters and follow extension parameters All parameters are normalized, with values ​​ranging from [0,1]. Then, hierarchical influence values ​​are constructed. The calculation formula is as follows: Where 'a' is the preset hierarchical influence coefficient, and the hierarchical influence value is... This is used to characterize the impact of the corresponding public opinion text's position in the dissemination chain on the degree of amplification, and then constructs the response activity value. The calculation formula is as follows: Where b is the preset response influence coefficient, c is the preset response following correlation coefficient, and the response activity value is... To characterize the impact of the level of activity in the dissemination feedback of the corresponding public opinion text on the degree of dissemination amplification, a follow-up extension value is then constructed. The calculation formula is as follows: Where d is the preset following influence coefficient, e is the preset following compensation coefficient, and the following extension value is... This is used to characterize the extent to which related public opinion texts continue to spread and form connections, and then based on hierarchical position parameters. Response active parameters and follow extension parameters Constructing the basic correction value for propagation The calculation formula is as follows: Where p, q, and r are the preset level weight, preset response weight, and preset follow weight, respectively, and satisfy the following conditions: Finally, based on the basic propagation correction value Hierarchical influence value Response activity value and follow extension value Determine the propagation correction value The calculation formula is as follows: And will propagate the correction value The propagation correction values ​​were determined for each piece of public opinion text.

[0123] S305: Determine the negative sentiment contribution value corresponding to each public opinion text based on the basic negative sentiment value, distortion correction value, and dissemination correction value.

[0124] Specifically, when determining the negative sentiment contribution value corresponding to each public opinion text based on the basic negative sentiment value, distortion correction value, and propagation correction value, the basic negative sentiment value corresponding to each public opinion text is first recorded as follows: The distortion correction value corresponding to each public opinion text is recorded as . The propagation correction value corresponding to each public opinion text is denoted as Among them, the basic negative sentiment value The distortion correction value E_i is used to characterize the degree of negative emotion in the corresponding public opinion text itself, while the propagation correction value is used to characterize the degree of emotional deviation generated by the corresponding public opinion text during its dissemination. This is used to characterize the degree of amplification of the corresponding public opinion text in the dissemination chain, and then based on the distortion correction value. and propagation correction value The synergistic contribution value is constructed from the interaction of the two parties. The calculation formula is as follows: ,in, A preset smoothing coefficient is used to avoid the denominator being zero, and the collaborative contribution value is... Used to characterize distortion correction values and propagation correction value The degree of joint enhancement of negative sentiment contribution at the same high level, based on the distortion correction value. and propagation correction value Constructing a difference modifier based on the differences between them The calculation formula is as follows: ,in, To preset the difference adjustment coefficient, the difference adjustment value Used to characterize distortion correction values and propagation correction value The degree of moderating contribution to negative sentiment when there are large differences, and then based on the baseline negative sentiment value. Collaborative contribution value and difference adjustment value Determine the negative sentiment contribution value The calculation formula is as follows: ,in, A preset collaborative contribution coefficient is used to adjust the collaborative contribution value. The proportion of influence in the negative sentiment contribution value, and finally the negative sentiment contribution value The negative sentiment contribution value corresponding to each public opinion text was determined.

[0125] In one embodiment, step S40 involves performing cumulative calculations and attenuation corrections on each negative sentiment contribution value in chronological order to obtain the accumulated negative sentiment potential energy corresponding to the target topic, including:

[0126] S401: Sort the negative sentiment contribution values ​​of each public opinion text according to the release time order to obtain the time-series negative sentiment contribution sequence corresponding to the target topic.

[0127] In this embodiment, the time-series negative sentiment contribution sequence refers to the sequence result formed by arranging the negative sentiment contribution values ​​corresponding to each public opinion text in the order of publication time, which is used to characterize the sequential distribution of the negative sentiment contribution corresponding to the target topic in the time dimension.

[0128] Specifically, when sorting the negative sentiment contribution values ​​corresponding to each public opinion text according to their release time to obtain the time-series negative sentiment contribution sequence corresponding to the target topic, the release time corresponding to each public opinion text is first recorded as... The negative sentiment contribution value corresponding to each public opinion text is recorded as follows: Then by release time Based on the main sorting criteria, for each group The negative sentiment contribution values ​​are sorted in ascending order so that those published earlier are at the beginning of the sequence and those published later are at the end. In cases where multiple sentiment texts correspond to the same publication time, a priority value for the same time slot is further constructed. The calculation formula is as follows: ,in, To preset the simultaneous propagation influence coefficient, This is the correction value for the dissemination of the corresponding public opinion text, and the priority value at the same moment. This is used to characterize the priority of negative sentiment contribution values ​​in the propagation chain under the same release time condition. Then, among the public opinion texts with the same release time, they are ranked according to the priority value at the same moment. The values ​​are then sorted in descending order, and finally, each negative sentiment contribution value is written into its corresponding time series position to obtain the time series negative sentiment contribution sequence for the target topic, denoted as: ,in, This represents the negative sentiment contribution value located at the first time position after sorting. This represents the negative sentiment contribution value located at the last time sequence position after sorting, and m represents the total number of public opinion texts corresponding to the target topic.

[0129] S402: Determine the time decay value corresponding to each negative sentiment contribution value based on the time interval between adjacent negative sentiment contribution values ​​in the time-series negative sentiment contribution sequence.

[0130] In this embodiment, the time interval refers to the interval between the release times of adjacent negative sentiment contribution values ​​in the time-series negative sentiment contribution sequence, and the time decay value refers to the result value formed after characterizing the degree of the continuous effect of the corresponding negative sentiment contribution value in the subsequent time sequence position according to the time interval, which is used to characterize the degree of weakening of the effect of the corresponding negative sentiment contribution value over time.

