Public service data analysis system and method based on artificial intelligence

Abstract

The invention discloses a public service data analysis system and method based on artificial intelligence, relates to the field of data analysis, and is used for solving the problem that information to be retrieved of a user is arranged later due to a relatively close group relationship. Establishing a data analysis model, generating a change evaluation coefficient, comparing the change evaluation coefficient with a change threshold value to obtain a larger change attribute and a smaller change attribute, subtracting the change evaluation coefficient of the larger change attribute from the change evaluation coefficient of the smaller change attribute to obtain a change trend, comparing the change trend with a trend threshold value, dividing a group with a larger change trend, and determining the group with the larger change trend. The method comprises the following steps: selecting a group with a relatively large change trend, collecting residual groups corresponding to the group with the relatively large change trend to obtain a similar trend among the groups, and performing group category prediction analysis through a time sequence model to ensure that recommended contents better meet actual requirements of the groups and meet the requirement that each user has personalized and real-time retrieval information arrangement.

Description

Technical Field

[0001] The present invention relates to the field of data analysis, and more specifically, to a public service data analysis system and method based on artificial intelligence. Background Art

[0002] With the acceleration of the urbanization process, the improvement of the public service system has become the key to enhancing urban management efficiency and people's livelihood. A large amount of public service data such as transportation, medical care, education, and security is continuously generated. Traditional algorithms can efficiently collect, analyze, and classify this data and provide accurate public service information to users.

[0003] The existing technology has the following deficiencies:

[0004] Currently, with the expansion of the urban area and the increase in the number of citizens (i.e., users), the existing technology includes the initial large base user group in the group analysis. However, according to the matching degree between the existing group types and different public service information, it is always fixed. The system will default to divide the group corresponding to the user, and the group relationship is relatively close. Although this can make the dissemination of public service information more rapid, for individual users, the information they want to retrieve will be ranked behind due to the group relationship, reducing the user experience and service efficiency. Therefore, a public service data analysis system and method based on artificial intelligence are proposed.

[0005] The above information disclosed in the background art section is only used to strengthen the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the existing technology, embodiments of the present invention provide a public service data analysis system and method based on artificial intelligence, and solve the problems proposed in the above background art by using different product inspection methods.

[0007] To achieve the above object, the present invention provides the following technical solution. A public service data analysis method based on artificial intelligence includes S1: collecting user characteristic information and retrieval characteristic information, and obtaining the similarity of group user historical retrieval information, the group historical retrieval difference value, and the group user retrieval correlation through data processing;

[0008] S2: obtaining the similarity of group user historical retrieval information, the group historical retrieval difference value, and the group user retrieval correlation, establishing a data analysis model (weighted calculation), and generating a change evaluation coefficient;

[0009] S3: Divide the change evaluation coefficient according to a preset change threshold to obtain a larger change attribute and a smaller change attribute, calculate the difference between the larger change attribute and the smaller change attribute to obtain a change trend, and compare the change trend with a preset trend threshold to divide out a group with a larger change trend;

[0010] S4: Obtain the group information corresponding to the group with a larger change trend, and collect the remaining groups corresponding to the group with a larger change trend, and calculate the similarity between groups to obtain the similarity trend between groups;

[0011] S5: According to the group information of the group with a larger change trend and the similarity trend between groups, perform group category prediction analysis through a time series model, further refine and optimize the adjusted group category, and generate a personalized recommended list plan.

[0012] In a preferred embodiment, the user characteristic information includes the similarity of the historical retrieval information of group users and the relevance of group user retrievals; the retrieval characteristic information includes the difference value of group historical retrievals;

[0013] Obtain the historical retrieval information and public service data of group users;

[0014] Perform vector representation on the historical retrieval information to obtain retrieval vectors, and then calculate the similarity between the retrieval vectors through cosine similarity to obtain the similarity Sy of the historical retrieval information of group users. i ; where i represents the i-th user;

[0015] Integrate the relevance between each group user and the public service data through weighted averaging to obtain the retrieval relevance Rg of group users. j ; j represents the j-th public service data;

[0016] Compare the retrieval vectors of each group user pairwise to obtain the degree of difference, and integrate the degree of difference through methods such as weighting or averaging to obtain the difference value Dg of group historical retrievals.

