Object scoring method and device based on feature information, electronic equipment and storage medium

Abstract

The invention discloses an object scoring method and device based on feature information, electronic equipment and a storage medium, and the method comprises the steps: obtaining a training sample set, determining a first object scoring set based on an original scoring model and the training sample set, obtaining a new scoring rule set corresponding to a new scoring theme, and obtaining a second object scoring rule set corresponding to the new scoring theme; constructing a preset scoring model based on the original scoring model and the newly added scoring rule set, determining a second object scoring set based on the preset scoring model and the training sample set, and performing iteration processing on the preset scoring data according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain target scoring data, and determining an updated new theme rule set based on the target scoring data. According to the embodiment of the invention, the new scoring rule set is formed by adding the iterative scoring score to the original scoring model, and then the final target scoring model is formed, so that the model forming process can be simplified, and the stability of the existing risk control strategy can be continuously maintained.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device, electronic device and storage medium for scoring an object based on feature information. Background Art

[0002] With the rapid development of computer technology and Internet technology, online payment has become a daily payment method for people. Merchant is a common concept in electronic payment or online payment, also known as merchant number, merchant account number or merchant number. At present, some illegal persons use merchants as a disguise to carry out illegal activities. Such merchants are abnormal merchants, so it is necessary to identify abnormal merchants so as to take corresponding measures.

[0003] Currently, we can use scoring models to score objects and judge whether the objects are abnormal based on the scores. However, with the increase of object features or the different interpretation dimensions of features, it is obviously not enough to still use the original scoring model to score objects and judge whether the objects are abnormal based on the scores. Based on this, we need to use some methods to obtain a new scoring model to score objects with features and judge whether the objects are abnormal based on the scores. Summary of the invention

[0004] In order to solve the problems of the prior art, the embodiments of the present invention provide a method, device, electronic device and storage medium for scoring an object based on feature information. The technical solution is as follows:

[0005] In one aspect, a method for scoring an object based on feature information is provided, the method comprising:

[0006] Acquire a training sample set; the training sample set includes a training object set and a first feature set; the first feature data set includes a plurality of features and feature data of each training object in the training object set under each feature;

[0007] Determine a first object scoring set based on the original scoring model and the training sample set; the first object scoring set includes object scoring data of each training object under the original scoring rule set included in the original scoring model;

[0008] Obtain a newly added scoring rule set corresponding to the newly added scoring topic; the newly added scoring rule set at least includes rule description information and preset scoring data;

[0009] Build a preset scoring model based on the original scoring model and the newly added scoring rule set;

[0010] Determine a second object scoring set based on a preset scoring model and a training sample set; the second object scoring set includes object scoring data of each training object under the preset scoring model;

[0011] Iteratively process the preset scoring data according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain target scoring data;

[0012] An updated set of newly added subject rules is determined based on the target scoring data; the updated set of newly added subject rules is used to construct a target scoring model together with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

[0013] In another aspect, a device for scoring an object based on feature information is provided, the device comprising:

[0014] The sample acquisition module is used to acquire a training sample set; the training sample set includes a training object set and a first feature set; the first feature data set includes a plurality of features and feature data of each training object in the training object set under each feature;

[0015] A first scoring determination module is used to determine a first object scoring set based on the original scoring model and the training sample set; the first object scoring set includes object scoring data of each training object under the original scoring rule set included in the original scoring model;

[0016] A rule set acquisition module is used to acquire a newly added scoring rule set corresponding to a newly added scoring topic; the newly added scoring rule set at least includes rule description information and preset scoring data;

[0017] A model building module is used to build a preset scoring model based on the original scoring model and the newly added scoring rule set;

[0018] A second scoring determination module is used to determine a second object scoring set based on a preset scoring model and a training sample set; the second object scoring set includes object scoring data of each training object under the preset scoring model;

[0019] A target score determination module, configured to iteratively process preset score data according to the first object score set, the second object score set, the initial score data and the preset score model to obtain target score data;

[0020] The rule set determination module is used to determine the updated new subject rule set based on the target scoring data; the updated new subject rule set is used to construct a target scoring model with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

[0021] In some possible embodiments, the target score determination module is used to:

[0022] Constructing a comprehensive evaluation function based on the function of the first object scoring set, the function of the second object scoring set, the function of the preset scoring data, and the function of the preset scoring model;

[0023] Determine the first scoring data corresponding to K newly added scoring rule sets by using a preset search rule; K is an integer greater than or equal to 1;

[0024] Determine K+1 newly added scoring rule sets based on the K first scoring data and the preset scoring data and the rule description information;

[0025] Use the original scoring model and K+1 newly added scoring rule sets to build K+1 preset scoring models;

[0026] Determine K+1 second object scoring sets based on K+1 preset scoring models and training sample sets;

[0027] Inputting the first object scoring set, K+1 second object scoring sets, K first scoring data, preset scoring data and K+1 preset scoring models into the comprehensive evaluation function to obtain K+1 comprehensive evaluation data;

[0028] The scoring data are iterated using K+1 comprehensive evaluation data, K first scoring data and preset scoring data to obtain target scoring data.

[0029] In some possible embodiments, the target score determination module is used to:

[0030] Classifying the K first scoring data and the preset scoring data into a first data category and a second data category based on the K+1 comprehensive evaluation data;

[0031] Determine a first probability density function and a second probability density function corresponding to the first data set and the second data set, respectively;

[0032] determining an acquisition function based on the first probability density function and the second probability density function;

[0033] Determining second scoring data using a prime number acquisition function;

[0034] Determine a preset scoring model and a second object scoring set corresponding to the second scoring data;

[0035] Inputting the first object scoring set, the second scoring data, the preset scoring model corresponding to the second scoring data, and the second object scoring set into the comprehensive evaluation function to obtain the comprehensive evaluation data corresponding to the second scoring data;

[0036] The scoring data are iterated using K+1 comprehensive evaluation data, K first scoring data, preset scoring data, second scoring data, and comprehensive evaluation data corresponding to the second scoring data to obtain target scoring data.

[0037] In some possible embodiments, a comprehensive evaluation function is constructed based on a function of the first object scoring set, a function of the second object scoring set, a function of preset scoring data, and a function of a preset scoring model, including:

[0038] constructing a score distribution change function based on the function of the first object score set and the function of the second object score set;

[0039] Constructing a model performance evaluation function based on a function of preset scoring data and a function of a preset scoring model;

[0040] A comprehensive evaluation function is constructed based on the score distribution change function and the model performance evaluation function.

[0041] In some possible embodiments, the rule set determination module is used to:

[0042] Using the target scoring data to update the preset scoring data, obtaining an updated set of newly added scoring rules;

[0043] Build a target scoring model based on the updated new scoring rule set and the original scoring model.

[0044] In some possible embodiments, the device further includes a testing module, wherein the testing module is configured to:

[0045] Acquire a test sample set; the training sample set includes a test object set and a second feature set; the second feature data set includes a plurality of features and feature data of each test object in the test object set under each feature;

[0046] Determine a third object scoring set based on the original scoring model and the test sample set; the third object scoring set includes object scoring data of each test object under the original scoring rule set included in the original scoring model;

[0047] Determine a fourth object scoring set based on the target scoring model and the test sample set; the fourth object scoring set includes object scoring data of each test object under the target scoring model;

[0048] Inputting the third object score set and the fourth object score set into the score distribution change function to determine the score distribution change data;

[0049] When the rating distribution change data indicates that the target rating model meets the application conditions, the target rating model is used to rate the object to be rated based on the feature set corresponding to the object to be rated.

[0050] In some possible embodiments, the test module is used to:

[0051] Mapping each object scoring data in the fourth object scoring set to obtain an updated fourth object scoring set; each object scoring data in the updated fourth object scoring set is within a preset data range;

[0052] The third object score set and the updated fourth object score set are input into the score distribution change function to determine the score distribution change data.

[0053] In some possible embodiments, the apparatus further includes an original scoring model training module, which is used to:

[0054] Acquire an original sample set; the original sample set includes an original object set and a third feature set; each original object in the original object set carries an object label, and the object label is used to indicate whether the original object is an abnormal object or a non-abnormal object; the third feature set includes multiple features and feature data of each original object under each feature; the multiple features included in the third feature data set are subsets of the multiple features included in the first feature data set;

[0055] Determine at least one original scoring rule set corresponding to at least one original scoring theme based on the original sample set; each original scoring rule set includes at least rule description information and original scoring data;

[0056] An original scoring model is constructed based on at least one original scoring rule set.

