CN110163761B - Suspicious item member identification method and device based on image processing - Google Patents
Suspicious item member identification method and device based on image processing Download PDFInfo
- Publication number
- CN110163761B CN110163761B CN201910238911.0A CN201910238911A CN110163761B CN 110163761 B CN110163761 B CN 110163761B CN 201910238911 A CN201910238911 A CN 201910238911A CN 110163761 B CN110163761 B CN 110163761B
- Authority
- CN
- China
- Prior art keywords
- suspicious
- image
- case review
- project
- project members
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Accounting & Taxation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Finance (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Development Economics (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Technology Law (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
技术领域technical field
本申请涉及图像处理技术领域,特别涉及一种基于图像处理的可疑项目成员识别方法。本申请同时涉及一种基于图像处理的可疑项目成员识别装置,一种计算设备,以及一种计算机可读存储介质。The present application relates to the technical field of image processing, in particular to a method for identifying suspicious project members based on image processing. The present application also relates to a suspicious project member identification device based on image processing, a computing device, and a computer-readable storage medium.
背景技术Background technique
随着大众抗风险意识增强,保险成为越来越多人的投资选项。风控是保险业务正常发展的重要环节,成长于互联网环境下的保险风控更为重要,保险风控中的核赔环节是指判断理赔是否符合保险保障条款,是防控骗保的核心手段之一。保险理赔环节一般会需要用户上传理赔材料,现在有很多非法获利组织通过招募志愿者形成骗保团伙,而骗保团伙上传的理赔材料间大多具有某种共性。As the public's awareness of anti-risk increases, insurance has become an investment option for more and more people. Risk control is an important part of the normal development of the insurance business. Risk control is even more important for insurance that grows in the Internet environment. The claim verification link in insurance risk control refers to judging whether the claims are in compliance with the insurance protection clauses, which is the core means of preventing and controlling fraudulent insurance one. Insurance claims generally require users to upload claim materials. Now many illegal profit-making organizations form insurance fraud gangs by recruiting volunteers, and most of the claim materials uploaded by fraud gangs have some commonality.
现有技术中,风控场景的团伙识别方法是基于事实性关系进行判断,事实性关系是指参与保险项目的项目成员之间的资金转账关系、通信关系、通讯录关系、设备关系等,通过判断项目成员之间是否存在事实性关系,达到确定骗保团伙的目的。但是,现在很多骗保团伙成员之间以网络作为联络方式,从而不能被基于事实性关系的团伙识别方案识别,存在一定的缺陷。In the prior art, the group identification method in the risk control scene is judged based on the factual relationship. The factual relationship refers to the fund transfer relationship, communication relationship, address book relationship, equipment relationship, etc. between project members participating in the insurance project. Judging whether there is a factual relationship between project members, to achieve the purpose of identifying fraudulent insurance gangs. However, many fraudulent insurance gang members now use the Internet as a means of communication, so they cannot be identified by gang identification schemes based on factual relationships, and there are certain defects.
发明内容Contents of the invention
有鉴于此,本申请实施例提供了一种基于图像处理的可疑项目成员识别方法,以解决现有技术中存在的技术缺陷。本申请实施例同时提供了一种基于图像处理的可疑项目成员识别装置,一种计算设备,以及一种计算机可读存储介质。In view of this, the embodiment of the present application provides a method for identifying suspicious project members based on image processing, so as to solve the technical defects existing in the prior art. The embodiment of the present application also provides an image processing-based suspicious project member identification device, a computing device, and a computer-readable storage medium.
本申请实施例公开了一种基于图像处理的可疑项目成员识别方法,包括:The embodiment of the present application discloses a method for identifying suspicious project members based on image processing, including:
获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集;Obtain a case review image set composed of case review images submitted by project members participating in the project;
提取所述案件审核图像集中所述案件审核图像的特征向量;Extracting the feature vector of the case review image in the case review image set;
根据所述特征向量确定所述案件审核图像之间的图像相似度;determining the image similarity between the case review images according to the feature vector;
基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类;clustering the case review images based on the similarity weight determined by the image similarity;
根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员。Suspicious project members among the project members are determined according to the clustering results of the clustering.
可选的,所述提取所述案件审核图像集中所述案件审核图像的特征向量,包括:Optionally, the extracting the feature vector of the case review image in the case review image set includes:
将所述案件审核图像输入预先训练好的深度学习模型进行图像向量化处理,输出所述案件审核图像的特征向量。The case review image is input into a pre-trained deep learning model for image vectorization processing, and the feature vector of the case review image is output.
可选的,所述根据所述特征向量确定所述案件审核图像之间的图像相似度,包括:Optionally, the determining the image similarity between the case review images according to the feature vectors includes:
根据所述特征向量利用词频逆文本频率指数算法计算所述案件审核图像之间的图像相似度。The image similarity between the case review images is calculated by using the word frequency inverse text frequency index algorithm according to the feature vector.
可选的,所述根据所述特征向量确定所述案件审核图像之间的图像相似度,包括:Optionally, the determining the image similarity between the case review images according to the feature vectors includes:
基于所述案件审核图像的特征向量之间的向量距离,确定所述案件审核图像之间的图像相似度。An image similarity between the case review images is determined based on a vector distance between feature vectors of the case review images.
可选的,所述基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类,包括:Optionally, the clustering of the case review images based on the similarity weight determined based on the image similarity includes:
构建连通图,将所述案件审核图像与所述连通图中的节点建立一一对应的关系,并将所述案件审核图像之间的图像相似度取对数值作为所述案件审核图像对应节点之间的边权重;Constructing a connected graph, establishing a one-to-one correspondence relationship between the case review image and the nodes in the connected graph, and taking the logarithmic value of the image similarity between the case review images as one of the corresponding nodes of the case review image The edge weight between;
将所述连通图输入聚类模型进行聚类,输出所述案件审核图像的聚类标识。The connectivity graph is input into a clustering model for clustering, and the cluster identification of the case review image is output.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员,包括:Optionally, the determining a suspicious project member among the project members according to the clustering result of the clustering includes:
确定具有相同聚类标识的所述案件审核图像为可疑图像;Determining that the case review images with the same cluster identification are suspicious images;
确定所述可疑图像的提交人为所述项目成员中的可疑项目成员。It is determined that the submitter of the suspicious image is a suspicious project member among the project members.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员步骤执行之后,包括:Optionally, after the step of determining suspicious project members among the project members according to the clustering result of the clustering is executed, the step includes:
确定所述可疑图像中具有同一聚类标识的所述案件审核图像的提交人的集合为可疑团伙。Determining that the set of submitters of the case review image with the same cluster identifier in the suspicious image is a suspicious gang.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员步骤执行之后,还包括:Optionally, after the step of determining suspicious project members among the project members according to the clustering results of the clustering is performed, it further includes:
获取所述可疑项目成员的事实性关系;Obtain the factual relationship of the suspicious project members;
基于所述事实性关系在所述可疑项目成员中确定具有实际事实性关系的项目成员为不可信项目成员;determining, among the suspicious project members, a project member with an actual factual relationship as an untrusted project member based on the factual relationship;
确定所述不可信项目成员中提交具有同一聚类标识的所述案件审核图像的成员集合为可疑团伙。Determining a set of members among the untrustworthy project members who submitted the case review image with the same cluster ID as a suspicious gang.
