Information processing method and device, equipment, storage medium and program product

By extracting and abnormal detection of target information, identifying abnormal features related to the behavior of replacement resource, and generating abnormal scores, it solves the problem that banks find it difficult to assess and manage the risk of replacement resource in pure online credit services, and achieves efficient risk control and improve service efficiency.

CN118917938APending Publication Date: 2024-11-08CHINA CONSTRUCTION BANK +1
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410989492.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-23
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

In pure online credit service scenarios, it is difficult for banks to effectively evaluate and manage the risk of replacement resources. The existing technology relies on historical data, has poor adaptability, and has failed to fully consider other factors affecting the risk of replacement resources.

Method used

By extracting the target information, using anomaly detection strategies (such as isolated forests and time-recurring neural networks) to identify abnormal features related to the behavior of the permutation resource, generate exception scores, and generate request results based on the score and preset thresholds to achieve the permutation resource Automatic processing of requests.

Benefits of technology

It improves the efficiency and risk control capabilities of replacement resource services, reduces the potential risk of financial loss, adapts to market changes and user behavior, and enhances the multi-dimensional evaluation and management of replacement resource transactions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118917938A_ABST
    Figure CN118917938A_ABST
Patent Text Reader

Abstract

The invention discloses an information processing method and device, equipment, a storage medium and a program product, and is applied to the technical field of financial information services. The method comprises the steps of obtaining target information in response to a first replacement resource obtaining request; performing feature extraction on the target information to obtain a feature extraction result; under the condition that the feature extraction result comprises a plurality of first feature sets, performing abnormal value detection on the first feature sets according to a first detection strategy to obtain a first abnormal score; determining the first abnormal score as an abnormal score corresponding to the first replacement resource acquisition request; and generating a first request result corresponding to the first replacement resource acquisition request based on the abnormal score corresponding to the first replacement resource acquisition request and a preset abnormal score threshold. According to the embodiment of the invention, the resource replacement risk can be effectively evaluated and managed, and the resource replacement request can be automatically processed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the field of financial information service technology, and in particular, relates to an information processing method, device, equipment, storage medium and program product. Background Art

[0002] With the rapid development of Internet finance, banks and financial institutions are increasingly adopting purely online credit services to improve user experience and credit service processing efficiency.

[0003] In a purely online credit service scenario, customers apply for replacement resources through the Internet without going to a physical bank branch. This requires banks to be able to effectively assess and manage replacement resource risks without face-to-face communication. Summary of the invention

[0004] The embodiments of the present application provide an information processing method, apparatus, device and computer storage medium, which can effectively evaluate and manage the risk of replacing resources and realize automatic processing of replacement resource requests.

[0005] In a first aspect, an embodiment of the present application provides an information processing method, including:

[0006] In response to the first replacement resource acquisition request, acquiring target information, the target information including market data, transaction information and personal information of a target requester corresponding to the first replacement resource acquisition request;

[0007] Perform feature extraction on the target information to obtain feature extraction results;

[0008] When the feature extraction result includes a plurality of first feature sets, performing outlier detection on the first feature set according to a first detection strategy to obtain a first anomaly score, the first feature set including a plurality of first features related to the resource replacement behavior;

[0009] Determine the first anomaly score as the anomaly score corresponding to the first replacement resource acquisition request;

[0010] Based on the anomaly score corresponding to the first replacement resource acquisition request and a preset anomaly score threshold, a first request result corresponding to the first replacement resource acquisition request is generated.

[0011] In a possible implementation, the first detection strategy includes:

[0012] Acquire a preset number of feature set samples from the multiple first feature sets, each feature set sample including a portion of the first feature sets in the multiple first feature sets;

[0013] For each feature set sample, the following steps are performed respectively: determine a target segmentation feature from a plurality of first features of the first feature set as a node feature; according to the value of the target segmentation feature in the first feature set and the segmentation threshold corresponding to the target segmentation feature, split part of the first feature set into two subtrees; for each subtree, return to determine the target segmentation feature from the plurality of first features of the first feature set until the subtree contains the target number of first feature sets or the tree structure reaches the target height, thereby obtaining an isolated tree corresponding to the feature set sample;

[0014] Calculate the abnormal score of each first feature set in the feature set sample according to the path length of each first feature set in the isolated tree to obtain an abnormal score set of the feature set sample;

[0015] A first anomaly score is calculated according to an anomaly score set of a preset number of feature set samples and a preset first anomaly score calculation rule.

[0016] In a possible implementation, the method further includes:

[0017] When the feature extraction result does not include the plurality of first feature sets, performing anomaly analysis on the second feature according to the second detection strategy to obtain a second anomaly score, wherein the second feature includes a time series data feature extracted from the time series data;

[0018] The second anomaly score is used to determine an anomaly score corresponding to the first replacement resource acquisition request.

[0019] In a possible implementation, the second detection strategy includes:

[0020] Extracting target time series data features from the second feature according to a preset time window and a preset step size;

[0021] Fit the target time series data features to obtain the corresponding relationship between the time series data features and the anomaly score;

[0022] According to the correspondence between the time series data feature and the anomaly score, a second anomaly score corresponding to the target time series data is determined.

[0023] In a possible implementation, feature extraction is performed on target information to obtain a feature extraction result, including:

[0024] Perform data cleaning on the target information to obtain the target data;

[0025] According to the preset feature selection rule, a plurality of preset features are extracted from the target data, wherein the plurality of preset features include features related to the abnormal amount of replacement resources;

[0026] According to a preset feature combination rule, at least some of the multiple preset features are combined to obtain multiple combined features, where the multiple combined features include at least one of a ratio of the target requester's assets to the amount of the replacement resources and a time interval between application and return of the replacement resources;

[0027] A first feature set and / or a second feature is extracted from a plurality of preset features and combined features to obtain a feature extraction result.

[0028] In a possible implementation, the method further includes:

[0029] Acquire feedback information corresponding to a target request result, where the target request result includes multiple request results corresponding to multiple replacement resource acquisition requests within a target time period;

[0030] When the feedback information does not meet the preset conditions, prompt information is output to prompt adjustment of at least one of the preset feature selection rules, feature combination rules or abnormal score thresholds.