[0057] In some possible embodiments, the original scoring model training module is used to:

[0058] Determine a fourth feature set corresponding to each original rating topic from the third feature set; the fourth feature set includes at least one feature and feature data of each original object under each feature in the at least one feature; at least one feature in the fourth feature set is a subset of multiple features included in the third feature data set, and different fourth feature sets contain different features;

[0059] Building a decision tree model corresponding to each original scoring topic based on the original object set and the fourth feature set corresponding to each original scoring topic;

[0060] An original scoring rule set corresponding to each original scoring topic is determined based on the decision tree model corresponding to each original scoring topic.

[0061] In some possible embodiments, the original scoring model training module is used to:

[0062] Determine multiple feature rules based on the nodes on the decision tree model corresponding to each original scoring topic;

[0063] Determine an abnormality level of each feature rule in a plurality of feature rules based on an object label carried by each original object set and a feature corresponding to each original object;

[0064] The feature rules with the same abnormal level are merged to obtain the rule description information in the original scoring rule set;

[0065] The original scoring data of the rule description information in the original scoring rule set is determined based on the anomaly level.

[0066] In some possible embodiments, the original scoring model training module is used to:

[0067] Determine the accuracy of each feature rule in the plurality of feature rules based on the object label carried by each original object set and the feature corresponding to each original object;

[0068] The abnormality level of each feature rule is determined according to the accuracy of each feature rule. On the other hand, an electronic device is provided, including a processor and a memory, wherein at least one instruction or at least one program is stored in the memory, and the at least one instruction or at least one program is loaded and executed by the processor to implement the above-mentioned object scoring method based on feature information.

[0069] On the other hand, a computer-readable storage medium is provided, in which at least one instruction or at least one program is stored, and the at least one instruction or at least one program is loaded and executed by a processor to implement the object scoring method based on feature information as described above.

[0070] On the other hand, a computer program product or a computer program is provided, the computer program product or the computer program includes computer instructions, the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the above-mentioned object scoring method based on feature information.

[0071] In an embodiment of the present invention, a training sample set is obtained, wherein the training sample set includes a training object set and a first feature set, wherein the first feature data set includes multiple features and feature data of each training object in the training object set under each feature, a first object scoring set is determined based on an original scoring model and the training sample set, wherein the first object scoring set includes object scoring data of each training object under an original scoring rule set included in the original scoring model, a new scoring rule set corresponding to a new scoring topic is obtained, wherein the new scoring rule set includes at least rule description information and preset scoring data, a preset scoring model is constructed based on the original scoring model and the new scoring rule set, a second object scoring set is determined based on the preset scoring model and the training sample set, wherein the second object scoring set includes object scoring data of each training object under the preset scoring model, the preset scoring data is iteratively processed according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain target scoring data, an updated new subject rule set is determined based on the target scoring data, the updated new subject rule set is used to construct a target scoring model with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored. The embodiment of the present application forms a new scoring rule set by adding an iterative scoring score to the original scoring model, and then forms a final target scoring model. This not only simplifies the model formation process, but also ensures that it can continue to maintain the stability of the existing risk control strategy, thereby reducing management costs and improving management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0073] Figure 1 is a schematic diagram of an implementation environment provided by an embodiment of the present invention;

[0074] Figure 2 is a flow chart of an object scoring method based on feature information provided by an embodiment of the present invention;

[0075] Figure 3 It is a flowchart of a method for determining an original scoring model provided by an embodiment of the present invention;

[0076] Figure 4 is a flowchart of a method for determining at least one original scoring rule set provided by an embodiment of the present invention;

[0077] Figure 5It is a flowchart of a method for determining an original scoring rule set corresponding to each original scoring topic provided by an embodiment of the present invention;

[0078] Figure 6 is a schematic diagram of a leaf node of a decision tree provided by an embodiment of the present invention;

[0079] Figure 7 is a flow chart of a method for determining scoring data for each target provided by an embodiment of the present invention;

[0080] Figure 8 It is a flowchart of a method for determining model verification provided by an embodiment of the present invention;

[0081] Fig. 9 is a structural block diagram of an object scoring device based on feature information provided by an embodiment of the present invention;

[0082] Fig.10 The present invention provides a hardware structure block diagram of an electronic device. DETAILED DESCRIPTION

[0083] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0084] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention 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 server that includes 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.

[0085] It is understandable that in the specific implementation of this application, related data such as user information is involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0086] In order to facilitate understanding of the above-mentioned technical solutions and the technical effects produced by the embodiments of the present disclosure, a brief introduction is given to the terms involved in the embodiments of the present disclosure:

[0087] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0088] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0089] Computer Vision (CV) is a science that studies how to make machines "see". To put it more specifically, it refers to machine vision such as using cameras and computers to replace human eyes to identify and measure targets, and further perform graphic processing so that computer processing becomes an image that is more suitable for human eye observation or transmission to instrument detection. As a scientific discipline, computer vision studies related theories and technologies, and attempts to establish an artificial intelligence system that can obtain information from images or multi-dimensional data. Computer vision technology usually includes image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content / behavior recognition, three-dimensional object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous positioning and map construction, and other technologies, as well as common biometric recognition technologies such as face recognition and fingerprint recognition.

[0090] Natural language processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that can achieve effective communication between people and computers using natural language. Natural language processing is a science that integrates linguistics, computer science, and mathematics. Therefore, research in this field will involve natural language, that is, the language people use in daily life, so it is closely related to the study of linguistics. Natural language processing technology usually includes text processing, semantic understanding, machine translation, robot question answering, knowledge graph and other technologies.

[0091] Machine Learning (ML) is a multi-disciplinary interdisciplinary subject involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning. Machine learning can be divided into supervised machine learning, unsupervised machine learning, and semi-supervised machine learning.

[0092] Among them, cloud technology refers to a hosting technology that unifies hardware, software, network and other resources in a wide area network or local area network to realize data computing, storage, processing and sharing. Cloud technology is a general term for network technology, information technology, integration technology, management platform technology, application technology, etc. based on the cloud computing business model application, which can form a resource pool and be used on demand, flexible and convenient. Cloud computing technology will become an important support. The background services of the technical network system require a large amount of computing and storage resources, such as video websites, picture websites and more portal websites. With the high development and application of the Internet industry, in the future, each item may have its own identification mark, and all need to be transmitted to the background system for logical processing. Data of different levels will be processed separately. All kinds of industry data require strong system backing support, which can only be achieved through cloud computing.

[0093] Cloud storage is a new concept that extends and develops from the concept of cloud computing. A distributed cloud storage system (hereinafter referred to as storage system) refers to a storage system that uses cluster applications, grid technology, and distributed storage file systems to bring together a large number of storage devices of various types (storage devices are also called storage nodes) in the network to work together through application software or application interfaces to provide data storage and business access functions. At present, the storage method of the storage system is to create a logical volume. When creating a logical volume, a physical storage space is allocated for each logical volume. The physical storage space may be composed of disks of a storage device or several storage devices. The client stores data on a logical volume, that is, the data is stored on the file system. The file system divides the data into many parts, each of which is an object. The object contains not only data but also additional information such as data identification (ID, ID entity). The file system writes each object into the physical storage space of the logical volume, and the file system records the storage location information of each object. Therefore, when the client requests to access the data, the file system can allow the client to access the data according to the storage location information of each object. The process of the storage system allocating physical storage space to a logical volume is as follows: based on the estimated capacity of the objects stored in the logical volume (this estimate often has a large margin relative to the actual capacity of the objects to be stored) and the grouping of independent redundant disk arrays (RAID, Redundant array of Independent Disks), the physical storage space is divided into stripes in advance. A logical volume can be understood as a stripe, thereby allocating physical storage space to the logical volume.

[0094] In short, a database can be seen as an electronic filing cabinet - a place where electronic files are stored, and users can add, query, update, delete, and other operations on the data in the files. The so-called "database" is a collection of data that is stored together in a certain way, can be shared with multiple users, has as little redundancy as possible, and is independent of the application.

[0095] See also Figure 1 , which is a schematic diagram of an implementation environment provided by an embodiment of the present invention, and the implementation environment may include a client 110, a server 120 and a database 130.

[0096] The client 110 and the server 120 as well as the server 120 and the database 130 may be connected and communicated via a network.

[0097] The client 110 includes but is not limited to mobile phones, computers, intelligent voice interaction devices, smart home appliances, vehicle-mounted clients, aircraft, etc. An application with a human-computer interaction function may be run in the client 110, and the application may launch virtual item distribution activities for different business scenarios, such as flash sales, lucky draws, and reward activities for completing tasks.