可选的,所述获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集步骤执行之前,还包括:Optionally, before the step of obtaining the case review image set composed of the case review images submitted by the project members participating in the project is executed, it may further include:
获取所述项目成员的事实性关系;obtain de facto relationships of said project members;
基于所述事实性关系确定具有实际事实性关系的项目成员;determining project members with actual factual relationships based on the factual relationships;
相应的,所述获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集步骤中,获取所述具有实际事实性关系的项目成员的案件审核图像,组成所述案件审核图像集。Correspondingly, in the step of obtaining the case review image set composed of the case review images submitted by the project members participating in the project, the case review images of the project members with actual factual relationship are obtained to form the case review image set.
可选的,所述项目成员的事实性关系,包括下述至少一项:Optionally, the de facto relationship of the project members includes at least one of the following:
所述项目成员之间的资金转账关系、通信关系、通讯录关系、设备关系。Fund transfer relationship, communication relationship, address book relationship, and equipment relationship among project members.
可选的,所述深度学习模型,基于下述任意一种神经网络构建:Optionally, the deep learning model is constructed based on any of the following neural networks:
卷积神经网络、深度神经网络。Convolutional Neural Networks, Deep Neural Networks.
可选的,所述聚类模型,基于下述任意一种算法:Optionally, the clustering model is based on any of the following algorithms:
标签传播算法、最大连通图算法。Label Propagation Algorithm, Maximum Connected Graph Algorithm.
本申请提供一种基于图像处理的可疑项目成员识别装置,包括:This application provides a device for identifying suspicious project members based on image processing, including:
案件审核图像集获取模块,被配置为获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集;The case review image set acquisition module is configured to obtain a case review image set composed of case review images submitted by project members participating in the project;
特征向量提取模块,被配置为提取所述案件审核图像集中所述案件审核图像的特征向量;A feature vector extraction module configured to extract feature vectors of the case review images in the case review image set;
图像相似度确定模块,被配置为根据所述特征向量确定所述案件审核图像之间的图像相似度;An image similarity determination module configured to determine the image similarity between the case review images according to the feature vectors;
聚类模块,被配置为基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类;A clustering module configured to cluster the case review images based on the similarity weight determined by the image similarity;
可疑项目成员确定模块,被配置为根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员。The suspicious project member determining module is configured to determine suspicious project members among the project members according to the clustering result of the clustering.
本申请提供一种计算设备,包括:The application provides a computing device, comprising:
存储器和处理器;memory and processor;
所述存储器用于存储计算机可执行指令,所述处理器用于执行所述计算机可执行指令:The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions:
获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集;Obtain a case review image set composed of case review images submitted by project members participating in the project;
提取所述案件审核图像集中所述案件审核图像的特征向量;Extracting the feature vector of the case review image in the case review image set;
根据所述特征向量确定所述案件审核图像之间的图像相似度;determining the image similarity between the case review images according to the feature vector;
基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类;clustering the case review images based on the similarity weight determined by the image similarity;
根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员。Suspicious project members among the project members are determined according to the clustering results of the clustering.
本申请提供一种计算机可读存储介质,其存储有计算机指令,该指令被处理器执行时实现所述基于图像处理的可疑项目成员识别方法的步骤。The present application provides a computer-readable storage medium, which stores computer instructions, and when the instructions are executed by a processor, the steps of the method for identifying suspicious project members based on image processing are realized.
与现有技术相比,本申请具有如下优点:Compared with the prior art, the present application has the following advantages:
本申请提供一种基于图像处理的可疑项目成员识别方法,包括:获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集;提取所述案件审核图像集中所述案件审核图像的特征向量;根据所述特征向量确定所述案件审核图像之间的图像相似度;基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类;根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员。This application provides a method for identifying suspicious project members based on image processing, including: obtaining a case review image set composed of case review images submitted by project members participating in the project; extracting the feature vector of the case review image in the case review image set ; Determine the image similarity between the case review images according to the feature vector; cluster the case review images based on the similarity weight determined based on the image similarity; determine according to the clustering results of the clustering Suspicious project members among said project members.
本申请提供的基于图像处理的可疑项目成员识别方法,通过对案件审核图像集中的案件审核图像进行图像相似度计算,依据图像相似度计算结果将案件审核图像进行图聚类操作,使得具有相似性的案件审核图像被标记相同标识,从而根据标识来识别出具有团伙骗保可能性的可疑项目成员,降低团伙骗保发生的概率。The suspicious project member identification method based on image processing provided by this application calculates the image similarity of the case review images in the case review image set, and performs graph clustering operations on the case review images according to the image similarity calculation results, so that they have similarity The case review images are marked with the same logo, so as to identify suspicious project members with the possibility of gang fraud according to the logo, and reduce the probability of gang fraud.
附图说明Description of drawings
图1是本申请实施例提供的一种基于图像处理的可疑项目成员识别方法流程图;Fig. 1 is a flow chart of a suspicious project member identification method based on image processing provided by the embodiment of the present application;
图2是本申请实施例提供的一种基于图像处理的可疑项目成员识别过程的示意图;Fig. 2 is a schematic diagram of a suspicious project member identification process based on image processing provided by the embodiment of the present application;
图3是本申请实施例提供的一种基于图像处理的可疑项目成员识别装置的示意图;Fig. 3 is a schematic diagram of a suspicious project member identification device based on image processing provided by an embodiment of the present application;
图4是本申请实施例提供的一种计算设备的结构框图。Fig. 4 is a structural block diagram of a computing device provided by an embodiment of the present application.
具体实施方式Detailed ways
在下面的描述中阐述了很多具体细节以便于充分理解本申请。但是本申请能够以很多不同于在此描述的其它方式来实施,本领域技术人员可以在不违背本申请内涵的情况下做类似推广,因此本申请不受下面公开的具体实施的限制。In the following description, numerous specific details are set forth in order to provide a thorough understanding of the application. However, the present application can be implemented in many other ways different from those described here, and those skilled in the art can make similar promotions without violating the connotation of the present application. Therefore, the present application is not limited by the specific implementation disclosed below.
在本说明书一个或多个实施例中使用的术语是仅仅出于描述特定实施例的目的,而非旨在限制本说明书一个或多个实施例。在本说明书一个或多个实施例和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式,除非上下文清楚地表示其他含义。还应当理解,本说明书一个或多个实施例中使用的术语“和/或”是指并包含一个或多个相关联的列出项目的任何或所有可能组合。Terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only, and are not intended to limit one or more embodiments of this specification. As used in one or more embodiments of this specification and the appended claims, the singular forms "a", "the", and "the" are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and/or" used in one or more embodiments of the present specification refers to and includes any or all possible combinations of one or more associated listed items.
