An information cocoon effect detection method based on dynamic regression and model screening

Abstract

The application discloses an information cocoon effect detection method based on dynamic regression and model screening, relates to the technical field of information cocoon effect, and constructs an information independent difference equation and searches for an information independent difference equation to obtain an initial test group; based on the information independent difference equation and the searched information independent difference equation, a promotion coefficient, an inhibition coefficient and an interaction influence factor are introduced to obtain an information interaction difference equation and a searched information interaction difference equation; a target test group is constructed, a triple test result of each interaction influence factor under each target test group is obtained based on the target test group and regression analysis triple test, the coincidence rate of each interaction influence factor is determined, and thus a target information interaction difference equation and a target searched information interaction difference equation are determined; and the target information interaction difference equation and the target searched information interaction difference equation are used for information cocoon effect detection at a current discrete time. The application realizes accurate detection of the information cocoon effect.

Description

Technical Field

[0001] This application relates to the field of information cocoon technology, and in particular to a method for detecting information cocoon effects based on dynamic regression and model screening. Background Technology

[0002] In today's information explosion era, internet platforms, through intelligent algorithm recommendations, have drastically changed how users obtain information. However, while improving efficiency, they have also increasingly highlighted the social problem of "information cocoons," where users' perspectives become narrower due to algorithmic feeding, potentially leading to cognitive rigidity and social polarization. Therefore, scientifically and accurately quantitatively assessing the information cocoon effect, especially the inhibitory effect of information push on users' proactive search behavior, has become an urgent need for academia and industry. Research on information cocoons often employs questionnaires, macro-statistics based on user behavior big data, or qualitative analysis. Some studies have begun to attempt quantitative models, but these generally suffer from technical problems such as model structure relying on subjective assumptions, limited verification methods, and a lack of efficient intelligent screening tools. Specifically, firstly, related studies often rely on empirical "assumptions" of a simple model function form, ignoring the theoretically possible existence of a vast space of candidate models composed of different variable interactions, different function forms, and different error structures. Secondly, traditional methods often rely on a single goodness-of-fit rating to test the model, neglecting whether the model parameters remain stable over time, potentially leading to the adoption of statistically significant but structurally flawed models. Third, given the vast number of candidate model structures, relying on manual modeling and dynamic stability verification for each one is unrealistic and extremely inefficient.

[0003] In summary, how to automatically select the dynamically stable optimal model from a massive pool of candidate model structures in order to accurately detect the information cocoon effect is a pressing technical problem that needs to be solved. Summary of the Invention

[0004] The purpose of this application is to provide a method for detecting the information cocoon effect based on dynamic regression and model screening, so as to achieve accurate detection of the information cocoon effect.

[0005] To achieve the above objectives, this application provides the following solution.

[0006] This application provides a method for detecting the information cocoon effect based on dynamic regression and model screening, including: Obtain historical continuous time series within a preset historical period; the historical continuous time series includes: the cumulative information volume time series of information pushed by the platform, the cumulative information volume time series of search generated by user search behavior, the information index time series, and the search index time series; Construct independent difference equations for information and search information; Based on historical continuous time series, parameter fitting is performed on the independent difference equations of information and search information to obtain an initial test set; the initial test set includes the first test values ​​of the test parameters at N discrete time points; N≥6 and N is an even number; the test parameters include: cumulative information content, cumulative search information content, information index, and search index; Define the promoting coefficient and 728 first-influence factors of the first-influence function, and define the inhibiting coefficient and 728 second-influence factors of the second-influence function; Any first impact factor and any second impact factor constitute an interactive impact factor, and any interactive impact factor is determined as the current interactive impact factor; By introducing the promotion coefficient and the first influence factor in the current interaction influence factors into the information independent difference equation, we obtain the information interaction difference equation corresponding to the first influence factor in the current interaction influence factors. By introducing the inhibition coefficient and the second influence factor in the current interaction influence factors into the search information independent difference equation, we obtain the search information interaction difference equation corresponding to the second influence factor in the current interaction influence factors. According to the time sequence of each discrete moment corresponding to the initial test group, construct N / 2 target test groups and determine any target test group as the current target test group; wherein, the i-th target test group contains the first test value of the test parameter from the first discrete moment to the first test value of the test parameter at the (N / 2+i-1)-th discrete moment, i=1,2,...,N / 2; Based on the first test value of the cumulative information content and the first test value of the information index at each discrete moment in the current target test group, the information information interaction difference equation corresponding to the first influence factor in the current interaction influence factors is fitted with parameters to obtain the information fitting parameters and promotion coefficient of the current interaction influence factor under the current target test group. Based on the first test value of the cumulative search information and the first test value of the search index at each discrete moment in the current target test group, the parameters of the search information interaction difference equation corresponding to the second influence factor in the current interaction influence factor are fitted to obtain the search fitting parameters and suppression coefficients of the current interaction influence factor under the current target test group. Based on the historical values ​​of the information index at each discrete moment in the current target test group, the information fitting parameters, promotion coefficient, search fitting parameters and inhibition coefficient of the current interaction factor under the current target test group, a triple test of regression analysis is performed to obtain the triple test results of the current interaction factor under the current target test group. The consistency rate of the current interactive influence factor is determined based on the triple test results of the current interactive influence factor under each target test group; The target values ​​of the optimal model parameters are determined based on the consistency rate of all interactive influence factors. The optimal model parameters include: the first model parameters and the second model parameters. The first model parameters include: the optimal first influence factor, the optimal promotion coefficient, the optimal intrinsic growth rate of information content, and the optimal upper limit of information content. The second model parameters include: the optimal second influence factor, the optimal inhibition coefficient, the optimal intrinsic growth rate of search information content, and the optimal upper limit of search information content. Substituting the target value of the first model parameter into the information interaction difference equation corresponding to the optimal first influence factor, we obtain the target information interaction difference equation; substituting the target value of the second model parameter into the search information interaction difference equation corresponding to the optimal second influence factor, we obtain the target search information interaction difference equation. Obtain the measured values ​​of the cumulative information content, cumulative search information content, information index, and search index at the current discrete moment; Substitute the measured value of the cumulative information amount at the current discrete moment into the target information interaction difference equation to obtain the calculated value of the information index at the current discrete moment. Substitute the measured value of the cumulative search information at the current discrete moment into the target information interaction difference equation to obtain the calculated value of the search index at the current discrete moment. Based on the measured values ​​of the information index, search index, and calculated values ​​of the information index and search index at the current discrete moment, the information cocoon effect index at the current discrete moment is determined, thus completing the detection of the information cocoon effect at the current discrete moment.