[0031] In a possible implementation, the method further includes:

[0032] Acquire preset information of a plurality of second replacement resource acquisition requests at a preset time, the plurality of second replacement resource acquisition requests including a plurality of second replacement resource acquisition requests within a preset time period, the preset information including market data corresponding to the second replacement resource acquisition requests, transaction information and personal information of the requesting party corresponding to the second replacement resource acquisition requests, and a second request result corresponding to the second replacement resource acquisition requests;

[0033] Extracting features from preset information of the plurality of second replacement resource acquisition requests respectively to obtain a plurality of third feature sets and a fourth feature corresponding to the second replacement resource acquisition requests, wherein the third feature set includes a plurality of third features related to the replacement resource behavior, and the fourth feature includes a time series data feature extracted from the time series data;

[0034] updating the first detection strategy according to the third feature set corresponding to each of the plurality of second replacement resource acquisition requests and the second request result;

[0035] The second detection strategy is updated according to the fourth characteristics corresponding to each of the plurality of second replacement resource acquisition requests and the second request results.

[0036] In a second aspect, an embodiment of the present application provides an information processing device, including:

[0037] An acquisition module, configured to acquire target information in response to the first replacement resource acquisition request, wherein the target information includes market data, transaction information of a target requester corresponding to the first replacement resource acquisition request, and personal information;

[0038] An extraction module is used to extract features from target information and obtain feature extraction results;

[0039] a detection module, configured to, when the feature extraction result includes a plurality of first feature sets, perform outlier detection on the first feature set according to a first detection strategy to obtain a first anomaly score, wherein the first feature set includes a plurality of first features related to the resource replacement behavior;

[0040] A determination module, configured to determine the first anomaly score as an anomaly score corresponding to the first replacement resource acquisition request;

[0041] A generating module is used to generate a first request result corresponding to the first replacement resource acquisition request based on the anomaly score corresponding to the first replacement resource acquisition request and a preset anomaly score threshold.

[0042] In a third aspect, an embodiment of the present application provides an information processing device, the device comprising:

[0043] A processor and a memory storing computer program instructions; when the processor executes the computer program instructions, any one of the above information processing methods is implemented.

[0044] In a fourth aspect, an embodiment of the present application provides a computer storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, any one of the above-mentioned information processing methods is implemented.

[0045] In a fifth aspect, an embodiment of the present application provides a computer program product. When instructions in the computer program product are executed by a processor of an electronic device, the electronic device is enabled to perform any one of the above-mentioned information processing methods.

[0046] The information processing method, device, equipment and computer storage medium of the embodiment of the present application can respond to the first replacement resource acquisition request and obtain target information, and the target information includes market data, transaction information of the target requester corresponding to the first replacement resource acquisition request, and personal information. The target information contains a variety of structured data and unstructured data, which is conducive to a more comprehensive analysis of the replacement resource acquisition request in combination with market changes and the behavior of the requester. The embodiment of the present application extracts features from the target information to obtain feature extraction results. Then, when the feature extraction result includes multiple first feature sets, the first feature set is detected for outliers according to the first detection strategy to obtain a first anomaly score, and the first anomaly score is determined as the anomaly score corresponding to the first replacement resource acquisition request. The first feature set includes multiple first features related to the replacement resource behavior. In this way, the corresponding anomaly score can be obtained by processing the feature set related to the replacement resource behavior, so that the anomaly existing in the replacement resource acquisition request can be accurately identified in multiple dimensions. The embodiment of the present application can also generate a first request result corresponding to the first replacement resource acquisition request based on the anomaly score corresponding to the first replacement resource acquisition request and a preset anomaly score threshold. In this way, the risks in resource replacement transactions can be effectively evaluated and managed based on the abnormality score, and the automatic processing of resource replacement requests can be realized. In this way, the efficiency of resource replacement services can be improved while the risk control ability can be improved to reduce the potential risk of financial losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0048] Figure 1 It is a flowchart of an information processing method provided by an embodiment of the present application;

[0049] Figure 2 is a flowchart of an information processing method provided by another embodiment of the present application;

[0050] Figure 3 is a flowchart of an information processing method provided by yet another embodiment of the present application;

[0051] Figure 4 is a flowchart of an information processing method provided by yet another embodiment of the present application;

[0052] Figure 5 is a structural diagram of an information processing device provided by yet another embodiment of the present application;

[0053] Figure 6It is a structural diagram of an information processing device provided in yet another embodiment of the present application. DETAILED DESCRIPTION

[0054] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0055] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0056] It should be noted that the acquisition, storage, use and processing of data in the embodiments of the present application are in compliance with the relevant provisions of national laws and regulations.

[0057] It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned, and they should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0058] As described in the background technology, a purely online credit service scenario requires banks to be able to effectively assess and manage the risks of replacement resources without face-to-face communication.

[0059] The relevant technology involves using user portraits to build anomaly detection models and credit scoring models, which aim to achieve preliminary screening of users by excluding abnormal customers with high fraud and low repayment ability, and to score and quantify users' credit scores.

[0060] However, the above solutions have certain data dependence and are overly dependent on historical data and labeled data, resulting in poor adaptability of the model to new situations. Some models may perform well on training data, but have poor generalization ability on unseen test data, resulting in poor practical application results. In addition, due to the large differences in business scenarios and data characteristics of different financial institutions, some solutions may be difficult to adapt to the needs of pure online credit replacement resource scenarios. In addition, the above solutions usually pay more attention to the credit of the borrower and do not fully consider other factors that affect the risk of replacement resources. It can be seen that how to effectively evaluate and manage the risk of replacement resources and realize the automatic processing of replacement resource requests is still a technical problem that needs to be solved.

[0061] In order to solve the problems of the prior art, the embodiments of the present application provide an information processing method, apparatus, device and computer storage medium.

[0062] The following first introduces the information processing method of the embodiment of the present application with reference to the drawings.

[0063] Figure 1 FIG. 1 shows a schematic diagram of a process for obtaining replacement resources provided by an embodiment of the present application. Figure 1 As shown, the replacement resource acquisition may include steps S110 to S150.

[0064] S110, the target requester clicks on the resource replacement acquisition application.

[0065] S120, the information processing device executes a regular access procedure.

[0066] S130, the information processing device performs resource replacement anomaly detection on the target requester.

[0067] S140 , when the replacement resource anomaly detection passes, issuing the replacement resource to the target requester through the replacement resource acquisition request of the target requester.

[0068] S150: If the replacement resource abnormality detection fails, reject the replacement resource acquisition request of the target requester.

[0069] In the above step S110, the target requester may include an individual or enterprise applying for replacement resources. The replacement resources may include credit replacement resources. The requested party may include an institution providing replacement resources, such as a bank. In one embodiment, the replacement resource acquisition request may include the amount of the replacement resources, the application time, the return time, the relevant materials provided by the target requester, etc.

[0070] In the above step S120, the regular access process may include a preliminary review of the target requester's application qualifications, material completeness, etc. The specific review content may be formulated based on relevant regulatory requirements and is not limited here.

