Communication service risk assessment method and device

Abstract

The invention discloses a communication service risk assessment method and device. The method comprises the following steps: acquiring first data of a target communication service, and determining a first service type and a second service type corresponding to the target communication service; respectively determining a first quantity corresponding to second data of the first service type and a second quantity corresponding to third data of the second service type in the preset data set; according to the first number and the second number, determining a risk level assessment strategy, the risk level assessment strategy including one of the following: a risk assessment strategy based on a dictionary rule, and a risk assessment strategy jointly determined according to the target classification model and the dictionary rule; and executing a risk level evaluation strategy on the first data of the target communication service to obtain a first target risk level of the target communication service. According to the method, the technical problems of overfitting and inaccurate assessment in risk level assessment of a traditional machine learning method are solved.

Description

Technical Field

[0001] The present application relates to the technical field of risk assessment, and in particular to a communication service risk assessment method and device. Background Art

[0002] In the modern communications industry, with the deepening of digital transformation and increasing business diversity, effectively assessing and managing potential risks in business processes has become increasingly important. Traditional communications business risk assessment relies heavily on manual review and fixed rule checks, which are insufficient for dealing with large, complex, and constantly changing business scenarios.

[0003] In recent years, the rapid development of artificial intelligence (AI) technology, particularly the widespread application of machine learning algorithms in data analysis, has provided new perspectives and solutions for risk assessment in the communications industry. Machine learning can automatically learn and identify patterns from historical data, demonstrating significant advantages in processing large, multi-dimensional datasets. However, machine learning is not a panacea. Especially when assessing risk in a small number of categories, algorithms can overfit due to insufficient training data, resulting in reduced classification accuracy. Furthermore, existing models may lack relevant experience for emerging business types, leaving the risk assessment process with blind spots. These issues have limited the breadth and depth of machine learning applications in communications business risk assessment, prompting the industry to consider how to optimize algorithms and introduce more flexible risk assessment mechanisms.

[0004] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0005] The embodiments of the present application provide a communication service risk assessment method and device to at least solve the technical problems of overfitting and inaccurate assessment in risk level assessment of traditional machine learning methods.

[0006] According to one aspect of an embodiment of the present application, a communication service risk assessment method is provided, including: obtaining first data of a target communication service, determining a first service type and a second service type corresponding to the target communication service, wherein the first service type is a subclass of the second service type; respectively determining a first quantity corresponding to the second data of the first service type and a second quantity corresponding to the third data of the second service type in a preset data set; determining a risk level assessment strategy based on the first quantity and the second quantity, wherein the risk level assessment strategy includes one of the following: a risk assessment strategy based on dictionary rules, a risk assessment strategy determined jointly based on a target classification model and dictionary rules, the dictionary rules are used to reflect the mapping relationship between each communication service and the first risk probability, and the target classification model is used to analyze the second risk probability corresponding to each communication service; executing the risk level assessment strategy on the first data of the target communication service to obtain a first target risk level of the target communication service.

[0007] Optionally, the preset data set is determined in the following manner: obtaining fourth data of multiple groups of historical communication services for multiple historical time periods; for each group of fourth data of historical communication services, determining the first business type and the second business type corresponding to the historical communication services; obtaining the second target risk level corresponding to the fourth data of the historical communication services; and generating a preset data set based on the fourth data of multiple groups of historical communication services and the second target risk level corresponding to the fourth data.

[0008] Optionally, determining the risk level assessment strategy based on the first quantity and the second quantity includes: determining a weight adjustment coefficient corresponding to the target communication service according to the following formula:

[0009]

[0010] In the formula, Δw represents the weight adjustment coefficient, n represents the first quantity corresponding to the second data of the first business type in the preset data set, and N represents the second quantity corresponding to the third data of the second business type in the preset data set; the risk level assessment strategy is determined based on the weight adjustment coefficient.

[0011] Optionally, the risk level assessment strategy is determined based on the weight adjustment coefficient, including: when the weight adjustment coefficient is equal to 0, determining the risk assessment strategy based on the dictionary rule as the risk level assessment strategy; when the weight adjustment coefficient is not equal to 0, determining the risk assessment strategy jointly determined by the target classification model and the dictionary rule as the risk level assessment strategy.

[0012] Optionally, a risk level assessment strategy is executed on the first data of the target communication service to determine the first target risk level of the target communication service, including: when the weight adjustment coefficient is equal to 0, analyzing the target communication service based on the dictionary rule; when the service type of the target communication service exists in the dictionary rule, querying the third risk probability corresponding to the target communication service based on the mapping relationship in the dictionary rule, and determining the third risk probability as the first target risk probability of the target communication service; when the service type of the target communication service does not exist in the dictionary rule, determining the preset risk probability as the first target risk probability of the target communication service; and determining the first target risk level of the target communication service based on the first target risk probability.

[0013] Optionally, executing a risk level assessment strategy on the first data of the target communication service to obtain a first target risk level of the target communication service includes: when the weight adjustment coefficient is not equal to 0, analyzing the target communication service in combination with the target classification model and the dictionary rule, respectively, to obtain a fourth risk probability output by the target classification model and a fifth risk probability determined based on the dictionary rule; and determining the first target risk probability of the target communication service according to the following formula:

[0014]

[0015] Where, is the first target risk probability, w is the weight factor, Δw is the weight adjustment coefficient, is the fourth risk probability output by the target classification model, The fifth risk probability is determined according to the dictionary rule; and the first target risk level of the target communication service is determined according to the first target risk probability.

[0016] Optionally, determining the first target risk level of the target communication service based on the first target risk probability includes: obtaining the first target risk probability, wherein the first target risk probability is used to reflect the probability that the target communication service is at different risk levels; comparing the probability values ​​of different risk levels in the first target risk probability, and determining the risk level corresponding to the maximum probability value as the initial risk level; when the initial risk level is multiple risk levels, selecting the highest risk level from the multiple risk levels as the first target risk level; when the initial risk level is one risk level, determining the initial risk level as the first target risk level.

[0017] Optionally, the above method also includes: when the service type of the target communication service does not exist in the dictionary rules, determining the preset risk probability as the seventh risk probability matching the target communication service, and updating the mapping relationship in the dictionary rules in combination with the seventh risk probability and the target communication service.

[0018] Optionally, the target classification model is trained in the following manner: dividing a preset data set into multiple training sets and multiple validation sets; constructing an initial classification model, wherein the initial classification model integrates a classification algorithm, and the classification algorithm includes one of the following: logistic regression algorithm, support vector machine algorithm, K-nearest neighbor algorithm and naive Bayes algorithm; inputting each training set into the initial classification model for iterative training; after each training batch, cross-validating the initial classification model in combination with the validation set, and adjusting the parameters of the initial classification model; after all training batches are completed, obtaining the target classification model.