[0017] In a preferred embodiment, obtain the similarity Sy of the historical retrieval information of group users. i 、the retrieval relevance Rg of group users j and the difference value Dg of group historical retrievals to generate a change evaluation coefficient. The specific formula is:

[0018]

[0019] In the formula, is the change evaluation coefficient, and They are respectively the preset proportionality coefficients of the similarity of the historical retrieval information of the group users, the retrieval relevance of the group users, and the group historical retrieval difference value, and and are all greater than 0.

[0020] In a preferred embodiment, the change evaluation coefficient is divided according to a preset change threshold to obtain a larger change attribute and a smaller change attribute, including:

[0021] Compare and analyze the change evaluation coefficient with the continuously iterated change threshold;

[0022] If the change evaluation coefficient is greater than or equal to the change threshold, mark the current group as a larger change attribute and generate a change signal;

[0023] If the change evaluation coefficient is less than the change threshold, mark the current group as a smaller change attribute and generate a hold signal.

[0024] In a preferred embodiment, the specific process of dividing the group with a larger change trend is as follows:

[0025] Collect the coefficient labels corresponding to all change evaluation coefficients, and count the first value corresponding to the change evaluation coefficient with the label of the larger change attribute and the second value corresponding to the change evaluation coefficient with the label of the smaller change attribute;

[0026] Subtract the second value from the first value to obtain the change trend of the current group;

[0027] Compare the change trend with the trend threshold. If the change trend is greater than or equal to the trend threshold, it is a group with a larger change trend.

[0028] In a preferred embodiment, the group information corresponding to the group with a larger change trend includes the similarity of the historical retrieval information of the group users in the corresponding group, the retrieval relevance of the group users, and the group historical retrieval difference value;

[0029] The similarity trend between groups is obtained by subtracting the similarity between the feature information of the group in the previous time period and the remaining groups from the similarity between the feature information of the group in the current time period and the remaining groups;

[0030] Through time series model for group category prediction analysis, a personalized recommendation list scheme is generated.

[0031] In a preferred embodiment, the time series model used is the ARIMAX model. The specific steps for adjusting the group category prediction analysis through the time series model are as follows:

[0032] Step A1, obtain the data for prediction;

[0033] Step A2, establish an ARIMAX model;

[0034] Step A3, use the maximum likelihood estimation (MLE) method to estimate the ARIMAX model parameters;

[0035] Step A4, verify the effect of the fitted model, and check the goodness of fit of the model through the residual analysis method;

[0036] Step A5, use the fitted model to predict the future group categories, and take the maximum value of the prediction results as the group category evaluated as the common preference type group in the future time period.

[0037] In a preferred embodiment, the exogenous variables in the ARIMAX model include the similarity of user historical retrieval information, the relevance of group user retrieval, the difference value of group historical retrieval, and the similarity trend between groups. The ARIMAX model formula is:

[0038]

[0039] In the formula, y t is the group category at the current time point, α is the constant term, is the i-th order autoregressive parameter, p is the order of the autoregressive term, θ j is the j-th order moving average parameter, q is the order of the moving average term, ∈ t-j is the white noise term lagged by j periods, ∈ t is the white noise term, representing the random error, β k is the coefficient of the exogenous variable X t-k , m is the lag order of the exogenous variable, X t-k is the exogenous variable lagged by k periods;

[0040] In step A3, α, θ1, β1, β2 are calculated and obtained through the maximum likelihood estimation method.

[0041] An artificial intelligence-based public service data analysis system includes a data collection module, a data processing module, a group analysis module, and a prediction and evaluation module;

[0042] The data collection module is used to collect user characteristic information and retrieval characteristic information, and obtain the similarity of group user historical retrieval information, the difference value of group historical retrieval, and the relevance of group user retrieval through data processing and send them to the data processing module;

[0043] The data processing module is used to obtain the similarity of group user historical retrieval information, the difference value of group historical retrieval, and the relevance of group user retrieval, establish a data analysis model (weighted calculation), and generate a change evaluation coefficient, and send it to the group analysis module;

[0044] The group analysis module is used to divide the change evaluation coefficient according to a preset change threshold, obtain the large-change attributes and small-change attributes, calculate the difference between the large-change attributes and the small-change attributes to obtain the change trend, and compare the change trend with the preset trend threshold to divide the groups with a large change trend and send them to the prediction and evaluation module;

[0045] The prediction and evaluation module is used to obtain the group information with a large change trend, collect the remaining groups corresponding to the groups with a large change trend, calculate the similarity between the groups to obtain the similar trend between the groups, perform group category prediction analysis through a time series model, further refine and optimize the adjusted group category, and generate a personalized recommended list scheme.