[0098] The client 110 may obtain a training sample set, the training sample set includes a training object set and a first feature set, the first feature data set includes multiple features and feature data of each training object in the training object set under each feature, determine a first object scoring set based on the original scoring model and the training sample set, the first object scoring set includes object scoring data of each training object under an original scoring rule set included in the original scoring model, obtain a new scoring rule set corresponding to a new scoring topic, the new scoring rule set includes at least rule description information and preset scoring data, build a preset scoring model based on the original scoring model and the new scoring rule set, determine a second object scoring set based on the preset scoring model and the training sample set, the second object scoring set includes object scoring data of each training object under the preset scoring model, iteratively process the preset scoring data according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain target scoring data, determine an updated new subject rule set based on the target scoring data, the updated new subject rule set is used to build a target scoring model with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored. The embodiment of the present application forms a new scoring rule set by adding an iterative scoring score to the original scoring model, and then forms a final target scoring model. This not only simplifies the model formation process, but also ensures that it can continue to maintain the stability of the existing risk control strategy, thereby reducing management costs and improving management efficiency.

[0099] The server 120 may obtain a training sample set, the training sample set including a training object set and a first feature set, the first feature data set including multiple features and feature data of each training object in the training object set under each feature, determine a first object scoring set based on the original scoring model and the training sample set, the first object scoring set including object scoring data of each training object under an original scoring rule set included in the original scoring model, obtain a new scoring rule set corresponding to a new scoring topic, the new scoring rule set including at least rule description information and preset scoring data, build a preset scoring model based on the original scoring model and the new scoring rule set, determine a second object scoring set based on the preset scoring model and the training sample set, the second object scoring set including object scoring data of each training object under the preset scoring model, iteratively process the preset scoring data according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain target scoring data, determine an updated new subject rule set based on the target scoring data, the updated new subject rule set is used to build a target scoring model with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored. The embodiment of the present application forms a new scoring rule set by adding an iterative scoring score to the original scoring model, and then forms a final target scoring model. This not only simplifies the model formation process, but also ensures that it can continue to maintain the stability of the existing risk control strategy, thereby reducing management costs and improving management efficiency.

[0100] The database 130 may include an in-memory database and a relational database. It should be noted that the servers, databases, nodes, etc. of the embodiments of the present invention may be independent physical servers, or they may be server clusters or distributed systems composed of multiple physical servers. They may also be cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0101] In an exemplary embodiment, the client 110, the server 120 and the database 130 can all be node devices in the blockchain system, and can share the acquired and generated information with other node devices in the blockchain system to achieve information sharing between multiple node devices. Multiple node devices in the blockchain system can be configured with the same blockchain, which is composed of multiple blocks, and the adjacent blocks have an association relationship, so that when the data in any block is tampered, it can be detected by the next block, thereby preventing the data in the blockchain from being tampered with, and ensuring the security and reliability of the data in the blockchain.

[0102] In the embodiment of the present application, the original scoring model is a model that previously scores objects carrying features and judges whether the objects are abnormal based on the scores. However, with the increase of features or the different interpretation dimensions of features, it is obviously not possible to meet the current requirements to still use the original scoring model to score objects and judge whether the objects are abnormal based on the scores. Based on this, it is necessary to obtain a new scoring model through some methods to score objects carrying features and judge whether the objects are abnormal based on the scores.

[0103] In some possible embodiments, some traditional modeling methods, such as logistic regression, decision tree, or random forest, can be used to train on a newly provided sample set containing objects and features to obtain a new scoring model.

[0104] However, although the new scoring model can make full use of new features and analyze and score objects from other abnormal dimensions, it will cause major changes in the scoring scale and affect the stability of existing risk control strategies.

[0105] Among them, a significant change in the scoring scale means that, for example, the original scoring scale is within a range of 0 to 100. When others see that an object is scored 95, they can easily understand that the score is high and the object is likely to be abnormal. However, when the new scoring model scores the object 135, it is not difficult for others to understand the actual meaning of this score.

[0106] The stability of existing risk control strategies means that in the process of continuous optimization of the original scoring model, each feature or feature under each abnormal dimension has just the right impact on the object judgment, and will not be over-magnified or ignored, so that a stable risk control strategy can be maintained. However, in the new scoring model, the newly emerged features or features under the newly added abnormal dimension may have an over-magnified impact on the object, thereby affecting the stability of the existing risk control strategy.

[0107] In some possible embodiments, the scoring rules under the newly added abnormal dimension can be directly added to the original scoring model. However, although this method is simple to operate, the lack of optimization of the weight of the new feature may cause the impact of the new feature to be over-amplified or ignored, affecting the risk identification ability of the scorecard. In addition, this method may also cause a significant change in the score distribution.

[0108] Based on the above reasons, the present application provides an object scoring method based on feature information, which can maintain the scoring scale while minimizing the impact on the stability of existing risk control strategies and reducing management costs and complexity.

[0109] See also Figure 2, which is a flow chart of an object scoring method based on feature information provided by an embodiment of the present invention, which can be applied to Figure 1 In the implementation environment shown, the execution subject of this method can be Figure 1 The server that executes object scoring can also be the client or other server node that executes object scoring. It should be noted that this specification provides method operation steps as described in the embodiments or flow charts, but more or fewer operation steps may be included based on routine or non-creative labor. The order of steps listed in the embodiments is only one way of executing the steps among many orders, and does not represent the only order of execution. When the system or product is actually executed, it can be executed sequentially or in parallel according to the method shown in the embodiments or drawings (for example, in a parallel processor or multi-threaded processing environment). Specifically, Figure 2 As shown, the method may include:

[0110] S201, obtaining a training sample set; the training sample set includes a training object set and a first feature set; the first feature data set includes multiple features and feature data of each training object in the training object set under each feature.

[0111] In the embodiment of the present application, the client may obtain a training sample set. Optionally, the training sample set may include a training object set and a first feature set.

[0112] In the embodiment of the present application, the training object may be any object that carries features and can be used to identify or process the object.

[0113] In some possible embodiments, the training object set may be a collection of some merchants. Therefore, the characteristics of the merchant may be its own attribute characteristics, such as merchant type, merchant years, merchant location, etc. Optionally, the merchant characteristics may be transaction characteristics associated with customers, such as the proportion of general-whole-hundred transactions, the number of general-whole-hundred transactions, the amount of general-whole-hundred transactions, the proportion of counterparties with abnormal labels, the number of counterparties, the proportion of male counterparties, the proportion of female counterparties, the daily repurchase rate, the proportion of sensitive information in transaction notes, the total amount of the transaction day, the average amount of the transaction day, and the cross-ratio of paying users and receiving users, etc.

[0114] In an embodiment of the present application, each training object in the training object set carries an object label, and the object label is used to indicate whether the training object is an abnormal object or a non-abnormal object. In some possible embodiments, the object label can be represented by the elements "1" and "0". Among them, the object label "1" can represent the training object as an abnormal object or a suspicious object, and the object label "0" can represent the training object as a non-abnormal object or a non-suspicious object. This application does not limit the specific representation of the object label.

[0115] In an embodiment of the present application, the first feature set may include multiple features and feature data of each training object in the training object set under each feature. Optionally, the multiple features included in the first feature set may be multiple attribute features. Optionally, the multiple features included in the first feature set may be multiple transaction features. Optionally, the multiple features included in the first feature set may be a combination of attribute features and transaction features.

[0116] In some possible embodiments, the training sample set may be represented by an object vector, a feature matrix, and a label vector.

[0117] The following is described by an example. Assuming that the training object set includes 10,000 training objects, it can be represented by the object vector [00000 00001 00002 00003...09999]. Assuming that the first feature set includes 10 features, the feature matrix is ​​a 10,000 by 10 matrix. The elements of each row in the feature matrix are 10 feature data under the 10 features corresponding to each training object. For example, the feature data of the feature "general whole hundred transaction ratio" is 0.6. The label vector of 10,000 training objects is [1 0 0 0 1 0 1...1], and the label vector includes 10,000 elements.

[0118] S203, determining a first object scoring set based on the original scoring model and the training sample set; the first object scoring set includes object scoring data of each training object under the original scoring rule set included in the original scoring model.

[0119] In an embodiment of the present application, the client may determine an original scoring model based on the original sample set.

[0120] See also Figure 3 , which is a flow chart of a method for determining an original scoring model provided by an embodiment of the present invention, such as Figure 3 As shown, including:

[0121] S301, obtaining an original sample set; the original sample set includes an original object set and a third feature set; each original object in the original object set carries an object label, and the object label is used to indicate whether the original object is an abnormal object or a non-abnormal object; the third feature set includes multiple features and feature data of each original object under each feature; the multiple features included in the third feature data set are a subset of the multiple features included in the first feature data set.

[0122] In the embodiment of the present application, the client may obtain an original sample set. Optionally, the original sample set may include an original object set and a third feature set.

[0123] In the embodiment of the present application, the original object may be any object that carries features and can be identified or processed using the features carried.

