Network public opinion multivariable relation detection method and system based on data driving

Abstract

The invention discloses an online public opinion multivariable relation detection method and system based on data driving, and belongs to the technical field of electric data processing. The method comprises the following steps: constructing a basic evolution model for N variables of network public opinions based on a Logistic model, wherein the basic evolution model comprises N basic equations; adding a function item representing an interaction relationship of N variables in each basic equation; performing static regression analysis and dynamic regression analysis on the network public opinion data, and setting M indexes; and sorting the mutual combination relationship of the N variables represented by the function item according to the M indexes to obtain a common structure factor, and replacing the function item representing the interaction relationship of the N variables with a linear combination of the common structure factor, so as to obtain a symbiotic evolution model containing the relationship of the N variables of the network public opinions. According to the method, the dynamic coupling relationship among multiple variables can be accurately quantified; the optimal interaction item can be scientifically and objectively screened out, and the subjectivity of relying on artificial experience is avoided.

Description

Technical Field

[0001] The present invention relates to a data-driven method and system for detecting multivariate relationships in online public opinion, belonging to the field of artificial intelligence technology. Background Art

[0002] Traditional independent evolution models assume that a single variable evolves solely based on its own historical data (for example, modeling solely through the target variable's growth rate and capacity limit), completely ignoring the dynamic interactions with other correlated variables. For example, when analyzing a two-variable system with synergistic effects (such as market demand and user activity), traditional models cannot account for the impact of correlated variables on the core indicator through feedback mechanisms.

[0003] To overcome the above technical problems, the existing technology provides a linear regression model, but its interaction terms are mainly manually preset, rely on experience, and are highly subjective.

[0004] Existing technologies also offer single-variable time series models (such as ARIMA), but these rely solely on the historical trends of the target variable itself, requiring the manual addition of exogenous variables to construct an extended model. Furthermore, they are unable to quantify the real-time coupling strength between variables. For example, when analyzing the "lagged reinforcing effect of user activity on market demand," the ARIMA model requires manual setting of the lag order.

[0005] The existing technology also provides a machine learning model, but its black box characteristics make it impossible to explain the interaction paths between variables. Summary of the Invention

[0006] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a data-driven multivariate relationship method and system for online public opinion, which can accurately quantify the dynamic coupling relationship between multiple variables; can scientifically and objectively screen out the optimal interaction terms, and avoid relying on the subjectivity of human experience.

[0007] To overcome the shortcomings of the prior art, the present invention provides a data-driven method for detecting multivariate relationships in online public opinion, which includes the following steps: S1: Constructing a basic evolution model for N variables of online public opinion based on a Logistic model, wherein the basic evolution model includes N basic equations; S2: Add a function term representing the interaction relationship of N variables to each basic equation; S3: Perform static regression analysis and dynamic regression analysis on online public opinion data and establish M indicators; S4: Sort the mutual combination relationship of N variables represented by the function terms according to M indicators to obtain the common structural factor, and use the linear combination of the common structural factor to replace the function term representing the interactive relationship of N variables, so as to obtain a symbiotic evolution model containing the relationship of N variables of online public opinion, where N and M are both positive integers greater than or equal to 2.

[0008] In order to overcome the shortcomings of the prior art, the present invention also provides a system comprising a storage medium and one or more processors, wherein the storage medium is used to store code for compiling the above-mentioned data-driven network public opinion multivariate relationship detection method into a computer program using a computer language, and the computer program can be called and executed by one or more processors.

[0009] Compared to existing technologies, the data-driven nonlinear detection method provided by this invention, by incorporating high-order nonlinear interaction terms, can accurately quantify the dynamic coupling relationships between multiple variables. This data-driven approach and the establishment of multiple evaluation criteria allow for the scientific and objective selection of optimal interaction terms, avoiding the subjectivity inherent in relying on human experience. Furthermore, this invention is highly interpretable and easy to understand. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a flow chart of the relationship detection method based on data-driven network public opinion provided by the present invention.

[0011] Figure 2 It is a coefficient line graph obtained by regression analysis of the first equation provided by the present invention.