[0007] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application discloses a method for detecting the information cocoon effect based on dynamic regression and model screening. It constructs a historical continuous time series containing both information and search sequences. After rigorously obtaining an initial test set that passes triple testing, it systematically traverses a candidate model space consisting of 728×728 interactive influencing factors. Furthermore, by combining dynamic window regression and a fit rate screening mechanism, it automatically identifies the statistically optimal model structure and parameters under cross-window conditions. Finally, it uses this optimal model to jointly calculate and evaluate the information index and search index at the current discrete moment. This process abandons the subjective assumptions about the model function form in traditional methods, overcomes the technical shortcomings of manual traversal of massive model structures and the potential for model misselection due to single goodness-of-fit tests, and achieves highly robust and accurate quantitative detection of the information cocoon effect, thus enabling precise detection of the information cocoon effect. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 A schematic diagram of the information cocoon effect detection method based on dynamic regression and model screening provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0010] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0011] The purpose of this application is to provide a method for detecting the information cocoon effect based on dynamic regression and model screening, aiming to achieve accurate detection of the information cocoon effect.

[0012] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0013] In one exemplary embodiment, such as Figure 1 As shown, a method for detecting the information cocoon effect based on dynamic regression and model screening is provided, including the following steps.

[0014] Step 1: Determine the initial test group. Step 1 specifically includes the following steps.

[0015] Step 11: Obtain the historical continuous time series in the preset historical period; the historical continuous time series includes: the cumulative information volume time series of information pushed by the platform, the cumulative information volume time series of search generated by user search behavior, the information index time series, and the search index time series.

[0016] Step 12: Construct independent difference equations for information and search information.

[0017] As an optional implementation, in step 12, the independent difference equation for the information is: ; in, Discrete time in the information independent difference equation The information index is used to measure the growth rate of the cumulative amount of information. The intrinsic growth rate of information volume; This represents the maximum amount of information that can be conveyed. Discrete time The cumulative amount of information in the following news items; The independent difference equation for searching information is: ; in, To search for discrete time intervals in the independent difference equations The search index is used to measure the growth rate of the cumulative amount of information searched. The intrinsic growth rate of the amount of information searched; This is the upper limit for the amount of information that can be searched; Discrete time The cumulative amount of information searched below.

[0018] Step 13: Based on the historical continuous time series, perform parameter fitting on the independent difference equations of information and search information to obtain the initial test set; wherein, the initial test set includes the first test values ​​of the test parameters at N discrete time points; N≥6 and N is an even number; the test parameters include: cumulative information content, cumulative search information content, information index and search index.

[0019] As an optional implementation, step 13 includes the following steps.

[0020] Step 131: Randomly sample the historical continuous time series to obtain the initial sampling sequence; the initial sampling sequence includes the historical values ​​of the cumulative information content, the cumulative search information content, the information index, and the search index at N discrete time points.

[0021] Step 132: Using the binary linear regression method and the triple test of regression analysis, the independent difference equations of information and search information are processed in multiple rounds based on the initial sampling sequence to obtain the initial test group; wherein, the processing in any current round includes the following steps.

[0022] Step 1321: Using the binary linear regression method, fit the parameters of the independent difference equation of information based on the historical values ​​of the cumulative information content and the historical values ​​of the information index at N discrete times in the sampling sequence under the current round, and obtain the information fitting parameters corresponding to the sampling sequence under the current round; the information fitting parameters include: the intrinsic growth rate of information content, the upper limit of information content, and the fitted value of the information index at each discrete time; when the current round is the initial round, the sampling sequence under the current round is the initial sampling sequence.