[0131] Specifically, when determining the time decay value corresponding to each negative sentiment contribution value based on the time interval between adjacent negative sentiment contribution values ​​in the time-series negative sentiment contribution sequence, the release time corresponding to the j-th time position in the time-series negative sentiment contribution sequence is first denoted as... And the time interval between the j-th time position and the (j-1)-th time position is denoted as The calculation formula is as follows: ,in, Reconstruct interval retention values The calculation formula is as follows: ,in, The preset interval attenuation coefficient and the interval retention value. This is used to characterize the degree to which the preceding negative sentiment contribution value is retained in the current time series position as the time interval between adjacent release times increases, and then a timeout adjustment value is constructed. The calculation formula is as follows: ,in, The preset timeout adjustment coefficient is used. The timeout adjustment value is the average time interval between adjacent release times in the time-series negative sentiment contribution sequence. This is used to characterize the additional attenuation when the current time interval exceeds the average time interval, and then based on the interval retention value. and timeout adjustment value Determine the time decay value corresponding to the j-th timing position. The calculation formula is as follows: Among them, time decay value The value range is [0,1). Smaller time decay values ​​correspond to shorter time intervals, while larger time decay values ​​correspond to longer time intervals exceeding the average time interval. Finally, the time decay values ​​corresponding to each time sequence position are... Each negative emotional contribution value was determined as its corresponding time decay value.

[0132] S403: Determine the incremental reinforcement value corresponding to each negative sentiment contribution value based on the growth difference between the current negative sentiment contribution value and the previous negative sentiment contribution value in the time-series negative sentiment contribution sequence.

[0133] In this embodiment, the growth difference refers to the growth change of the current negative sentiment contribution value relative to the previous negative sentiment contribution value in the time-series negative sentiment contribution sequence. The incremental enhancement value refers to the result value formed by characterizing the degree of enhancement of the current negative sentiment contribution value in the time-series accumulation process based on the growth difference, and is used to characterize the incremental amplification degree of the current negative sentiment contribution value relative to the previous negative sentiment contribution value.

[0134] Specifically, when determining the incremental reinforcement value corresponding to each negative sentiment contribution value based on the growth difference between the current negative sentiment contribution value and the previous negative sentiment contribution value in the temporal negative sentiment contribution sequence, the negative sentiment contribution value corresponding to the j-th temporal position in the temporal negative sentiment contribution sequence is first denoted as... The negative sentiment contribution value corresponding to the (j-1)th time position is denoted as ,in, Then, the difference between the current negative sentiment contribution value and the previous negative sentiment contribution value is recorded as... The calculation formula is as follows: Then construct the effective growth value. The calculation formula is as follows: Among them, effective growth value This is used to characterize the positive growth portion of the current negative sentiment contribution value relative to the previous negative sentiment contribution value, and then constructs the relative growth value. The calculation formula is as follows: ,in, To preset the relative growth coefficient, The preset smoothing coefficient, Used to avoid a denominator of zero, relative growth value This is used to characterize the relative growth rate of the current negative sentiment contribution value compared to the previous negative sentiment contribution value, and then a super-average growth value is constructed. The calculation formula is as follows: ,in, To preset the super-average growth coefficient, The average of all effective growth values ​​in the time-series negative sentiment contribution sequence, and the super-average growth value. This is used to characterize the additional enhancement corresponding to the current effective growth value exceeding the average growth level, and finally based on the effective growth value. Relative growth value and super-average growth value Determine the incremental enhancement value corresponding to the j-th time sequence position. The calculation formula is as follows: and incremental enhancement value The incremental reinforcement value is determined as the value corresponding to each negative emotional contribution.

[0135] S404: Determine the stage accumulation value corresponding to each time sequence position based on each negative emotion contribution value, each time decay value, and each incremental reinforcement value.

[0136] In this embodiment, the stage accumulation value refers to the stage accumulation result value formed by the combined effect of the corresponding negative sentiment contribution value, time decay value and incremental reinforcement value at each time position, which is used to characterize the degree of accumulation of negative sentiment at the corresponding time position in the current stage.

[0137] Specifically, when determining the stage accumulation value corresponding to each time sequence position based on each negative sentiment contribution value, each time decay value, and each incremental reinforcement value, the negative sentiment contribution value corresponding to the j-th time sequence position is used as the basis. Time decay value and incremental enhancement value To perform phase accumulation calculations, first, based on the time decay value... Build the current retained value The calculation formula is as follows: , where the current retained value This is used to characterize the degree to which the negative sentiment contribution value corresponding to the j-th time sequence position is retained after the time decay effect, and then based on the incremental reinforcement value. With negative sentiment contribution value The proportional relationship between them constructs the incremental amplification value The calculation formula is as follows: ,in, To preset the incremental amplification factor, The preset smoothing coefficient and incremental amplification value This is used to characterize the amplification effect of the incremental reinforcement value on the current negative sentiment contribution value, and then based on the time decay value. With incremental enhancement value The combined effect of these factors constructs the phase adjustment value. The calculation formula is as follows: ,in, The preset stage adjustment coefficient and stage adjustment value are... This is used to characterize the moderating effect of the combined effects of time decay and incremental reinforcement on the accumulation level at the current time-series position, and finally based on the negative sentiment contribution value. Current retained values Incremental amplification value and stage adjustment value Determine the stage accumulation value corresponding to the j-th time position. The calculation formula is as follows: and the accumulated value in stages The stage accumulation value is determined as the corresponding time position.

[0138] S405: Continuously accumulate the accumulated values ​​of each stage in chronological order to obtain the negative emotional potential energy corresponding to the target theme.

[0139] Specifically, the accumulated values ​​of each stage are continuously accumulated in chronological order. When the negative emotional potential energy corresponding to the target topic is obtained, the stage accumulated value corresponding to the j-th time position is recorded as... Let the negative emotional accumulation potential energy corresponding to the j-th time sequence position be denoted as And the negative emotional accumulation potential energy corresponding to the first time sequence position is determined as: For the j-th time sequence position, where Based on the negative emotional energy accumulated in the previous time sequence position The stage accumulation value corresponding to the current time sequence position and the stage accumulation value corresponding to the previous time series position. Determine the negative emotional accumulation potential corresponding to the j-th time sequence position. The calculation formula is as follows: ,in, As a historical retention coefficient, and The historical retention coefficient is used to characterize the degree to which the negative sentiment accumulation potential corresponding to the previous time position is retained in the current time position. The adjacent linkage coefficient is used to characterize the degree of joint influence of the stage accumulation values ​​corresponding to two adjacent time positions on the current negative sentiment accumulation potential energy. The stage leap coefficient is used to characterize the degree of enhancement of the current stage accumulation value to the current negative emotional accumulation potential energy when the current stage accumulation value is higher than the previous stage accumulation value. To pre-determine smoothing coefficients and avoid zero denominators, each step is determined sequentially according to its timing position. The negative emotional accumulation potential energy corresponding to the final time sequence position is determined as the negative emotional accumulation potential energy corresponding to the target theme.