[0071] Figure 2 FIG. 1 is a flow chart showing an information processing method provided by an embodiment of the present application. Figure 1 As shown, the method may include the following steps S210 to S250.

[0072] S210, in response to the first replacement resource acquisition request, acquiring target information, the target information including market data, transaction information and personal information of a target requester corresponding to the first replacement resource acquisition request.

[0073] In step S210, the first replacement resource may include any credit replacement resource. Market data may include real-time market data, for example, including but not limited to interest rates and exchange rate changes. Transaction information may include at least one of the replacement resource amount, application time, return time, account status information of the target requester, historical behavior data and historical performance data of the target requester of the first replacement resource acquisition request. Among them, the historical behavior of the target requester may include the transaction behavior data of the target requester, and the historical performance data may include the behavior log, credit report, etc. of the target requester on the requested party platform. In one embodiment, the first replacement resource acquisition request may be a non-first replacement resource acquisition request of the target requester, and the above-mentioned transaction behavior data may also include data related to the historical replacement resource acquisition request of the target requester. Personal information may include basic information of the target requester and activity data on social media.

[0074] S220, extracting features from the target information to obtain feature extraction results.

[0075] In step S220, feature extraction may include performing feature selection according to preset rules to extract features useful for anomaly detection. Specific rules may be formulated by business personnel based on experience and are not limited here.

[0076] S230, when the feature extraction result includes multiple first feature sets, perform outlier detection on the first feature set according to a first detection strategy to obtain a first anomaly score, the first feature set including multiple first features related to the resource replacement behavior.

[0077] In step S230, multiple first feature sets may include multiple feature sets obtained by feature extraction and conversion of multiple data related to the behavior of replacing resources. The multiple data related to the behavior of replacing resources may include, but are not limited to, one or more of the number of historical replacement resource acquisition requests, quotas, and application times of the target requester. As an example, each type of data may be feature extracted and converted to obtain a first feature set, and the feature set may include features of multiple dimensions. The features of the multiple dimensions are the multiple first features mentioned above. The first detection strategy may include an algorithm or a model, for example, may include a machine learning algorithm, such as an autoencoder.

[0078] S240: Determine the first abnormality score as the abnormality score corresponding to the first replacement resource acquisition request.

[0079] S250: Generate a first request result corresponding to the first replacement resource acquisition request based on the anomaly score corresponding to the first replacement resource acquisition request and a preset anomaly score threshold.

[0080] In one embodiment, when the anomaly score corresponding to the first replacement resource acquisition request is lower than a preset anomaly score threshold, the generated first request result may include that the first replacement resource acquisition request is passed; when the anomaly score corresponding to the first replacement resource acquisition request is not lower than the preset anomaly score threshold, the generated first request result may include that the first replacement resource acquisition request is not passed. The anomaly score threshold can be set according to actual needs. As an example, a higher anomaly score threshold can be set to reduce the risk of fraudulent replacement resource acquisition requests passing. As an example, a lower anomaly score threshold can be set to reduce the risk of normal replacement resource acquisition requests being misjudged as abnormal, so as to improve the pass rate of normal user replacement resource acquisition requests.

[0081] The information processing method of the embodiment of the present application can respond to the first replacement resource acquisition request and obtain target information, and the target information includes market data, transaction information of the target requester corresponding to the first replacement resource acquisition request, and personal information. The target information contains a variety of structured data and unstructured data, which is conducive to a more comprehensive analysis of the replacement resource acquisition request in combination with market changes and the behavior of the requester. The embodiment of the present application extracts features from the target information to obtain feature extraction results. Then, when the feature extraction result includes multiple first feature sets, the first feature set is detected for anomalies according to the first detection strategy to obtain a first anomaly score, and the first anomaly score is determined as the anomaly score corresponding to the first replacement resource acquisition request. The first feature set includes multiple first features related to the replacement resource behavior. In this way, the corresponding anomaly score can be obtained by processing the feature set related to the replacement resource behavior, so that the anomaly existing in the replacement resource acquisition request can be accurately identified in multiple dimensions. The embodiment of the present application can also generate a first request result corresponding to the first replacement resource acquisition request based on the anomaly score corresponding to the first replacement resource acquisition request and the preset anomaly score threshold. In this way, the risks existing in the replacement resource transaction can be effectively evaluated and managed according to the anomaly score, and the automatic processing of the replacement resource request can be realized. In this way, the efficiency of resource replacement services can be improved while risk control capabilities can be enhanced to reduce potential financial loss risks.

[0082] In the above embodiment, the target information may include historical data (such as historical behavior data and historical performance data, etc.) and real-time data (such as interest rates and exchange rates in the current market, etc.). The target information may be extracted from different systems. In one embodiment, an integrated application programming interface (API) may be developed to obtain the above target information from different systems. The API can support real-time data transmission and ensure the security and integrity of the data.

[0083] In one embodiment, a data pipeline can also be constructed, which can process real-time data streams from different sources. Specifically, the data pipeline can use a message queue (the technical component used can be, for example, the open source stream processing platform Kafka) and a stream processing system (the component used can be, for example, Flink Apache Flink) to access real-time data. The data pipeline can also perform necessary conversions on the data to match the input requirements of the algorithm or model, for example, "date of birth" can be converted into "age". The data pipeline can also cache data in memory to reduce latency.

[0084] In one embodiment, the data acquired by the data pipeline can be temporarily stored in a database. As an example, the database may include an in-memory database, a time series database, and a traditional relational database. The in-memory database may apply a remote dictionary server (Remote Dictionary Server, Redis) component to store unstructured real-time data; the time series database may apply an open source time series database InfluxDB component to store time series data; the relational database may apply a PG database (PgSQL) component to store structured data and / or historical data. When it is necessary to obtain target information, you can directly query from the database or call an API to obtain real-time data to obtain the target information corresponding to the target requester.

[0085] In one embodiment, the first detection strategy may include:

[0086] A preset number of feature set samples are obtained from the multiple first feature sets, each feature set sample including a portion of the first feature sets in the multiple first feature sets.

[0087] For each feature set sample, the following steps are performed respectively: determine the target segmentation feature from multiple first features of the first feature set as the node feature; split part of the first feature set into two subtrees according to the value of the target segmentation feature in the first feature set and the segmentation threshold corresponding to the target segmentation feature; for each subtree, return to determine the target segmentation feature from multiple first features of the first feature set until the subtree contains the target number of first feature sets or the tree structure reaches the target height, thereby obtaining an isolated tree corresponding to the feature set sample.

[0088] The abnormal score of each first feature set in the feature set sample is calculated according to the path length of each first feature set in the isolated tree to obtain an abnormal score set of the feature set sample.