[0023] Step 1322: Using the binary linear regression method, fit the parameters of the independent difference equation of the search information using the historical values ​​of the cumulative search information and the historical values ​​of the search index at N discrete time points in the current round of the sampling sequence, and obtain the search fitting parameters corresponding to the sampling sequence in the current round; the search fitting parameters include: the intrinsic growth rate of the search information, the upper limit of the search information, and the fitted value of the search index at each discrete time point.

[0024] Step 1323: Based on the historical values ​​of the information index, information fitting parameters, and search fitting parameters of the N discrete time points in the sampling sequence under the current round, perform a triple test of regression analysis on the sampling sequence under the current round to obtain the triple test results under the current round.

[0025] As an optional implementation, step 1323 includes the following steps.

[0026] Step 13231: Based on the intrinsic growth rate of information content, the upper limit of information content, the intrinsic growth rate of search information content, and the upper limit of search information content corresponding to the sampling sequence in the current round, perform a structural test to obtain the structural test results for the current round.

[0027] Specifically, when and When the structural test result is "passed", the structural test is used to test whether the independent difference equations of information and search information meet the structural requirements in terms of behavior.

[0028] Step 13232: Based on the fitted values ​​and historical values ​​of the information index at N discrete times corresponding to the sampling sequence in the current round, determine the information relevance test results in the current round.

[0029] Specifically, the first goodness-of-fit index for the current round is determined based on the fitted values ​​and historical values ​​of the information index at N discrete time points corresponding to the sampling sequence in the current round. When the first goodness-of-fit index for the current round is greater than 0.25, the information relevance test result for the current round is considered to have passed the test. The information relevance test is used to test whether the independent difference equation of information has the explanatory power for the cumulative information content.

[0030] Step 13233: Based on the fitted values ​​and historical values ​​of the search index at N discrete times corresponding to the sampling sequence in the current round, determine the search relevance test results in the current round.

[0031] Specifically, the second goodness-of-fit index for the current round is determined based on the fitted values ​​and historical values ​​of the search index at N discrete time points corresponding to the sampling sequence in the current round. When the second goodness-of-fit index for the current round is greater than 0.25, the search relevance test result for the current round is considered to have passed the test. The search relevance test is used to test whether the independent difference equation of the search information has explanatory power for the cumulative amount of search information.

[0032] The calculation formulas for the first and second goodness-of-fit indices in steps 13232 and 13233 are uniformly expressed as follows: ; ; in, This can be either the first or second goodness-of-fit index. Discrete time Below or The fitted value; For N discrete time points Down The mean; Discrete time Below or The measured value.

[0033] Step 13234: Based on the intrinsic growth rate and upper limit of information content of the sampled sequence in the current round, obtain the information significance test results for the current round.

[0034] Specifically, based on the intrinsic growth rate and upper limit of the information content corresponding to the sampling sequence in the current round, the first significance index for the current round is determined. When the first significance index for the current round is less than 0.05, the information significance test result for the current round is considered to have passed the test.

[0035] Step 13235: Based on the intrinsic growth rate and upper limit of the search information corresponding to the sampling sequence in the current round, obtain the search significance test results for the current round.

[0036] Specifically, based on the intrinsic growth rate and upper limit of the information content corresponding to the sampling sequence in the current round, the second significance index for the current round is determined. When the second significance index for the current round is less than 0.05, the search significance test result for the current round is considered to have passed the test.

[0037] The calculation formulas for the first and second significance indices in steps 13234 and 13235 are uniformly expressed as follows: ; ; ; in, The first or second significance index; To obtain the maximum value; For the first regression coefficient to be tested, the sub-significance index is... When the first significance index is used, the first regression coefficient to be tested is ,when When the second significance index is used, the first regression coefficient to be tested is ; For the significance index corresponding to the second regression coefficient to be tested, when When the first significance index is used, the second regression coefficient to be tested is ,when When the second significance index is used, the second regression coefficient to be tested is ; For the first Sub-significance indices corresponding to the regression coefficients to be tested; The residual degrees of freedom are The t-distribution cumulative distribution function; for; For the first The t-statistic corresponding to each regression coefficient to be tested; For the first One regression coefficient to be tested; for The corresponding standard error.

[0038] Step 13236: When the structural test result, information relevance test result, search relevance test result, information significance test result, and search significance test result are all passed in the current round, the triple test result in the current round is determined to be passed.

[0039] Step 1324: If the triple test result in the current round is a pass, then the historical values ​​of the cumulative information content, the cumulative information content, the information index, and the search index at N discrete times in the current round are determined as the initial test group.

[0040] Step 1325: If the triple test result in the current round is a failure, then randomly sample the historical continuous time series to obtain the sampling sequence for the next round, and process it for the next round until the initial test group is obtained.

[0041] Step 2: Determine the information interaction difference equation and the search information interaction difference equation. Step 2 specifically includes the following steps.

[0042] Step 21: Define the promotion coefficient and 728 first influence factors of the first influence function, and define the inhibition coefficient and 728 second influence factors of the second influence function.

[0043] As an optional implementation, in step 21, the 728 first impact factors and 728 second impact factors are determined based on the cumulative information volume and the cumulative search information volume.

[0044] Specifically, 728 first impact factors It is about , , , , , The permutations and combinations of the six structural factors are: There are a total of 63 structural combinations; considering the positive and negative signs of the coefficients, have There are a total of 728 structures. Similarly, It is about , , , , , There are 63 possible structural combinations based on the six structural factors. Taking into account the sign of the coefficients... There are 728 possible structures.