[0019] Optionally, the above method also includes: periodically obtaining multiple groups of first data of target communication services, weight adjustment coefficients corresponding to the target communication services, and first target risk probabilities of the target communication services; for each group of target communication services, creating a data set according to the service type name, and dividing the first data of the target communication services corresponding to the weight adjustment coefficient equal to 0 into the data set; using the first data in the data set whose data volume is greater than a preset threshold to update the preset data set, using the updated preset data set to train the target classification model, and updating the target classification model.

[0020] Optionally, obtaining the first data of the target communication service includes: obtaining initial data of the target communication service; performing a preprocessing operation on the initial data to obtain the first data of the target communication service, wherein the preprocessing operation includes at least one of the following: data cleaning, and data format standardization.

[0021] According to another aspect of an embodiment of the present application, a communication service risk assessment device is also provided, including: an acquisition module, used to acquire first data of a target communication service, and determine a first service type and a second service type corresponding to the target communication service, wherein the first service type is a subclass of the second service type; a first determination module, used to respectively determine a first quantity corresponding to the second data of the first service type and a second quantity corresponding to the third data of the second service type in a preset data set; a second determination module, used to determine a risk level assessment strategy based on the first quantity and the second quantity, wherein the risk level assessment strategy includes one of the following: a risk assessment strategy based on dictionary rules, a risk assessment strategy determined jointly based on a target classification model and dictionary rules, the dictionary rules are used to reflect the mapping relationship between each communication service and the first risk probability, and the target classification model is used to analyze the second risk probability corresponding to each communication service; an assessment module, used to execute the risk level assessment strategy on the first data of the target communication service to obtain the first target risk level of the target communication service.

[0022] According to another aspect of an embodiment of the present application, a computer program product is further provided, the computer program product comprising: a computer program, wherein when the computer program is executed by a processor, the above-mentioned communication service risk assessment method is implemented.

[0023] According to another aspect of an embodiment of the present application, an electronic device is further provided, which includes: a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the above-mentioned communication service risk assessment method through the computer program.

[0024] In an embodiment of the present application, first data of a target communication service is obtained, and a first service type and a second service type corresponding to the target communication service are determined, wherein the first service type is a subclass of the second service type. For any communication service to be evaluated, accurately identifying its specific service type and subclass is the basis for subsequent risk assessment, ensuring that subsequent steps can perform refined risk analysis for specific service scenarios; a first quantity corresponding to the second data of the first service type and a second quantity corresponding to the third data of the second service type in a preset data set are respectively determined. This is particularly important for identifying potential minority services and new service types, as they are often difficult to accurately evaluate using machine learning algorithms due to insufficient sample size. A risk level assessment strategy is determined based on the first quantity and the second quantity, and a risk assessment strategy based on dictionary rules or a risk assessment strategy determined jointly by a target classification model and dictionary rules is selected. This process can more accurately assess the risk level of communication services by dynamically adjusting the weights of dictionary rules and target classification models. In particular, when dealing with newly emerging business types, it can effectively combine historical data and real-time analysis results to improve the comprehensiveness and accuracy of the assessment. Finally, the risk level assessment strategy is executed on the first data of the target communication service to obtain the first target risk level of the target communication service, thereby solving the technical problems of overfitting and inaccurate assessment in risk level assessment of traditional machine learning methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0026] Figure 1 This is a flow chart of an optional communication service risk assessment method according to an embodiment of the present application;

[0027] Figure 2 This is a flowchart of an optional and complete business risk assessment method according to an embodiment of the present application;

[0028] Figure 3 is a schematic diagram of a workflow of an optional adaptive risk assessment method according to an embodiment of the present application;

[0029] Figure 41 is a schematic structural diagram of an optional communication service risk assessment device according to an embodiment of the present application;

[0030] Figure 5 It is a schematic structural diagram of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0032] It should be noted that the terms "first", "second", etc. in the specification, claims, and drawings of the present application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.

[0033] In order to better understand the embodiments of the present application, some nouns or terms that appear in the description of the embodiments of the present application are first translated and explained as follows:

[0034] Logistic regression is a statistical method used to analyze the relationship between one or more independent variables and a binary (or categorical) dependent variable. Despite the name, logistic regression is often used for classification problems, particularly binary classification problems. It converts the output of linear regression into probability values ​​by fitting the data to a logarithmic probability function (also known as the sigmoid function), thereby determining the likelihood that a sample belongs to a particular category. Logistic regression is suitable for linearly separable binary classification problems.

[0035] The support vector machine algorithm is a supervised learning model used for classification and regression analysis. Its basic idea is to find a hyperplane in high-dimensional space that separates samples of different categories as clearly as possible. The support vector machine classifies new samples by finding the decision boundary (i.e., support vector) with the largest margin. This method is particularly effective for problems with high feature dimensions and relatively small sample sizes, achieving good generalization. Support vector machines excel at processing high-dimensional data and are relatively tolerant of noisy data.

[0036] The K-nearest neighbor algorithm is an instance-based learning method used to solve classification and regression problems. It works as follows: when a new unknown sample needs to be classified, the algorithm searches for the K training samples closest to the sample in the feature space. It then predicts the category of the new sample based on the majority class of these K samples. The K-nearest neighbor algorithm relies on the sample's "neighbors" for classification decisions, measuring similarity by calculating distances (such as Euclidean distance), making it an intuitive and easy-to-understand classification algorithm.

[0037] The Naive Bayes algorithm is a classification method based on probability theory, particularly suitable for scenarios where features are relatively independent. It calculates prior and conditional probabilities and applies Bayes' theorem to predict the posterior probability of a sample belonging to different categories, thereby determining the sample's category. The Naive Bayes algorithm assumes independence between features. Although this assumption often fails in practice, it is widely used due to its simplicity, efficiency, and surprisingly high accuracy in certain scenarios. It is effective in classification tasks where features are relatively independent.

[0038] BBS (Bullentin Board System): It is an early form of publishing and discussing information on the Internet, where users can post, read and reply to others' posts, similar to today's forums or message board systems.

[0039] The information collected in the embodiments of the present application is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant countries and regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0040] Example 1

[0041] According to an embodiment of the present application, a communication service risk assessment method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0042] Figure 1 FIG. 1 is a flow chart of a communication service risk assessment method provided in accordance with an embodiment of the present application. Figure 1 As shown, the method includes the following steps:

[0043] Step S102, obtaining first data of a target communication service, and determining a first service type and a second service type corresponding to the target communication service, wherein the first service type is a subclass of the second service type;

[0044] Step S104, determining a first quantity corresponding to the second data of the first business type and a second quantity corresponding to the third data of the second business type in the preset data set;

[0045] Step S106: determining a risk level assessment strategy based on the first quantity and the second quantity, wherein the risk level assessment strategy includes one of the following: a risk assessment strategy based on a dictionary rule, and a risk assessment strategy determined jointly by a target classification model and a dictionary rule, wherein the dictionary rule is used to reflect a mapping relationship between each communication service and the first risk probability, and the target classification model is used to analyze the second risk probability corresponding to each communication service;

[0046] Step S108: executing a risk level assessment strategy on the first data of the target communication service to obtain a first target risk level of the target communication service.