应当理解,尽管在本说明书一个或多个实施例中可能采用术语第一、第二等来描述各种信息,但这些信息不应限于这些术语。这些术语仅用来将同一类型的信息彼此区分开。例如,在不脱离本说明书一个或多个实施例范围的情况下,第一也可以被称为第二,类似地,第二也可以被称为第一。取决于语境,如在此所使用的词语“如果”可以被解释成为“在……时”或“当……时”或“响应于确定”。It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present specification, the information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, the first may also be referred to as the second, and similarly, the second may also be referred to as the first without departing from the scope of one or more embodiments of the present specification. Depending on the context, the word "if" as used herein may be interpreted as "at" or "when" or "in response to a determination."
本申请提供一种基于图像处理的可疑项目成员识别方法,本申请还提供一种基于图像处理的可疑项目成员识别装置,一种计算设备,以及一种计算机可读存储介质。以下分别结合本申请提供的实施例的附图逐一进行详细说明,并且对方法的各个步骤进行说明。The present application provides a method for identifying suspicious project members based on image processing. The present application also provides a device for identifying suspicious project members based on image processing, a computing device, and a computer-readable storage medium. The following describes in detail one by one with reference to the drawings of the embodiments provided in the present application, and describes each step of the method.
本申请提供的一种基于图像处理的可疑项目成员识别方法实施例如下:An embodiment of a suspicious project member identification method based on image processing provided by this application is as follows:
参照附图1,其示出了本实施例提供的一种基于图像处理的可疑项目成员识别方法流程图;参照附图2,其示出了本实施例提供的一种基于图像处理的可疑项目成员识别过程的示意图。Referring to accompanying drawing 1, it shows the flow chart of a method for identifying suspicious project members based on image processing provided by this embodiment; referring to accompanying drawing 2, it shows a kind of suspicious project member identification method based on image processing provided by this embodiment Schematic illustration of the member identification process.
步骤S102,获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集。Step S102, acquiring a case review image set composed of case review images submitted by project members participating in the project.
保险理赔环节一般会需要用户上传理赔材料,如实物图片,以车险为例,保险公司会要求用户提交受损车辆的外观图片,来判断用户提出的理赔申请是否符合保险保障条款。所述参与项目的项目成员即参与保险项目的用户,案件审核图像即参与保险项目的用户提交的理赔材料中的实物图片,如受损车辆的外观图片、交通事故现场图片等。实际应用中,骗保团伙上传的理赔图片在一定程度上具有相似性,因此可以通过分析理赔图片来识别骗保团伙,从而降低保险理赔环节中骗保团伙骗保情况发生的概率。In the process of insurance claims, users generally need to upload claim materials, such as pictures of real objects. Taking auto insurance as an example, insurance companies will require users to submit pictures of the appearance of damaged vehicles to determine whether the claims submitted by users meet the insurance protection terms. The project members participating in the project are the users participating in the insurance project, and the case review images are the physical pictures in the claim settlement materials submitted by the users participating in the insurance project, such as the appearance picture of the damaged vehicle, the scene picture of the traffic accident, etc. In practical applications, the claims pictures uploaded by fraudulent insurance gangs are similar to a certain extent, so the fraudulent insurance gangs can be identified by analyzing the claims pictures, thereby reducing the probability of insurance fraud by fraudulent insurance gangs in the insurance claim process.
本实施例通过对所述案件审核图像集中的案件审核图像进行图像识别,获得所述案件审核图像集中所有案件审核图像的广义相似关系,该相似关系表现为相似的案件审核图像归属于同一社区即被聚为同一类,从而通过追溯属于同一社区的案件审核图像的提交人来识别可疑项目成员,进一步,还可以在识别出可疑项目成员的基础上,根据可疑项目成员归属的社区来圈定可疑团伙,防止骗保情况的发生。In this embodiment, by performing image recognition on the case review images in the case review image set, the generalized similarity relationship of all the case review images in the case review image set is obtained. Be clustered into the same category, so as to identify suspicious project members by tracing the submitters of the case review images belonging to the same community. Further, on the basis of identifying suspicious project members, it is also possible to delineate suspicious groups according to the community to which the suspicious project members belong , to prevent the occurrence of insurance fraud.
出于减少图像识别的计算量的目的,可以在进行图像识别之前缩小图像识别对象的范围,本申请实施例提供的一种优选实施方式中,在所述获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集之前,通过下述实现方式来缩小图像识别对象的范围:For the purpose of reducing the amount of calculation for image recognition, the range of image recognition objects can be narrowed before image recognition. In a preferred implementation mode provided by the embodiment of this application, in the acquisition of the case review submitted by the project members participating in the project Before the case review image set composed of images, the scope of image recognition objects is narrowed by the following implementation methods:
获取所述项目成员的事实性关系;obtain de facto relationships of said project members;
基于所述事实性关系确定具有实际事实性关系的项目成员;determining project members with actual factual relationships based on the factual relationships;
相应的,在获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集的过程中,获取所述具有实际事实性关系的项目成员的案件审核图像,组成所述案件审核图像集。Correspondingly, in the process of obtaining the case review image set composed of the case review images submitted by the project members participating in the project, the case review images of the project members with actual factual relationship are obtained to form the case review image set.
上述提供的实现方式,首先基于是否具有实际事实性关系对所述项目成员进行筛选,然后获取具有实际事实性关系的项目成员提交的案件审核图像组成的所述案件审核图像集,并在组成的所述案件审核图像集的基础上进行图像识别,从而缩小了图像识别范围,同时,识别出可疑项目成员的可能性会更大一些。The implementation method provided above first screens the project members based on whether they have actual factual relationships, and then obtains the case review image set composed of case review images submitted by project members with actual factual relationships, and Image recognition is performed on the basis of the case review image set, thereby narrowing the scope of image recognition, and at the same time, the possibility of identifying suspicious project members is greater.
其中,所述项目成员的事实性关系,包括下述至少一项:所述项目成员之间的资金转账关系、通信关系、通讯录关系、设备关系。Wherein, the factual relationship of the project members includes at least one of the following: fund transfer relationship, communication relationship, address book relationship, and equipment relationship among the project members.
参与骗保团伙的团伙成员在实际应用可能会通过某种媒介进行交易往来和互通信息,具有资金转账交易记录、通过通讯设备通信、对方互为通讯录联系人、多方曾使用同一登录设备等都属于事实性关系,所述具有实际事实性关系是指所述项目成员被检测到确实具有上述事实性关系中的至少一种关系,具有事实性关系的项目成员在被识别出其提交的案件审核图像归属于相同社区的情况下,判断其为可疑项目成员会更加合理。The members of the gang participating in the fraudulent insurance gang may conduct transactions and exchange information through a certain medium in practical applications. They have fund transfer transaction records, communicate through communication equipment, the other party is a contact in the address book, and multiple parties have used the same login device. It belongs to a factual relationship. The actual factual relationship means that the project member is detected to have at least one of the above-mentioned factual relationships, and the project member with a factual relationship is identified when the case submitted by him is reviewed. When the images belong to the same community, it is more reasonable to judge them as members of suspicious projects.