[0124] Since the new scoring model is used to replace the original scoring model and score the characteristics of the merchant, the original object set in the original sample set used to construct the original scoring model can be a collection of some merchants. The characteristics of the merchant can be its own attribute characteristics, such as merchant type, merchant age, merchant location, etc. Optionally, the merchant's characteristics can be transaction characteristics associated with the customer, such as the proportion of general whole hundred transactions, the number of general whole hundred transactions, the amount of general whole hundred transactions, the proportion of counterparties with abnormal labels, the number of counterparties, the proportion of male counterparties, the proportion of female counterparties, the daily repurchase rate, the proportion of sensitive information in transaction notes, the total amount of the transaction day, and the average amount of the transaction day.

[0125] In an embodiment of the present application, each original object in the original object set carries an object label, and the object label is used to indicate whether the original object is an abnormal object or a non-abnormal object. In some possible embodiments, the object label can be represented by the elements "1" and "0". Among them, the object label "1" can represent the original object as an abnormal object or a suspicious object, and the object label "0" can represent the original object as a non-abnormal object or a non-suspicious object. This application does not limit the specific representation of the object label.

[0126] In an embodiment of the present application, the third feature set may include multiple features and feature data of each original object in the original object set under each feature. Optionally, the multiple features included in the third feature set may be multiple attribute features. Optionally, the multiple features included in the third feature set may be multiple transaction features. Optionally, the multiple features included in the third feature set may be a combination of attribute features and transaction features.

[0127] In some possible embodiments, the generation of a new scoring model may be due to the emergence of new features, and therefore, the multiple features included in the third feature data set may be a subset of the multiple features included in the first feature data set. In some possible embodiments, the multiple features included in the third feature data set are the multiple features included in the first feature data set. In other possible embodiments, the multiple features included in the third feature data set are some of the multiple features included in the first feature data set, for example, the first feature data set includes 10 features, and the multiple features included in the third feature data set are 9 of the 10 features.

[0128] In some possible embodiments, the original sample set may be represented by an object vector, a feature matrix, and a label vector.

[0129] The following is a description using an example. Assuming that the original object set includes 100,000 original objects, it can be represented by the object vector [00000 00001 00002 00003…19999]. Assuming that the third feature set includes 9 features, the feature matrix is ​​a 100,000 by 9 matrix. The elements of each row in the feature matrix are the 9 feature data under the 9 features corresponding to each original object. For example, the feature data of the feature "general whole hundred transaction ratio" is 0.6. The label vector of 100,000 original objects is [1 0 0 0 1 0 1…1], and the label vector includes 100,000 elements.

[0130] S303: Determine at least one original scoring rule set corresponding to at least one original scoring theme based on the original sample set; each original scoring rule set includes at least rule description information and original scoring data.

[0131] In an embodiment of the present application, the client can determine at least one original scoring rule set corresponding to at least one original scoring theme based on the original sample set, wherein each original scoring rule set includes at least rule description information and original scoring data.

[0132] In the embodiment of the present application, at least one original scoring topic may be obtained first. Optionally, the number of original scoring topics may be one or more.

[0133] In the embodiment of the present application, obtaining an original scoring theme is to obtain the features associated with the original scoring theme. In the embodiment of the present application, an original scoring theme can be associated with multiple features, wherein the multiple features associated with an original scoring theme can be part of the multiple features included in the original sample set.

[0134] For example, suppose there are two original scoring topics, one of which has a theme name of "integer amount transactions" and the other has a theme name of "counterparties". Among the multiple features included in the original sample set, the features related to "integer amount transactions" can be divided into the original scoring topic. In this way, the features included in the original scoring topic "integer amount transactions" may include the proportion of pan-integer transactions, the number of pan-integer transactions, the amount of pan-integer transactions and the total amount of the transaction day. Similarly, the features included in the original scoring topic "counterparties" may include the number of counterparties, the proportion of male counterparties and the average amount of the transaction day. The above original scoring topics are all examples and can be adjusted according to actual conditions. This application does not impose any restrictions.

[0135] See also Figure 4 , which is a flow chart of a method for determining at least one original scoring rule set provided by an embodiment of the present invention, such as Figure 4 As shown, including:

[0136] S401, determining a fourth feature set corresponding to each original rating topic from the third feature set; the fourth feature set includes at least one feature and feature data of each original object under each feature in the at least one feature; at least one feature in the fourth feature set is a subset of multiple features included in the third feature data set, and different fourth feature sets contain different features.

[0137] In the embodiment of the present application, at least one feature corresponding to each original scoring theme can be determined from the third feature set. For example, the features included in the original scoring theme "integer amount transactions" cited above may include the proportion of general whole-hundred transactions, the number of general whole-hundred transactions, the amount of general whole-hundred transactions, and the total amount of the transaction day. The features included in the original scoring theme "counterparties" may include the number of counterparties, the proportion of male counterparties, the proportion of female counterparties, and the average amount of the transaction day.

[0138] Optionally, the feature data of each original object under each feature involved in the original scoring theme can be determined from the feature data of each original object under each feature included in the third feature set. For example, for the original scoring theme "integer amount transaction", the feature data that can be obtained are the feature data of each original object under the four features of the general whole hundred transaction ratio, the number of general whole hundred transactions, the general whole hundred transaction amount and the total amount of the transaction day. In this way, a feature matrix of 100,000 times 4 can be obtained.

[0139] Since each original scoring topic is a different topic, the features it contains are different. In this way, a fourth feature set corresponding to each original scoring topic can be obtained.

[0140] S403: construct a decision tree model corresponding to each original scoring topic based on the original object set and the fourth feature set corresponding to each original scoring topic.

[0141] In the embodiment of the present application, a decision tree model corresponding to each original scoring topic can be built according to the object label carried by each original object and the fourth feature set corresponding to each original scoring topic.

[0142] In the embodiment of the present application, the number of layers of the decision tree can be a preset number of layers, such as 4 layers, and the present application does not limit this.

[0143] In the embodiment of the present application, a feature may be reused when the decision tree is split, and the present application does not limit the number of repetitions.

[0144] S405 , determining an original scoring rule set corresponding to each original scoring topic based on the decision tree model corresponding to each original scoring topic.

[0145] In the embodiment of the present application, the original scoring rule set corresponding to each original scoring topic can be determined based on the decision tree model corresponding to each original scoring topic.

[0146] See also Figure 5 , which is a flow chart of a method for determining an original scoring rule set corresponding to each original scoring topic provided by an embodiment of the present invention, such as Figure 5 As shown, including:

[0147] S501, determining multiple feature rules based on the nodes on the decision tree model corresponding to each original scoring topic.

[0148] In the embodiment of the present application, the leaf nodes on the decision tree model corresponding to each original scoring topic can be used, such as Figure 6 The end node (node ​​A) shown in FIG. 1 determines multiple feature rules. That is, there are as many feature rules as there are leaf nodes on the decision tree.

[0149] S503: Determine the abnormality level of each feature rule in a plurality of feature rules based on the object label carried by each original object set and the feature corresponding to each original object.

[0150] In an embodiment of the present application, the accuracy of each feature rule in multiple feature rules can be determined based on the object label carried by each original object set and the features corresponding to each original object, and the abnormality level of each feature rule can be determined according to the accuracy of each feature rule.

[0151] Optionally, the features corresponding to each original object may be evaluated using multiple feature rules contained in the decision tree model corresponding to each original scoring topic to obtain an evaluation label.

[0152] Optionally, an evaluation label and an object label of the original object corresponding to each feature in a plurality of feature rules are obtained, and the precision is determined based on the evaluation label and the object label of the original object corresponding to each feature.

[0153] Here, precision refers to the quotient between the sum of true positives and the sum of true positives and false positives.

[0154] In this way, the precision of each rule feature in the multiple feature rules corresponding to each original scoring topic can be determined. Optionally, the corresponding relationship between the precision and the abnormality level is obtained.

[0155] In the present application, the correspondence between the accuracy and the abnormality level can be set in advance, for example, accuracy ≥ 0.8 is "high risk", 0.6 ≤ accuracy < 0.8 is "higher risk", 0.4 ≤ accuracy < 0.6 is "medium risk", 0.2 ≤ accuracy < 0.4 is "lower risk", and accuracy ≤ 0.2 is "low risk".

[0156] Optionally, the present application may determine the abnormality level of each rule feature based on the corresponding relationship between the precision rate and the abnormality level.

[0157] S505: Merge the feature rules of the same abnormality level to obtain rule description information in the original scoring rule set.

[0158] In some possible embodiments, feature rules of the same abnormality level may be combined to obtain description information in the original scoring rule set.