[0046] The technical effects and advantages of the present invention:

[0047] 1. By collecting user characteristic information and retrieval characteristic information, establishing a data analysis model, generating a change evaluation coefficient and comparing it with the change threshold, the present invention obtains large-change attributes and small-change attributes. By subtracting the change evaluation coefficient of the large-change attribute from the change evaluation coefficient of the small-change attribute, the change trend is obtained and compared with the trend threshold to divide the groups with a large change trend, accurately identify the changes in the groups, optimize the allocation of data resources, and facilitate user retrieval.

[0048] 2. By obtaining the group information with a large change trend and the similar trend between the groups, and performing group category prediction analysis according to the time series model, the present invention generates a personalized recommended list scheme. According to the maximum value, minimum value and average value of the prediction results, the group category is dynamically adjusted, so as to ensure that the recommended content more meets the actual needs of the group, make the group category more flexible, meet the personalized and real-time retrieval information arrangement for each user, and enhance the user experience. Description of the Drawings

[0049] Figure 1 It is a method flow chart of a public service data analysis method based on artificial intelligence of the present invention.

[0050] Figure 2 It is a module schematic diagram of a public service data analysis system based on artificial intelligence of the present invention. Detailed Embodiments

[0051] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0052] Through group recommendation based on data analysis, the present invention preferentially divides users (i.e., citizens) into pre-planned groups. For example, groups can be set according to different regions (public service data in different communities, districts, townships, etc. are different), and multiple public service groups are obtained;

[0053] Using predefined heuristic methods, according to the retrieval data of group users, personalized recommendation lists are generated for each group user respectively. By clustering the personalized recommendation lists of group users, filtering and ranking the recommendation results, a group personalized recommendation list is generated and the correlation threshold with public service data is used to evaluate the change situation and trend of the group;

[0054] Embodiment 1

[0055] Please refer to Figure 1 , a method for analyzing public service data based on artificial intelligence, and the specific operation process is as follows:

[0056] S1: Collect user characteristic information and retrieval characteristic information, and obtain the similarity of group user historical retrieval information, the difference value of group historical retrieval, and the retrieval correlation of group users through data processing;

[0057] Among them, user characteristic information includes the similarity of group user historical retrieval information and the retrieval correlation of group users; retrieval characteristic information includes the difference value of group historical retrieval;

[0058] The similarity of group user historical retrieval information refers to the similarity degree between the user's past search behaviors and query records to determine whether there are significant changes in the user portrait; that is, the similarity between the user and public service data. Its acquisition logic is to vectorize the user's historical retrieval information (such as retrieval keywords, query records, etc.) through natural language processing (NLP) technology, and then calculate the similarity of group user historical retrieval information through cosine similarity. Its formula expression is:

[0059]

[0060] Among them, Sy i is the similarity of group user historical retrieval information, and respectively represent the retrieval information vectors of two times. For it is defined as the most recent retrieval information, and for it is defined as the remaining historical retrieval information except the most recent retrieval information; among them, i represents the i-th user; multiple calculations are performed to obtain the similarity of group user historical retrieval information for each group;

[0061] Specifically, the quantity of the remaining historical retrieval information excluding the most recent retrieval information is not limited, but is obtained by the system based on the historical retrieval information retained by different users, which will not be elaborated here;

[0062] It should be noted that a user profile refers to a user characteristic description model formed by collecting and analyzing various behavioral, attribute, and preference data of users; a user profile usually includes data such as basic information, retrieval hobbies, and online behaviors of users;

[0063] The relevance of group user retrieval refers to the degree of relevance between users within a group and public service data; that is, the relevance between the group and public service data, and its acquisition logic is to integrate the relevance between each group user and public service data through weighted average. The specific formula is expressed as:

[0064]

[0065] Among them, Rg j represents the relevance between group g and public service data j, that is, the relevance of group user retrieval. n is the total number of people in the group, and P(R i =1│i,j) represents the relevance between user i and public service data j, and j represents the jth public service data;

[0066] Among them, for the public service data corresponding to the groups in which the system preferentially divides users (i.e., citizens) into pre-planned groups is limited (i.e., the above-mentioned jth public service data). The public service data of this group is set based on the category differences of users in the group according to regional differences. For example, for the citizens in a certain jurisdiction, the public service data of this jurisdiction is preferentially recommended (i.e., j represents the jth public service data of this jurisdiction), and then the public service data of other jurisdictions is recommended. Therefore, groups have different categories. Among them, the initial group is default defined as a shared preference type group, that is, by default, most group users (for example, at least 50% of group users) are users of similar types. Group users do not need to have common preferences, that is, most group users do not object to the recommended items;