[0047] Figure 2 A more complete communication service risk assessment method is shown in the following flow chart. Figure 2 Explain the steps of the communication business risk assessment method.

[0048] The preset data set serves as the empirical reference data for risk level assessment, so it is of great significance to obtain high-quality preset data sets.

[0049] As an optional implementation, the preset data set is determined in the following manner: obtaining fourth data of multiple groups of historical communication services for multiple historical time periods; for each group of fourth data of historical communication services, determining the first business type and the second business type corresponding to the historical communication services; obtaining the second target risk level corresponding to the fourth data of the historical communication services; and generating a preset data set based on the fourth data of multiple groups of historical communication services and the second target risk level corresponding to the fourth data.

[0050] First, we extracted a large amount of historical service order data covering a wide range of services from telecommunications company databases to ensure the diversity and integrity of the dataset. This data includes not only secondary service types such as new installations, speed upgrades, silent single-service suspension, subscriptions, and sales product delivery, but also primary service types for converged products, including new mobile service installations, new fixed-line phone installations, mobile service speed upgrades, broadband speed upgrades, silent single-service suspension for mobile services, smart home sales product delivery, and dedicated broadband subscriptions. These primary service types are subtypes of the secondary service types.

[0051] Since the same business data may come from different channels, for example, when business data comes from offline channels or electronic channels, the data may have inconsistent formats, and even the integrity of the data may be affected. Considering the above problems with the collected data, it is necessary to pre-process the collected raw data. Figure 3 This paper illustrates the workflow of an adaptive risk assessment method. Data from various sources is first preprocessed and formatted uniformly. The preprocessed data is then input into an adaptive risk assessment system. Business data and corresponding risk levels are then output to an audit system or BBS for archiving and display. Business data sources include, but are not limited to, BBSs, audit systems, blacklist systems, and employee ID systems.

[0052] For the fourth data of each set of historical communication services, it is necessary to carefully identify the corresponding first business type (such as new installation of mobile services, new installation of fixed-line telephones, mobile business speed-up, broadband speed-up, silent single suspension of mobile services, smart home sales products delivered to the back-end, dedicated broadband subscription, etc.) and the second business type (new installation, speed-up, silent single suspension, subscription, sales products delivered to the back-end, etc.), and combine various risk points in the business acceptance process, such as positioning authentication, secondary authentication, customer blacklist records, etc., as well as feedback information from business backward audits, to extract the true risk level of each business. These risk levels are the second target risk levels, which are divided into four levels: no risk, low risk, medium risk and high risk.

[0053] By systematically matching and analyzing the fourth data of historical communication services with the corresponding second target risk level, we use the Naive Bayes algorithm in machine learning as the main training method, but it does not rule out the use of other classification algorithms such as support vector machines or neural networks as substitutes in certain specific scenarios, such as when the correlation between factors is strong.

[0054] Each training session aims to learn underlying business risk patterns from historical data and optimize the model's ability to identify and predict risk. After multiple iterations of training and validation, a highly accurate business risk dataset was formed as the default dataset. This dataset not only encompasses a rich set of business types and risk points, but also accurately maps the true risk level of each risk scenario, providing a solid foundation for subsequent risk assessments.

[0055] After determining the preset data set, a risk level assessment is performed on the acquired data of the target communication business. Similarly, since the data of the target communication business may come from different channels, for example, when the business data comes from offline channels or electronic channels, it is also necessary to pre-process the data.

[0056] As an optional implementation, obtaining the first data of the target communication service includes: obtaining the initial data of the target communication service; performing a preprocessing operation on the initial data to obtain the first data of the target communication service, wherein the preprocessing operation includes at least one of the following: data cleaning, data format standardization.

[0057] Initial data may contain errors, missing data, or records in inconsistent formats. Data cleaning can remove or correct these errors and missing data, improving data accuracy. Data format standardization ensures that all data is presented in a unified format for easier model processing. For example, all timestamps can be converted to a unified time format, and all numerical data can be converted to the same numerical range. This preprocessing operation not only improves data quality but also enhances the model's processing capabilities, ensuring the reliability of the assessment results. Implementing data preprocessing is an indispensable step in the assessment process. It provides a clean, unified data foundation for subsequent risk assessments and is a key factor in improving assessment accuracy and efficiency.

[0058] As an optional implementation manner, determining the risk level assessment strategy based on the first quantity and the second quantity includes: determining a weight adjustment coefficient corresponding to the target communication service according to the following formula:

[0059]

[0060] In the formula, Δw represents the weight adjustment coefficient, n represents the first quantity corresponding to the second data of the first business type in the preset data set, and N represents the second quantity corresponding to the third data of the second business type in the preset data set; the risk level assessment strategy is determined based on the weight adjustment coefficient.

[0061] The weight adjustment coefficient is based on the ratio of the amount of business type data n for the fused product type to the amount of data N for a broader range of business types, reflecting the representativeness of a specific business type. When n / N is close to 0, indicating that the amount of data for a specific business type is extremely small, the evaluation strategy is more inclined to use dictionary rules, as dictionary rules are based on a wide range of business type data and can provide more stable evaluation results. Conversely, when n / N is large, the evaluation strategy relies more on the target classification model, as the model can learn more specific risk characteristics from a large amount of data on a specific business type. This strategy selection mechanism ensures the accuracy and reliability of the evaluation results by dynamically adjusting the weights of the dictionary rules and the target classification model. This is particularly important when dealing with new business types with smaller data volumes, as it can avoid model overfitting and ensure evaluation stability.

[0062] As an optional implementation method, the risk level assessment strategy is determined based on the weight adjustment coefficient, including: when the weight adjustment coefficient is equal to 0, determining the risk assessment strategy based on the dictionary rule as the risk level assessment strategy; when the weight adjustment coefficient is not equal to 0, determining the risk assessment strategy jointly determined by the target classification model and the dictionary rule as the risk level assessment strategy.

[0063] When the weight adjustment coefficient Δw equals 0, it means that there is no historical data for a specific business type in the preset dataset. In this case, risk assessment relies entirely on dictionary rules. Dictionary rules, based on a wide range of business type data, provide a basic risk assessment framework and ensure the feasibility of evaluating new business types. When Δw is not equal to 0, it indicates that historical data for a specific business type exists. In this case, an assessment strategy that combines a target classification model with dictionary rules can integrate the static risk characteristics of historical data and the dynamic risk characteristics of real-time data, improving the accuracy and comprehensiveness of the assessment. This strategy choice under these conditions solves the difficult problem of risk assessment for new business types by flexibly adjusting the assessment method, while maintaining the accuracy of the assessment of mature business types.

[0064] The dictionary rules are formed by arranging and combining all scenarios of different risk points and giving corresponding risk levels. Table 1 shows some examples of dictionary rules.