[0159] For example, assuming that among the multiple rule features corresponding to an original scoring topic, the anomaly levels of rule feature 1 (f1>t1andf2>t2) and rule feature 2 (f1>t1and f2<=t2) are both high risks, they can be merged into (f1>t1andf2>t2)or(f1>t1and f2<=t2), so that (f1>t1and f2>t2)or(f1>t1and f2<=t2) can become a rule description information corresponding to the original scoring topic.

[0160] Optionally, assuming that among multiple rule features corresponding to an original scoring theme, a certain rule feature is of medium risk, the rule feature can be determined as a rule description information corresponding to the original scoring theme.

[0161] Similarly, different rule description information corresponding to the original scoring topic can be obtained.

[0162] S507: Determine original scoring data of the rule description information in the original scoring rule set based on the abnormality level.

[0163] In some possible embodiments, the correspondence between the score and the abnormal level corresponding to each original scoring theme can be obtained. Optionally, different original scoring themes have different correspondences between the score and the abnormal level. For example, the high risk corresponding score of some original scoring themes is 35, and the high risk corresponding score of some original scoring themes is 25.

[0164] In the present application, the correspondence between the scoring score and the abnormality level can be set in advance, for example, a scoring score of 30 corresponds to "high risk", a scoring score of 25 corresponds to "higher risk", a scoring score of 15 corresponds to "medium risk", a scoring score of 10 corresponds to "lower risk", and a scoring score of 5 corresponds to "low risk".

[0165] Optionally, the original scoring score of each rule description information may be determined according to the abnormality level of each rule description information corresponding to each original scoring topic from the corresponding relationship between the scoring score corresponding to the original scoring topic and the abnormality level.

[0166] In this way, all rule description information corresponding to each original scoring topic and the original scoring score corresponding to each rule description information can be sorted to obtain an original scoring rule set corresponding to each original scoring topic.

[0167] S305: construct an original scoring model based on at least one original scoring rule set.

[0168] In the embodiment of the present application, an original scoring model can be constructed according to each original scoring rule set.

[0169] In an embodiment of the present application, each original scoring rule set may include not only all rule description information and the original scoring score corresponding to each rule description information, but also the abnormality level corresponding to each rule description information, the rule sequence number corresponding to each rule description information, and the original scoring theme name.

[0170] The following table shows an original scoring model including the original scoring subject "integer amount transaction" and the original scoring subject "counterparty". The following original scoring model is an example and does not limit this application.

[0171] Table: Original scoring model

[0172]

[0173]

[0174]

[0175] In an embodiment of the present application, the client can use the original original scoring model and training sample set to score each training object in the training object set, and obtain the object scoring data of each training object under the original scoring rule set contained in the original scoring model, thereby obtaining a first object scoring set, which can also be considered as an original scoring distribution data set.

[0176] Specifically, the client can substitute the feature set of each training object in the training object set into the original scoring model to obtain the object scoring data of each training object. For example, assuming that training object A is substituted into the original scoring rule set under the original scoring topic "Integer Amount Transaction" and the score is 45, and substituted into the original scoring rule set under the original scoring topic "Counterparty" and the score is 45, then the object scoring data of training object A is the sum of the two, 90.

[0177] S205, obtaining a newly added scoring rule set corresponding to the newly added scoring topic; the newly added scoring rule set at least includes rule description information and preset scoring data.

[0178] In an embodiment of the present application, the client can obtain a newly added scoring rule set corresponding to the newly added scoring topic. Optionally, the features involved in the newly added scoring topic may be some of the multiple features included in the third feature set. For example, assuming that the newly added scoring topic is "transaction note sensitive words", the features involved in the newly added scoring topic "transaction note sensitive words" may be the proportion of sensitive words involved in the transaction notes, the total amount of the transaction day, and the average amount of the transaction day.

[0179] Optionally, the features involved in the newly added scoring topic may be some of the multiple features included in the third feature set and a new feature. For example, assuming that the newly added scoring topic is "intersection of payer and payee users", the features involved in the newly added scoring topic "intersection of payer and payee users" may be the intersection ratio of payer and payee users and the average amount of transaction days.

[0180] In the embodiment of the present application, the newly added scoring rule set includes at least rule description information and preset scoring data. Optionally, the newly added scoring rule set may include all rule description information, the preset scoring score corresponding to each rule description information, the abnormal level corresponding to each rule description information, the rule sequence number corresponding to each rule description information, and the original scoring topic name. In this way, the following Table 2 can be obtained.

[0181] Table 2: New scoring rule set

[0182]

[0183] In the embodiment of the present application, the scoring score of each rule description information included in the newly added scoring rule set is uncertain, but is randomly set, and needs to be adjusted later to avoid a large change in the scoring scale and affect the stability of the existing risk control strategy.

[0184] S207, constructing a preset scoring model based on the original scoring model and the newly added scoring rule set.

[0185] In the embodiment of the present application, the client can build a preset scoring model based on the original scoring model and the newly added scoring rule set, that is, the preset scoring model includes three scoring rule sets.

[0186] S209, determining a second object scoring set based on the preset scoring model and the training sample set; the second object scoring set includes object scoring data of each training object under the preset scoring model.

[0187] In an embodiment of the present application, the client can use a preset scoring model and a training sample set to score each training object in the training object set, obtain object scoring data for each training object under multiple scoring rule sets included in the preset scoring model, and thus obtain a second object scoring set.

[0188] S211, iteratively processing the preset scoring data according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain target scoring data.

[0189] In the embodiment of the present application, the client may iteratively process the preset scoring data according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain the target scoring data.

[0190] See also Figure 7 , which is a flow chart of a method for determining scoring data of each target provided by an embodiment of the present invention, such as Figure 7 As shown, including:

[0191] S701, constructing a comprehensive evaluation function based on a function of a first object scoring set, a function of a second object scoring set, a function of preset scoring data, and a function of a preset scoring model;

[0192] In an embodiment of the present application, the client constructs a score distribution change function based on the function of the first object score set and the function of the second object score set, constructs a model performance evaluation function based on the function of the preset score data and the function of the preset score model, and constructs a comprehensive evaluation function based on the score distribution change function and the model performance evaluation function.

[0193] f(w)=-KL(s0,s new )+λ·Perf(M * (w))......Formula (1)

[0194]

[0195] Where, in formula (1) and formula (2), f(w) is the comprehensive evaluation function; -KL(s0,s new ) is the score distribution change function; Perf(M * (w)) is the model performance evaluation function. s0 is the function of the first object score set; s new is the function of the second object score set, M * is a function of the preset scoring model; w is a function of the preset scoring data; is the indicator function of each rule description information included in the newly added scoring rule set; λ is the weighing factor.

[0196] In the embodiment of the present application, KL(s0,s new) is used to measure the score change, specifically, to measure the distribution similarity between the first object score set and the second object score set. In the embodiment of the present application, in order to reduce the impact on the stability of the existing risk control strategy after adding a scoring rule set, -KL(s0,S new ) can be as small as possible.

[0197] In the present application embodiment, Perf(M * (w)) is used to evaluate the performance of the new scoring model obtained based on the original scoring model. In order to make the performance of the new scoring model feasible, Perf(M * The value of (w)) can be as large as possible. Optionally, the performance of the new scoring model can be judged using AUC values ​​and KS curves.

[0198] S703: Determine first scoring data corresponding to K newly added scoring rule sets by using a preset search rule; K is an integer greater than or equal to 1.

[0199] It can be known from the above that the scoring score of each rule description information included in the newly added scoring rule set is uncertain, but is randomly set. Therefore, the scoring score of each rule description information included in the newly added scoring rule set can be obtained by searching, and then the new scoring model and the second scoring data set under the new scoring model can be determined by the scoring scores obtained by searching.

[0200] Based on this, the first scoring data corresponding to K newly added scoring rule sets may be determined using a preset search rule, where K is an integer greater than or equal to 1.

[0201] In the embodiment of the present application, the scoring score can be searched using the TPE algorithm.

[0202] Specifically, we can first define the prior distribution of w, such as the uniform distribution w~U(0,10), and select K initial points w from it. (1) ,w (2) ,…,w (k) , that is, the first scoring data corresponding to the K newly added scoring rule sets. Each first scoring data includes the first scoring data corresponding to the newly added scoring rule set.

[0203] S705: Determine K+1 newly added scoring rule sets based on the K first scoring data and the preset scoring data and the rule description information.

[0204] Optionally, K+1 newly added scoring rule sets may be determined based on the K first scoring data and the preset scoring data and the rule description information. Table (2) lists the newly added scoring rule sets composed of the preset scoring data and the rule description information.

[0205] S707, construct K+1 preset scoring models using the original scoring model and the K+1 newly added scoring rule sets.

[0206] In the embodiment of the present application, K+1 preset scoring models can be constructed by using the original scoring model and K+1 newly added scoring rule sets.