[0067] Among them, the definition of group categories is limited here. Three group categories are defined, namely, common preference type group, shared preference type group, and split preference type group. For the common preference type group, the recommended public service data is relevant to all users in the group, that is, all group users are relevant to this public service data, and at least in the case where there are no users who dislike it, a consensus can be reached; for the split preference type group, at least one group user is relevant to this public service data, then the group users of this type may be users of different types;

[0068] The group historical retrieval difference value refers to the degree of difference in the historical retrieval behaviors and records among different users within a group; that is, the difference value between the group and the users; its acquisition logic is that the historical retrieval information of each user is converted into a vector representation through natural language processing technology, the retrieval vectors of each group user are compared, the degree of difference is obtained and integrated through methods such as weighting or averaging, and its formula is expressed as follows:

[0069]

[0070] Among them, Dg represents the group historical retrieval difference value, n is the total number of people in the group, and d(i,g) represents the difference value between the retrieval information vectors of the i-th user and the g-th user;

[0071] To sum up, the three parameters collected (the similarity Sy of the historical retrieval information of group users i , the retrieval relevance Rg of group users j and the group historical retrieval difference value Dg) are all analyzed around the strength of the relationship among users, public service data, and groups to determine the degree of change of the group;

[0072] Specifically, the similarity between users within the group or the similarity between public service data within the group is not used as a reference basis because the focus of the technology of this application is to analyze the relationship between users and public service data to evaluate the change situation of the group. The similarity between users pays more attention to the individual differences between users rather than the matching degree of the group as a whole with public service data; although the similarity between public service data can help discover similar service data, for the group system, the needs of users are often diverse, and the service items concerned may vary greatly, which is likely to mislead the calculation results;

[0073] S2: Obtain the similarity of the historical retrieval information of group users, the group historical retrieval difference value, and the retrieval relevance of group users, establish a data analysis model, and generate a change evaluation coefficient;

[0074] Among them, the data analysis model refers to a weighted analysis model, and a change evaluation coefficient is generated through weighted calculation;

[0075] Obtain the similarity Sy of the historical retrieval information of group users i , the retrieval relevance Rg of group users j and the group historical retrieval difference value Dg, and generate a change evaluation coefficient According to the formula:

[0076]

[0077] In the formula, is the change evaluation coefficient, and They are respectively the preset proportionality coefficients of the similarity of the historical retrieval information of group users, the retrieval relevance of group users, and the difference value of the historical retrieval of the group, and and are all greater than 0;

[0078] Among them, the similarity of the historical retrieval information of group users, the retrieval relevance of group users, and the difference value of the historical retrieval of the group are all digital manifestations directly expressing the current change trend of the group;

[0079] Specifically, since the number of users is not the same as the number of public service data, the calculation order is limited. Let the current user set be {n1, n2, …, n i}, and the current public service data set be {m1, m2, …, m j}, then there exists {i > j or i < j or i = j}. Among them, for {i = j}, the calculation is performed according to the corresponding numerical sorting; for {i > j}, the remaining number of users is j - i, and the retrieval relevance of group users of the public service data most relevant to the remaining users is substituted for calculation. Conversely, for {i < j}, the remaining number of public service data is i - j, and the similarity of the historical retrieval information of group users most similar to the remaining public service data is substituted for calculation;

[0080] It can be seen from the above formula that when the similarity of the historical retrieval information of group users and the retrieval relevance of group users are larger, it indicates that the change trend of the current group is small, the change trend of the user retrieval content is small, and the relevance between the retrieval content within the group and the public service data is strong, then the change evaluation coefficient is smaller. Conversely, the larger the difference value of the historical retrieval of the group, the larger the change evaluation coefficient;

[0081] S3: Divide the change evaluation coefficient according to the preset change threshold to obtain the larger change attribute and the smaller change attribute, calculate the difference between the larger change attribute and the smaller change attribute to obtain the change trend, and compare the change trend with the preset trend threshold to divide the group with a larger change trend;

[0082] The acquisition logic of the change threshold is to collect the distribution sample set of the historical group division information database, then divide the data set into a training set and a test set, set the evaluation index and clustering algorithm, train the model on the training set and evaluate the model performance on the test set in each round of iteration of cross-validation, and then adjust the change threshold according to the performance of the validation set. Therefore, the change threshold is constantly iteratively updated;