[0140] In one embodiment, step S50, namely, determining the negative diffusion inflection point corresponding to the target topic based on the change in negative emotional accumulation potential energy over a continuous time period, includes:

[0141] S501: Divide the accumulated potential energy of negative emotions into intervals according to continuous time periods to obtain multiple continuous potential energy intervals.

[0142] In this embodiment, the continuous potential energy interval refers to the result of multiple consecutive intervals formed by dividing the time process corresponding to the accumulation of negative emotional potential energy into continuous time periods. It is used to characterize the stage distribution of the accumulation of negative emotional potential energy in different continuous time ranges.

[0143] Specifically, when dividing the accumulated negative emotional energy into intervals according to continuous time periods to obtain multiple continuous energy intervals, the release time corresponding to the j-th time sequence position is denoted as... Let the negative emotional accumulation potential energy corresponding to the j-th time sequence position be denoted as For two adjacent time positions, the time proximity value is first determined based on the interval between the release times. The calculation formula is as follows: ,in, The preset time proximity coefficient, time proximity value This is used to characterize the temporal proximity of adjacent time-series locations, and then the potential adjacency value is determined based on the change in the potential energy between adjacent negative emotional accumulations. The calculation formula is as follows: ,in, Preset potential energy adjacency coefficient, potential energy adjacency value This is used to characterize the adjacency of adjacent temporal locations in terms of changes in negative sentiment accumulation potential, and then based on temporal proximity values. Adjacency value of potential energy Determine the interval merge value The calculation formula is as follows: Among them, the interval merged value This is used to characterize whether the (j-1)th time series position and the jth time series position are suitable to be included in the same continuous potential energy interval, and then the interval merging value is used. With preset merging threshold In comparison, In this case, the j-th time position is assigned to the same continuous potential energy interval as the (j-1)-th time position. In the case of the j-th time sequence position, the starting position of the new continuous potential energy interval is taken as the starting position of each continuous potential energy interval. Finally, the negative emotional accumulation potential energy is sequentially divided according to the starting position of each continuous potential energy interval to obtain multiple continuous potential energy intervals.

[0144] S502: Based on the negative emotional accumulation potential energy change value corresponding to each continuous potential energy interval, determine the potential energy growth value corresponding to each continuous potential energy interval.

[0145] In this embodiment, the negative emotion accumulation potential energy change value refers to the change result value of the negative emotion accumulation potential energy in each continuous potential energy interval relative to the starting position of the corresponding interval, and the potential energy growth value refers to the result value determined based on the negative emotion accumulation potential energy change value corresponding to each continuous potential energy interval, used to characterize the degree of growth of negative emotion accumulation potential energy in the corresponding continuous potential energy interval.

[0146] Specifically, when determining the potential energy growth value corresponding to each continuous potential energy interval based on the change value of negative emotional accumulation potential energy corresponding to each continuous potential energy interval, the negative emotional accumulation potential energy corresponding to the starting position of the k-th continuous potential energy interval is denoted as... Let the negative emotional accumulation potential energy corresponding to the end position of the kth continuous potential energy interval be denoted as Let the negative emotional accumulation potential energy corresponding to the u-th interval position within the k-th continuous potential energy interval be denoted as ,in, , Let represent the total number of positions within the k-th consecutive potential energy interval. Then, based on the difference in potential energy between the beginning and end of the k-th consecutive potential energy interval and the continuous increase within the interval, determine the change in potential energy corresponding to the k-th consecutive potential energy interval. The calculation formula is as follows: ,in, The preset range variation adjustment coefficient, The interval potential energy change is set to a preset smoothing coefficient. This is used to characterize the overall change in negative emotional accumulation potential energy within the k-th continuous potential energy interval, and then based on the continuous upward portion within the k-th continuous potential energy interval relative to the interval potential energy change. The degree of concentration is used to determine the interval rising concentration value corresponding to the k-th continuous potential energy interval. The calculation formula is as follows: ,in, Preset rising concentration coefficient, interval rising concentration value This is used to characterize the degree of concentration of negative emotional accumulation potential energy when it shows a continuous upward trend within the k-th consecutive potential energy interval. Finally, it is based on the change in potential energy within the interval. and interval rising concentration value Determine the interval potential energy growth value corresponding to the k-th continuous potential energy interval. The calculation formula is as follows: And the interval potential energy growth value The potential energy growth value corresponding to each continuous potential energy interval is determined.

[0147] S503: Determine the slope offset value corresponding to each continuous potential energy interval based on the difference in the change of potential energy growth value between adjacent continuous potential energy intervals.

[0148] In this embodiment, the slope offset value refers to the result value formed after the potential energy growth value corresponding to adjacent continuous potential energy intervals changes in the interval growth trend, and is used to characterize the degree of offset of the corresponding continuous potential energy interval in the growth slope.