步骤S104,提取所述案件审核图像集中所述案件审核图像的特征向量。Step S104, extracting feature vectors of the case review images in the case review image set.
实际应用中,保险公司在核保过程中的保险理赔环节获取的理赔图片数量比较大,为提高图像处理效率,本申请实施例提供的一种优选实施方式中,所述提取所述案件审核图像集中所述案件审核图像的特征向量,包括:将所述案件审核图像输入预先训练好的深度学习模型进行图像向量化处理,输出所述案件审核图像的特征向量。In practical applications, the number of claim settlement images acquired by insurance companies in the insurance claim settlement process in the underwriting process is relatively large. In order to improve image processing efficiency, in a preferred implementation mode provided by the embodiment of the present application, the extraction of the case review images Aggregating the feature vectors of the case review images includes: inputting the case review images into a pre-trained deep learning model for image vectorization processing, and outputting the feature vectors of the case review images.
在计算案件审核图像的相似度的过程中,需要将图像进行向量化处理,将图像转化成向量的过程基于深度学习模型实现,所述深度学习模型是预先被训练样本训练好的模型,将所述案件审核图像作为输入量,输入所述深度学习模型中,该模型的输出即为所述案件审核图像的特征向量。In the process of calculating the similarity of case review images, the images need to be vectorized, and the process of converting images into vectors is based on a deep learning model. The deep learning model is a model trained by training samples in advance. The above-mentioned case review image is used as an input quantity and input into the deep learning model, and the output of the model is the feature vector of the case review image.
所述深度学习模型,优选基于卷积神经网络(CNN,Convolutional NeuralNetwork)或者深度神经网络(DNN,Deep Neural Network)构建,除此之外,还可以采用其他神经网络,本实施例在此不做限定。The deep learning model is preferably constructed based on a convolutional neural network (CNN, Convolutional NeuralNetwork) or a deep neural network (DNN, Deep Neural Network). In addition, other neural networks can also be used, which is not done in this embodiment. limited.
步骤S106,根据所述特征向量确定所述案件审核图像之间的图像相似度。Step S106, determining the image similarity between the case review images according to the feature vectors.
本申请实施例提供的一种优选实施方式中,所述根据所述特征向量确定所述案件审核图像之间的图像相似度,包括:根据所述特征向量利用词频逆文本频率指数(TF-IDF,Term Frequency-Inverse Document Frequency)算法计算所述案件审核图像之间的图像相似度。In a preferred embodiment provided by the embodiment of the present application, the determining the image similarity between the case review images according to the feature vectors includes: using the term frequency inverse text frequency index (TF-IDF) according to the feature vectors , Term Frequency-Inverse Document Frequency) algorithm calculates the image similarity between described case review images.
所述词频逆文本频率指数算法是一种统计方法,应用对象多用于文本,用以评估一个字词对于一个文件集或一个语料库中的其中一份文件的重要程度,本实施例将所述案件审核图像转化为特征向量后,采用所述词频逆文本频率指数算法进行图像相似度计算。The term frequency inverse text frequency index algorithm is a statistical method, and its application objects are mostly used in texts to evaluate the importance of a word for a document in a document collection or a corpus. In this embodiment, the case After the audit image is converted into a feature vector, the word frequency inverse text frequency index algorithm is used to calculate the image similarity.
本申请实施例提供的一种优选实施方式中,所述根据所述特征向量确定所述案件审核图像之间的图像相似度,包括:基于所述案件审核图像的特征向量之间的向量距离,确定所述案件审核图像之间的图像相似度。In a preferred implementation manner provided by the embodiment of the present application, the determining the image similarity between the case review images according to the feature vectors includes: based on the vector distance between the feature vectors of the case review images, An image similarity between the case review images is determined.
其中,所述案件审核图像的特征向量之间的向量距离,包括下述至少一项:欧式距离、余弦距离;余弦距离的特点是余弦值接近1,夹角趋于0,表明两个向量越相似,图像相似度越高,欧氏距离的特点是欧氏距离越小,两个向量越相似,图像相似度越高。Wherein, the vector distance between the feature vectors of the case review image includes at least one of the following: Euclidean distance and cosine distance; the characteristic of cosine distance is that the cosine value is close to 1, and the included angle tends to 0, indicating that the two vectors are closer Similarity, the higher the image similarity, the characteristic of Euclidean distance is that the smaller the Euclidean distance, the more similar the two vectors are, the higher the image similarity.
步骤S108,基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类。Step S108, clustering the case review images based on the similarity weight determined by the image similarity.
本申请实施例提供的一种优选实施方式中,所述基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类,包括:In a preferred implementation manner provided in the embodiment of the present application, the clustering of the case review images based on the similarity weight determined based on the image similarity includes:
构建连通图,将所述案件审核图像与所述连通图中的节点建立一一对应的关系,并将所述案件审核图像之间的图像相似度取对数值作为所述案件审核图像对应节点之间的边权重;Constructing a connected graph, establishing a one-to-one correspondence relationship between the case review image and the nodes in the connected graph, and taking the logarithmic value of the image similarity between the case review images as one of the corresponding nodes of the case review image The edge weight between;
将所述连通图输入聚类模型进行聚类,输出所述案件审核图像的聚类标识。The connectivity graph is input into a clustering model for clustering, and the cluster identification of the case review image is output.
优选的,所述聚类模型,优选基于标签传播算法(LPA,Label PropagationAlgorithm)或者最大连通图算法实现,除此之外,还可以采用其他算法,本实施例在此不做限定。Preferably, the clustering model is preferably implemented based on the Label Propagation Algorithm (LPA, Label Propagation Algorithm) or the maximum connected graph algorithm. In addition, other algorithms can also be used, which are not limited in this embodiment.
以所述聚类模型基于标签传播算法实现为例,对所述聚类步骤进行说明:Taking the implementation of the clustering model based on the label propagation algorithm as an example, the clustering steps are described:
聚类模型的输入为基于案件审核图像以及图像相似度构建的连通图,具体在构建连通图的过程中,将连通图中包含的节点与所述案件审核图像建立一一对应关系,每个节点分别包含各自对应案件审核图像的图像编号,并且,将上述求得的所述案件审核图像之间的图像相似度取对数值作为所述案件审核图像对应节点之间的边权重,完成所述聚类模型所需连通图的构建。The input of the clustering model is a connected graph constructed based on the case review image and image similarity. Specifically, in the process of constructing the connected graph, a one-to-one correspondence is established between the nodes contained in the connected graph and the case review image. Each node Include the image numbers of the corresponding case review images respectively, and take the logarithmic value of the image similarity between the case review images obtained above as the edge weight between the corresponding nodes of the case review images, and complete the aggregation Construction of connectivity graphs required for class models.