[0045] Step 22: Combine any first impact factor and any second impact factor to form an interactive impact factor, and determine any interactive impact factor as the current interactive impact factor.

[0046] Step 23: Introduce the promotion coefficient and the first influence factor in the current interaction influence factors into the information independent difference equation to obtain the information interaction difference equation corresponding to the first influence factor in the current interaction influence factors. Introduce the inhibition coefficient and the second influence factor in the current interaction influence factors into the search information independent difference equation to obtain the search information interaction difference equation corresponding to the second influence factor in the current interaction influence factors.

[0047] As an optional implementation, in step 23, the information interaction difference equation corresponding to any first influence factor is: ; in, Discrete time in the information interaction difference equation Information index below; The promotion coefficient; It is the number one impact factor; The search information interaction difference equation corresponding to any second influence factor is: ; in, To search for discrete moments in the information exchange difference equation Information index below; The inhibition coefficient; It is the second most influential factor.

[0048] Step 3: Determine the concordance rate. Step 3 specifically includes the following steps.

[0049] Step 31: Construct N / 2 target test groups according to the time sequence of each discrete moment corresponding to the initial test group, and determine any target test group as the current target test group; wherein, the i-th target test group contains the first test value of the test parameter from the first discrete moment to the first test value of the test parameter at the (N / 2+i-1)-th discrete moment, i=1,2,…,N / 2.

[0050] Step 32: Based on the first test value of the cumulative information content and the first test value of the information index at each discrete moment in the current target test group, perform parameter fitting on the information interaction difference equation corresponding to the first influence factor in the current interaction influence factors to obtain the information fitting parameters and promotion coefficient of the current interaction influence factor under the current target test group.

[0051] Step 33: Based on the first test value of the cumulative search information and the first test value of the search index at each discrete time in the current target test group, perform parameter fitting on the search information interaction difference equation corresponding to the second influence factor in the current interaction influence factor to obtain the search fitting parameters and suppression coefficients of the current interaction influence factor under the current target test group.

[0052] Step 34: Perform a triple regression analysis based on the historical values ​​of the information index at each discrete moment in the current target test group, the information fitting parameters, promotion coefficient, search fitting parameters, and inhibition coefficient of the current interaction factor under the current target test group, and obtain the triple test results of the current interaction factor under the current target test group.

[0053] As an optional implementation, step 34 includes the following steps.

[0054] Step 341: Based on the intrinsic growth rate, upper limit of information content, intrinsic growth rate, and upper limit of search information content of the current interactive influence factor under the current target test group, perform a structural test to obtain the structural test results of the current interactive influence factor under the current target test group.

[0055] Specifically, similar to steps 13231, in step 341, when and When the structural test result is "passed", the structural test is used to test whether the information interaction difference equation and the search information interaction difference equation meet the structural requirements in terms of behavior.

[0056] Step 342: Based on the fitted value and historical value of the information index of the current interactive influence factor under the current target test group, determine the information relevance test result of the current interactive influence factor under the current target test group.

[0057] Specifically, similar to steps 13232, in step 342, based on the fitted value and historical value of the information index of the current interactive influence factor under the current target test group, the first goodness-of-fit index of the current interactive influence factor under the current target test group is determined. When the first goodness-of-fit index of the current interactive influence factor under the current target test group is greater than 0.25, the information relevance test result of the current interactive influence factor under the current target test group is passed. The information relevance test is used to test whether the information interaction difference equation has the explanatory power for the cumulative information amount.

[0058] Step 343: Based on the fitted value and historical value of the search index of the current interactive influencing factor under the current target test group, determine the search relevance test result of the current interactive influencing factor under the current target test group.

[0059] Specifically, similar to steps 13233, in step 343, based on the fitted value and historical value of the search index of the current interactive influence factor under the current target test group, the second goodness-of-fit index of the current interactive influence factor under the current target test group is determined. When the second goodness-of-fit index of the current interactive influence factor under the current target test group is greater than 0.25, the search relevance test result of the current interactive influence factor under the current target test group is passed. The search relevance test is used to test whether the search information interaction difference equation has explanatory power for the cumulative search information.

[0060] The calculation formulas for the first and second goodness-of-fit indices in steps 342 and 343 are similar to those in steps 13232 and 13233, except that N fitness is adjusted to the total number of discrete moments in the current target test group.

[0061] Step 344: Based on the intrinsic growth rate and upper limit of the information content of the current interactive influence factor under the current target test group, obtain the information significance test results of the current interactive influence factor under the current target test group.

[0062] Specifically, similar to steps 13234, in step 344, the first significance index of the current interactive influence factor under the current target test group is determined based on the intrinsic growth rate and upper limit of the information content of the current interactive influence factor under the current target test group. When the first significance index of the current interactive influence factor under the current target test group is less than 0.05, the information significance test result of the current interactive influence factor under the current target test group is considered to have passed the test.

[0063] Step 345: Based on the intrinsic growth rate and upper limit of the search information of the current interactive influencing factor under the current target test group, obtain the search significance test results of the current interactive influencing factor under the current target test group.