[0149] Specifically, when determining the slope offset value corresponding to each continuous potential energy interval based on the difference in the change of potential energy growth values ​​corresponding to adjacent continuous potential energy intervals, the interval potential energy growth value corresponding to the k-th continuous potential energy interval is denoted as... Let the total number of positions contained in the k-th continuous potential energy interval be denoted as . Then, based on the potential energy growth value of the interval Total number of positions in the interval Determine the interval growth rate corresponding to the k-th continuous potential energy interval. The calculation formula is as follows: ,in, A preset smoothing coefficient is used to avoid the denominator being zero, and the interval growth rate is... This is used to characterize the potential energy growth rate corresponding to a unit interval position within the k-th continuous potential energy interval, and then the interval growth rate corresponding to the k-th continuous potential energy interval is... The interval growth rate corresponding to the (k-1)th continuous potential energy interval Perform difference calculations to obtain the difference in growth rates. The calculation formula is as follows: Then based on the difference in growth rate Determine the positive offset value The calculation formula is as follows: ,in, Preset positive offset coefficient, positive offset value This is used to characterize the degree of growth amplification when the growth rate of the current continuous potential energy interval is higher than that of the previous continuous potential energy interval. The adjacent offset value is then determined based on the change in the growth rate of the intervals corresponding to two adjacent continuous potential energy intervals. The calculation formula is as follows: ,in, Preset adjacent offset coefficients, adjacent offset values This is used to characterize the overall change in the growth rate between adjacent continuous potential energy intervals, and finally based on the difference in growth rate. Positive offset value and adjacent offset values Determine the slope offset value corresponding to the k-th continuous potential energy interval. The calculation formula is as follows: and slope offset value The slope offset value is determined as the value corresponding to each continuous potential energy interval.

[0150] S504: Based on the amplified difference between the potential energy growth value corresponding to the current continuous potential energy interval and the potential energy growth value corresponding to the previous continuous potential energy interval, determine the sudden shift value corresponding to each continuous potential energy interval.

[0151] In this embodiment, the amplification difference refers to the amplified change result formed by the potential energy growth value corresponding to the current continuous potential energy interval relative to the potential energy growth value corresponding to the previous continuous potential energy interval. The sudden increase offset value refers to the result value formed after characterizing the sudden increase in potential energy growth of the corresponding continuous potential energy interval based on the amplification difference. It is used to characterize the degree of offset of the corresponding continuous potential energy interval in the process of negative emotion diffusion, showing an abnormal amplification trend.

[0152] Specifically, when determining the surge offset value for each continuous potential energy interval based on the amplified difference between the potential energy growth value corresponding to the current continuous potential energy interval and the potential energy growth value corresponding to the previous continuous potential energy interval, the interval potential energy growth value corresponding to the k-th continuous potential energy interval is denoted as... Let the interval potential energy growth value corresponding to the (k-1)th continuous potential energy interval be denoted as Then, the difference between the potential energy growth value corresponding to the current continuous potential energy interval and the potential energy growth value corresponding to the previous continuous potential energy interval is denoted as... The calculation formula is as follows: Then based on the difference Building an effective amplification volume The calculation formula is as follows: Among them, the effective amplification volume This is used to characterize the effective amplification portion of the interval potential energy growth value corresponding to the current continuous potential energy interval, which is higher than the interval potential energy growth value corresponding to the previous continuous potential energy interval. Then, based on the effective amplification portion... Interval potential energy growth value corresponding to the preceding continuous potential energy interval The proportional relationship between them is used to construct the relative magnification value. The calculation formula is as follows: ,in, A preset smoothing coefficient is used to avoid the denominator being zero, resulting in a relative amplification value. This is used to characterize the relative amplification of the interval potential energy growth value corresponding to the current continuous potential energy interval compared to the interval potential energy growth value corresponding to the previous continuous potential energy interval. Then, based on the relative amplification value... The difference between the average relative amplification values ​​corresponding to multiple consecutive potential energy intervals is used to construct the super-average amplification value. The calculation formula is as follows: ,in, This is the preset super-average amplification factor. The average of the relative amplification values ​​corresponding to multiple continuous potential energy intervals, and the super-average amplification value. This is used to characterize the additional enhancement when the relative amplification value corresponding to the current continuous potential energy range exceeds the overall average amplification level, and then based on the effective amplification amount. The interval potential energy growth value corresponding to the current continuous potential energy interval The proportion in, construct the new proportion value The calculation formula is as follows: ,in, To preset the percentage increase coefficient, the percentage increase value is... This is used to characterize the proportion of the newly amplified portion in the current continuous potential energy range within the range's potential energy growth value. Finally, based on the effective amplification... Super-average amplification value and the percentage of new additions Determine the sudden shift value corresponding to the k-th continuous potential energy interval. The calculation formula is as follows: and the sudden increase in offset value The sudden shift value is determined as the value corresponding to each continuous potential energy interval.

[0153] S505: Determine the continuous growth value corresponding to each continuous potential energy interval based on the direction of change of the potential energy growth value corresponding to multiple consecutive potential energy intervals.

[0154] In this embodiment, the direction of change of potential energy growth value refers to the direction of increase or decrease of potential energy growth value corresponding to multiple consecutive potential energy intervals in the interval order. The continuous growth value refers to the result value formed after characterizing the degree of continuation of the corresponding continuous potential energy interval in the continuous growth state based on the direction of change of potential energy growth value corresponding to multiple consecutive potential energy intervals. It is used to characterize the degree to which the corresponding continuous potential energy interval maintains a continuous growth trend during the spread of negative emotions.

[0155] Specifically, when determining the continuous growth value corresponding to each continuous potential energy interval based on the direction of change of the potential energy growth value corresponding to multiple consecutive potential energy intervals, the interval potential energy growth value corresponding to the k-th consecutive potential energy interval is denoted as... The direction of the k-th continuous potential energy interval relative to the previous continuous potential energy interval is denoted as . The calculation formula is as follows: Among them, the interval direction marker This is used to characterize the direction of change of the interval potential energy growth value corresponding to the current continuous potential energy interval relative to the interval potential energy growth value corresponding to the previous continuous potential energy interval, and then construct the number of consecutive rising segments. , among which, when When, it is determined recursively according to the following formula: ,when At that time, the number of consecutive ascending segments Set to the preset initial value of 1, number of consecutive ascending segments This is used to characterize the number of consecutive intervals where the potential energy growth value maintains a positive increase up to the k-th consecutive potential energy interval, and then constructs the upward continuation value. The calculation formula is as follows: ,in, The preset upward continuation coefficient and upward continuation value are used to define the upward continuation value. Used to characterize the degree of growth continuation corresponding to an increase in the number of consecutive upward segments, and then constructing the direction-maintaining value. The calculation formula is as follows: ,in, Here, h is the preset direction retention coefficient, and h is the preset number of lookback intervals. Preset smoothing coefficient, direction retention value This characterizes the stability of the potential energy growth value within a preset number of consecutive potential energy intervals preceding the current consecutive potential energy interval, maintaining a positive change. Finally, it is determined based on the number of consecutive upward segments. Continuation of the rise and direction retention value Determine the sustained growth value corresponding to the k-th continuous potential energy interval. The calculation formula is as follows: And will continue to grow in value It is determined as the continuous growth value corresponding to each continuous potential energy interval.