[0065] Table 1

[0066]

[0067] As an optional implementation method, a risk level assessment strategy is executed on the first data of the target communication service to determine the first target risk level of the target communication service, including: when the weight adjustment coefficient is equal to 0, analyzing the target communication service based on the dictionary rule; when the service type of the target communication service exists in the dictionary rule, querying the third risk probability corresponding to the target communication service based on the mapping relationship in the dictionary rule, and determining the third risk probability as the first target risk probability of the target communication service; when the service type of the target communication service does not exist in the dictionary rule, determining the preset risk probability as the first target risk probability of the target communication service; and determining the first target risk level of the target communication service based on the first target risk probability.

[0068] As can be seen from Table 1, the output of the dictionary rule Represents the risk probability vector obtained by querying the dictionary, and the result set is {[1, 0, 0, 0] T ,[0,1,0,0] T ,[0,0,1,0] T ,[0,0,0,1] T}, corresponding to no risk, low risk, medium risk and high risk respectively, that is, the risk result given by the dictionary is a definite risk probability vector.

[0069] Determining the risk level under the dictionary rules is the safeguard mechanism for this application when processing new service types. When the weight adjustment coefficient Δw is equal to 0, it means that the amount of data for a specific service type is insufficient. At this time, the dictionary rules become the primary evaluation tool. The dictionary rules contain a wide range of mappings between service types and their risk probabilities. If the service type of the target communication service exists in the dictionary rules, the evaluation strategy will directly query and apply the corresponding third risk probability as the first target risk probability, quickly providing a risk assessment result.

[0070] Specifically, assuming that the target communication service is the smart home product sales delivery backend, Table 2 is the pre-processed data of the smart home product sales delivery backend.

[0071] Table 2

[0072]

[0073] When the first service type of the communication service is smart home sales product delivery to the back-end, the data volume of this service type in the preset data set is 0, and the dictionary rules can be directly consulted. When the dictionary rules contain the risk probability corresponding to the service type of smart home sales product delivery to the back-end, the risk level corresponding to the queried risk probability can be directly used as the target risk level.

[0074] Conversely, if the business type doesn't exist in the dictionary rules, a preset risk probability is used as the first target risk probability. This preset risk probability can typically be set as needed based on empirical evaluation, avoiding biased risk assessments for new business types. In this way, dictionary rules not only enable rapid risk assessments for known business types, but also provide preliminary risk assessments for new business types, ensuring comprehensiveness and initial accuracy.

[0075] As an optional implementation, executing a risk level assessment strategy on the first data of the target communication service to obtain a first target risk level of the target communication service includes: when the weight adjustment coefficient is not equal to 0, analyzing the target communication service in combination with the target classification model and the dictionary rule, respectively, to obtain a fourth risk probability output by the target classification model and a fifth risk probability determined based on the dictionary rule; and determining the first target risk probability of the target communication service according to the following formula:

[0076]

[0077] Where, is the first target risk probability, w is the weight factor, Δw is the weight adjustment coefficient, is the fourth risk probability output by the target classification model, The fifth risk probability is determined according to the dictionary rule; and the first target risk level of the target communication service is determined according to the first target risk probability.

[0078] Assume that in the mobile new installation business scenario, the acceptance records from the system are cleaned, junk data is eliminated, and standardized according to a unified format to obtain the relevant data of the formatted mobile new installation business order. The pre-processed data of the mobile new installation business order is shown in Table 3.

[0079] Table 3

[0080]

[0081] This scenario is a relatively common new installation case, so there is a large sample size in the preset data set. The sample size of new installation business is 100,203, of which the sample size of mobile new installation business is 38,102, so its weight adjustment coefficient is:

[0082]

[0083] Input the preprocessed data into the target classification model to obtain Input the preprocessed data into the dictionary rules for query and obtain Assuming that the weight factor w is 0.5, since Δw≠0, the first target risk probability in this mobile new installation service scenario is:

[0084]

[0085] Assuming a mobile silent single-stop service scenario, the preprocessed data is shown in Table 4.

[0086] Table 4

[0087]

[0088] The total number of samples of the silent single stop service type in the preset data set is 80124, while the number of sample data of the mobile silent single stop service in the preset data set is 10015. The calculation process is the same as the above process. First, the weight adjustment coefficient is calculated as follows:

[0089]

[0090] The preprocessed data is input into the target classification model, and the risk probability determined based on the target classification model is:

[0091]

[0092] Next, the formatted data is input into the dictionary rules for query. The risk probability determined based on the dictionary rules is:

[0093]

[0094] Since Δw≠0, when the weight factor w is 0.5, the final target risk probability in the moving silent single stop scenario is calculated based on the weight adjustment coefficient, the risk probability output by the target classification model, and the risk probability determined based on the dictionary rule:

[0095]

[0096] Specifically, the weight factor w can be set as needed. When the data volume corresponding to the target communication service type accounts for a large proportion of the preset dataset, the weight adjustment coefficient will also be larger, and risk assessment will be guided by machine learning results. Considering that the small sample size in some scenarios may lead to bias in the machine learning training set, the basic approach in this case is to reduce w, that is, to reduce the weight of machine learning results and increase the weight of the query dictionary. Risk assessment is mainly guided by the query dictionary results. This calculation method fully utilizes the advantages of machine learning and avoids the overfitting problem under small sample conditions.

[0097] When Δw≠0, the combination of target classification model and dictionary rules is the core of this application to improve the accuracy of evaluation. When the weight adjustment coefficient Δw is not equal to 0, it indicates that there is historical data of a specific business type in the preset data set. At this time, the target classification model can learn the risk characteristics of the specific business type from these data and output the fourth risk probability At the same time, dictionary rules provide the fifth risk probability based on a wide range of business type data. Through the above solution Based on the formula, this application can dynamically adjust the weights of the model output and the dictionary rule output to ensure that the evaluation results take into account both the real-time risk characteristics of specific business types and the historical risk patterns of a wide range of business types.

[0098] As an optional implementation, determining the first target risk level of the target communication service based on the first target risk probability includes: obtaining the first target risk probability, wherein the first target risk probability is used to reflect the probability that the target communication service is at different risk levels; comparing the probability values ​​of different risk levels in the first target risk probability, and determining the risk level corresponding to the maximum probability value as the initial risk level; when the initial risk level is multiple risk levels, selecting the highest risk level from the multiple risk levels as the first target risk level; when the initial risk level is one risk level, determining the initial risk level as the first target risk level.

[0099] It can be seen that the target risk level is a risk probability vector, which can be expressed as:

[0100] [p none , p low , p medium , p high ] T

[0101] Then, p none represents the probability that the target communication service is at a risk-free level, p low Indicates the probability that the target communication service is at a low risk level, p medium Indicates the probability that the target communication service is at a medium risk level, p high Indicates the probability that the target communication service is at a high risk level.

[0102] Based on the above example, we can see that for the mobile new installation business scenario, the target risk probability is [0.9912, 0.0088, 0, 0] T, the risk level corresponding to 0.9912 can be used as the initial risk level. Furthermore, 0.9912 corresponds to the level of "no risk". Therefore, the target risk level of the new mobile service is no risk. For the mobile silent single stop service scenario, the target risk probability is [0.5, 0.5, 0, 0] T , the risk levels corresponding to a probability value of 0.5 are "no risk" and "low risk" respectively. Therefore, the initial risk levels are set as "no risk" and "low risk". Furthermore, the "low risk" level is higher than the risk level of "no risk". Therefore, the mobile silent single stop service scenario is directly evaluated as the "low risk" level.