[0064] Specifically, similar to steps 13235, in step 345, the second significance index of the current interactive influence factor under the current target test group is determined based on the intrinsic growth rate and upper limit of the search information of the current interactive influence factor under the current target test group. When the second significance index of the current interactive influence factor under the current target test group is less than 0.05, the search significance test result of the current interactive influence factor under the current target test group is considered to have passed the test.

[0065] The calculation formulas for the first and second significance indices in steps 343 and 345 are the same as those for steps 13234 and 13235.

[0066] Step 346: When the structural test result, information relevance test result, search relevance test result, information significance test result, and search significance test result of the current interactive influence factor under the current target test group are all passed, the triple test result of the current interactive influence factor under the current target test group is determined to be passed.

[0067] Step 35: Determine the consistency rate of the current interaction factor based on the triple test results of the current interaction factor under each target test group.

[0068] As an optional implementation, step 35 includes: Step 351: Using the consistency rate calculation formula, calculate the consistency rate of the current interactive influence factor based on the triplet test results of the current interactive influence factor under each target test group; the consistency rate calculation formula is: ; in, This represents the consistency rate of the current interactive influence factors; The number of times the current interactive influence factor passes the triplet test under each target test group; The total number of the target test groups.

[0069] Step 4: Determine the target information interaction difference equation and the target search information interaction difference equation. Step 4 specifically includes the following steps.

[0070] Step 41: Determine the target values ​​of the optimal model parameters based on the consistency rate of all interactive influencing factors; the optimal model parameters include: the first model parameters and the second model parameters; the first model parameters include: the optimal first influencing factor, the optimal promotion coefficient, the optimal intrinsic growth rate of information content, and the optimal upper limit of information content; the second model parameters include: the optimal second influencing factor, the optimal inhibition coefficient, the optimal intrinsic growth rate of search information content, and the optimal upper limit of search information content.

[0071] As an optional implementation, step 41 includes: Step 411: Determine the interaction factor with the highest matching rate as the optimal interaction factor; Step 412: Determine the first influence factor in the optimal interaction influence factors as the optimal first influence factor, and determine the second influence factor in the optimal interaction influence factors as the optimal second influence factor; Step 413: Determine the facilitation coefficient corresponding to the optimal interaction influence factor as the optimal facilitation coefficient, determine the intrinsic growth rate of the information content corresponding to the optimal interaction influence factor as the optimal intrinsic growth rate of the information content, and determine the upper limit of the information content corresponding to the optimal interaction influence factor as the optimal upper limit of the information content. Step 414: Determine the inhibition coefficient corresponding to the optimal interaction influence factor as the optimal inhibition coefficient, determine the intrinsic growth rate of the search information volume corresponding to the optimal interaction influence factor as the optimal intrinsic growth rate of the search information volume, and determine the upper limit of the search information volume corresponding to the optimal interaction influence factor as the optimal upper limit of the search information volume.

[0072] Step 42: Substitute the target value of the first model parameter into the information interaction difference equation corresponding to the optimal first influence factor to obtain the target information interaction difference equation; substitute the target value of the second model parameter into the search information interaction difference equation corresponding to the optimal second influence factor to obtain the target search information interaction difference equation.

[0073] Step 5: Obtain the measured values ​​of the cumulative information content, cumulative search information content, information index, and search index at the current discrete time.

[0074] Step 6: Substitute the measured value of the cumulative information amount at the current discrete moment into the target information interaction difference equation to obtain the calculated value of the information index at the current discrete moment.

[0075] Step 7: Substitute the measured value of the cumulative search information at the current discrete moment into the target information interaction difference equation to obtain the calculated value of the search index at the current discrete moment.

[0076] Step 8: Based on the measured values ​​of the information index, search index, calculated information index, and calculated search index at the current discrete moment, determine the information cocoon effect index at the current discrete moment, and complete the detection of the information cocoon effect at the current discrete moment.

[0077] As an optional implementation, step 8 includes the following steps.

[0078] Step 81: Based on the measured values ​​of the information index and search index at the current discrete moment, determine the measured value of the push dominance at the current discrete moment.

[0079] Specifically, the formula for calculating the measure of push dominance is as follows: ; in, Discrete time The measure of push dominance; Discrete time The measured value of the information index; Discrete time The measurement value of the search index; This is a preset minimum constant used to prevent the denominator from being zero, and .

[0080] Step 82: Based on the measured value of the search index at the current discrete time, determine the measured value of the search inhibition at the current discrete time.

[0081] Specifically, the formula for calculating the search inhibition measure is as follows: ; in, Discrete time The measurement of search inhibition.

[0082] Step 83: Based on the measured values ​​of push dominance and search inhibition at the current discrete moment, determine the measured value of information cocoon strength at the current discrete moment.

[0083] Specifically, the formula for calculating the measurement value of information cocoon intensity is as follows: ; in, Discrete time The measured value of the information cocoon strength; These are weighting coefficients used to adjust the contribution weights of push dominance and search inhibition in the observations; the default is... This indicates that both are equally important; The truncation function indicates that the calculation result will be restricted to a certain range. Inside.

[0084] Step 84: Based on the calculated values ​​of the information index and search index at the current discrete moment, determine the calculated value of the push dominance at the current discrete moment.