[0156] S506: Based on each slope offset value, each sudden increase offset value, and each continuous growth value, determine the inflection point trigger value corresponding to each continuous potential energy interval.

[0157] In this embodiment, the inflection point trigger value refers to the result value formed by jointly judging the slope offset value, sudden increase offset value and continuous growth value corresponding to each continuous potential energy interval, which is used to characterize the strength of triggering the negative diffusion inflection point in the corresponding continuous potential energy interval.

[0158] Specifically, when determining the inflection point trigger value corresponding to each continuous potential energy interval based on each slope offset value, each sudden increase offset value, and each sustained increase value, the slope offset value corresponding to the k-th continuous potential energy interval is denoted as... Let the sudden increase offset value corresponding to the k-th continuous potential energy interval be denoted as Let the continuous growth value corresponding to the k-th continuous potential energy interval be denoted as Then based on the slope offset value and sudden increase in offset value The interaction relationship constructs the offset synergistic value The calculation formula is as follows: ,in, A preset smoothing coefficient is used to avoid the denominator being zero and to offset the co-factor value. Used to characterize slope offset value and sudden increase in offset value The degree of combined enhancement of the inflection point trigger strength when both are at a high level, and then based on the sustained growth value. Constructing continuous weighted values The calculation formula is as follows: ,in, To preset the continuous weighting coefficients, the continuous weighting value Used to characterize sustained growth value The weighted influence of the inflection point trigger intensity, and then based on the slope offset value. and sudden increase in offset value The differences between them construct the difference identification value The calculation formula is as follows: ,in, To preset the difference recognition coefficient, the initial trigger value is then determined based on the slope offset value, sudden increase offset value, offset coordination value, continuous weighting value, and difference recognition value. k is calculated using the following formula: Then, construct persistent constraint values ​​based on the persistent growth values. The calculation formula is as follows: ,in, To preset the continuous constraint coefficient, the inflection point trigger value corresponding to the k-th continuous potential energy interval is finally determined based on the initial trigger value and the continuous constraint value. The calculation formula is as follows: and the inflection point trigger value The inflection point trigger value is determined as the value corresponding to each continuous potential energy interval.

[0159] S507: When the inflection point trigger value meets the preset inflection point conditions, the corresponding continuous potential energy range is determined as the negative diffusion inflection point of the target theme.

[0160] In this embodiment, the preset inflection point condition refers to the conditional rule used to determine whether the continuous potential energy range has reached the requirements for identifying the negative diffusion inflection point.

[0161] Specifically, when the inflection point trigger value meets the preset inflection point conditions, and the corresponding continuous potential energy interval is determined as the negative diffusion inflection point of the target theme, the inflection point trigger value corresponding to the k-th continuous potential energy interval is recorded as Yk. Then, based on the inflection point trigger values ​​corresponding to multiple continuous potential energy intervals, the inflection point judgment benchmark Hk is determined, and its calculation formula is as follows: ,in, This is the average of the inflection point trigger values ​​corresponding to multiple consecutive potential energy intervals. The standard deviation of the inflection point trigger values ​​corresponding to multiple consecutive potential energy intervals. To preset the discrete adjustment coefficient, Preset preceding reference coefficients, inflection point determination criteria This serves as a comparison benchmark for whether the k-th continuous potential energy interval meets the inflection point identification requirements. Then, based on the proportional relationship between the inflection point trigger value and the inflection point determination benchmark, the trigger exceedance value is determined. The calculation formula is as follows: ,in, To preset the smoothing coefficient, the inflection point trigger value Yk corresponding to the current continuous potential energy interval is compared with the inflection point trigger value corresponding to the previous continuous potential energy interval. The relationship between the changes is used to determine the interval salience value Nk, which is calculated using the following formula: ,in, A preset interval salience coefficient is used, and then the inflection point confirmation value is determined based on the trigger exceedance value and the interval salience value. The calculation formula is as follows: Then, the inflection point confirmation value Jk is compared with the preset confirmation threshold J0. In this case, the k-th continuous potential energy interval is determined to satisfy the preset inflection point condition. Finally, when multiple continuous potential energy intervals satisfy the preset inflection point condition, the continuous potential energy interval with the largest corresponding inflection point confirmation value Jk is determined as the negative diffusion inflection point corresponding to the target theme.

[0162] In one embodiment, step S50, namely generating the target public opinion sentiment analysis result based on the negative diffusion inflection point, includes:

[0163] S601: Based on the inflection point trigger value, occurrence time and location, and continuous potential energy interval location corresponding to the negative diffusion inflection point, determine the inflection point level information corresponding to the target theme.

[0164] In this embodiment, the occurrence time position refers to the time position of the negative diffusion inflection point in the analysis process corresponding to the target topic; the continuous potential energy interval position refers to the arrangement position of the negative diffusion inflection point in multiple continuous potential energy intervals; and the inflection point level information refers to the level result information formed based on the trigger intensity and position characteristics corresponding to the negative diffusion inflection point, which is used to characterize the strength of the diffusion inflection point.

[0165] Specifically, when determining the inflection point level information of the target topic based on the inflection point trigger value, occurrence time position, and continuous potential energy interval position corresponding to the negative diffusion inflection point, the inflection point trigger value corresponding to the negative diffusion inflection point is first read and used as the basic trigger basis for the negative diffusion inflection point. Then, based on the time position of the negative diffusion inflection point in the analysis process corresponding to the target topic, the time position feature of the negative diffusion inflection point is determined. Based on the order of the negative diffusion inflection point in multiple continuous potential energy intervals, the interval position feature of the negative diffusion inflection point is determined. Then, the basic trigger basis, time position feature, and interval position feature are jointly judged to determine the level classification result of the negative diffusion inflection point in the diffusion transition process, and the level classification result is determined as the inflection point level information of the target topic.