[0012] Figure 3 The coefficient line diagram is obtained by regression splitting the second equation provided by the present invention. DETAILED DESCRIPTION

[0013] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0014] Figure 1 This is a flow chart of the relationship detection method based on data-driven network public opinion provided by the present invention. Figure 1 As shown, the relationship detection method based on data-driven network public opinion provided by the present invention includes the following steps: S1: Constructing a basic evolution model for N variables of online public opinion based on a Logistic model, wherein the basic evolution model includes N basic equations; S2: Add a function term representing the interaction relationship of N variables to each basic equation; S3: Perform static regression analysis and dynamic regression analysis on online public opinion data and establish M indicators; S4: Sort the mutual combination relationship of N variables represented by the function terms according to M indicators to obtain the common structural factor, and use the linear combination of the common structural factor to replace the function term representing the interactive relationship of N variables, so as to obtain a symbiotic evolution model containing the relationship of N variables of online public opinion, where N and M are both positive integers greater than or equal to 2.

[0015] The network public opinion data in the present invention is, for example, movie box office data. The first variable and the second variable are, for example, the box office number of a movie and the search index of the movie.

[0016] For example, when N=2, the basic evolution model is: , Where, The first variable representing the i-th type of online public opinion The difference of The second variable representing the i-th type of online public opinion The difference of ; n represents the discrete variable of time; , are the first variables representing the i-th type of online public opinion and the second variable The intrinsic growth rate of , are the first variables representing the i-th type of online public opinion and the second variable The upper limit of i, for example, i=50, represents 50 movies.

[0017] In the present invention, after adding a function term representing the interaction relationship of N variables to each basic equation, the basic evolution model is transformed into a symbiotic evolution model: , Where, Represents the second variable For the first variable The impact relationship; Represents the first variable For the second variable influence relationship.

[0018] The data-driven multivariate relationship detection method for online public opinion provided by the present invention also includes: Get About the first variable and the second variable Various linear combinations of the first and second terms of There are six possible structures: , , , , , The six items can be combined with each other, so there are A combination of situations.

[0019] The data-driven multivariate relationship detection method for online public opinion provided by the present invention also includes: Get About the first variable and the second variable Various linear combinations of the first and second terms of There are six possible structures: , , , , , The six items can be combined with each other, so there are A combination of situations.

[0020] The data-driven multivariate relationship detection method for online public opinion provided by the present invention also includes: using M indicators to test each equation of the symbiotic evolution model to obtain K groups of common factor groups whose M indicators are higher than the set threshold of each indicator.

[0021] In the present invention, M=5, and the five indicators are the structural test pass rate STR, the fitting test pass rate FTR, the significance test pass rate SGTR, the three test pass rate ATR and the dynamic data pass rate DTR, among which the structural test pass rate STR represents the number of static data that pass the structural test divided by the total number of static data; the fitting test pass rate FTR represents the number of static data that pass the fitting test divided by the total number of static data; the three test pass rate ATR represents the number of static data that pass the three tests divided by the total number of static data; the dynamic data pass rate DTR represents the number of dynamic data that pass the three tests divided by the total number of dynamic data. Among the five indicators, ATR takes various situations into consideration and should be the indicator to be considered first. Among the other three tests, the structural test is a guarantee of the basic structure of the model and should be given priority among the three tests. In addition, the coefficient of the common structural factor is also an object we need to consider. We need to select a common structural factor with a stable positive and negative coefficient. A stable coefficient means a stable interaction relationship.

[0022] For example, to filter out The common structural factors of 50 groups of data were analyzed by static and dynamic regression, and five groups of common factors with higher ATR, STR, FTR, SGTR and DTR were obtained, as shown in Table 1:

[0023] Table 1

[0024] To filter out The common structural factors of 50 groups of data were analyzed by static and dynamic regression, and five groups of common factors with higher ATR, STR, FTR, SGTR and DTR were obtained, as shown in Table 2:

[0025] Table 2

[0026] The data-driven multivariate relationship detection method for online public opinion provided by the invention also includes: performing stability analysis on K groups of common factor groups, and using common factor groups with stability higher than the stability threshold to replace the influence relationship, where K is a positive integer greater than or equal to 1.