[0103] Determining the initial risk level is a key step in the assessment process. The first target risk probability reflects the possibility that the target communication service is at different risk levels. By comparing these probability values, the initial risk level corresponding to the maximum probability can be determined. This probability-based assessment method can provide a more detailed risk level division, such as no risk, low risk, medium risk, high risk, etc. When determining the initial risk level, if there are multiple risk levels with the same or similar probability values, this application will automatically select the highest risk level as the first target risk level. This processing method ensures that in cases of high uncertainty, a more conservative risk management strategy is adopted to avoid underestimation of potential risks. This probability-based hierarchical assessment not only improves the refinement of risk assessment, but also ensures that more cautious risk management measures can be taken when facing new business types or businesses with high uncertainty, providing a guarantee for the safe and stable operation of the communication network.

[0104] Optionally, the above method also includes: when the service type of the target communication service does not exist in the dictionary rules, determining the preset risk probability as the seventh risk probability matching the target communication service, and updating the mapping relationship in the dictionary rules in combination with the seventh risk probability and the target communication service.

[0105] The dynamic update of dictionary rules is the key to the adaptation of this application to new business types. When the business type of the target communication business does not exist in the dictionary rules, this application will combine the preset risk probability (seventh risk probability) with the characteristics of the target communication business, and update the mapping relationship in the dictionary rules, wherein the preset risk probability can be set artificially based on experience and an assessment of the business scenario. This update mechanism enables the dictionary rules to be continuously enriched with the emergence of new business types, thereby improving the adaptability and comprehensiveness of the rules. When the new business type is queried next time, the risk probability corresponding to the new business type already exists in the dictionary rules, and the risk level can be determined directly by querying the dictionary rules to give the risk probability.

[0106] Optionally, the target classification model is trained in the following manner: dividing a preset data set into multiple training sets and multiple validation sets; constructing an initial classification model, wherein the initial classification model integrates a classification algorithm, and the classification algorithm includes one of the following: logistic regression algorithm, support vector machine algorithm, K-nearest neighbor algorithm and naive Bayes algorithm; inputting each training set into the initial classification model for iterative training; after each training batch, cross-validating the initial classification model in combination with the validation set, and adjusting the parameters of the initial classification model; after all training batches are completed, obtaining the target classification model.

[0107] The training of the target classification model is the technical basis for the technical solution of this application to improve the accuracy of risk assessment. The preset data set is first divided into multiple training sets and validation sets. The training set is used for model learning, and the validation set is used for model verification. The initial classification model can adopt a variety of classification algorithms, such as logistic regression algorithm, support vector machine algorithm, K nearest neighbor algorithm and naive Bayes algorithm, which can learn the characteristic patterns of business risks from the data. Through iterative training, the model can continuously optimize its parameters and improve its ability to identify risk characteristics. After each training batch, cross-validation is performed using the validation set to evaluate the generalization ability of the model, avoid overfitting, and ensure the prediction accuracy of the model on new data. This training and verification process based on a preset data set not only improves the accuracy of the target classification model, but also ensures the stability and generalization ability of the model, providing strong technical support for the real-time assessment of communication business risks.

[0108] Optionally, the above method also includes: periodically obtaining multiple groups of first data of target communication services, weight adjustment coefficients corresponding to the target communication services, and first target risk probabilities of the target communication services; for each group of target communication services, creating a data set according to the service type name, and dividing the first data of the target communication services corresponding to the weight adjustment coefficient equal to 0 into the data set; using the first data in the data set whose data volume is greater than a preset threshold to update the preset data set, using the updated preset data set to train the target classification model, and updating the target classification model.

[0109] Assuming that the sample data volume of the "Smart Home Sales Product Delivery Backend" business type in the preset data set is 0, that is, the corresponding weight adjustment coefficient is 0, the "Smart Home Sales Product Delivery Backend" business type can be treated as a new business type, and a data set named "Smart Home Sales Product Delivery Backend" business type can be established. The data for the current target communication service, "Smart Home Sales Product Delivery Backend," can be stored in this established data set. If the data volume in the data set exceeds a preset threshold, the data in this data set can be used as new sample data to train and update the target classification model. Specifically, assuming the preset threshold is 10,000 data items, and the data volume for the current target communication service, "Smart Home Sales Product Delivery Backend," is 212, this indicates that the threshold has not been reached, and training of the target classification model will not be triggered.

[0110] The periodic update of the target classification model is the key to maintaining the timeliness and adaptability of the evaluation model of this application. By periodically collecting the first data, weight adjustment coefficient and first target risk probability of the target communication business, a data set for a specific business type can be created. These sets contain the latest business data and risk assessment results. For business types whose data volume exceeds the preset threshold, this application will use these data to update the preset data set, thereby providing the latest data support for model training. The updated target classification model can not only learn the latest risk characteristics, but also adapt to changes in the business environment, such as the emergence of new business types, changes in user behavior patterns, etc., ensuring the timeliness and adaptability of the evaluation model. This periodic update mechanism not only improves the accuracy of risk assessment, but also ensures that the model can be continuously optimized and adapt to the ever-changing communication business environment, providing communication operators with a dynamic risk management tool.

[0111] Through the above steps, first data of the target communications service is obtained, and a first service type and a second service type corresponding to the target communications service are determined, wherein the first service type is a subclass of the second service type. For any communications service to be evaluated, accurately identifying its specific service type and subtype is the basis for subsequent risk assessment, ensuring that subsequent steps can conduct refined risk analysis for specific business scenarios. A first quantity corresponding to the second data of the first service type and a second quantity corresponding to the third data of the second service type in the preset data set are respectively determined. This is particularly important for identifying potential minority services and new service types, as these are often difficult to accurately evaluate using machine learning algorithms due to insufficient sample size. A risk level assessment strategy is determined based on the first and second quantities, selecting a risk assessment strategy based on dictionary rules or a risk assessment strategy determined jointly by a target classification model and dictionary rules. This process can more accurately assess the risk level of the communications service by dynamically adjusting the weights of the dictionary rules and the target classification model. In particular, when dealing with emerging service types, it can effectively combine historical data and real-time analysis results to improve the comprehensiveness and accuracy of the assessment. Finally, the risk level assessment strategy is executed on the first data of the target communications service to obtain a first target risk level for the target communications service, thereby solving the technical problems of overfitting and inaccurate assessment in risk level assessments of traditional machine learning methods.