[0083] In the present invention, clustering algorithm is a kind of unsupervised learning algorithm, which is used to divide the data points in the data set into groups or clusters with similarity; a common one is K-means clustering, which divides the data points in the data set into K clusters, so that the distance between the data point of each curve and the center point (centroid) of the cluster to which it belongs is minimized, and finally the effect of the adjusted change threshold is measured by the Euclidean distance, so as to set the change threshold;

[0084] After obtaining the change assessment coefficient, the change assessment coefficient is compared and analyzed with the continuously iterated change threshold;

[0085] If the change assessment coefficient is greater than or equal to the change threshold, the current group is marked as a large change attribute and a change signal is generated;

[0086] If the change evaluation coefficient is less than the change threshold, the current group is marked as a small change attribute and a hold signal is generated;

[0087] Through the above comparison, the current group obtains multiple label marks. Due to the continuous iteration of the change threshold, even the change evaluation coefficients with the same value may have different labels due to different calculation orders;

[0088] Specifically, the specific process of dividing the groups with larger change trends is as follows:

[0089] Collect coefficient labels corresponding to all change assessment coefficients, and the statistical coefficient labels are the first values corresponding to the change assessment coefficients of the larger change attribute and the second values corresponding to the change assessment coefficients of the smaller change attribute;

[0090] Subtract the first value from the second value to obtain the change trend of the current group;

[0091] The change trend is compared with the trend threshold. If the change trend is greater than or equal to the trend threshold, it is a group with a larger change trend.

[0092] Among them, the trend threshold is obtained through the current change assessment coefficient and the historical group change information;

[0093] Specifically, the groups with larger change trends refer to the groups that have undergone significant changes, and the public service data corresponding to the groups need to be deleted or added to meet the most efficient recommendation of the groups;

[0094] The present invention collects user characteristic information and retrieval characteristic information, establishes a data analysis model, generates a change evaluation coefficient and compares it with a change threshold to obtain a large-change attribute and a small-change attribute. By subtracting the change evaluation coefficient of the large-change attribute from the change evaluation coefficient of the small-change attribute, a change trend is obtained and compared with a trend threshold to divide groups with a large change trend, accurately identify the change situation of the groups, optimize the allocation of data resources, and facilitate user retrieval.

[0095] Embodiment 2

[0096] In Embodiment 1 of the present invention, emphasis is placed on exemplifying the collection of user characteristic information and retrieval characteristic information, establishing a data analysis model, generating a change evaluation coefficient and comparing it with a change threshold to obtain a large-change attribute and a small-change attribute. By subtracting the change evaluation coefficient of the large-change attribute from the change evaluation coefficient of the small-change attribute, a change trend is obtained and compared with a trend threshold to divide groups with a large change trend; however, in Embodiment 1, only starting from obtaining the change trend of the groups, obviously, although those skilled in the art can adjust the public service information recommended preferentially for the groups according to the specific change trend, relatively speaking, the efficiency of manual adjustment is low, still causing the problem of low service efficiency; in view of the above problems, Embodiment 2 of the present invention is further refined;

[0097] S4: Obtain the group information corresponding to the groups with a large change trend, and collect the remaining groups corresponding to the groups with a large change trend, and calculate the similarity to obtain the similarity trend between groups;

[0098] The group information of the groups with a large change trend includes the similarity of the historical retrieval information of the group users in the corresponding group, the relevance of the group users' retrieval, and the difference value of the historical retrievals of the group;

[0099] The remaining groups corresponding to the groups with a large change trend refer to the other groups except the groups with a large change trend. The similarity trend between groups is obtained by subtracting the similarity between groups in the previous time period from the similarity between groups in the current time period;

[0100] Specifically, the similarity between groups in the previous time period refers to the similarity between the characteristic information of the groups in the previous time period (the characteristic information includes the similarity of the historical retrieval information of the group users, the relevance of the group users' retrieval, and the difference value of the historical retrievals of the group) and the remaining groups; the "similarity between groups" in the similarity between groups in the current time period is the same as the above-mentioned similarity between groups in the previous time period;

[0101] Among them, the time period refers to the time interval for collecting parameters in Embodiment 1. Usually, the experimenter comprehensively analyzes information such as the change frequency of the groups to obtain the time interval for collecting parameters. It can be understood that the specific length of the time interval is not fixed, but is implemented by the experimenter according to the specific situation, which will not be elaborated here;