[0166] S602: Based on the inflection point level information and the changes in the negative sentiment accumulation potential energy before and after the negative diffusion inflection point, determine the diffusion status information corresponding to the target topic.

[0167] In this embodiment, diffusion state information refers to the result information formed based on inflection point level information and changes in the potential energy of negative emotion accumulation before and after the negative diffusion inflection point. It is used to characterize the diffusion change state of the negative emotion corresponding to the target topic at the negative diffusion inflection point and in the stages before and after it.

[0168] Specifically, based on the inflection point level information and the changes in negative emotional accumulation potential energy before and after the negative diffusion inflection point, when determining the diffusion state information corresponding to the target topic, the continuous potential energy interval corresponding to the negative diffusion inflection point is first used as the central interval for state determination. Then, the changes in negative emotional accumulation potential energy corresponding to the adjacent continuous potential energy intervals before and after the negative diffusion inflection point are determined separately. Finally, the inflection point level information is used as the basis for determining the intensity of diffusion change, and the changes in negative emotional accumulation potential energy before and after the negative diffusion inflection point are... The results serve as the basis for determining the direction and duration of diffusion. Then, the inflection point level information, the negative emotion accumulation potential energy change results corresponding to the adjacent continuous potential energy intervals before the negative diffusion inflection point, and the negative emotion accumulation potential energy change results corresponding to the adjacent continuous potential energy intervals after the negative diffusion inflection point are jointly determined to determine whether the negative emotion corresponding to the target topic belongs to the initial diffusion state, continuous diffusion state, accelerated diffusion state, or slowed diffusion state at the negative diffusion inflection point. Finally, the corresponding state determination results are determined as the diffusion state information corresponding to the target topic.

[0169] S603: Based on the diffusion state information and the negative sentiment accumulation potential energy corresponding to the target topic, determine the risk priority information corresponding to the target topic.

[0170] In this embodiment, risk priority information refers to the result information formed based on diffusion state information and the negative emotional accumulation potential energy corresponding to the target topic, which is used to characterize the degree of risk attention of the target topic among multiple targets to be analyzed.

[0171] Specifically, when determining the risk priority information for a target topic based on diffusion state information and the negative sentiment accumulation potential energy corresponding to the target topic, the diffusion state information is first used as the basis for judging the degree of diffusion change corresponding to the target topic, and the negative sentiment accumulation potential energy corresponding to the target topic is used as the basis for judging the degree of negative sentiment accumulation corresponding to the target topic. When the diffusion state information indicates that the target topic is in an accelerated diffusion state or a continuous diffusion state, the risk priority of the target topic is increased. When the diffusion state information indicates that the target topic is in an initial diffusion state or a slowing diffusion state, the risk priority of the target topic is adjusted in combination with the magnitude of the negative sentiment accumulation potential energy. Then, the result content corresponding to the adjusted risk priority is determined as the risk priority information for the target topic.

[0172] S604: Generate target public opinion sentiment analysis results based on inflection point level information, diffusion status information, and risk priority information.

[0173] Specifically, when generating the target public opinion sentiment analysis results based on inflection point level information, diffusion status information, and risk priority information, the inflection point level information is first used as the basis for determining the strength of the diffusion turning point corresponding to the target topic, the diffusion status information is used as the basis for determining the diffusion change status corresponding to the target topic, and the risk priority information is used as the basis for determining the priority of risk attention corresponding to the target topic. Then, the inflection point level information, diffusion status information, and risk priority information are correlated to form a comprehensive judgment result corresponding to the target topic. Then, based on the comprehensive judgment result, the results of the public opinion sentiment change status corresponding to the target topic are merged to obtain the result content corresponding to the target topic. Finally, the result content is determined as the target public opinion sentiment analysis result.

[0174] In one embodiment, step S604, namely generating target public opinion sentiment analysis results based on inflection point level information, diffusion status information, and risk priority information, includes:

[0175] S6041: Based on inflection point level information, determine the diffusion transition trend information corresponding to the target theme.

[0176] In this embodiment, diffusion transition trend information refers to the result information formed based on inflection point level information, which is used to characterize the direction of evolution of negative sentiment diffusion corresponding to the target topic from the current state to subsequent states.

[0177] Specifically, when determining the diffusion transition trend information corresponding to the target topic based on inflection point level information, the inflection point level information is first used as the basis for judging the strength of the diffusion inflection point corresponding to the target topic. Then, based on the level of the inflection point level information, the diffusion evolution trend of the target topic in the subsequent analysis stage is judged in layers. When the level of the inflection point level information is high, it is determined that the negative sentiment diffusion corresponding to the target topic has a tendency to transition to a higher diffusion intensity state. When the level of the inflection point level information is at an intermediate level, it is determined that the negative sentiment diffusion corresponding to the target topic has a tendency to extend to a continuous diffusion state. When the level of the inflection point level information is low, it is determined that the negative sentiment diffusion corresponding to the target topic has a tendency to change to a moderate diffusion state. Then, the corresponding trend judgment results are determined as the diffusion transition trend information of the target topic.

[0178] S6042: Determine the state continuation information corresponding to the target topic based on the diffusion state information.

[0179] In this embodiment, state continuation information refers to the result information formed based on diffusion state information, which characterizes whether the negative emotional diffusion state corresponding to the target topic continues to be maintained, strengthened, or gradually weakened in the subsequent analysis stage.

[0180] Specifically, when determining the state continuation information corresponding to the target topic based on the diffusion state information, the diffusion state information is first used as the basis for judging the current diffusion state of the target topic. Then, based on the state type corresponding to the diffusion state information, the direction of the target topic's state maintenance in the subsequent analysis stage is determined. If the diffusion state information indicates that the target topic is in a continuous diffusion state or an accelerated diffusion state, it is determined that the negative emotional diffusion state corresponding to the target topic has a tendency to continue. If the diffusion state information indicates that the target topic is in an initial diffusion state, it is determined that the negative emotional diffusion state corresponding to the target topic has a tendency to be further confirmed. If the diffusion state information indicates that the target topic is in a slowing diffusion state, it is determined that the negative emotional diffusion state corresponding to the target topic has a tendency to gradually weaken. Then, the corresponding trend determination result is determined as the state continuation information corresponding to the target topic.