[0089] A first anomaly score is calculated according to an anomaly score set of a preset number of feature set samples and a preset first anomaly score calculation rule.

[0090] In the above embodiment, the first detection strategy can be an isolation forest algorithm. The isolation forest algorithm is an efficient unsupervised learning algorithm. The isolation forest algorithm can be applied to outlier detection, and its basic idea is that there are very few outliers and they are very different from other values. If a data point is abnormal, it will be isolated faster than most normal data points. In other words, outliers are usually in the edge areas of the data, not in dense clusters.

[0091] In the above embodiment, each feature set may correspond to one data point, in other words, each feature set is obtained by feature conversion of one data point. Each feature set includes multiple first features, and each first feature corresponds to features of different dimensions of the data point. Obtaining a preset number of feature set samples from multiple first feature sets may specifically include: randomly extracting a certain number of first feature sets from the first feature sets corresponding to multiple data points to obtain feature set samples; repeating the above steps multiple times until a preset number of feature set samples are obtained. The number of first feature sets in the feature set samples may be much smaller than the number of first feature sets in the feature extraction results. In this way, the size of the isolated tree can be limited and the complexity of the calculation can be reduced.

[0092] The above embodiment can construct an isolated tree by randomly selecting features and randomly selecting segmentation points. Specifically, a first feature can be randomly selected from multiple first features of the first feature set as a target segmentation feature; the target segmentation feature can be used as a node feature; a value can be randomly selected between the maximum and minimum values ​​of the target segmentation feature in the feature sample set as a segmentation threshold; the first feature set whose target segmentation feature is less than the segmentation threshold is divided into a subtree; the first feature set whose target segmentation feature is not less than the segmentation threshold is divided into another subtree. Repeat the above process for each subtree, and recursively perform segmentation until the subtree contains the target number of first feature sets (for example, the subtree only contains a single first feature set, or the number of first feature sets contained in the subtree is less than a preset value) or the tree structure reaches the target height, and an isolated tree corresponding to the feature set sample is obtained. It should be noted that in the process of constructing an isolated tree, for multiple subtrees at the same level, the target segmentation features can be the same or different, and the segmentation thresholds corresponding to the target segmentation features can be the same or different. For subtrees at different levels, the target segmentation features can be different first features.

[0093] In the above embodiment, the path length of the first feature set in the isolated tree can represent the path length from the root node to the leaf node where the first feature set is located. The path length of the abnormal point is usually shorter. Calculating the abnormal score of each first feature set in the feature set sample according to the path length of each first feature set in the isolated tree can be achieved by methods known in the art. For example, an abnormal score can be assigned to each first feature set based on the path length. The height of the abnormal score can reflect the possibility that the first feature set is abnormal. As an example, the abnormal score can be the reciprocal of 2 to the power of n, where n can be the average value of the path length of the first feature set in all isolated trees divided by the average path length of the isolated tree.

[0094] In the above embodiment, according to the anomaly score set of a preset number of feature set samples and the preset first anomaly score calculation rule, the calculation of the first anomaly score can be achieved by a variety of methods. As an example, the average value of the anomaly scores of the preset number of feature set samples can be used as the first anomaly score. As another example, the anomaly scores of the preset number of feature set samples can be weighted summed to obtain the first anomaly score. As another example, the above average value or weighted summation result can also be standardized to obtain the first anomaly score. The specific calculation method can be combined with the adaptive adjustment of the anomaly score threshold, which is not limited in the embodiments of the present application.

[0095] The above embodiment performs outlier detection on the first feature set by using the isolation forest algorithm. In this way, the abnormal pattern of the first feature set can be automatically identified by an unsupervised learning method. In this way, not only can the impact of data noise and missing values ​​in the target information be reduced, but also the detection efficiency of outliers can be improved. Thus, while improving the efficiency of replacement resource services, the ability of risk control is improved and the potential risk of financial losses is reduced.

[0096] In one embodiment, the method may further include:

[0097] When the feature extraction result does not include the plurality of first feature sets, an abnormality analysis is performed on the second feature according to the second detection strategy to obtain a second abnormality score, wherein the second feature includes a time series data feature extracted from the time series data.

[0098] The second anomaly score is used to determine an anomaly score corresponding to the first replacement resource acquisition request.

[0099] In the above embodiment, the second detection strategy may include an algorithm or a model, for example, a time recurrent neural network, such as a long short-term memory network (LSTM), etc. The time series data may include historical transaction behavior data, social media behavior data, and other data containing time information of the target requester.

[0100] In the above embodiment, when the feature extraction result does not include multiple first feature sets, the time series data feature anomaly analysis can be performed according to the second detection strategy to obtain a second anomaly score, and the second anomaly score can be used to determine the anomaly score corresponding to the first replacement resource acquisition request. In this way, the long-term dependency of the replacement resource behavior can be captured, thereby accurately identifying whether the first replacement resource acquisition request is abnormal.

[0101] Therefore, the embodiments of the present application can use different detection strategies for anomaly identification according to the feature types contained in the feature extraction results. The first detection strategy can quickly identify anomalies and is suitable for processing high-dimensional data; while the second detection strategy can capture long-term dependencies in time series data and is suitable for analyzing customers' replacement resource behavior patterns. In this way, anomalies in replacement resource acquisition requests can be accurately identified in multiple dimensions. In this way, the accuracy and robustness of replacement resource processing can be improved, thereby improving risk control capabilities and reducing potential financial loss risks while improving replacement resource service efficiency.

[0102] In one embodiment, the second detection strategy may include:

[0103] According to the preset time window and the preset step size, the target time series data feature is extracted from the second feature.

[0104] The target time series data features are fitted to obtain the corresponding relationship between the time series data features and the anomaly scores.

[0105] According to the correspondence between the time series data feature and the anomaly score, a second anomaly score corresponding to the target time series data is determined.

[0106] In the above embodiment, the second detection strategy may include LSTM. LSTM is a special type of recurrent neural network (RNN), which can solve the gradient vanishing and gradient exploding problems of traditional RNN by introducing a gating mechanism. LSTM can determine how much information in the second feature can flow into the current cell state through the input gate; it can determine which historical information should be discarded in the current state through forgetting; the cell state of LSTM can carry information across multiple time steps and is controlled by the input gate and forgetting gate; the output gate of LSTM can determine what the hidden state of the next time step should be, based on the current cell state and output weights. As a result, LSTM can learn long-term dependencies and capture important events with longer time distances in time series data; the gating mechanism enables the network to learn important patterns in the data and ignore unimportant information. In this way, LSTM can be applied to various complex sequence prediction tasks.