[0102] S5: Based on the group information with a large change trend and the similar trends among groups, perform group category prediction analysis through a time series model, further refine and optimize the adjusted group categories, and generate a personalized recommended list solution;

[0103] As can be seen from the above, since the time interval of the acquisition parameters is continuously recorded, the change trend of the group is determined by past data, and the group characteristics will change over time. Therefore, the time series model can effectively predict the future change trend of the group;

[0104] It should be noted that the time series model adopted in this embodiment is the ARIMAX model. Refining and optimizing the adjusted group categories means that when it is determined that the current group belongs to the group with a large change trend, through further analysis and adjustment, the generated personalized recommended list solution is made more scientific and reasonable, so as to realize the control of the corresponding public service data within the group;

[0105] Furthermore, the specific steps of performing group category prediction analysis by adjusting through the time series model are as follows:

[0106] Step A1, obtain the data for prediction;

[0107] Step A2, establish an ARIMAX model;

[0108] Step A3, use the maximum likelihood estimation (MLE) method to estimate the parameters of the ARIMAX model;

[0109] Step A4, verify the effect of the fitted model, and check the goodness of fit of the model through the residual analysis method;

[0110] Step A5, use the fitted model to predict the future group categories, and take the maximum value of the prediction result as the group category evaluated as the common preference type group within the future time period;

[0111] Among them, for adjusting the group categories, there are three categories. The common preference type group, the shared preference type group, and the split preference type group respectively correspond to the maximum value, the average value, and the minimum value of the prediction result;

[0112] Specifically, the group categories have been described in Embodiment 1 and will not be elaborated here;

[0113] Specifically, the data for prediction includes the similarity of the historical retrieval information of group users corresponding to each time point, the retrieval correlation of group users, the difference value of group historical retrievals, the similar trends among groups, and the group change trend data; among them, the similarity of the historical retrieval information of group users corresponding to each time point, the retrieval correlation of group users, and the difference value of group historical retrievals have been exemplified in Embodiment 1 and will not be elaborated here;

[0114] The group change trend data refers to the historical group change trend data, which is used as the main variable of the time series;

[0115] Furthermore, the basic form of the ARIMAX model is:

[0116]

[0117] In the formula, y t is the group category at the current time point, α is the constant term, is the autoregressive parameter of the i-th order, p is the order of the autoregressive term, θ j is the moving average parameter of the j-th order, q is the order of the moving average term, ∈ t-j is the white noise term lagged by j periods, ∈ t is the white noise term, representing the random error, β k is the coefficient of the exogenous variable X t-k m is the lag order of the exogenous variable, X t-k is the exogenous variable lagged by k periods;

[0118] It should be noted that the exogenous variable part can integrate the influence of other relevant variables (such as the similarity of historical retrieval information of group users, the retrieval relevance of group users, the difference value of group historical retrievals, and the similarity trend between groups) on the group change trend;

[0119] It should be noted that in step A3, α, θ1, β1, β2 are calculated and obtained by the maximum likelihood estimation (MLE) method. The specific steps are as follows:

[0120] The error term ∈ t obeys the normal distribution N(0,σ 2 ), then the likelihood function is:

[0121]

[0122] Taking the logarithm of the likelihood function, the log-likelihood function is obtained:

[0123]

[0124] By maximizing the log-likelihood function, the parameter estimates α, θ1, β1, β2 are obtained;

[0125] By the maximum value, minimum value, and average value of the prediction results, the group category strategy is dynamically adjusted;

[0126] The following is an example of this embodiment:

[0127] The maximum value of the prediction result It represents the basis for evaluating the group category as a common preference group in the future time period. The group is classified into the common preference group. The correlation probability between the group and the item is expressed by the product of the correlations between all users and the item. The correlation probability between each public service data and the common preference group is calculated, and then filtered and ranked according to the correlation to generate a personalized recommendation list for the common preference group;

[0128] According to the minimum value of the prediction result It represents the basis for evaluating the group category as a split preference group in the future time period. The group is classified into the split preference group. By calculating the probability with the smallest correlation between the group users and the public service data, the correlation probability between each public service data and the split preference group is obtained, and then filtered and ranked according to the correlation to generate a personalized recommendation list for the split preference group;

[0129] According to the average value of the prediction result It represents that the group category maintains the current category in the future time period to meet the usage needs of the users within the group;

[0130] Among them, the setting of the group category is specifically implemented and divided by the experimenter, and the boundary of this division is not limited;

[0131] The present invention generates a personalized recommendation list solution by obtaining group information with large change trends and similar trends between groups, and performing group category prediction analysis according to the time series model. According to the maximum value, minimum value and average value of its prediction results, the group category is dynamically adjusted, so as to ensure that the recommended content better meets the actual needs of the group, making the group category more flexible, satisfying the personalized and real-time retrieval information arrangement for each user, and enhancing the user experience.