[0181] S6043: Based on risk priority information, determine the risk clustering direction information corresponding to the target theme.

[0182] In this embodiment, risk clustering direction information refers to the result information formed based on risk priority information, which is used to characterize the direction of change of the degree of risk concern corresponding to the target topic in subsequent analysis stages, such as clustering towards high risk, medium risk, or low risk.

[0183] Specifically, when determining the risk clustering direction information corresponding to a target topic based on risk priority information, the risk priority information is first used as the basis for determining the order of risk attention for the target topic. Then, based on the priority of the risk priority information, the direction of risk clustering change for the target topic in the subsequent analysis stage is determined. If the priority of the risk priority information is high, it is determined that the risk attention for the target topic tends to change towards high-risk clustering. If the priority of the risk priority information is at an intermediate level, it is determined that the risk attention for the target topic tends to change towards medium-risk clustering. If the priority of the risk priority information is low, it is determined that the risk attention for the target topic tends to change towards low-risk clustering. The corresponding direction determination results are then determined as the risk clustering direction information for the target topic.

[0184] S6044: Construct a negative diffusion evolution path corresponding to the target theme based on diffusion transition trend information, state continuity information, and risk aggregation direction information.

[0185] In this embodiment, the negative diffusion evolution path refers to the result information formed based on diffusion transition trend information, state continuation information, and risk aggregation direction information, which is used to characterize the evolution direction and process of negative sentiment diffusion corresponding to the target topic in the subsequent analysis stage.

[0186] Specifically, when constructing the negative diffusion evolution path corresponding to the target theme based on diffusion transition trend information, state continuation information, and risk aggregation direction information, the diffusion transition trend information is first used as the basis for determining the direction of diffusion intensity change corresponding to the target theme, the state continuation information is used as the basis for determining the continuous change of diffusion state corresponding to the target theme, and the risk aggregation direction information is used as the basis for determining the risk concern direction corresponding to the target theme. Then, the diffusion transition trend information, state continuation information, and risk aggregation direction information are correlated to form the diffusion evolution relationship of the target theme in the subsequent analysis stage. Then, based on the diffusion evolution relationship, the path of the negative sentiment diffusion process corresponding to the target theme is organized according to the combination order of diffusion intensity change direction, continuous change of diffusion state, and risk concern direction, thus obtaining the negative diffusion evolution path corresponding to the target theme.

[0187] S6045: Generate target public opinion sentiment analysis results based on the negative diffusion evolution path.

[0188] Specifically, when generating target public opinion sentiment analysis results based on the negative diffusion evolution path, the negative diffusion evolution path is first used as the basis for determining the negative sentiment diffusion evolution process corresponding to the target theme. Then, based on the direction of diffusion intensity change, the continuous change of diffusion status, and the direction of risk concern represented in the negative diffusion evolution path, the results of the public opinion sentiment change process of the target theme in the subsequent analysis stage are merged to form a comprehensive analysis result corresponding to the target theme. Finally, the comprehensive analysis result is determined as the target public opinion sentiment analysis result.

[0189] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0190] In one embodiment, a public opinion dissemination distortion analysis device is provided, which corresponds one-to-one with the public opinion dissemination distortion analysis method described in the above embodiments. For example... Figure 2 As shown, this public opinion dissemination distortion analysis device includes a public opinion association acquisition module, an emotion dissemination distortion determination module, a negative emotion contribution determination module, a negative emotion accumulation potential energy determination module, a negative diffusion inflection point determination module, and a public opinion emotion analysis result generation module. Detailed descriptions of each functional module are as follows:

[0191] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0192] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for analyzing distortion in public opinion dissemination, characterized in that, The aforementioned method for analyzing distortion in public opinion dissemination includes: Obtain multiple public opinion texts corresponding to the target topic and the dissemination correlation information between these texts; Based on the offset between the sentiment results of the preceding public opinion texts that are related to the current public opinion text and the sentiment results of the current public opinion text, the sentiment dissemination distortion information corresponding to the current public opinion text is determined; Based on the text sentiment results and the sentiment propagation distortion information, determine the negative sentiment contribution value corresponding to each of the aforementioned public opinion texts; The negative sentiment contribution values ​​are cumulatively calculated and attenuated in chronological order to obtain the negative sentiment accumulation potential energy corresponding to the target topic. Based on the changes in the negative emotional accumulation potential energy over a continuous time period, the negative diffusion inflection point corresponding to the target topic is determined; Based on the negative diffusion inflection point, the target public opinion sentiment analysis results are generated.

2. The method for analyzing distortion in public opinion dissemination according to claim 1, characterized in that, The step of determining the emotional propagation distortion information corresponding to the current public opinion text based on the offset between the text sentiment results corresponding to previous public opinion texts that are related to the current public opinion text and the text sentiment results corresponding to the current public opinion text includes: Extract the preceding sentiment polarity information and preceding sentiment intensity information corresponding to the preceding public opinion text, and extract the current sentiment polarity information and current sentiment intensity information corresponding to the current public opinion text; Based on the preceding sentiment polarity information and the current sentiment polarity information, determine the sentiment direction offset information corresponding to the current public opinion text; Based on the preceding sentiment intensity information and the current sentiment intensity information, determine the sentiment intensity offset information corresponding to the current public opinion text; The core stance expression information is extracted from the preceding public opinion text and the current public opinion text respectively. Based on the difference between the core stance expression information in the preceding public opinion text and the core stance expression information in the current public opinion text, the stance deviation information corresponding to the current public opinion text is determined. Based on the emotional direction offset information, the emotional intensity offset information, and the stance deviation information, the emotional propagation distortion information corresponding to the current public opinion text is determined.