[0107] In the above embodiment, the target time series data feature is extracted from the second feature according to the preset time window and the preset step size, which may specifically include extracting the target time series data feature from multiple time series data features according to the preset time window and the preset step size. For example, the time series data feature may include the time series data feature related to the historical transaction behavior data of the target requester, the time series data feature related to the social media behavior data, etc., and the time series data feature related to the historical transaction behavior data and the time series data feature related to the social media behavior data may be extracted according to the preset time window and the preset step size, respectively, to obtain the target time series data feature.

[0108] The above embodiment can accurately capture the dependency of resource replacement behavior within a specific time period, thereby accurately predicting the target requester's resource replacement repayment ability and default risk, and identifying whether the first resource replacement acquisition request is abnormal. In this way, the accuracy of resource replacement processing can be improved and the potential financial loss risk can be reduced.

[0109] In one embodiment, feature extraction is performed on the target information to obtain feature extraction results, which may specifically include:

[0110] Perform data cleaning on the target information to obtain the target data.

[0111] According to preset feature selection rules, multiple preset features are extracted from the target data, and the multiple preset features include features related to abnormal replacement resource quotas.

[0112] According to the preset feature combination rules, at least part of the multiple preset features are combined to obtain multiple combined features, and the multiple combined features include at least one of the ratio of the target requester's assets to the replacement resource quota and the time interval between the application and return of the replacement resources.

[0113] A first feature set and / or a second feature is extracted from a plurality of preset features and combined features to obtain a feature extraction result.

[0114] In the above embodiment, data cleaning can be performed by methods known in the art. As an example, data cleaning can include but is not limited to at least one of deleting duplicate records, processing missing values, and identifying and processing outliers. Processing missing values ​​can be a technique known in the art, for example, it can be filled using the mean, median, or a prediction model. Identifying and processing outliers can be a technique known in the art, for example, it can be identified and processed by methods such as box plot analysis or standard deviation. In one example, data conversion can also be performed after data cleaning, and target data can be obtained after data conversion. Data conversion can include encoding conversion of non-numeric data, specifically, discrete data encoding conversion method can be used for encoding conversion. For example, one-hot encoding can be used to process categorical variables. Data conversion can also include time formatting of time series data, for example, the time of time series data can be converted into a Unix timestamp. In the above embodiment, the preset feature selection rule can include features related to abnormal replacement resource quotas pre-set by business personnel. As an example, the preset feature selection rule can include selecting specific types of features, for example, it can include extracting replacement resource quotas, replacement resource application frequency, replacement resource return delay times, etc. from target data.

[0115] In the above embodiment, the preset feature combination rules may include combining some features directly extracted from the target data to obtain a pre-set combination feature related to the abnormal replacement resource quota. For example, the ratio of the target requester's assets (such as asset income or total assets) to the replacement resource quota corresponding to the first replacement resource acquisition request can be calculated to obtain a combination feature; the time interval between the application time and the return time of the same replacement resource acquisition request of the target requester can also be calculated to obtain a combination feature. It should be noted that the combination feature may also include other features, and those skilled in the art may set them according to actual needs, and the embodiments of the present application are not limited to this.

[0116] In one example, the features may also be transformed. For example, a logarithmic transformation may be performed on a numerical feature to reduce the impact of outliers.

[0117] The above embodiment can perform data cleaning on the target information, and then perform feature selection and feature combination, and extract the first feature set and / or the second feature. The first feature set and / or the second feature obtained in this way have a high correlation with the abnormal amount of replacement resources, which is conducive to improving the accuracy and efficiency of abnormal value detection or abnormal analysis, thereby improving the accuracy and efficiency of replacement resource processing.

[0118] In one embodiment, the method may further include:

[0119] Feedback information corresponding to a target request result is obtained, where the target request result includes multiple request results corresponding to multiple replacement resource acquisition requests within a target time period.

[0120] When the feedback information does not meet the preset conditions, prompt information is output to prompt adjustment of at least one of the preset feature selection rules, feature combination rules or abnormal score thresholds.

[0121] In the above embodiment, the feedback information may include feedback information from business-related personnel and the requesting party corresponding to the target request result, for example, the opinions of business-related personnel on the target request result, the appeal of the requesting party on the failure of the replacement resource acquisition request, etc. In the case where the feedback information is used to characterize the inaccuracy of the target request result, the above prompt information may be output. After receiving the prompt information, the relevant personnel may adjust at least one of the preset feature selection rules, feature combination rules, or abnormal score thresholds. Exemplarily, the preset feature selection rules and feature combination rules may be adjusted in combination with the opinions of business-related personnel on the target request result, for example, new features or new combinations may be added. Exemplarily, in the case where the feedback shows that a fraudulent replacement resource acquisition request has been passed, the relevant personnel may increase the abnormal score threshold after verification. Exemplarily, in the case where the feedback shows that a normal replacement resource acquisition request has been misjudged as abnormal, resulting in the failure of the replacement resource acquisition request, the relevant personnel may lower the abnormal score threshold after verification.

[0122] The above embodiment sets a feedback mechanism, which can adjust the feature selection rules, feature combination rules or abnormal score thresholds in response to feedback information. In this way, not only can the accuracy of resource replacement processing be improved, but also a feedback channel can be provided to users, user needs can be responded to in a timely manner, and user experience can be improved.

[0123] In one embodiment, the method may further include:

[0124] Obtain preset information of multiple second replacement resource acquisition requests at a preset time, the multiple second replacement resource acquisition requests include multiple second replacement resource acquisition requests within a preset time period, the preset information includes market data corresponding to the second replacement resource acquisition request, transaction information and personal information of the requesting party corresponding to the second replacement resource acquisition request, and second request results corresponding to the second replacement resource acquisition request.

[0125] Feature extraction is performed on the preset information of multiple second replacement resource acquisition requests respectively to obtain multiple third feature sets and fourth features corresponding to the second replacement resource acquisition requests. The third feature set includes multiple third features related to the replacement resource behavior, and the fourth feature includes the time series data feature extracted from the time series data.

[0126] The first detection strategy is updated according to the third feature sets corresponding to each of the plurality of second replacement resource acquisition requests and the second request results.

[0127] The second detection strategy is updated according to the fourth characteristics corresponding to each of the plurality of second replacement resource acquisition requests and the second request results.

[0128] In the above embodiment, the meanings of the third feature set and the fourth feature are respectively the same as those of the first feature set and the second feature, and the only difference is the requesting party and time, which will not be elaborated here.

[0129] The above embodiment can update the first detection strategy and the second detection strategy at a specific moment. In this way, it is possible to continuously learn and optimize according to market changes and the evolution of user behavior, and realize dynamic adjustment of the first detection strategy and the second detection strategy. In this way, the accuracy of outlier detection and anomaly analysis can be further improved, thereby improving the accuracy of replacement resource processing.