[0207] S709: Determine K+1 second object scoring sets based on the K+1 preset scoring models and the training sample sets.

[0208] In an embodiment of the present application, the client can use K+1 preset scoring models and training sample sets to score each training object in the training object set respectively, and obtain the object scoring data of each training object under the multiple scoring rule sets included in each preset scoring model, so as to obtain K+1 second object scoring sets.

[0209] S711, input the first object scoring set, K+1 second object scoring sets, K first scoring data, preset scoring data and K+1 preset scoring models into a comprehensive evaluation function to obtain K+1 comprehensive evaluation data.

[0210] Specifically, the first object scoring set, a second object scoring set, the first scoring data corresponding to the second object scoring set, or the preset scoring score and the preset scoring model can be input into the comprehensive evaluation function to obtain the comprehensive evaluation data corresponding to the second object scoring set. In this way, K+1 comprehensive evaluation data can be obtained.

[0211] S713, using K+1 comprehensive evaluation data, K first scoring data and preset scoring data to iterate the scoring data to obtain target scoring data.

[0212] In the embodiment of the present application, the K first scoring data and the preset scoring data can be classified into the first data class and the second data class according to the K+1 comprehensive evaluation data and the evaluation threshold, and the first probability density function and the second probability density function corresponding to the first data set and the second data set can be determined respectively. Subsequently, the acquisition function can be determined based on the first probability density function and the second probability density function, and the second scoring data can be determined using the prime number acquisition function.

[0213] Optionally, a preset scoring model and a second object scoring set corresponding to the second scoring data can be determined, and the first object scoring set, the second scoring data, the preset scoring model and the second object scoring set corresponding to the second scoring data can be input into the comprehensive evaluation function to obtain the comprehensive evaluation data corresponding to the second scoring data, and the scoring data can be iterated using K+1 comprehensive evaluation data, K first scoring data, the preset scoring data, the second scoring data, and the comprehensive evaluation data corresponding to the second scoring data until the iteration termination condition is met, and the target scoring data can be obtained from the preset scoring scores, the second scoring scores, the second scoring scores...

[0214] S213, determining an updated new subject rule set based on the target scoring data; the updated new subject rule set is used to construct a target scoring model together with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

[0215] In an embodiment of the present application, the client can use the target scoring data to update the preset scoring data, obtain an updated new scoring rule set, and construct a target scoring model based on the updated new scoring rule set and the original scoring model.

[0216] In some possible embodiments, in order to enable the target scoring model to meet the stability of the existing risk control strategy, the test level can be used to verify the changes in the scoring distribution of the two.

[0217] See also Figure 8 , which is a flow chart of a method for determining model verification provided by an embodiment of the present invention, such as Figure 8 As shown, including:

[0218] S801, obtaining a test sample set; the training sample set includes a test object set and a second feature set; the second feature data set includes multiple features and feature data of each test object in the test object set under each feature.

[0219] In an embodiment of the present application, the client may obtain a test sample set. Optionally, the test sample set may include a test object set and a second feature set. Optionally, the test sample set may be a training sample set. Optionally, the test sample set may include a training sample set.

[0220] In the embodiment of the present application, the test object may be any object that carries features and can be identified or processed using the features carried.

[0221] In some possible embodiments, the test object set may be a collection of some merchants. Therefore, the characteristics of the merchant may be its own attribute characteristics, such as merchant type, merchant years, merchant location, etc. Optionally, the merchant characteristics may be transaction characteristics associated with customers, such as the proportion of general-entire-hundred transactions, the number of general-entire-hundred transactions, the amount of general-entire-hundred transactions, the proportion of counterparties with abnormal labels, the number of counterparties, the proportion of male counterparties, the proportion of female counterparties, the daily repurchase rate, the proportion of sensitive information in transaction notes, the total amount of the transaction day, the average amount of the transaction day, and the cross-ratio of paying users and receiving users, etc.

[0222] In an embodiment of the present application, each test object in the test object set carries an object label, and the object label is used to indicate whether the test object is an abnormal object or a non-abnormal object. In some possible embodiments, the object label can be represented by the elements "1" and "0". Among them, the object label "1" can represent the test object as an abnormal object or a suspicious object, and the object label "0" can represent the test object as a non-abnormal object or a non-suspicious object. This application does not limit the specific representation of the object label.

[0223] In an embodiment of the present application, the second feature set may include multiple features and feature data of each test object in the test object set under each feature. Optionally, the multiple features included in the first feature set may be multiple attribute features. Optionally, the multiple features included in the first feature set may be multiple transaction features. Optionally, the multiple features included in the first feature set may be a combination of attribute features and transaction features.

[0224] In the embodiment of the present application, the multiple features included in the second feature set are the multiple features included in the first feature set.

[0225] S803, determining a third object scoring set based on the original scoring model and the test sample set; the third object scoring set includes object scoring data of each test object under the original scoring rule set included in the original scoring model.

[0226] Optionally, the client can substitute the feature set of each test object in the test sample set into the original scoring model to obtain the object scoring data of each test object, i.e., the third object scoring set. For example, assuming that the test object B is substituted into the original scoring rule set under the original scoring subject "Integer Amount Transaction" and obtains a score of 30, and is substituted into the original scoring rule set under the original scoring subject "Counterparty" and obtains a score of 45, then the object scoring data of the test object B is the sum of the two, 75.

[0227] S805, determining a fourth object scoring set based on the target scoring model and the test sample set; the fourth object scoring set includes object scoring data of each test object under the target scoring model.

[0228] Optionally, the client can substitute the feature set of each test object in the test sample set into the target scoring model to obtain the object scoring data of each test object, i.e., the fourth object scoring set. For example, assuming that the test object B is substituted into the original scoring rule set under the original scoring subject "Integer Amount Transaction", the score is 30, substituted into the original scoring rule set under the original scoring subject "Counterparty", the score is 45, and substituted into the newly added scoring rule set under the newly added scoring subject "Intersection of Paying User and Payee User", the score is 20, then the object scoring data of the test object B is the sum of the three, 95.

[0229] S807: Input the third object score set and the fourth object score set into a score distribution change function to determine score distribution change data.

[0230] Optionally, the client may input the third object scoring set and the fourth object scoring set into the scoring distribution change function -KL(s0,s new ) to obtain the score distribution change data.

[0231] In the embodiment of the present application, each object scoring data in the fourth object scoring set can be mapped as shown in formula (3) to obtain an updated fourth object scoring set, and each object scoring data in the updated fourth object scoring set is within a preset data range. The third object scoring set and the updated fourth object scoring set are input into the scoring distribution change function to determine the scoring distribution change data.

[0232]

[0233] Among them, s new Indicates the fourth object score set; max(s new ) is the largest object score data in the fourth object score set, s * Indicates the fourth object score set mapped to the range [0,100].

[0234] Through the above method, the requirement that the rating scale will not change significantly can be met, so that users can more easily understand the meaning of the rating scores.

[0235] S809: When the score distribution change data indicates that the target score model meets the application condition, the target score model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

[0236] In an embodiment of the present application, if the score distribution change data indicates that the target scoring model meets the application conditions, such as being less than a certain threshold, the target scoring model can be tried, that is, the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

[0237] In summary, the embodiment of the present application forms a new scoring rule set by adding an iterative scoring score to the original scoring model, thereby forming a final target scoring model, which not only simplifies the model formation process, but also ensures that it can continue to maintain the stability of the existing risk control strategy, thereby reducing management costs and improving management efficiency. At the same time, the mapping process can meet the requirement that the scoring scale will not change significantly, so that users can more easily understand the meaning of the scoring score.

[0238] See also Fig. 9 , which is a schematic diagram of the structure of a feature information-based object scoring device provided in an embodiment of the present invention. The device has the function of implementing the feature information-based object scoring method in the above method embodiment. The function can be implemented by hardware or by hardware executing corresponding software.