3. The method for analyzing distortion in public opinion dissemination according to claim 1, characterized in that, The step of determining the negative sentiment contribution value corresponding to each of the aforementioned public opinion texts based on the text sentiment results and the sentiment propagation distortion information includes: Based on the text sentiment results, determine the basic negative sentiment value corresponding to each of the aforementioned public opinion texts; The distortion information of emotional transmission is evaluated item by item to obtain the direction distortion value, intensity distortion value and stance distortion value corresponding to each of the public opinion texts; Based on the directional distortion value, the intensity distortion value, and the position distortion value, determine the distortion correction value corresponding to each of the public opinion texts; Based on the propagation level information, propagation response information, and subsequent follow-up information of each of the aforementioned public opinion texts in the propagation association information, determine the propagation correction value corresponding to each of the aforementioned public opinion texts; Based on the basic negative sentiment value, the distortion correction value, and the propagation correction value, the negative sentiment contribution value corresponding to each of the aforementioned public opinion texts is determined.

4. The method for analyzing distortion in public opinion dissemination according to claim 1, characterized in that, The step of performing cumulative calculation and attenuation correction on each of the negative sentiment contribution values ​​in chronological order to obtain the negative sentiment accumulation potential energy corresponding to the target topic includes: The negative sentiment contribution values ​​corresponding to each of the aforementioned public opinion texts are sorted according to the publication time order to obtain the time-series negative sentiment contribution sequence corresponding to the target topic; Based on the time interval between adjacent negative sentiment contribution values ​​in the time-series negative sentiment contribution sequence, determine the time decay value corresponding to each negative sentiment contribution value; Based on the growth difference between the current negative sentiment contribution value and the previous negative sentiment contribution value in the time-series negative sentiment contribution sequence, the incremental reinforcement value corresponding to each negative sentiment contribution value is determined; Based on the negative emotion contribution value, the time decay value, and the incremental reinforcement value, determine the stage accumulation value corresponding to each time position; By continuously accumulating the accumulated values ​​of each stage in chronological order, the negative emotional potential energy corresponding to the target topic is obtained.

5. The method for analyzing distortion in public opinion dissemination according to claim 1, characterized in that, The step of determining the negative diffusion inflection point corresponding to the target topic based on the change in the negative emotional accumulation potential energy over a continuous time period includes: The accumulated potential energy of negative emotions is divided into intervals according to continuous time periods to obtain multiple continuous potential energy intervals; Based on the negative emotional accumulation potential energy change value corresponding to each of the continuous potential energy intervals, the potential energy growth value corresponding to each of the continuous potential energy intervals is determined. Based on the difference in the change between the potential energy growth values ​​corresponding to adjacent continuous potential energy intervals, the slope offset value corresponding to each continuous potential energy interval is determined. Based on the amplification difference between the potential energy growth value corresponding to the current continuous potential energy interval and the potential energy growth value corresponding to the previous continuous potential energy interval, the sudden shift value corresponding to each continuous potential energy interval is determined. Based on the direction of change of potential energy growth value corresponding to multiple consecutive potential energy intervals, determine the continuous growth value corresponding to each consecutive potential energy interval; Based on the slope offset value, the sudden increase offset value, and the continuous growth value, the inflection point trigger value corresponding to each continuous potential energy interval is determined. If the inflection point trigger value meets the preset inflection point condition, the corresponding continuous potential energy interval is determined as the negative diffusion inflection point corresponding to the target theme.

6. The method for analyzing distortion in public opinion dissemination according to claim 1, characterized in that, The step of generating target public opinion sentiment analysis results based on the negative diffusion inflection point includes: Based on the inflection point trigger value, occurrence time location, and continuous potential energy interval location corresponding to the negative diffusion inflection point, the inflection point level information corresponding to the target theme is determined; Based on the inflection point level information and the changes in negative emotional accumulation potential energy before and after the negative diffusion inflection point, the diffusion state information corresponding to the target topic is determined. Based on the diffusion state information and the negative sentiment accumulation potential energy corresponding to the target topic, the risk priority information corresponding to the target topic is determined; Based on the inflection point level information, the diffusion status information, and the risk priority information, the target public opinion sentiment analysis result is generated.

7. The method for analyzing distortion in public opinion dissemination according to claim 1, characterized in that, The step of generating the target public opinion sentiment analysis result based on the inflection point level information, the diffusion status information, and the risk priority information includes: Based on the inflection point level information, determine the diffusion transition trend information corresponding to the target topic; Based on the diffusion state information, determine the state continuation information corresponding to the target topic; Based on the risk priority information, determine the risk clustering direction information corresponding to the target topic; Based on the diffusion transition trend information, the state continuation information, and the risk aggregation direction information, a negative diffusion evolution path corresponding to the target theme is constructed; Based on the negative diffusion evolution path, the target public opinion sentiment analysis results are generated.

8. A device for analyzing distortion in public opinion dissemination, characterized in that, The aforementioned public opinion dissemination distortion analysis device includes: The public opinion association acquisition module is used to acquire multiple public opinion texts corresponding to the target topic and the dissemination association information between the public opinion texts; The emotion propagation distortion determination module is used to determine the emotion propagation distortion information corresponding to the current public opinion text based on the offset between the text emotion result corresponding to the previous public opinion text that has a propagation relationship with the current public opinion text and the text emotion result corresponding to the current public opinion text. The negative sentiment contribution determination module is used to determine the negative sentiment contribution value corresponding to each of the aforementioned public opinion texts based on the text sentiment results and the sentiment propagation distortion information. The negative emotion accumulation potential energy determination module is used to perform cumulative calculation and attenuation correction on each negative emotion contribution value in chronological order to obtain the negative emotion accumulation potential energy corresponding to the target topic. The negative diffusion inflection point determination module is used to determine the negative diffusion inflection point corresponding to the target topic based on the change results of the negative emotion accumulation potential energy over a continuous time period. The public opinion sentiment analysis result generation module is used to generate target public opinion sentiment analysis results based on the negative diffusion inflection point.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the public opinion dissemination distortion analysis method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the public opinion dissemination distortion analysis method as described in any one of claims 1 to 7.

Images