[0112] Example 2

[0113] According to an embodiment of the present application, a communication service risk assessment device for implementing the communication service risk assessment method in embodiment 1 is also provided. Figure 4 As shown, the communication service risk assessment device at least includes: an acquisition module 41, a first determination module 42, a second determination module 43 and an assessment module 44, wherein:

[0114] The acquisition module 41 may acquire first data of a target communication service and determine a first service type and a second service type corresponding to the target communication service, wherein the first service type is a subclass of the second service type;

[0115] The first determining module 42 may determine a first quantity corresponding to the second data of the first business type and a second quantity corresponding to the third data of the second business type in the preset data set;

[0116] A second determination module 43 may be configured to determine a risk level assessment strategy based on the first quantity and the second quantity, wherein the risk level assessment strategy includes one of the following: a risk assessment strategy based on a dictionary rule, and a risk assessment strategy determined jointly by a target classification model and a dictionary rule, wherein the dictionary rule is configured to reflect a mapping relationship between each communication service and the first risk probability, and the target classification model is configured to analyze the second risk probability corresponding to each communication service;

[0117] The evaluation module 44 may be configured to execute a risk level evaluation strategy on the first data of the target communication service to obtain a first target risk level of the target communication service.

[0118] The functions of each module of the communication service risk assessment device are described below in conjunction with a specific implementation process.

[0119] The preset data set serves as the empirical reference data for risk level assessment, so it is of great significance to obtain high-quality preset data sets.

[0120] As an optional implementation, the preset data set is determined in the following manner: obtaining fourth data of multiple groups of historical communication services for multiple historical time periods; for each group of fourth data of historical communication services, determining the first business type and the second business type corresponding to the historical communication services; obtaining the second target risk level corresponding to the fourth data of the historical communication services; and generating a preset data set based on the fourth data of multiple groups of historical communication services and the second target risk level corresponding to the fourth data.

[0121] As an optional implementation, the acquisition module obtains the first data of the target communication service, which can be achieved in the following ways: obtaining the initial data of the target communication service; performing a preprocessing operation on the initial data to obtain the first data of the target communication service, wherein the preprocessing operation includes at least one of the following: data cleaning, data format standardization.

[0122] Initial data may contain errors, missing data, or records in inconsistent formats. Data cleaning can remove or correct these errors and missing data, improving data accuracy. Data format standardization ensures that all data is presented in a unified format for easier model processing. For example, all timestamps can be converted to a unified time format, and all numerical data can be converted to the same numerical range. This preprocessing operation not only improves data quality but also enhances the model's processing capabilities, ensuring the reliability of the assessment results. Implementing data preprocessing is an indispensable step in the assessment process. It provides a clean, unified data foundation for subsequent risk assessments and is a key factor in improving assessment accuracy and efficiency.

[0123] As an optional implementation, the second determination module determines the risk level assessment strategy based on the first quantity and the second quantity, which can be specifically implemented in the following manner: determining the weight adjustment coefficient corresponding to the target communication service according to the following formula:

[0124]

[0125] In the formula, Δw represents the weight adjustment coefficient, n represents the first quantity corresponding to the second data of the first business type in the preset data set, and N represents the second quantity corresponding to the third data of the second business type in the preset data set; the risk level assessment strategy is determined based on the weight adjustment coefficient.

[0126] The weight adjustment coefficient is based on the ratio of the amount of business type data n for the fused product type to the amount of data N for a broader range of business types, reflecting the representativeness of a specific business type. When n / N is close to 0, indicating that the amount of data for a specific business type is extremely small, the evaluation strategy is more inclined to use dictionary rules, as dictionary rules are based on a wide range of business type data and can provide more stable evaluation results. Conversely, when n / N is large, the evaluation strategy relies more on the target classification model, as the model can learn more specific risk characteristics from a large amount of data on a specific business type. This strategy selection mechanism ensures the accuracy and reliability of the evaluation results by dynamically adjusting the weights of the dictionary rules and the target classification model. This is particularly important when dealing with new business types with smaller data volumes, as it can avoid model overfitting and ensure evaluation stability.

[0127] As an optional implementation, the second determination module determines the risk level assessment strategy based on the weight adjustment coefficient, which can be specifically implemented in the following way: when the weight adjustment coefficient is equal to 0, the risk assessment strategy based on the dictionary rule is determined to be the risk level assessment strategy; when the weight adjustment coefficient is not equal to 0, the risk assessment strategy jointly determined by the target classification model and the dictionary rule is determined to be the risk level assessment strategy.

[0128] When the weight adjustment coefficient Δw equals 0, it means that there is no historical data for a specific business type in the preset dataset. In this case, risk assessment relies entirely on dictionary rules. Dictionary rules, based on a wide range of business type data, provide a basic risk assessment framework and ensure the feasibility of evaluating new business types. When Δw is not equal to 0, it indicates that historical data for a specific business type exists. In this case, an assessment strategy that combines a target classification model with dictionary rules can integrate the static risk characteristics of historical data and the dynamic risk characteristics of real-time data, improving the accuracy and comprehensiveness of the assessment. This strategy choice under these conditions solves the difficult problem of risk assessment for new business types by flexibly adjusting the assessment method, while maintaining the accuracy of the assessment of mature business types.

[0129] As an optional implementation, the evaluation module executes a risk level evaluation strategy on the first data of the target communication service to determine the first target risk level of the target communication service. This can be specifically achieved in the following manner: when the weight adjustment coefficient is equal to 0, the target communication service is analyzed based on the dictionary rules; when the service type of the target communication service exists in the dictionary rules, based on the mapping relationship in the dictionary rules, the third risk probability corresponding to the target communication service is queried, and the third risk probability is determined as the first target risk probability of the target communication service; when the service type of the target communication service does not exist in the dictionary rules, the preset risk probability is determined as the first target risk probability of the target communication service; and the first target risk level of the target communication service is determined based on the first target risk probability.

[0130] As an optional implementation, the evaluation module executes a risk level evaluation strategy on the first data of the target communication service to obtain a first target risk level of the target communication service. This can be specifically achieved by: when the weight adjustment coefficient is not equal to 0, analyzing the target communication service in combination with the target classification model and the dictionary rule, respectively, to obtain a fourth risk probability output by the target classification model and a fifth risk probability determined based on the dictionary rule; and determining the first target risk probability of the target communication service according to the following formula:

[0131]

[0132] Where, is the first target risk probability, w is the weight factor, Δw is the weight adjustment coefficient, is the fourth risk probability output by the target classification model, The fifth risk probability is determined according to the dictionary rule; and the first target risk level of the target communication service is determined according to the first target risk probability.

[0133] Among them, the combination of target classification model and dictionary rules is the core of this application to improve the accuracy of assessment. When the weight adjustment coefficient Δw is not equal to 0, it indicates that there is historical data of a specific business type in the preset data set. At this time, the target classification model can learn the risk characteristics of the specific business type from these data and output the fourth risk probability At the same time, dictionary rules provide the fifth risk probability based on a wide range of business type data. Through the above solution The formula of this application can dynamically adjust the weights of the model output and the dictionary rule output to ensure that the evaluation results take into account both the real-time risk characteristics of specific business types and the historical risk patterns of a wide range of business types.