所述聚类模型基于标签传播算法,标签传播算法的应用场景为:社区发现,传统意义上的社区指的是网络中的一组节点间具有较大的相似性,从而形成的一种内部连接紧密,而外部稀疏的群体结构,对给定的网络图寻找其社区结构的过程称为社区发现,大体上看,社区发现的过程就是一种聚类的过程;标签传播算法的基本思想是:将一个节点的邻居节点的标签中数量最多的标签作为该节点自身的标签,具体为给每个节点添加标签以代表它所属的社区,并通过标签的传播形成同一标签的社区结构。The clustering model is based on the label propagation algorithm. The application scenario of the label propagation algorithm is: community discovery. A community in the traditional sense refers to a group of nodes in the network that have a relatively large similarity, thereby forming an internal connection The process of finding the community structure for a given network graph is called community discovery. Generally speaking, the process of community discovery is a clustering process; the basic idea of the label propagation algorithm is: The label with the largest number among the labels of a node's neighbor nodes is used as the label of the node itself, specifically adding a label to each node to represent the community it belongs to, and forming a community structure of the same label through label propagation.
标签传播算法的传播过程可以概括为:The propagation process of the label propagation algorithm can be summarized as:
1)初始时,给每个节点一个唯一的标签;1) Initially, give each node a unique label;
2)每个节点使用其邻居节点的标签中最多的标签来更新自身的标签;2) Each node updates its own label with the most label among the labels of its neighbor nodes;
3)反复执行步骤2),直到每个节点的标签都不再发生变化为止。3) Repeat step 2) until the label of each node does not change.
将所述连通图输入所述聚类模型运行标签传播算法进行聚类,所述案件审核图像被划分社区,输出划分后的代表每个社区的标签,该标签即是指所述案件审核图像的聚类标识。Input the connected graph into the clustering model and run the label propagation algorithm for clustering, the case review image is divided into communities, and the divided label representing each community is output, and the label refers to the case review image Cluster ID.
步骤S110,根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员。Step S110, determining suspicious project members among the project members according to the clustering results of the clustering.
本申请实施例提供的一种优选实施方式中,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员,包括:In a preferred implementation manner provided by an embodiment of the present application, the determining the suspicious project members among the project members according to the clustering results of the clustering includes:
确定具有相同聚类标识的所述案件审核图像为可疑图像;Determining that the case review images with the same cluster identification are suspicious images;
确定所述可疑图像的提交人为所述项目成员中的可疑项目成员。It is determined that the submitter of the suspicious image is a suspicious project member among the project members.
在上述构建连通图的步骤中,将所述案件审核图像之间的图像相似度取对数值作为所述案件审核图像对应节点之间的边权重,就使得所述案件审核图像之间相似度高的案件审核图像,连接其在连通图中对应节点的边的权重也高,节点连接越紧密,从而更易形成同一社区,所述聚类模型输出的所述案件审核图像的聚类标识可作为划分社区的标准,具有相同聚类标识的所述案件审核图像属于相同社区,从而确定其为可疑图像,相应的,所述可疑图像的提交人为所述项目成员中的可疑项目成员。In the above step of constructing the connected graph, the logarithmic value of the image similarity between the case review images is used as the edge weight between the corresponding nodes of the case review images, so that the similarity between the case review images is high The weight of the edge connecting its corresponding nodes in the connected graph is also high, and the closer the nodes are connected, the easier it is to form the same community. The cluster identification of the case review image output by the clustering model can be used as a division According to community standards, the case review images with the same cluster ID belong to the same community, so they are determined to be suspicious images, and correspondingly, the submitter of the suspicious images is a suspicious project member among the project members.
优选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员步骤执行之后,包括:确定所述可疑图像中具有同一聚类标识的所述案件审核图像的提交人的集合为可疑团伙。Preferably, after the step of determining suspicious project members among the project members according to the clustering results of the clustering is performed, it includes: determining the submitters of the case review images with the same cluster identification in the suspicious images A collection of suspicious gangs.
在确定所述可疑项目成员之后,根据所述聚类结果,具有同一聚类标识的所述案件审核图像形成同一社区,同一社区内的案件审核图像对应的提交人组成的群体则很有可能是可疑团伙。After the suspicious project members are identified, according to the clustering results, the case review images with the same cluster ID form the same community, and the group composed of the submitters corresponding to the case review images in the same community is likely to be Suspicious gang.
本申请实施例提供的一种优选实施方式中,根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员之后,在聚类获得的聚类结果的基础上,还可以通过分析事实性关系来判断可疑项目成员是否出自于同一骗保团伙,具体实现如下:In a preferred implementation mode provided by the embodiment of the present application, after the suspicious project members among the project members are determined according to the clustering results of the clustering, on the basis of the clustering results obtained by clustering, it is also possible to analyze The factual relationship is used to judge whether the suspicious project members are from the same insurance fraud group. The specific realization is as follows:
1)获取所述可疑项目成员的事实性关系;1) Obtain the factual relationship of the suspicious project members;
2)基于所述事实性关系在所述可疑项目成员中确定具有实际事实性关系的项目成员为不可信项目成员;2) Determining project members with actual factual relationships among the suspicious project members based on the factual relationship as untrustworthy project members;
3)确定所述不可信项目成员中提交具有同一聚类标识的所述案件审核图像的成员集合为可疑团伙。3) Determining a set of members among the untrustworthy project members who submitted the case review image with the same cluster ID as a suspicious gang.
在确定所述可疑项目成员之后,基于获取的可疑项目成员的事实性关系筛选出具有实际事实性关系的项目成员为不可信项目成员的好处在于,首先图聚类操作实现的是第一轮筛选,筛选出被划分为同一社区的案件审核图像,该案件审核图像的提交人具有团伙骗保的嫌疑,其次通过判断是否具有实际事实性关系进行第二轮筛选,来判断可疑项目成员在现实生活中是否具有实际联系,经过这两轮筛选,筛选出的项目成员同时具备提交的案件审核图片为所述可疑图片且该项目成员间具有事实性关系这两个条件,在此基础上确定所述不可信项目成员集合构成的骗保团伙的准确性更高。After the suspicious project members are determined, the advantage of screening project members with actual factual relationships as untrustworthy project members based on the obtained factual relationship of suspicious project members is that the graph clustering operation realizes the first round of screening , screen out the case review images that are divided into the same community, and the submitter of the case review images is suspected of gang fraud, and then conduct a second round of screening by judging whether there is an actual factual relationship to determine whether the suspicious project members are in real life After these two rounds of screening, the selected project members have the two conditions that the picture submitted for case review is the above-mentioned suspicious picture and that there is a factual relationship between the project members. On this basis, the above-mentioned The accuracy of fraudulent insurance gangs composed of a collection of untrustworthy project members is higher.