[0085] Specifically, the formula for calculating the push dominance value is as follows: ; in, Discrete time The calculated value of the push dominance; Discrete time The calculated value of the information index; Discrete time The calculated value of the search index.

[0086] Step 85: Based on the calculated value of the search index at the current discrete time, determine the calculated value of the search inhibition degree at the current discrete time.

[0087] Specifically, the formula for calculating the search inhibition degree is as follows: ; in, Discrete time The calculated value of the search inhibition degree.

[0088] Step 86: Based on the calculated values ​​of push dominance and search inhibition at the current discrete moment, determine the calculated value of information cocoon strength at the current discrete moment.

[0089] Specifically, the formula for calculating the intensity of information cocoons is as follows: ; in, Discrete time The calculated value of the information cocoon strength.

[0090] Step 87: Determine the consistency weight for the current discrete moment based on the measured and calculated values ​​of the information cocoon strength at the current discrete moment.

[0091] Specifically, the formula for calculating the consistency weight is as follows: ; in, Discrete time Consistency weight; For consistency scale parameters, Measurement values ​​used to regulate the strength of information cocoons Compared with calculated values The deviation between them affects the consistency weight The degree of sensitivity to the impact, due to and Both are constrained to the interval [0, 1] by the clip function, therefore the absolute deviation between the two is... It is also located in the interval [0, 1]. The larger the value of , the more sensitive the consistency weight is to the deviation between the measured value and the calculated value; that is, even a small deviation can significantly reduce the consistency weight. The smaller the value of , the higher the tolerance of the consistency weight for deviation.

[0092] Based on the preset acceptable deviation And the target consistency weight under this deviation Confirmed, the calculation formula is: ; in, To pre-set acceptable deviation, ; for Expected consistency weight at time Therefore, when the deviation between the measured value and the calculated value reaches... When the consistency weight satisfies Thus The value of has a clear physical meaning and a data meaning. Take from the historical sample set The mean, median, or root mean square error, Use values ​​between 0.5 and 0.9; when a more rigorous evaluation of the consistency between the model's calculated values ​​and the measured values ​​is required, Take the smaller value or Take the larger value; when it is necessary to increase the tolerance to noise fluctuations, Take the larger value or Take the smaller value.

[0093] exist and When all values ​​are normalized to the interval [0, 1], The value range is from 0.1 to 20, preferably from 1 to 10. Unless otherwise specified, a value of can be used. , ,but This default value indicates that when the measured value and the calculated value of the information cocoon strength differ by 0.1, the consistency weight is approximately 0.8.

[0094] Step 88: Determine the information cocoon effect index at the current discrete moment based on the measured value of the information cocoon strength, the calculated value of the information cocoon strength, and the consistency weight.

[0095] Specifically, the formula for calculating the information cocoon effect index is as follows: in, Discrete time The information cocoon effect index.

[0096] In one exemplary embodiment, a computer device is provided, including: 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 an information cocoon effect detection method based on dynamic regression and model screening.

[0097] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements an information cocoon effect detection method based on dynamic regression and model screening.

[0098] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements an information cocoon effect detection method based on dynamic regression and model screening.

[0099] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 2 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements an information cocoon effect detection method based on dynamic regression and model screening.