[0027] In the present invention, , Where, The second variable representing the i-th type of online public opinion For the first variable Linear influence coefficient; The second variable representing the i-th type of online public opinion For the first variable The quadratic influence coefficient, and like Figure 2 As shown. Figure 2 It can be seen that the interaction factors of 50 groups of data were observed All are positive, the interaction factors of 49 groups of data A negative number indicates that the interaction is stable.

[0028] In the present invention, , The first variable representing the i-th type of online public opinion For the second variable The quadratic influence coefficient, such as Figure 3 As shown. Figure 3 Observation: Interaction Factor The positive and negative results show that only 3 groups of data have negative coefficients, while the other 47 groups are positive. 94% of the interaction factors are positive, and the common factors are The DTR is 88%, even slightly higher than its ATR. The interaction between box office and search index is stable.

[0029] The present invention also provides a system comprising a storage medium and one or more processors, wherein the storage medium is used to store code for compiling the above-mentioned data-driven network public opinion multivariate relationship detection method into a computer program using a computer language, and the computer program can be called and executed by one or more processors.

[0030] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined. "Several" means one or more, unless otherwise specifically defined.

[0031] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A data-driven method for detecting multivariate relationships in online public opinion, characterized in that: The steps include: S1: Constructing a basic evolution model for N variables of online public opinion based on a Logistic model, wherein the basic evolution model includes N basic equations; S2: Add a function term representing the interaction relationship of N variables to each basic equation; S3: Perform static regression analysis and dynamic regression analysis on online public opinion data and establish M indicators; S4: Sort the mutual combination relationship of N variables represented by the function terms according to M indicators to obtain the common structural factor, and use the linear combination of the common structural factor to replace the function term representing the interactive relationship of N variables, so as to obtain a symbiotic evolution model containing the relationship of N variables of online public opinion, where N and M are both positive integers greater than or equal to 2.

2. The data-driven multivariate relationship detection method for online public opinion according to claim 1 is characterized in that: The N=2, the basic evolution model is: ， Where, The first variable representing the i-th type of online public opinion The difference of The second variable representing the i-th type of online public opinion The difference of ; n represents the discrete variable of time; , are the first variables representing the i-th type of online public opinion and the second variable The intrinsic growth rate of , are the first variables representing the i-th type of online public opinion and the second variable upper limit.

3. The data-driven multivariate relationship detection method for online public opinion according to claim 2 is characterized in that: After adding a function term representing the interaction relationship of N variables to each basic equation, the basic evolution model is transformed into a symbiotic evolution model: ， Where, Represents the second variable For the first variable The impact relationship; Represents the first variable For the second variable influence relationship.

4. The data-driven multivariate relationship detection method for online public opinion according to claim 3 is characterized in that: Get About the first variable and the second variable Various linear combinations of the first and second terms of ; obtain About the first variable and the second variable Various linear combinations of the linear and quadratic terms of .

5. The data-driven multivariate relationship detection method for online public opinion according to claim 4 is characterized in that: Also includes: Each equation of the symbiotic evolution model is tested using M indicators to obtain K groups of common factor groups whose M indicators are higher than the set threshold of each indicator.

6. The data-driven multivariate relationship detection method for online public opinion according to claim 5 is characterized in that: Also includes: Perform stability analysis on K groups of common factor groups, and use the common factor groups with stability higher than the stability threshold to replace the influence relationship, where K is a positive integer greater than or equal to 1.

7. The data-driven multivariate relationship detection method for online public opinion according to claim 6 is characterized in that: ， Where, The second variable representing the i-th type of online public opinion For the first variable Linear influence coefficient; The second variable representing the i-th type of online public opinion For the first variable Quadratic influence coefficient.

8. The data-driven multivariate relationship detection method for online public opinion according to claim 4 is characterized in that: ， Where, The first variable representing the i-th type of online public opinion For the second variable Quadratic influence coefficient.

9. The data-driven multivariate relationship detection method for online public opinion according to any one of claims 1 to 8, characterized in that: M=5, and the five indicators are the structural test pass rate, the fitting test pass rate, the significance test pass rate, the three test pass rates and the dynamic data pass rate.

10. A system comprising a storage medium and one or more processors, wherein the storage medium is used to store code that uses a computer language to compile the data-driven network public opinion multivariate relationship detection method described in any one of claims 1 to 9 into a computer program, and the computer program can be called and executed by one or more processors.

Images