[0239] like Fig. 9 As shown, the object scoring device 900 based on feature information may include:

[0240] The sample acquisition module 901 is used to acquire a training sample set; the training sample set includes a training object set and a first feature set; the first feature data set includes multiple features and feature data of each training object in the training object set under each feature;

[0241] A first scoring determination module 902 is used to determine a first object scoring set based on the original scoring model and the training sample set; the first object scoring set includes object scoring data of each training object under the original scoring rule set included in the original scoring model;

[0242] A rule set acquisition module 903 is used to acquire a newly added scoring rule set corresponding to a newly added scoring topic; the newly added scoring rule set at least includes rule description information and preset scoring data;

[0243] A model building module 904 is used to build a preset scoring model based on the original scoring model and the newly added scoring rule set;

[0244] A second scoring determination module 905 is used to determine a second object scoring set based on a preset scoring model and a training sample set; the second object scoring set includes object scoring data of each training object under the preset scoring model;

[0245] A target score determination module 906 is used to iteratively process the preset score data according to the first object score set, the second object score set, the initial score data and the preset score model to obtain target score data;

[0246] The rule set determination module 907 is used to determine the updated new theme rule set based on the target scoring data; the updated new theme rule set is used to construct a target scoring model with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

[0247] In some possible embodiments, the target score determination module is used to:

[0248] Constructing a comprehensive evaluation function based on the function of the first object scoring set, the function of the second object scoring set, the function of the preset scoring data, and the function of the preset scoring model;

[0249] Determine the first scoring data corresponding to K newly added scoring rule sets by using a preset search rule; K is an integer greater than or equal to 1;

[0250] Determine K+1 newly added scoring rule sets based on the K first scoring data and the preset scoring data and the rule description information;

[0251] Use the original scoring model and K+1 newly added scoring rule sets to build K+1 preset scoring models;

[0252] Determine K+1 second object scoring sets based on K+1 preset scoring models and training sample sets;

[0253] Inputting the first object scoring set, K+1 second object scoring sets, K first scoring data, preset scoring data and K+1 preset scoring models into the comprehensive evaluation function to obtain K+1 comprehensive evaluation data;

[0254] The scoring data are iterated using K+1 comprehensive evaluation data, K first scoring data and preset scoring data to obtain target scoring data.

[0255] In some possible embodiments, the target score determination module is used to:

[0256] Classifying the K first scoring data and the preset scoring data into a first data category and a second data category based on the K+1 comprehensive evaluation data;

[0257] Determine a first probability density function and a second probability density function corresponding to the first data set and the second data set, respectively;

[0258] determining an acquisition function based on the first probability density function and the second probability density function;

[0259] Determining second scoring data using a prime number acquisition function;

[0260] Determine a preset scoring model and a second object scoring set corresponding to the second scoring data;

[0261] Inputting the first object scoring set, the second scoring data, the preset scoring model corresponding to the second scoring data, and the second object scoring set into the comprehensive evaluation function to obtain the comprehensive evaluation data corresponding to the second scoring data;

[0262] The scoring data are iterated using K+1 comprehensive evaluation data, K first scoring data, preset scoring data, second scoring data, and comprehensive evaluation data corresponding to the second scoring data to obtain target scoring data.

[0263] In some possible embodiments, a comprehensive evaluation function is constructed based on a function of the first object scoring set, a function of the second object scoring set, a function of preset scoring data, and a function of a preset scoring model, including:

[0264] constructing a score distribution change function based on the function of the first object score set and the function of the second object score set;

[0265] Constructing a model performance evaluation function based on a function of preset scoring data and a function of a preset scoring model;

[0266] A comprehensive evaluation function is constructed based on the score distribution change function and the model performance evaluation function.

[0267] In some possible embodiments, the rule set determination module is used to:

[0268] Using the target scoring data to update the preset scoring data, obtaining an updated set of newly added scoring rules;

[0269] Build a target scoring model based on the updated new scoring rule set and the original scoring model.

[0270] In some possible embodiments, the device further includes a testing module, wherein the testing module is configured to:

[0271] Acquire a test sample set; the training sample set includes a test object set and a second feature set; the second feature data set includes a plurality of features and feature data of each test object in the test object set under each feature;

[0272] Determine a third object scoring set based on the original scoring model and the test sample set; the third object scoring set includes object scoring data of each test object under the original scoring rule set included in the original scoring model;

[0273] Determine a fourth object scoring set based on the target scoring model and the test sample set; the fourth object scoring set includes object scoring data of each test object under the target scoring model;

[0274] Inputting the third object score set and the fourth object score set into the score distribution change function to determine the score distribution change data;

[0275] When the rating distribution change data indicates that the target rating model meets the application conditions, the target rating model is used to rate the object to be rated based on the feature set corresponding to the object to be rated.

[0276] In some possible embodiments, the test module is used to:

[0277] Mapping each object scoring data in the fourth object scoring set to obtain an updated fourth object scoring set; each object scoring data in the updated fourth object scoring set is within a preset data range;

[0278] The third object score set and the updated fourth object score set are input into the score distribution change function to determine the score distribution change data.

[0279] In some possible embodiments, the apparatus further includes an original scoring model training module, which is used to:

[0280] Acquire an original sample set; the original sample set includes an original object set and a third feature set; each original object in the original object set carries an object label, and the object label is used to indicate whether the original object is an abnormal object or a non-abnormal object; the third feature set includes multiple features and feature data of each original object under each feature; the multiple features included in the third feature data set are subsets of the multiple features included in the first feature data set;

[0281] Determine at least one original scoring rule set corresponding to at least one original scoring theme based on the original sample set; each original scoring rule set includes at least rule description information and original scoring data;

[0282] An original scoring model is constructed based on at least one original scoring rule set.

[0283] In some possible embodiments, the original scoring model training module is used to:

[0284] Determine a fourth feature set corresponding to each original rating topic from the third feature set; the fourth feature set includes at least one feature and feature data of each original object under each feature in the at least one feature; at least one feature in the fourth feature set is a subset of multiple features included in the third feature data set, and different fourth feature sets contain different features;

[0285] Building a decision tree model corresponding to each original scoring topic based on the original object set and the fourth feature set corresponding to each original scoring topic;

[0286] An original scoring rule set corresponding to each original scoring topic is determined based on the decision tree model corresponding to each original scoring topic.

[0287] In some possible embodiments, the original scoring model training module is used to:

[0288] Determine multiple feature rules based on the nodes on the decision tree model corresponding to each original scoring topic;

[0289] Determine an abnormality level of each feature rule in a plurality of feature rules based on an object label carried by each original object set and a feature corresponding to each original object;

[0290] The feature rules with the same abnormal level are merged to obtain the rule description information in the original scoring rule set;

[0291] The original scoring data of the rule description information in the original scoring rule set is determined based on the anomaly level.

[0292] In some possible embodiments, the original scoring model training module is used to:

[0293] Determine the accuracy of each feature rule in the plurality of feature rules based on the object label carried by each original object set and the feature corresponding to each original object;

[0294] The abnormality level of each feature rule is determined according to the accuracy of each feature rule. It should be noted that the device provided in the above embodiment, when realizing its function, only uses the division of the above functional modules as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0295] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program with a predetermined function, and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0296] An embodiment of the present invention provides an electronic device, which includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement an object scoring method based on feature information as provided in the above method embodiment.

[0297] The memory can be used to store software programs and modules, and the processor executes various functional applications and clustering results by running the software programs and modules stored in the memory. The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, application programs required for functions, etc.; the data storage area may store data created according to the use of the device, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. Accordingly, the memory may also include a memory controller to provide the processor with access to the memory.

[0298] The method embodiments provided in the embodiments of the present invention can be executed in a computer terminal, a server or a similar computing device. Taking running on a server as an example, Fig. 9 is a hardware structure block diagram of a server running an object scoring method based on feature information provided by an embodiment of the present invention, such as Fig. 9 As shown, the server 2000 may have relatively large differences due to different configurations or performances, and may include one or more central processing units (CPU) 2010 (the processor 2010 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 2030 for storing data, and one or more storage media 2020 (such as one or more mass storage devices) for storing application programs 2023 or data 2022. Among them, the memory 2030 and the storage medium 2020 can be short-term storage or permanent storage. The program stored in the storage medium 2020 may include one or more modules, each of which may include a series of instruction operations on the server. Furthermore, the central processing unit 2010 can be configured to communicate with the storage medium 2020 and execute a series of instruction operations in the storage medium 220 on the server 2000. The server 2000 may also include one or more power supplies 2060, one or more wired or wireless network interfaces 2050, one or more input and output interfaces 2040, and / or one or more operating systems 2021, such as Windows ServerTM, MacOS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0299] The input / output interface 2040 may be used to receive or send data via a network. The specific example of the network may include a wireless network provided by a communication provider of the server 2000. In one example, the input / output interface 2040 includes a network adapter (Network Interface Controller, NIC), which may be connected to other network devices via a base station so as to communicate with the Internet. In one example, the input / output interface 2040 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0300] It can be understood by those skilled in the art that Fig. 9 The structure shown is only for illustration and does not limit the structure of the above electronic device. Fig. 9 More or fewer components as shown, or with Fig. 9 Different configurations shown.

[0301] An embodiment of the present invention also provides a computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to implementing an object scoring method based on feature information. The at least one instruction or the at least one program is loaded and executed by the processor to implement the object scoring method based on feature information provided by the above method embodiment.

[0302] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0303] It should be noted that the sequence of the embodiments of the present invention described above is for description only and does not represent the advantages and disadvantages of the embodiments. The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0304] The embodiment of the present invention also provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the above-mentioned object scoring method based on feature information.