[0100] Those skilled in the art will understand that Figure 2 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0101] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0102] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0103] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0104] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0105] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for detecting the information cocoon effect based on dynamic regression and model screening, characterized in that, The information cocoon effect detection method based on dynamic regression and model screening includes: Obtain historical continuous time series within a preset historical period; the historical continuous time series includes: the cumulative information volume time series of information pushed by the platform, the cumulative information volume time series of search generated by user search behavior, the information index time series, and the search index time series; Construct independent difference equations for information and search information; Based on historical continuous time series, parameter fitting is performed on the independent difference equations of information and search information to obtain an initial test set; the initial test set includes the first test values ​​of the test parameters at N discrete time points; N≥6 and N is an even number; the test parameters include: cumulative information content, cumulative search information content, information index, and search index; Define the promoting coefficient and 728 first-influence factors of the first-influence function, and define the inhibiting coefficient and 728 second-influence factors of the second-influence function; Any first impact factor and any second impact factor constitute an interactive impact factor, and any interactive impact factor is determined as the current interactive impact factor; By introducing the promotion coefficient and the first influence factor in the current interaction influence factors into the information independent difference equation, we obtain the information interaction difference equation corresponding to the first influence factor in the current interaction influence factors. By introducing the inhibition coefficient and the second influence factor in the current interaction influence factors into the search information independent difference equation, we obtain the search information interaction difference equation corresponding to the second influence factor in the current interaction influence factors. According to the time sequence of each discrete moment corresponding to the initial test group, construct N / 2 target test groups and determine any target test group as the current target test group; wherein, the i-th target test group contains the first test value of the test parameter from the first discrete moment to the first test value of the test parameter at the (N / 2+i-1)-th discrete moment, i=1,2,...,N / 2; Based on the first test value of the cumulative information content and the first test value of the information index at each discrete moment in the current target test group, the information information interaction difference equation corresponding to the first influence factor in the current interaction influence factors is fitted with parameters to obtain the information fitting parameters and promotion coefficient of the current interaction influence factor under the current target test group. Based on the first test value of the cumulative search information and the first test value of the search index at each discrete moment in the current target test group, the parameters of the search information interaction difference equation corresponding to the second influence factor in the current interaction influence factor are fitted to obtain the search fitting parameters and suppression coefficients of the current interaction influence factor under the current target test group. Based on the historical values ​​of the information index at each discrete moment in the current target test group, the information fitting parameters, promotion coefficient, search fitting parameters and inhibition coefficient of the current interaction factor under the current target test group, a triple test of regression analysis is performed to obtain the triple test results of the current interaction factor under the current target test group. The consistency rate of the current interactive influence factor is determined based on the triple test results of the current interactive influence factor under each target test group; The target values ​​of the optimal model parameters are determined based on the consistency rate of all interactive influence factors. The optimal model parameters include: the first model parameters and the second model parameters. The first model parameters include: the optimal first influence factor, the optimal promotion coefficient, the optimal intrinsic growth rate of information content, and the optimal upper limit of information content. The second model parameters include: the optimal second influence factor, the optimal inhibition coefficient, the optimal intrinsic growth rate of search information content, and the optimal upper limit of search information content. Substituting the target value of the first model parameter into the information interaction difference equation corresponding to the optimal first influence factor, we obtain the target information interaction difference equation; substituting the target value of the second model parameter into the search information interaction difference equation corresponding to the optimal second influence factor, we obtain the target search information interaction difference equation. Obtain the measured values ​​of the cumulative information content, cumulative search information content, information index, and search index at the current discrete moment; Substitute the measured value of the cumulative information amount at the current discrete moment into the target information interaction difference equation to obtain the calculated value of the information index at the current discrete moment. Substitute the measured value of the cumulative search information at the current discrete moment into the target information interaction difference equation to obtain the calculated value of the search index at the current discrete moment. Based on the measured values ​​of the information index, search index, and calculated values ​​of the information index and search index at the current discrete moment, the information cocoon effect index at the current discrete moment is determined, thus completing the detection of the information cocoon effect at the current discrete moment.

2. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 1, characterized in that, The independent difference equation for the information is: ； in, Discrete time in the information independent difference equation Information index below; The intrinsic growth rate of information volume; This represents the maximum amount of information that can be conveyed. Discrete time The cumulative amount of information in the following news items; The independent difference equation for searching information is: ； in, To search for discrete time intervals in the independent difference equations The search index below; The intrinsic growth rate of the amount of information searched; This is the upper limit for the amount of information that can be searched; Discrete time The cumulative amount of information searched below.

3. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 1, characterized in that, Based on historical continuous time series data, parameter fitting was performed on the independent difference equations for information and search information to obtain the initial test set, including: Random sampling is performed on the historical continuous time series to obtain the initial sampling sequence; the initial sampling sequence includes the historical values ​​of the cumulative information content, the cumulative search information content, the information index, and the search index at N discrete time points; Using a binary linear regression method and a triple test of regression analysis, the independent difference equations of information and search information are processed in multiple rounds based on the initial sampling sequence to obtain the initial test set; wherein, the processing procedure in any current round includes: Using the binary linear regression method, the independent difference equation of information is fitted with parameters based on the historical values ​​of the cumulative information content and the historical values ​​of the information index at N discrete time points in the current round of sampling sequence. The information fitting parameters are: the intrinsic growth rate of information content, the upper limit of information content, and the fitted value of the information index at each discrete time point. When the current round is the initial round, the sampling sequence in the current round is the initial sampling sequence. Using the binary linear regression method, the historical values ​​of the cumulative search information and the historical values ​​of the search index at N discrete moments in the current round are used to fit the parameters of the independent difference equation of the search information, so as to obtain the search fitting parameters corresponding to the sampling sequence in the current round. The search fitting parameters include: the intrinsic growth rate of the search information, the upper limit of the search information, and the fitted value of the search index at each discrete moment. Based on the historical values ​​of the information index, information fitting parameters, and search fitting parameters of the N discrete moments in the sampling sequence of the current round, a triple test of regression analysis is performed on the sampling sequence of the current round to obtain the triple test results of the current round. If the triple test result in the current round is a pass, then the historical values ​​of the cumulative information content, the cumulative information content, the information index, and the search index at N discrete times in the current round are determined as the initial test group. If the triple test result in the current round fails, random sampling is performed on the historical continuous time series to obtain the sampling sequence for the next round, and the next round of processing is carried out until the initial test group is obtained.

4. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 3, characterized in that, Based on the historical values ​​of the information index, information fitting parameters, and search fitting parameters of the N discrete time points in the sampling sequence of the current round, a triple test of regression analysis is performed on the sampling sequence of the current round to obtain the triple test results for the current round, including: Based on the intrinsic growth rate of information content, the upper limit of information content, the intrinsic growth rate of search information content, and the upper limit of search information content corresponding to the sampling sequence in the current round, a structural test is performed to obtain the structural test results for the current round. Based on the fitted values ​​and historical values ​​of the information index at N discrete moments corresponding to the sampling sequence in the current round, the information relevance test results in the current round are determined. Based on the fitted values ​​and historical values ​​of the search index at N discrete moments corresponding to the sampling sequence in the current round, the search relevance test results in the current round are determined. Based on the intrinsic growth rate and upper limit of information content of the sampled sequence in the current round, the significance test results of information in the current round are obtained; Based on the intrinsic growth rate and upper limit of the search information corresponding to the sampling sequence in the current round, the search significance test results in the current round are obtained. If the structural test result, information relevance test result, search relevance test result, information significance test result, and search significance test result are all passed in the current round, then the triple test result for the current round is determined to be passed.

5. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 1, characterized in that, The 728 first impact factors and 728 second impact factors were determined based on the cumulative amount of information and the cumulative amount of information searched.

6. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 2, characterized in that, The information interaction difference equation corresponding to any first impact factor is: ； in, Discrete time in the information interaction difference equation Information index below; The promotion coefficient; It is the number one impact factor; The search information interaction difference equation corresponding to any second influence factor is: ； in, To search for discrete moments in the information exchange difference equation Information index below; The inhibition coefficient; It is the second most influential factor.

7. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 6, characterized in that, Based on the historical values ​​of the information index at each discrete time point in the current target test group, the information fit parameters, facilitation coefficient, search fit parameters, and inhibition coefficient of the current interaction factor under the current target test group, a triple regression analysis is performed to obtain the triple test results of the current interaction factor under the current target test group, including: Based on the intrinsic growth rate, upper limit of information content, intrinsic growth rate and upper limit of search information content of the current interactive influence factor under the current target test group, a structural test is conducted to obtain the structural test results of the current interactive influence factor under the current target test group. Based on the fitted value and historical value of the information index of the current interactive influence factor under the current target test group, determine the information relevance test result of the current interactive influence factor under the current target test group; Based on the fitted value and historical value of the search index of the current interactive influencing factor under the current target test group, determine the search relevance test result of the current interactive influencing factor under the current target test group; Based on the intrinsic growth rate and upper limit of the information content of the current interactive influence factor under the current target test group, the information significance test results of the current interactive influence factor under the current target test group are obtained. Based on the intrinsic growth rate and upper limit of the search information of the current interactive influencing factors under the current target test group, the search significance test results of the current interactive influencing factors under the current target test group are obtained. When the current interactive influence factor passes all four tests (structural, information relevance, search relevance, information significance, and search significance) under the current target test group, the triple test result of the current interactive influence factor under the current target test group is determined to be a pass.

8. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 6, characterized in that, The consistency rate of the current interaction factor is determined based on the triplet test results of the current interaction factor under each target test group, including: Using the concordance rate calculation formula, based on the triplet test results of the current interactive influence factor under each target test group, the concordance rate of the current interactive influence factor is calculated; the concordance rate calculation formula is: ； in, This represents the consistency rate of the current interactive influence factors; The number of times the current interactive influence factor passes the triplet test under each target test group; The total number of the target test groups.

9. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 1, characterized in that, The target values ​​of the optimal model parameters are determined based on the consistency rate of all interactive influencing factors, including: The interaction factor with the highest matching rate is determined as the optimal interaction factor. The first influence factor in the optimal interaction influence factors is determined as the optimal first influence factor, and the second influence factor in the optimal interaction influence factors is determined as the optimal second influence factor. The facilitation coefficient corresponding to the optimal interaction factor is determined as the optimal facilitation coefficient, and the intrinsic growth rate of the information content corresponding to the optimal interaction factor is determined as the optimal intrinsic growth rate of the information content. The upper limit of the information content corresponding to the optimal interaction factor is determined as the upper limit of the information content. The inhibition coefficient corresponding to the optimal interaction influence factor is determined as the optimal inhibition coefficient, the intrinsic growth rate of the search information volume corresponding to the optimal interaction influence factor is determined as the optimal intrinsic growth rate of the search information volume, and the upper limit of the search information volume corresponding to the optimal interaction influence factor is determined as the upper limit of the optimal search information volume.

10. The method for detecting the information cocoon effect based on dynamic regression and model screening according to claim 1, characterized in that, Based on the measured values ​​of the information index, search index, calculated information index, and calculated search index at the current discrete moment, the information cocoon effect index at the current discrete moment is determined, including: Based on the measured values ​​of the information index and search index at the current discrete moment, determine the measured value of the push dominance at the current discrete moment; Based on the measured value of the search index at the current discrete time, determine the measured value of the search inhibition at the current discrete time; Based on the measured values ​​of push dominance and search inhibition at the current discrete moment, the measured value of information cocoon strength at the current discrete moment is determined. Based on the calculated values ​​of the information index and search index at the current discrete moment, the calculated value of the push dominance at the current discrete moment is determined. Based on the calculated value of the search index at the current discrete moment, determine the calculated value of the search inhibition degree at the current discrete moment; Based on the calculated values ​​of push dominance and search inhibition at the current discrete moment, the calculated value of information cocoon strength at the current discrete moment is determined. Based on the measured and calculated values ​​of the information cocoon strength at the current discrete moment, the consistency weight at the current discrete moment is determined. Based on the measured value of the information cocoon strength at the current discrete moment, the calculated value of the information cocoon strength, and the consistency weight, the information cocoon effect index at the current discrete moment is determined.

Images