下述结合附图2,对本申请提供的基于图像处理的可疑项目成员识别方法进行进一步说明,具体实现如下:The method for identifying suspicious project members based on image processing provided by the present application will be further described in conjunction with accompanying drawing 2, and the specific implementation is as follows:
步骤S202,获取保险理赔环节要求用户上传的N张理赔图片。Step S202, obtaining N claims pictures that are required to be uploaded by the user during the insurance claim process.
保险理赔环节一般会需要用户上传理赔材料,如理赔图片,保险公司会对理赔材料进行分析来判断用户是否满足理赔条件。本实施例是基于图像处理技术来识别可疑项目成员,首先需要获取用户上传的N张理赔图片,N张理赔图片分别表示为:Image1,Image2,Image3,……,ImageN。The insurance claim process generally requires users to upload claim materials, such as claim pictures, and the insurance company will analyze the claim materials to determine whether the user meets the claim conditions. This embodiment is based on image processing technology to identify suspicious project members. First, it is necessary to obtain N claims pictures uploaded by users, and the N claims pictures are represented as: Image1, Image2, Image3, ..., ImageN.
步骤S204,利用深度学习模型提取N张理赔图片各自的特征向量。Step S204, using the deep learning model to extract the respective feature vectors of the N claims pictures.
基于理赔图片数量大、占用计算空间大的问题,N张理赔图片统一采用模型来处理,以提高图像处理效率。本方法是根据理赔图片之间的图像相似度来进行图聚类操作,在计算图像相似度的过程中,需要先将图像进行向量化处理,将图像转化成向量的过程基于开源的深度学习框架来实现,具体的,在深度学习框架下,利用卷积神经网络预先构建并且训练好深度学习模型,通过将Image1到ImageN这N张理赔图片输入深度学习模型获取Image1到ImageN各自的特征向量。Due to the large number of claims pictures and the large calculation space, the N claims pictures are uniformly processed by the model to improve the image processing efficiency. This method is to perform graph clustering operation based on the image similarity between claims pictures. In the process of calculating image similarity, the image needs to be vectorized first, and the process of converting the image into a vector is based on an open source deep learning framework. To achieve this, specifically, under the deep learning framework, the convolutional neural network is used to pre-build and train the deep learning model, and the respective feature vectors of Image1 to ImageN are obtained by inputting N claims pictures from Image1 to ImageN into the deep learning model.
步骤S206,利用词频逆文本频率指数算法计算N张理赔图片之间的图像相似度。Step S206, using the word frequency inverse text frequency index algorithm to calculate the image similarity between the N claims pictures.
对N张理赔图片进行向量化操作得到其对应的特征向量后,即可采用词频逆文本频率指数算法计算理赔图片之间的图像相似度,作为图聚类操作的依据。After vectorizing the N claims pictures to obtain their corresponding feature vectors, the word frequency inverse text frequency index algorithm can be used to calculate the image similarity between the claims pictures as the basis for the graph clustering operation.
步骤S208,利用标签传播算法对N张理赔图片进行聚类。Step S208, using the label propagation algorithm to cluster the N claims pictures.
具体的,对理赔图片的聚类操作由聚类模型来进行,该聚类模型在聚类过程中采用的算法为标签传播算法,步骤如下:Specifically, the clustering operation of claims images is performed by a clustering model, and the algorithm used by the clustering model in the clustering process is the label propagation algorithm, and the steps are as follows:
1)构建连通图,N张理赔图片Image1到ImageN与连通图中的节点一一对应,相应的,节点可以被标记为Image1,Image2,Image3,……,ImageN,将N张理赔图片之间的图像相似度取对数值作为理赔图片对应节点之间的边权重;1) Construct a connected graph, and the N claims pictures Image1 to ImageN correspond to the nodes in the connected graph one by one. Correspondingly, the nodes can be marked as Image1, Image2, Image3, ..., ImageN, and the N claims pictures The logarithmic value of the image similarity is used as the edge weight between the corresponding nodes of the claim image;
2)将连通图输入聚类模型,利用标签传播算法获得节点的标签,具有较高相似度的理赔图片对应的节点具有相同的标签,被聚为同类。2) Input the connected graph into the clustering model, use the label propagation algorithm to obtain the labels of the nodes, and the nodes corresponding to the claim pictures with a high similarity have the same label and are clustered into the same category.
步骤S210,确定可疑项目成员。Step S210, determining suspicious project members.
将聚类模型输出的理赔图片的标识作为划分社区的标准,具有同一聚类标识的理赔图片在某一方面具有相似度,被划分为同一社区,确定具有社区属性的理赔图片为可疑图片,确定可疑图片的提交人为可疑项目成员。The identification of the claim settlement picture output by the clustering model is used as the standard for dividing the community. The claim settlement pictures with the same cluster identification have similarity in a certain aspect and are classified into the same community. The claim settlement pictures with community attributes are determined to be suspicious pictures. Suspicious images are submitted by members of the Suspicious Project.
步骤S212,获取可疑项目成员的事实性关系。Step S212, obtaining the factual relationship of the suspicious project members.
确定可疑项目成员后,还可通过可疑项目成员之间的事实性关系进一步判断该可疑项目成员是否具有骗保嫌疑,事实性关系包括项目成员之间的资金转账关系、通信关系、通讯录关系、设备关系等,具有资金转账交易记录、通过通讯设备通信、对方互为通讯录联系人、多方曾使用同一登录设备等行为的可疑项目成员都被认定为具有事实性关系。After the suspicious project members are identified, it is also possible to further judge whether the suspicious project members are suspected of fraudulent insurance through the factual relationship between the suspicious project members. The factual relationship includes the fund transfer relationship, communication relationship, address book relationship, Equipment relationship, etc. Suspicious project members who have fund transfer transaction records, communicate through communication equipment, the other party is a contact in the address book, and multiple parties have used the same login device are all identified as having a factual relationship.
步骤S214,在可疑项目成员中确定具有实际事实性关系的项目成员为不可信项目成员。Step S214, among the suspicious project members, it is determined that the project members with actual factual relationship are untrusted project members.
对理赔图片的聚类操作筛选出了提交的理赔图片具有相似性的可疑项目成员后,如果可疑项目成员还同时具有事实性关系,则该可疑项目成员参与骗保的可能性更大,据此可以基于事实性关系在可疑项目成员中筛选出具有事实性关系的项目成员作为不可信项目成员。After the clustering operation of the claim pictures screens out suspicious project members whose submitted claim pictures are similar, if the suspicious project members also have factual relationships, the possibility of the suspicious project members participating in insurance fraud is greater. According to this Project members with factual relationships can be screened out among suspicious project members based on factual relationships as untrusted project members.