[0130] In one embodiment, updating the first detection strategy according to the third feature set corresponding to each of the plurality of second replacement resource acquisition requests and the second request result may specifically include:

[0131] An outlier detection is performed on the third feature set according to the first detection strategy to obtain a third anomaly score corresponding to the second replacement resource acquisition request.

[0132] Based on a preset anomaly score threshold, a third request result corresponding to the third anomaly score is determined.

[0133] The first detection strategy is updated according to errors between the third request results corresponding to each of the plurality of second replacement resource acquisition requests and the second request results.

[0134] In one embodiment, updating the second detection strategy according to the fourth characteristics corresponding to each of the plurality of second replacement resource acquisition requests and the second request results may specifically include:

[0135] An anomaly analysis is performed on the fourth feature according to the second detection strategy to obtain a fourth anomaly score corresponding to the second replacement resource acquisition request.

[0136] Based on a preset anomaly score threshold, a fourth request result corresponding to a fourth anomaly score is determined.

[0137] The second detection strategy is updated according to errors between fourth request results corresponding to each of the plurality of second replacement resource acquisition requests and the second request results.

[0138] In one embodiment, the first detection strategy and the second detection strategy may be the first detection model and the second detection model, respectively. The above-mentioned process of updating the first detection strategy and the second detection strategy may be the process of retraining the first detection model and the second detection model. The first detection model and the second detection model may be integrated in an anomaly detection module, and the anomaly detection module may be integrated in a stream processing system for replacement resource acquisition, or may not be integrated in a stream processing system for replacement resource acquisition, and acquire data through API calls for outlier detection and anomaly analysis.

[0139] In one embodiment, the performance of the anomaly detection module can also be monitored, for example, the accuracy rate, recall rate and other indicators of the model can be monitored to maintain the stability and accuracy of the model. As an example, the first detection model and the second detection model can be integrated into the anomaly detection module as an anomaly detection model. The process of anomaly detection by the anomaly detection model can be as follows: Figure 3 Specifically, the following steps may be included:

[0140] S310, acquiring data of the target requester in real time.

[0141] The target requester's data may include data including target information.

[0142] S320, inputting the data of the target requester into the anomaly detection model.

[0143] S330, running anomaly detection model.

[0144] S340, output model results.

[0145] Specifically, the model result may be an anomaly score.

[0146] S350, judging whether it is passed.

[0147] Step S350 may include determining whether the request is approved based on the anomaly score and a preset anomaly score threshold.

[0148] In one embodiment, Figure 4 As shown, before executing the information processing method of the embodiment of the present application, it is also necessary to train the anomaly detection model. Specifically, the model training may include the following steps.

[0149] S410, data collection and integration.

[0150] S420, data preprocessing and feature engineering.

[0151] S430, model training and optimization.

[0152] After model training and optimization, a trained anomaly detection model can be obtained. In one embodiment, after the anomaly detection model training is completed, step S440 may be included, anomaly detection rule setting. Specifically, the anomaly detection rule may include a case where the anomaly score is lower than a preset anomaly score threshold, through the replacement resource acquisition request; when the anomaly score is not lower than the preset anomaly score threshold, the replacement resource acquisition request is rejected. In some embodiments, the anomaly detection rule may be loaded into the anomaly detection model, or it may be independent of the anomaly detection model. In some embodiments, the anomaly detection rule may not be loaded into the anomaly detection model, which is conducive to flexible adjustment of the anomaly detection rule.

[0153] In some embodiments, after the abnormality detection model training is completed, the method may further include: S450, user feedback and continuous optimization. Specifically, the model performance obtained by user feedback and real-time monitoring may be returned to step S430 to retrain and optimize the model.

[0154] The following takes the isolation forest model as the first detection model and the LSTM model as the second detection model as an example to introduce the model training process in detail.

[0155] First, data can be collected from different data sources by means of database query or API call, etc. The data may include market data corresponding to multiple historical replacement resource acquisition requests, transaction information and personal information of the requesting party corresponding to multiple historical replacement resource acquisition requests.

[0156] Then, data cleaning and data conversion can be performed. The implementation methods of data cleaning and data conversion can be as described above, and will not be repeated here.

[0157] Next, feature engineering can be performed. Specifically, based on business knowledge and domain experience, features related to abnormal replacement resource quotas can be selected, such as replacement resource quotas, replacement resource application frequency, number of replacement resource return delays, and so on. New combined features can also be created. For example, the ratio of the requester's assets (such as asset income or total assets) to the replacement resource quota corresponding to the replacement resource acquisition request can be calculated to obtain a combined feature; the time interval between the application time and the return time of the same replacement resource acquisition request of the requester can also be calculated to obtain a combined feature. Features can also be transformed. For example, numerical features can be logarithmically transformed to reduce the impact of outliers.

[0158] The dataset after feature engineering is split into training set, validation set and test set. The split ratio can be 70% training set, 15% validation set and 15% test set. In this way, the randomness and representativeness of the data can be ensured and data leakage can be avoided.

[0159] The isolation forest model is trained using the training set, and the model can detect outliers on the first target feature set in the training set by randomly selecting features and randomly selecting split points. The first target feature set can be a feature set extracted from data related to the behavior of replacing resources, and the first target feature set can include multiple first target features related to the behavior of replacing resources.

[0160] The time series data in the training set and the anomaly score labels corresponding to the time series data are input into the LSTM model to train the LSTM model. The specific training of the isolation forest model and the LSTM model can be implemented by methods known in the art, which will not be described in detail here.

[0161] After training the isolation forest model and LSTM model using the training set, you can use the test set data to tune the model to adjust the model parameters. For example, you can adjust the number of trees in the isolation forest, the number of neurons in the hidden layer of the LSTM, etc. You can use grid search, random search, or Bayesian optimization techniques to perform automated parameter tuning. You can also evaluate the performance of the model on the validation set. Specifically, you can use indicators such as accuracy, recall, F1 score, and receiver operating characteristic curve (ROC curve) to compare the performance of different models and select the best model. This completes the model training.

[0162] The trained model can be deployed to the production environment. As an example, the Predictive Model Markup Language (PMML) of the model service platform can be used to achieve rapid deployment and reasoning of the model.

[0163] See first Figure 5 The information processing device 500 provided in the embodiment of the present application includes an acquisition module 501, an extraction module 502, a detection module 503, a determination module 504, and a generation module 505.

[0164] The acquisition module 501 is used to acquire target information in response to the first replacement resource acquisition request, where the target information includes market data, transaction information and personal information of the target requester corresponding to the first replacement resource acquisition request.