[0134] As an optional implementation, the evaluation module determines the first target risk level of the target communication service based on the first target risk probability, including: obtaining the first target risk probability, wherein the first target risk probability is used to reflect the probability that the target communication service is at different risk levels; comparing the probability values ​​of different risk levels in the first target risk probability, and determining the risk level corresponding to the maximum probability value as the initial risk level; when the initial risk level is multiple risk levels, selecting the highest risk level from the multiple risk levels as the first target risk level; when the initial risk level is one risk level, determining the initial risk level as the first target risk level.

[0135] It can be seen that the target risk level is a risk probability vector, which can be expressed as:

[0136] [p none , p low , p medium , p high ] T

[0137] Then, p none represents the probability that the target communication service is at a risk-free level, p low Indicates the probability that the target communication service is at a low risk level, p medium Indicates the probability that the target communication service is at a medium risk level, p high Indicates the probability that the target communication service is at a high risk level.

[0138] As an optional implementation, the communication service risk assessment device further includes a training module and an updating module.

[0139] As an optional implementation, when the service type of the target communication service does not exist in the dictionary rules, the update module can determine the preset risk probability as the seventh risk probability matching the target communication service, and update the mapping relationship in the dictionary rules based on the seventh risk probability and the target communication service.

[0140] The dynamic update of dictionary rules is the key to the adaptation of this application to new business types. When the business type of the target communication business does not exist in the dictionary rules, this application will combine the preset risk probability (seventh risk probability) with the characteristics of the target communication business, and update the mapping relationship in the dictionary rules, wherein the preset risk probability can be set artificially based on experience and an assessment of the business scenario. This update mechanism enables the dictionary rules to be continuously enriched with the emergence of new business types, thereby improving the adaptability and comprehensiveness of the rules. When the new business type is queried next time, the risk probability corresponding to the new business type already exists in the dictionary rules, and the risk level can be determined directly by querying the dictionary rules to give the risk probability.

[0141] As an optional implementation, the target classification model is obtained by training in combination with the training module in the following manner: dividing the preset data set into multiple training sets and multiple validation sets; constructing an initial classification model, wherein the initial classification model integrates a classification algorithm, and the classification algorithm includes one of the following: logistic regression algorithm, support vector machine algorithm, K nearest neighbor algorithm and naive Bayes algorithm; inputting each training set into the initial classification model for iterative training; after each training batch, cross-validating the initial classification model in combination with the validation set, and adjusting the parameters of the initial classification model; after all training batches are completed, the target classification model is obtained.

[0142] The training of the target classification model is the technical basis for the technical solution of this application to improve the accuracy of risk assessment. The preset data set is first divided into multiple training sets and validation sets. The training set is used for model learning, and the validation set is used for model verification. The initial classification model can adopt a variety of classification algorithms, such as logistic regression algorithm, support vector machine algorithm, K nearest neighbor algorithm and naive Bayes algorithm, which can learn the characteristic patterns of business risks from the data. Through iterative training, the model can continuously optimize its parameters and improve its ability to identify risk characteristics. After each training batch, cross-validation is performed using the validation set to evaluate the generalization ability of the model, avoid overfitting, and ensure the prediction accuracy of the model on new data. This training and verification process based on a preset data set not only improves the accuracy of the target classification model, but also ensures the stability and generalization ability of the model, providing strong technical support for the real-time assessment of communication business risks.

[0143] As an optional implementation, the training module can also periodically obtain multiple groups of first data of target communication services, weight adjustment coefficients corresponding to the target communication services, and first target risk probabilities of the target communication services; for each group of target communication services, create a data set according to the service type name, and divide the first data of the target communication services corresponding to the weight adjustment coefficient equal to 0 into the data set; use the first data in the data set whose data volume is greater than a preset threshold to update the preset data set, use the updated preset data set to train the target classification model, and update the target classification model.

[0144] It should be noted that each module in the communication business risk assessment device in the embodiment of the present application corresponds one-to-one to each implementation step of the communication business risk assessment method in Example 1. Since a detailed description has been given in Example 1, some details not reflected in this embodiment can be referred to Example 1 and will not be elaborated here.

[0145] Example 3

[0146] According to an embodiment of the present application, a computer program product is also provided, which includes a computer program, wherein when the computer program is executed by a processor, the communication service risk assessment method in Example 1 is implemented.

[0147] According to an embodiment of the present application, a non-volatile storage medium is also provided, which includes a stored computer program, wherein the device where the non-volatile storage medium is located executes the communication service risk assessment method in Example 1 by running the computer program.

[0148] According to an embodiment of the present application, a processor is also provided, which is used to run a computer program, wherein the communication service risk assessment method in Example 1 is executed when the computer program is running.

[0149] According to an embodiment of the present application, an electronic device is also provided, which includes: a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the communication service risk assessment method in Example 1 through the computer program.

[0150] Specifically, when the computer program is running, the following steps are executed: obtaining the first data of the target communication service, determining the first service type and the second service type corresponding to the target communication service, wherein the first service type is a subclass of the second service type; respectively determining the first quantity corresponding to the second data of the first service type and the second quantity corresponding to the third data of the second service type in the preset data set; determining a risk level assessment strategy based on the first quantity and the second quantity, wherein the risk level assessment strategy includes one of the following: a risk assessment strategy based on dictionary rules, a risk assessment strategy determined jointly based on a target classification model and dictionary rules, the dictionary rules are used to reflect the mapping relationship between each communication service and the first risk probability, and the target classification model is used to analyze the second risk probability corresponding to each communication service; executing the risk level assessment strategy on the first data of the target communication service to obtain the first target risk level of the target communication service.

[0151] As an optional implementation, the electronic device may be in the form of a mobile terminal, a computer terminal or a similar computing device. Figure 5 The hardware structure block diagram of an electronic device for implementing a communication service risk assessment method is shown. Figure 5As shown, the electronic device 50 may include one or more (502a, 502b, ..., 502n are shown in the figure) processors 502 (the processor 502 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 504 for storing data, and a transmission device 506 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It will be understood by those skilled in the art that Figure 5 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 5 More or fewer components than shown, or with Figure 5 Different configurations shown.

[0152] It should be noted that the one or more processors 502 and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry". The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuitry may be a single independent processing module, or may be incorporated in whole or in part into any of the other components of the electronic device 50. As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0153] The memory 504 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the communication service risk assessment method in the embodiments of the present application. The processor 502 executes the software programs and modules stored in the memory 504 to perform various functional applications and data processing, thereby implementing the vulnerability detection method for the application described above. The memory 504 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 504 may further include memory remotely located relative to the processor 502, and these remote memories may be connected to the electronic device 50 via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0154] The transmission device 506 is used to receive or send data via a network. Specific examples of the aforementioned network may include a wireless network provided by the communications provider of the electronic device 50. In one embodiment, the transmission device 506 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, the transmission device 506 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0155] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the electronic device 50 .

[0156] The serial numbers of the above embodiments are for description only and do not represent the advantages or disadvantages of the embodiments.