步骤S216,确定不可信项目成员中提交的具有同一标签的理赔图像的成员集合为可疑团伙。Step S216, determining that the set of members of the untrustworthy project members who submitted the claim image with the same label is a suspicious group.
项目成员中被确定的不可信项目成员具有很大的骗保嫌疑,而属于同一骗保团伙的项目成员提交的理赔图片大多在某一方面具有相似性,易被标识同一标签,据此,根据之前得到的聚类结果,可以确定不可信项目成员中提交的具有同一标签的理赔图像的成员组成的集合为可疑团伙。The identified untrustworthy project members among the project members are highly suspected of fraudulent insurance, and most of the claim settlement pictures submitted by project members belonging to the same fraudulent gang are similar in one aspect and are easy to be labeled with the same label. Accordingly, according to According to the clustering results obtained before, it can be determined that the set composed of members who submitted claim images with the same label among the untrustworthy project members is a suspicious group.
本申请提供的一种基于图像处理的可疑项目成员识别装置实施例如下:An embodiment of a suspicious project member identification device based on image processing provided by this application is as follows:
在上述的实施例中,提供了一种基于图像处理的可疑项目成员识别方法,与之相对应的,本申请还提供了一种基于图像处理的可疑项目成员识别装置,下面结合附图进行说明。In the above-mentioned embodiments, a method for identifying suspicious project members based on image processing is provided. Correspondingly, the present application also provides a device for identifying suspicious project members based on image processing, which will be described below with reference to the accompanying drawings .
参照附图3,其示出了本申请提供的一种基于图像处理的可疑项目成员识别装置实施例的示意图。Referring to FIG. 3 , it shows a schematic diagram of an embodiment of an image processing-based member identification device for suspicious projects provided by the present application.
由于装置实施例基本相似于方法实施例,所以描述得比较简单,相关的部分请参见上述提供的方法实施例的对应说明即可。下述描述的装置实施例仅仅是示意性的。Since the device embodiment is basically similar to the method embodiment, the description is relatively simple. For relevant parts, please refer to the corresponding description of the method embodiment provided above. The device embodiments described below are illustrative only.
本申请提供一种基于图像处理的可疑项目成员识别装置,包括:This application provides a device for identifying suspicious project members based on image processing, including:
案件审核图像集获取模块302,被配置为获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集;The case review image set
特征向量提取模块304,被配置为提取所述案件审核图像集中所述案件审核图像的特征向量;The feature
图像相似度确定模块306,被配置为根据所述特征向量确定所述案件审核图像之间的图像相似度;The image
聚类模块308,被配置为基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类;The
可疑项目成员确定模块310,被配置为根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员。The suspicious project
可选的,所述特征向量提取模块304,具体被配置为将所述案件审核图像输入预先训练好的深度学习模型进行图像向量化处理,输出所述案件审核图像的特征向量。Optionally, the feature
可选的,所述图像相似度确定模块306,具体被配置为根据所述特征向量利用词频逆文本频率指数算法计算所述案件审核图像之间的图像相似度。Optionally, the image
可选的,所述图像相似度确定模块306,具体被配置为基于所述案件审核图像的特征向量之间的向量距离,确定所述案件审核图像之间的图像相似度。Optionally, the image
可选的,所述聚类模块308,包括:Optionally, the
连通图构建子模块,被配置为构建连通图,将所述案件审核图像与所述连通图中的节点建立一一对应的关系,并将所述案件审核图像之间的图像相似度取对数值作为所述案件审核图像对应节点之间的边权重;The connected graph construction sub-module is configured to construct a connected graph, establish a one-to-one correspondence relationship between the case review image and the nodes in the connected graph, and take a logarithmic value for the image similarity between the case review images As the edge weight between the corresponding nodes of the case review image;
聚类标识输出模块,被配置为将所述连通图输入聚类模型进行聚类,输出所述案件审核图像的聚类标识。The cluster identification output module is configured to input the connectivity graph into a clustering model for clustering, and output the cluster identification of the case review image.
可选的,所述可疑项目成员确定模块310,包括:Optionally, the suspicious project
可疑图像确定子模块,被配置为确定具有相同聚类标识的所述案件审核图像为可疑图像;The suspicious image determination submodule is configured to determine that the case review images with the same cluster identification are suspicious images;
可疑项目成员确定子模块,被配置为确定所述可疑图像的提交人为所述项目成员中的可疑项目成员。The suspicious project member determination submodule is configured to determine that the submitter of the suspicious image is a suspicious project member among the project members.
可选的,所述基于图像处理的可疑项目成员识别装置,还包括:Optionally, the device for identifying suspicious project members based on image processing further includes:
第一可疑团伙确定模块,被配置为确定所述可疑图像中具有同一聚类标识的所述案件审核图像的提交人的集合为可疑团伙。The first suspicious gang determination module is configured to determine that a set of submitters of the case review image with the same cluster identifier in the suspicious image is a suspicious gang.
可选的,所述基于图像处理的可疑项目成员识别装置,还包括:Optionally, the device for identifying suspicious project members based on image processing further includes:
第二事实性关系获取模块,被配置为获取所述可疑项目成员的事实性关系;The second factual relationship acquisition module is configured to acquire the factual relationship of the suspicious project members;
不可信项目成员确定模块,被配置为基于所述事实性关系在所述可疑项目成员中确定具有实际事实性关系的项目成员为不可信项目成员;An untrustworthy project member determination module configured to determine, among the suspicious project members, a project member with an actual factual relationship as an untrustworthy project member based on the factual relationship;
第二可疑团伙确定模块,被配置为确定所述不可信项目成员中提交具有同一聚类标识的所述案件审核图像的成员集合为可疑团伙。The second suspicious group determination module is configured to determine a set of members among the untrustworthy project members who submitted the case review image with the same cluster ID as a suspicious group.
可选的,所述基于图像处理的可疑项目成员识别装置,还包括:Optionally, the device for identifying suspicious project members based on image processing further includes:
第一事实性关系获取模块,被配置为获取所述项目成员的事实性关系;A first factual relationship acquisition module configured to acquire the factual relationship of the project members;
案件审核图像获取对象确定模块,被配置为基于所述事实性关系确定具有实际事实性关系的项目成员;A case review image acquisition object determination module configured to determine project members with actual factual relationships based on the factual relationships;
相应的,所述案件审核图像集获取模块302,具体被配置为获取所述具有实际事实性关系的项目成员的案件审核图像,组成所述案件审核图像集。Correspondingly, the case review image set
可选的,所述项目成员的事实性关系,包括下述至少一项:Optionally, the de facto relationship of the project members includes at least one of the following:
所述项目成员之间的资金转账关系、通信关系、通讯录关系、设备关系。Fund transfer relationship, communication relationship, address book relationship, and equipment relationship among project members.
可选的,所述深度学习模型,基于下述任意一种神经网络构建:Optionally, the deep learning model is constructed based on any of the following neural networks:
卷积神经网络、深度神经网络。Convolutional Neural Networks, Deep Neural Networks.