[0165] The extraction module 502 is used to extract features from the target information to obtain feature extraction results.

[0166] The detection module 503 is used to perform outlier detection on the first feature set according to the first detection strategy to obtain a first anomaly score when the feature extraction result includes multiple first feature sets, and the first feature set includes multiple first features related to the resource replacement behavior.

[0167] The determination module 504 is configured to determine the first abnormality score as the abnormality score corresponding to the first replacement resource acquisition request.

[0168] The generating module 505 is configured to generate a first request result corresponding to the first replacement resource acquisition request based on the anomaly score corresponding to the first replacement resource acquisition request and a preset anomaly score threshold.

[0169] The information processing device of the embodiment of the present application can respond to the first replacement resource acquisition request and obtain target information, and the target information includes market data, transaction information of the target requester corresponding to the first replacement resource acquisition request, and personal information. The target information contains a variety of structured data and unstructured data, which is conducive to a more comprehensive analysis of the replacement resource acquisition request in combination with market changes and the behavior of the requester. The embodiment of the present application extracts features from the target information to obtain feature extraction results. Then, when the feature extraction result includes multiple first feature sets, the first feature set is detected for anomalies according to the first detection strategy to obtain a first anomaly score, and the first anomaly score is determined as the anomaly score corresponding to the first replacement resource acquisition request. The first feature set includes multiple first features related to the replacement resource behavior. In this way, the corresponding anomaly score can be obtained by processing the feature set related to the replacement resource behavior, so that the anomaly existing in the replacement resource acquisition request can be accurately identified in multiple dimensions. The embodiment of the present application can also generate a first request result corresponding to the first replacement resource acquisition request based on the anomaly score corresponding to the first replacement resource acquisition request and a preset anomaly score threshold. In this way, the risks existing in the replacement resource transaction can be effectively evaluated and managed according to the anomaly score, and the automatic processing of the replacement resource request can be realized. In this way, the efficiency of resource replacement services can be improved while risk control capabilities can be enhanced to reduce potential financial loss risks.

[0170] In one embodiment, the first detection strategy may include:

[0171] The acquisition module is used to acquire a preset number of feature set samples from multiple first feature sets, each feature set sample including a part of the first feature sets in the multiple first feature sets.

[0172] The execution module is used to respectively perform the following steps for each feature set sample: determine a target segmentation feature from multiple first features of a first feature set as a node feature; split part of the first feature set into two subtrees according to the value of the target segmentation feature in the first feature set and the segmentation threshold corresponding to the target segmentation feature; for each subtree, return to determine the target segmentation feature from multiple first features of the first feature set until the subtree contains a target number of first feature sets or the tree structure reaches a target height, thereby obtaining an isolated tree corresponding to the feature set sample.

[0173] The calculation module is used to calculate the abnormal score of each first feature set in the feature set sample according to the path length of each first feature set in the isolated tree, so as to obtain the abnormal score set of the feature set sample.

[0174] The calculation module is further used to calculate a first anomaly score according to an anomaly score set of a preset number of feature set samples and a preset first anomaly score calculation rule.

[0175] In one embodiment, the apparatus may further include:

[0176] The detection module is used to perform an anomaly analysis on the second feature according to the second detection strategy to obtain a second anomaly score when the feature extraction result does not include multiple first feature sets, wherein the second feature includes a time series data feature extracted from the time series data.

[0177] The determination module is used to determine the anomaly score corresponding to the first replacement resource acquisition request based on the second anomaly score.

[0178] In one embodiment, the second detection strategy may include:

[0179] The extraction module is used to extract the target time series data feature from the second feature according to a preset time window and a preset step size.

[0180] The fitting module is used to fit the target time series data features to obtain the corresponding relationship between the time series data features and the anomaly scores.

[0181] The determination module is used to determine a second anomaly score corresponding to the target time series data according to the corresponding relationship between the time series data feature and the anomaly score.

[0182] In one embodiment, the extraction module is used to extract features from the target information to obtain feature extraction results, which may specifically include:

[0183] The cleaning module is used to clean the target information to obtain the target data.

[0184] The extraction module is used to extract multiple preset features from the target data according to preset feature selection rules, and the multiple preset features include features related to abnormal replacement resource quotas.

[0185] The combination module is used to combine at least part of the multiple preset features according to the preset feature combination rules to obtain multiple combined features, and the multiple combined features include at least one of the ratio of the target requester's assets to the replacement resource quota and the time interval between the application and return of the replacement resources.

[0186] The extraction module is used to extract the first feature set and / or the second feature from a plurality of preset features and combined features to obtain a feature extraction result.

[0187] In one embodiment, the apparatus may further include:

[0188] The acquisition module is used to acquire feedback information corresponding to a target request result, where the target request result includes multiple request results corresponding to multiple replacement resource acquisition requests within a target time period.

[0189] The output module is used to output prompt information when the feedback information does not meet the preset conditions, so as to prompt the user to adjust at least one of the preset feature selection rules, feature combination rules or abnormal score threshold.

[0190] In one embodiment, the apparatus may further include:

[0191] An acquisition module is used to acquire preset information of multiple second replacement resource acquisition requests at a preset time, wherein the multiple second replacement resource acquisition requests include multiple second replacement resource acquisition requests within a preset time period, and the preset information includes market data corresponding to the second replacement resource acquisition request, transaction information of the requesting party corresponding to the second replacement resource acquisition request, personal information, and a second request result corresponding to the second replacement resource acquisition request.

[0192] An extraction module is used to perform feature extraction on preset information of multiple second replacement resource acquisition requests respectively to obtain multiple third feature sets and fourth features corresponding to the second replacement resource acquisition requests, wherein the third feature set includes multiple third features related to the replacement resource behavior, and the fourth feature includes a time series data feature extracted from the time series data.

[0193] The updating module is used to update the first detection strategy according to the third feature set corresponding to each of the plurality of second replacement resource acquisition requests and the second request result.

[0194] The updating module is further used to update the second detection strategy according to the fourth characteristics corresponding to each of the plurality of second replacement resource acquisition requests and the second request results.

[0195] Each module of the information processing device provided in the embodiment of the present application can realize Figure 2 The functions of each step of the information processing method are provided, and the corresponding technical effects can be achieved. For the sake of concise description, they will not be repeated here.

[0196] Figure 6 A schematic diagram of the hardware structure of an information processing device provided in an embodiment of the present application is shown.

[0197] The information processing device may include a processor 601 and a memory 602 storing computer program instructions.

[0198] Specifically, the processor 601 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0199] The memory 602 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 602 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 602 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 602 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 602 is a non-volatile solid-state memory.