[0132] Embodiment 3

[0133] Please refer to Figure 2 , a public service data analysis system based on artificial intelligence, including a data collection module, a data processing module, a group analysis module, and a prediction evaluation module;

[0134] The data collection module is used to collect user characteristic information and retrieval characteristic information, and obtain the similarity of group user historical retrieval information, the group historical retrieval difference value, and the group user retrieval correlation through data processing, and send them to the data processing module;

[0135] The data processing module is used to obtain the similarity of group user historical retrieval information, the group historical retrieval difference value, and the group user retrieval correlation, establish a data analysis model (weighted calculation), and generate a change evaluation coefficient, and send it to the group analysis module;

[0136] The group analysis module is used to divide the change evaluation coefficient according to a preset change threshold, obtain the large-change attributes and small-change attributes, calculate the difference between the large-change attributes and the small-change attributes to obtain the change trend, and compare the change trend with a preset trend threshold to divide the groups with a large change trend and send them to the prediction and evaluation module;

[0137] The prediction and evaluation module is used to obtain the group information with a large change trend, collect the remaining groups corresponding to the groups with a large change trend, calculate the similarity between the groups to obtain the similarity trend between the groups, perform group category prediction and analysis through a time series model, further refine and optimize the adjusted group category, and generate a personalized recommendation list solution.

[0138] The above formulas are all dimensionless and take their numerical values for calculation. The formula is a formula obtained by collecting a large amount of data for software simulation to get the closest to the real situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0139] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired or wireless (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center that contains one or more collections of available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0140] It should be understood that in various embodiments of the present application, the magnitudes of the serial numbers of the above processes do not mean 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 to the implementation process of the embodiments of the present application.

[0141] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0142] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0143] In several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0144] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0145] In addition, the functional units in each embodiment of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0146] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0147] As described above, the above are only specific implementation manners of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A method for analyzing public service data based on artificial intelligence, characterized in that: Including S1: Collect user characteristic information and retrieval characteristic information, and obtain the similarity of historical retrieval information of group users, the difference value of group historical retrievals, and the retrieval correlation of group users through data processing; S2: Obtain the similarity of historical retrieval information of group users, the difference value of group historical retrievals, and the retrieval correlation of group users, establish a data analysis model, and generate a change evaluation coefficient; S3: Divide the change evaluation coefficient according to a preset change threshold to obtain a large change attribute and a small change attribute, calculate the difference between the large change attribute and the small change attribute to obtain a change trend, and compare the change trend with a preset trend threshold to divide out groups with a large change trend; S4: Obtain the group information corresponding to the groups with a large change trend, and collect the remaining groups corresponding to the groups with a large change trend, and calculate the similarity between groups to obtain the similarity trend between groups; S5: According to the group information of the groups with a large change trend and the similarity trend between groups, conduct a predictive analysis of group categories through a time series model, further refine and optimize the adjustment of group categories, and generate a personalized recommendation list plan.

2. The method for analyzing public service data based on artificial intelligence according to claim 1, wherein: The user characteristic information includes the similarity of historical retrieval information of group users and the retrieval correlation of group users; the retrieval characteristic information includes the difference value of group historical retrievals; Obtain the historical retrieval information and public service data of group users; Vectorize the historical retrieval information to obtain retrieval vectors, and then calculate the similarity between the retrieval vectors through cosine similarity to obtain the similarity Sy of the historical retrieval information of the group users i ; where i represents the i-th user; Integrate the correlation between each group of users and the public service data through weighted average to obtain the group user retrieval correlation Rg j ; j represents the j-th public service data; Compare the retrieval vectors of each group user pairwise to obtain the degree of difference, and integrate the degree of difference by means of weighting or averaging to obtain the group historical retrieval difference value Dg.

3. The method for analyzing public service data based on artificial intelligence according to claim 2, wherein: Obtain the similarity Sy of the historical retrieval information of group users i , the retrieval relevance Rg of group users j and the historical retrieval difference value Dg of the group to generate a change evaluation coefficient The specific formula is as follows: In the formula, is the change evaluation coefficient, and are respectively the preset proportionality coefficients of the similarity of the historical retrieval information of the group users, the retrieval relevance of the group users, and the group historical retrieval difference value, and and are both greater than 0.