[0157] In the above embodiments of the present application, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0158] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of units can be a logical function division. In actual implementation, there may be other division methods, such as 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 mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0159] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple units. Some or all of the units may be selected to achieve the purpose of the present embodiment according to actual needs.

[0160] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0161] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program code.

[0162] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A communication service risk assessment method, characterized in that: include: Acquire first data of a target communication service, and determine a first service type and a second service type corresponding to the target communication service, wherein the first service type is a subclass of the second service type; respectively determining a first quantity corresponding to the second data of the first business type and a second quantity corresponding to the third data of the second business type in a preset data set; Determining a risk level assessment strategy based on the first number and the second number, wherein the risk level assessment strategy includes one of the following: a risk assessment strategy based on a dictionary rule, and a risk assessment strategy determined jointly based on a target classification model and a dictionary rule, the dictionary rule being used to reflect a mapping relationship between each communication service and a first risk probability, and the target classification model being used to analyze a second risk probability corresponding to each communication service; The risk level assessment strategy is executed on the first data of the target communication service to obtain a first target risk level of the target communication service.

2. The method according to claim 1, characterized in that The preset data set is determined in the following manner: Acquiring fourth data of multiple groups of historical communication services in multiple historical time periods; For each set of fourth data of historical communication services, determining a first service type and a second service type corresponding to the historical communication services; Obtaining a second target risk level corresponding to fourth data of the historical communication service; The preset data set is generated based on multiple groups of fourth data of the historical communication services and second target risk levels corresponding to the fourth data.

3. The method according to claim 1, characterized in that Determining a risk level assessment strategy based on the first number and the second number includes: The weight adjustment coefficient corresponding to the target communication service is determined according to the following formula: Wherein, Δw represents the weight adjustment coefficient, n represents the first number corresponding to the second data of the first service type in the preset data set, and N represents the second number corresponding to the third data of the second service type in the preset data set; The risk level assessment strategy is determined based on the weight adjustment coefficient.

4. The method according to claim 3, characterized in that Determining the risk level assessment strategy based on the weight adjustment coefficient includes: When the weight adjustment coefficient is equal to 0, determining the risk assessment strategy based on dictionary rules as the risk level assessment strategy; When the weight adjustment coefficient is not equal to 0, the risk assessment strategy jointly determined by the target classification model and the dictionary rule is determined to be the risk level assessment strategy.

5. The method according to claim 4, characterized in that Executing the risk level assessment strategy on the first data of the target communication service to determine a first target risk level of the target communication service includes: When the weight adjustment coefficient is equal to 0, analyzing the target communication service based on the dictionary rule; When the service type of the target communication service exists in the dictionary rule, querying a third risk probability corresponding to the target communication service based on a mapping relationship in the dictionary rule, and determining the third risk probability as the first target risk probability of the target communication service; When the service type of the target communication service does not exist in the dictionary rule, determining a preset risk probability as a first target risk probability of the target communication service; A first target risk level of the target communication service is determined according to the first target risk probability.

6. The method according to claim 4, characterized in that Executing the risk level assessment strategy on the first data of the target communication service to obtain a first target risk level of the target communication service includes: When the weight adjustment coefficient is not equal to 0, analyzing the target communication service in combination with the target classification model and the dictionary rule, respectively, to obtain a fourth risk probability output by the target classification model and a fifth risk probability determined according to the dictionary rule; The first target risk probability of the target communication service is determined according to the following formula: Where, is the first target risk probability, w is the weight factor, Δw is the weight adjustment coefficient, is the fourth risk probability output by the target classification model, is a fifth risk probability determined according to the dictionary rule; A first target risk level of the target communication service is determined according to the first target risk probability.

7. The method according to claim 5 or 6, characterized in that Determining a first target risk level of the target communication service according to the first target risk probability includes: Obtaining the first target risk probability, wherein the first target risk probability is used to reflect the probability that the target communication service is at different risk levels; comparing the probability values ​​of different risk levels in the first target risk probability, and determining the risk level corresponding to the maximum probability value as the initial risk level; When the initial risk level is multiple risk levels, selecting the highest risk level from the multiple risk levels as the first target risk level; When the initial risk level is one risk level, the initial risk level is determined to be a first target risk level.

8. The method according to claim 5, characterized in that The method further comprises: When the service type of the target communication service does not exist in the dictionary rules, the preset risk probability is determined to be a seventh risk probability matching the target communication service, and the mapping relationship in the dictionary rules is updated in combination with the seventh risk probability and the target communication service.

9. The method according to claim 1, characterized in that The target classification model is trained in the following way: Dividing the preset data set into multiple training sets and multiple validation sets; Constructing an initial classification model, wherein the initial classification model integrates a classification algorithm, and the classification algorithm includes one of the following: a logistic regression algorithm, a support vector machine algorithm, a K-nearest neighbor algorithm, and a naive Bayes algorithm; Inputting each of the training sets into the initial classification model for iterative training; After each training batch is completed, cross-validating the initial classification model in combination with the validation set, and adjusting the parameters of the initial classification model; After all training batches are completed, the target classification model is obtained.

10. The method according to claim 9, characterized in that The method further comprises: Periodically acquiring a plurality of groups of first data of target communication services, weight adjustment coefficients corresponding to the target communication services, and first target risk probabilities of the target communication services; For each group of target communication services, create a data set according to the service type name, and divide the first data of the target communication service corresponding to the weight adjustment coefficient equal to 0 into the data set; The preset data set is updated using the first data in the data set whose data volume is greater than a preset threshold, and the target classification model is trained using the updated preset data set to update the target classification model.

11. The method according to claim 1, wherein Acquiring first data of a target communication service, including: Acquire initial data of a target communication service; perform a preprocessing operation on the initial data to obtain first data of the target communication service, wherein the preprocessing operation includes at least one of the following: data cleaning and data format standardization.

12. A communication service risk assessment device, characterized in that: include: an acquisition module, configured to acquire first data of a target communication service, and determine a first service type and a second service type corresponding to the target communication service, wherein the first service type is a subclass of the second service type; a first determining module, configured to respectively determine a first quantity corresponding to the second data of the first business type and a second quantity corresponding to the third data of the second business type in a preset data set; a second determination module, configured to determine a risk level assessment strategy based on the first quantity and the second quantity, wherein the risk level assessment strategy includes one of the following: a risk assessment strategy based on a dictionary rule, and a risk assessment strategy determined jointly by a target classification model and a dictionary rule, the dictionary rule being configured to reflect a mapping relationship between each communication service and a first risk probability, and the target classification model being configured to analyze a second risk probability corresponding to each communication service; An evaluation module is configured to execute the risk level evaluation strategy on the first data of the target communication service to obtain a first target risk level of the target communication service.

13. A computer program product, characterized in that include: A computer program, wherein when the computer program is executed by a processor, the communication service risk assessment method according to any one of claims 1 to 11 is implemented.

14. An electronic device, characterized in that: include: A memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the communication service risk assessment method according to any one of claims 1 to 11 through the computer program.

Images