[0200] The memory may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical or other physical / tangible memory storage devices. Thus, typically, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0201] The processor 601 implements any one of the information processing methods in the above embodiments by reading and executing computer program instructions stored in the memory 602 .

[0202] In one example, the information processing device may further include a communication interface 603 and a bus 610. Figure 6As shown, the processor 601, the memory 602, and the communication interface 603 are connected via a bus 610 and communicate with each other.

[0203] The communication interface 603 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0204] Bus 610 includes hardware, software or both, and the parts of online data flow billing equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front-end bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 710 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the application considers any suitable bus or interconnection.

[0205] The information processing device can execute the information processing method in the embodiment of the present application, thereby realizing the combination Figure 2 and Figure 5 The information processing method and device described herein. The information processing method in the above embodiments can be implemented by providing a computer storage medium in the present application embodiment. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any one of the information processing methods in the above embodiments is implemented.

[0206] An embodiment of the present application also provides a computer program product, including a computer program, which implements any one of the information processing methods in the above embodiments when the computer program is processed and executed.

[0207] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0208] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0209] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0210] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0211] The above are only specific implementation methods of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. An information processing method, characterized in that: include: In response to the first replacement resource acquisition request, acquiring target information, the target information including market data, transaction information and personal information of a target requester corresponding to the first replacement resource acquisition request; Performing feature extraction on the target information to obtain a feature extraction result; In a case where the feature extraction result includes a plurality of first feature sets, performing outlier detection on the first feature set according to a first detection strategy to obtain a first anomaly score, the first feature set including a plurality of first features related to the resource replacement behavior; Determining the first anomaly score as the anomaly score corresponding to the first replacement resource acquisition request; Based on the anomaly score corresponding to the first replacement resource acquisition request and a preset anomaly score threshold, a first request result corresponding to the first replacement resource acquisition request is generated.

2. The information processing method according to claim 1, characterized in that: The first detection strategy includes: Acquire a preset number of feature set samples from the plurality of first feature sets, each of the feature set samples including a portion of the first feature sets in the plurality of first feature sets; For each of the feature set samples, the following steps are performed respectively: determining a target segmentation feature from a plurality of first features of the first feature set as a node feature; dividing the part of the first feature set into two subtrees according to the value of the target segmentation feature in the first feature set and the segmentation threshold corresponding to the target segmentation feature; for each subtree, returning to the target segmentation feature determined from the plurality of first features of the first feature set until the subtree contains a target number of first feature sets or the tree structure reaches a target height, thereby obtaining an isolated tree corresponding to the feature set sample; Calculate the anomaly score of each first feature set in the feature set sample according to the path length of each first feature set in the isolated tree to obtain an anomaly score set of the feature set sample; The first anomaly score is calculated according to the anomaly score set of the preset number of feature set samples and a preset first anomaly score calculation rule.

3. The information processing method according to claim 1, characterized in that: The method further comprises: When the feature extraction result does not include the plurality of first feature sets, performing anomaly analysis on a second feature according to a second detection strategy to obtain a second anomaly score, wherein the second feature includes a time series data feature extracted from the time series data; The second anomaly score is used to determine an anomaly score corresponding to the first replacement resource acquisition request.

4. The information processing method according to claim 3, characterized in that: The second detection strategy includes: Extracting target time series data features from the second features according to a preset time window and a preset step size; Fitting the target time series data features to obtain a corresponding relationship between the time series data features and the anomaly score; According to the correspondence between the time series data feature and the anomaly score, a second anomaly score corresponding to the target time series data is determined.

5. The information processing method according to claim 3 or 4, characterized in that: Performing feature extraction on the target information to obtain feature extraction results includes: Performing data cleaning on the target information to obtain target data; According to a preset feature selection rule, a plurality of preset features are extracted from the target data, wherein the plurality of preset features include features related to abnormal replacement resource quota; According to a preset feature combination rule, at least some of the multiple preset features are combined to obtain multiple combined features, wherein the multiple combined features include at least one of a ratio of the target requester's assets to the amount of the replacement resources and a time interval between application and return of the replacement resources; The first feature set and / or the second feature are extracted from the multiple preset features and the combined features to obtain the feature extraction result.

6. The information processing method according to claim 5, characterized in that: The method further comprises: Acquire feedback information corresponding to a target request result, wherein the target request result includes multiple request results corresponding to multiple replacement resource acquisition requests within a target time period; When the feedback information does not meet the preset condition, prompt information is output to prompt adjustment of at least one of the preset feature selection rule, the feature combination rule or the abnormal score threshold.

7. The information processing method according to claim 3, characterized in that: The method further comprises: Acquire preset information of a plurality of second replacement resource acquisition requests at a preset time, wherein the plurality of second replacement resource acquisition requests include a plurality of second replacement resource acquisition requests within a preset time period, and the preset information includes market data corresponding to the second replacement resource acquisition request, transaction information and personal information of the requester corresponding to the second replacement resource acquisition request, and a second request result corresponding to the second replacement resource acquisition request; Performing feature extraction on the preset information of the plurality of second replacement resource acquisition requests respectively to obtain a plurality of third feature sets and a fourth feature corresponding to the second replacement resource acquisition requests, wherein the third feature set includes a plurality of third features related to the replacement resource behavior, and the fourth feature includes a time series data feature extracted from the time series data; updating the first detection strategy according to the third feature set corresponding to each of the plurality of second replacement resource acquisition requests and the second request result; The second detection strategy is updated according to the fourth characteristics corresponding to each of the multiple second replacement resource acquisition requests and the second request results.

8. An information processing device, characterized in that: include: an acquisition module, configured to acquire target information in response to a first replacement resource acquisition request, wherein the target information includes market data, transaction information and personal information of a target requester corresponding to the first replacement resource acquisition request; An extraction module, used to extract features from the target information to obtain feature extraction results; a detection module, configured to, when the feature extraction result includes a plurality of first feature sets, perform outlier detection on the first feature sets according to a first detection strategy to obtain a first anomaly score, wherein the first feature set includes a plurality of first features related to the resource replacement behavior; a determination module, configured to determine the first anomaly score as an anomaly score corresponding to the first replacement resource acquisition request; A generating module is used to generate a first request result corresponding to the first replacement resource acquisition request based on the anomaly score corresponding to the first replacement resource acquisition request and a preset anomaly score threshold.

9. An information processing device, characterized in that: The device comprises: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the information processing method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the information processing method according to any one of claims 1 to 7 is implemented.

11. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device is enabled to execute the information processing method according to any one of claims 1 to 7.

Citation Information

Cited By

  • Mobile phone number one-key login method and device, electronic equipment and storage medium

    CN119136199A