[0305] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0306] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0307] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An object scoring method based on feature information, characterized in that: The method comprises: Acquire a training sample set; the training sample set includes a training object set and a first feature set; the first feature data set includes a plurality of features and feature data of each training object in the training object set under each feature; Determine a first object scoring set based on the original scoring model and the training sample set; the first object scoring set includes object scoring data of each training object under the original scoring rule set included in the original scoring model; Acquire a newly added scoring rule set corresponding to the newly added scoring topic; the newly added scoring rule set at least includes rule description information and preset scoring data; Constructing a preset scoring model based on the original scoring model and the newly added scoring rule set; Determine a second object scoring set based on the preset scoring model and the training sample set; the second object scoring set includes the object scoring data of each training object under the preset scoring model; Iteratively process the preset scoring data according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain target scoring data; An updated set of newly added subject rules is determined based on the target scoring data; the updated set of newly added subject rules is used to construct a target scoring model together with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

2. The object scoring method based on feature information according to claim 1, characterized in that: The iterative processing of the initial scoring data according to the first object scoring set, the second object scoring set, the initial scoring data and the preset scoring model to obtain target scoring data includes: constructing a comprehensive evaluation function based on the function of the first object scoring set, the function of the second object scoring set, the function of the preset scoring data, and the function of the preset scoring model; Determine the first scoring data corresponding to K newly added scoring rule sets by using a preset search rule; wherein K is an integer greater than or equal to 1; Determine K+1 newly added scoring rule sets based on the K first scoring data and the preset scoring data and the rule description information respectively; Constructing the K+1 preset scoring models using the original scoring model and the K+1 newly added scoring rule sets respectively; Determine the K+1 second object scoring sets based on the K+1 preset scoring models and the training sample set; Inputting the first object scoring set, the K+1 second object scoring sets, the K first scoring data, the preset scoring data and the K+1 preset scoring model into the comprehensive evaluation function to obtain K+1 comprehensive evaluation data; The K+1 comprehensive evaluation data, the K first scoring data and the preset scoring data are used to iterate the scoring data to obtain the target scoring data.

3. The object scoring method based on feature information according to claim 2, characterized in that: The step of iterating the scoring data using the K+1 comprehensive evaluation data, the K first scoring data and the preset scoring data to obtain the target scoring data includes: Classifying the K first scoring data and the preset scoring data into a first data category and a second data category based on the K+1 comprehensive evaluation data; Determine a first probability density function and a second probability density function corresponding to the first data set and the second data set respectively; determining an acquisition function based on the first probability density function and the second probability density function; Determining second scoring data using a prime number acquisition function; Determine a preset scoring model and a second object scoring set corresponding to the second scoring data; Inputting the first object scoring set, the second scoring data, the preset scoring model corresponding to the second scoring data, and the second object scoring set into the comprehensive evaluation function to obtain the comprehensive evaluation data corresponding to the second scoring data; Scoring data iteration is performed using the K+1 comprehensive evaluation data, the K first scoring data, the preset scoring data, the second scoring data, and the comprehensive evaluation data corresponding to the second scoring data to obtain the target scoring data.

4. The object scoring method based on feature information according to claim 2, characterized in that: The constructing of a comprehensive evaluation function based on the function of the first object scoring set, the function of the second object scoring set, the function of the preset scoring data and the function of the preset scoring model includes: constructing a score distribution change function based on the function of the first object score set and the function of the second object score set; Constructing a model performance evaluation function based on a function of preset scoring data and a function of the preset scoring model; A comprehensive evaluation function is constructed based on the score distribution change function and the model performance evaluation function.

5. The object scoring method based on feature information according to any one of claims 1 to 4, characterized in that: The step of determining an updated set of newly added subject rules based on the target scoring data includes: Using the target scoring data to update the preset scoring data, to obtain an updated newly added scoring rule set; The target scoring model is constructed based on the updated newly added scoring rule set and the original scoring model.

6. The object scoring method based on feature information according to claim 5, characterized in that: The method further comprises: Acquire a test sample set; the training sample set includes a test object set and a second feature set; the second feature data set includes the multiple features and feature data of each test object in the test object set under each feature; Determine a third object scoring set based on the original scoring model and the test sample set; the third object scoring set includes object scoring data of each test object under the original scoring rule set included in the original scoring model; Determine a fourth object scoring set based on the target scoring model and the test sample set; the fourth object scoring set includes object scoring data of each test object under the target scoring model; Inputting the third object score set and the fourth object score set into a score distribution change function to determine score distribution change data; When the score distribution change data indicates that the target scoring model meets the application condition, the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

7. The object scoring method based on feature information according to claim 6, characterized in that: The step of inputting the third object score set and the fourth object score set into a score distribution change function to determine score distribution change data comprises: Mapping each object scoring data in the fourth object scoring set to obtain an updated fourth object scoring set; each object scoring data in the updated fourth object scoring set is within a preset data range; The third object score set and the updated fourth object score set are input into a score distribution change function to determine score distribution change data.

8. The object scoring method based on feature information according to any one of claims 1 to 4, characterized in that: The method further comprises the step of determining the original scoring model, comprising: Acquire an original sample set; the original sample set includes an original object set and a third feature set; each original object in the original object set carries an object label, and the object label is used to indicate whether the original object is an abnormal object or a non-abnormal object; the third feature set includes multiple features and feature data of each original object under each feature; the multiple features included in the third feature data set are subsets of the multiple features included in the first feature data set; Determine at least one original scoring rule set corresponding to at least one original scoring theme based on the original sample set; each original scoring rule set includes at least rule description information and original scoring data; The raw scoring model is constructed based on the at least one raw scoring rule set.

9. The object scoring method based on feature information according to claim 8, characterized in that: The determining, based on the original sample set, at least one original scoring rule set corresponding to at least one original scoring topic comprises: Determining a fourth feature set corresponding to each original rating topic from the third feature set; the fourth feature set includes at least one feature and feature data of each original object under each feature of the at least one feature; at least one feature in the fourth feature set is a subset of multiple features included in the third feature data set, and different fourth feature sets contain different features; Constructing a decision tree model corresponding to each original scoring topic based on the original object set and the fourth feature set corresponding to each original scoring topic; An original scoring rule set corresponding to each original scoring topic is determined based on the decision tree model corresponding to each original scoring topic.

10. The object scoring method based on feature information according to claim 9, characterized in that: The determining the original scoring rule set corresponding to each original scoring topic based on the decision tree model corresponding to each original scoring topic includes: Determine multiple feature rules based on the nodes on the decision tree model corresponding to each original scoring topic; Determine the abnormality level of each feature rule in the plurality of feature rules based on the object label carried by each original object set and the feature corresponding to each original object; Merging the feature rules of the same abnormality level to obtain the rule description information in the original scoring rule set; The original scoring data of the rule description information in the original scoring rule set is determined based on the abnormality level.

11. The object scoring method based on feature information according to claim 10, characterized in that: The determining the abnormality level of each feature rule in the plurality of feature rules based on the object label carried by each original object set includes: Determine the accuracy of each feature rule in the plurality of feature rules based on the object label carried by each original object set and the feature corresponding to each original object; The abnormality level of each feature rule is determined according to the accuracy of each feature rule.

12. An object scoring device based on feature information, characterized in that: The device comprises: A sample acquisition module, used to acquire a training sample set; the training sample set includes a training object set and a first feature set; the first feature data set includes a plurality of features and feature data of each training object in the training object set under each feature; A first scoring determination module is used to determine a first object scoring set based on the original scoring model and the training sample set; the first object scoring set includes object scoring data of each training object under the original scoring rule set included in the original scoring model; A rule set acquisition module, used to acquire a newly added scoring rule set corresponding to a newly added scoring topic; the newly added scoring rule set at least includes rule description information and preset scoring data; A model building module, used to build a preset scoring model based on the original scoring model and the newly added scoring rule set; A second scoring determination module is used to determine a second object scoring set based on the preset scoring model and the training sample set; the second object scoring set includes object scoring data of each training object under the preset scoring model; a target score determination module, configured to iteratively process the preset score data according to the first object score set, the second object score set, the initial score data and the preset score model to obtain target score data; A rule set determination module is used to determine an updated new subject rule set based on the target scoring data; the updated new subject rule set is used to construct a target scoring model with the original scoring model, and the target scoring model is used to score the object to be scored based on the feature set corresponding to the object to be scored.

13. An electronic device, characterized in that: It comprises a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the object scoring method based on feature information as described in any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the object scoring method based on feature information according to any one of claims 1 to 11.

15. A computer program, characterized in that When the computer program is executed by a processor, the object scoring method based on feature information according to any one of claims 1 to 11 is implemented.

Images