4. A method for analyzing public service data based on artificial intelligence according to claim 1, characterized in that: Dividing the change evaluation coefficient according to a preset change threshold to obtain a large change attribute and a small change attribute, including: Compare and analyze the change evaluation coefficient with the continuously iterated change threshold; If the change evaluation coefficient is greater than or equal to the change threshold, mark the current group as a large change attribute and generate a change signal; If the change evaluation coefficient is less than the change threshold, mark the current group as a small change attribute and generate a hold signal.

5. A method for analyzing public service data based on artificial intelligence according to claim 1, characterized in that: The specific process of dividing out groups with a large change trend is as follows: Collect the coefficient labels corresponding to all change evaluation coefficients, and count the first value corresponding to the change evaluation coefficient with the label of large change attribute and the second value corresponding to the change evaluation coefficient with the label of small change attribute; Subtract the first value and the second value to calculate the change trend of the current group; Compare the change trend with the trend threshold. If the change trend is greater than or equal to the trend threshold, it is a group with a large change trend.

6. The method for analyzing public service data based on artificial intelligence according to claim 1, wherein: The group information corresponding to the groups with a large change trend includes the similarity of historical retrieval information of group users, the retrieval correlation of group users, and the difference value of group historical retrievals in the corresponding groups; Obtain the similarity trend between groups by subtracting the similarity between the characteristic information of the groups in the previous period and the remaining groups from the similarity between the characteristic information of the groups in the current period and the remaining groups; Generate a personalized recommendation list plan through predictive analysis of group categories by a time series model.

7. A method for analyzing public service data based on artificial intelligence according to claim 6, characterized in that: The time series model used is the ARIMAX model. The specific steps for adjusting the predictive analysis of group categories through the time series model are as follows: Step A1, obtain the data for prediction; Step A2, establish an ARIMAX model; Step A3, use the maximum likelihood estimation (MLE) method to estimate the ARIMAX model parameters; Step A4, verify the effect of the fitted model, and check the goodness of fit of the model through residual analysis; Step A5, use the fitted model to predict the future group categories, and take the maximum value of the prediction results as the group category evaluated as a common preference group in the future time period.

8. The method for analyzing public service data based on artificial intelligence according to claim 7, characterized in that: The exogenous variables in the ARIMAX model include the similarity of the historical retrieval information of group users, the retrieval relevance of group users, the difference value of group historical retrievals, and the similarity trend between groups. The ARIMAX model formula is: where y t is the group category at the current time point, α is the constant term, is the autoregressive parameter of the i-th order, p is the order of the autoregressive term, θ j is the moving average parameter of the j-th order, q is the order of the moving average term, ∈ t-j is the white noise term lagged by j periods, ∈ t is the white noise term, representing the random error, β k is the coefficient of the exogenous variable X t-k , m is the lag order of the exogenous variable, X t-k is the exogenous variable lagged by k periods; In step A3, α, θ1, β1, and β2 are calculated and obtained by the maximum likelihood estimation method.

9. An artificial intelligence-based public service data analysis system for implementing an artificial intelligence-based public service data analysis method according to any one of claims 1-8, characterized in that: It includes a data collection module, a data processing module, a group analysis module, and a prediction evaluation module; The data collection module is used to collect user characteristic information and retrieval characteristic information, and obtain the similarity of the historical retrieval information of group users, the difference value of group historical retrievals, and the retrieval relevance of group users through data processing and send them to the data processing module; The data processing module is used to obtain the similarity of the historical retrieval information of group users, the difference value of group historical retrievals, and the retrieval relevance of group users, establish a data analysis model (weighted calculation), and generate a change evaluation coefficient, and send it to the group analysis module; The group analysis module is used to divide the change evaluation coefficient according to a preset change threshold to obtain a large change attribute and a small change attribute, calculate the difference between the large change attribute and the small change attribute to obtain a change trend, and compare the change trend with a preset trend threshold to divide out the groups with a large change trend and send them to the prediction evaluation module; The prediction evaluation module is used to obtain the group information with a large change trend, collect the remaining groups corresponding to the groups with a large change trend, calculate the similarity to obtain the similarity trend between groups, conduct group category prediction analysis through a time series model, further refine and optimize the adjustment of group categories, and generate a personalized recommendation list plan.

Images