可选的,所述聚类模型,基于下述任意一种算法:Optionally, the clustering model is based on any of the following algorithms:
标签传播算法、最大连通图算法。Label Propagation Algorithm, Maximum Connected Graph Algorithm.
本申请提供的一种计算设备实施例如下:An example of a computing device provided by this application is as follows:
图4是示出了根据本说明书一实施例的计算设备400的结构框图。该计算设备400的部件包括但不限于存储器410和处理器420。处理器420与存储器410通过总线430相连接,数据库450用于保存数据。FIG. 4 is a structural block diagram illustrating a computing device 400 according to an embodiment of the present specification. Components of the computing device 400 include, but are not limited to, a memory 410 and a processor 420 . The processor 420 is connected to the memory 410 through the bus 430, and the database 450 is used for storing data.
计算设备400还包括接入设备440,接入设备440使得计算设备400能够经由一个或多个网络460通信。这些网络的示例包括公用交换电话网(PSTN)、局域网(LAN)、广域网(WAN)、个域网(PAN)或诸如因特网的通信网络的组合。接入设备440可以包括有线或无线的任何类型的网络接口(例如,网络接口卡(NIC))中的一个或多个,诸如IEEE802.11无线局域网(WLAN)无线接口、全球微波互联接入(Wi-MAX)接口、以太网接口、通用串行总线(USB)接口、蜂窝网络接口、蓝牙接口、近场通信(NFC)接口,等等。Computing device 400 also includes an access device 440 that enables computing device 400 to communicate via one or more networks 460 . Examples of these networks include the Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or a combination of communication networks such as the Internet. Access device 440 may include one or more of any type of network interface (e.g., a network interface card (NIC)), wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, Worldwide Interoperability for Microwave Access ( Wi-MAX) interface, Ethernet interface, Universal Serial Bus (USB) interface, cellular network interface, Bluetooth interface, Near Field Communication (NFC) interface, etc.
在本说明书的一个实施例中,计算设备400的上述部件以及图4中未示出的其他部件也可以彼此相连接,例如通过总线。应当理解,图4所示的计算设备结构框图仅仅是出于示例的目的,而不是对本说明书范围的限制。本领域技术人员可以根据需要,增添或替换其他部件。In an embodiment of the present specification, the above-mentioned components of the computing device 400 and other components not shown in FIG. 4 may also be connected to each other, for example, through a bus. It should be understood that the structural block diagram of the computing device shown in FIG. 4 is only for the purpose of illustration, rather than limiting the scope of this description. Those skilled in the art can add or replace other components as needed.
计算设备400可以是任何类型的静止或移动计算设备,包括移动计算机或移动计算设备(例如,平板计算机、个人数字助理、膝上型计算机、笔记本计算机、上网本等)、移动电话(例如,智能手机)、可佩戴的计算设备(例如,智能手表、智能眼镜等)或其他类型的移动设备,或者诸如台式计算机或PC的静止计算设备。计算设备400还可以是移动式或静止式的服务器。Computing device 400 may be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile telephones (e.g., smartphones), ), wearable computing devices (eg, smart watches, smart glasses, etc.), or other types of mobile devices, or stationary computing devices such as desktop computers or PCs. Computing device 400 may also be a mobile or stationary server.
本申请提供一种计算设备,包括存储器410、处理器420及存储在存储器上并可在处理器上运行的计算机指令,所述处理器420用于执行如下计算机可执行指令:The present application provides a computing device, including a memory 410, a processor 420 and computer instructions stored in the memory and operable on the processor, the processor 420 is used to execute the following computer-executable instructions:
获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集;Obtain a case review image set composed of case review images submitted by project members participating in the project;
提取所述案件审核图像集中所述案件审核图像的特征向量;Extracting the feature vector of the case review image in the case review image set;
根据所述特征向量确定所述案件审核图像之间的图像相似度;determining the image similarity between the case review images according to the feature vector;
基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类;clustering the case review images based on the similarity weight determined by the image similarity;
根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员。Suspicious project members among the project members are determined according to the clustering results of the clustering.
可选的,所述提取所述案件审核图像集中所述案件审核图像的特征向量,包括:Optionally, the extracting the feature vector of the case review image in the case review image set includes:
将所述案件审核图像输入预先训练好的深度学习模型进行图像向量化处理,输出所述案件审核图像的特征向量。The case review image is input into a pre-trained deep learning model for image vectorization processing, and the feature vector of the case review image is output.
可选的,所述根据所述特征向量确定所述案件审核图像之间的图像相似度,包括:Optionally, the determining the image similarity between the case review images according to the feature vectors includes:
根据所述特征向量利用词频逆文本频率指数算法计算所述案件审核图像之间的图像相似度。The image similarity between the case review images is calculated by using the word frequency inverse text frequency index algorithm according to the feature vector.
可选的,所述根据所述特征向量确定所述案件审核图像之间的图像相似度,包括:Optionally, the determining the image similarity between the case review images according to the feature vectors includes:
基于所述案件审核图像的特征向量之间的向量距离,确定所述案件审核图像之间的图像相似度。An image similarity between the case review images is determined based on a vector distance between feature vectors of the case review images.
可选的,所述基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类,包括:Optionally, the clustering of the case review images based on the similarity weight determined based on the image similarity includes:
构建连通图,将所述案件审核图像与所述连通图中的节点建立一一对应的关系,并将所述案件审核图像之间的图像相似度取对数值作为所述案件审核图像对应节点之间的边权重;Constructing a connected graph, establishing a one-to-one correspondence relationship between the case review image and the nodes in the connected graph, and taking the logarithmic value of the image similarity between the case review images as one of the corresponding nodes of the case review image The edge weight between;
将所述连通图输入聚类模型进行聚类,输出所述案件审核图像的聚类标识。The connectivity graph is input into a clustering model for clustering, and the cluster identification of the case review image is output.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员,包括:Optionally, the determining a suspicious project member among the project members according to the clustering result of the clustering includes:
确定具有相同聚类标识的所述案件审核图像为可疑图像;Determining that the case review images with the same cluster identification are suspicious images;
确定所述可疑图像的提交人为所述项目成员中的可疑项目成员。It is determined that the submitter of the suspicious image is a suspicious project member among the project members.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员步骤执行之后,包括:Optionally, after the step of determining suspicious project members among the project members according to the clustering result of the clustering is executed, the step includes:
确定所述可疑图像中具有同一聚类标识的所述案件审核图像的提交人的集合为可疑团伙。Determining that the set of submitters of the case review image with the same cluster identifier in the suspicious image is a suspicious gang.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员步骤执行之后,还包括:Optionally, after the step of determining suspicious project members among the project members according to the clustering results of the clustering is performed, it further includes:
获取所述可疑项目成员的事实性关系;Obtain the factual relationship of the suspicious project members;
基于所述事实性关系在所述可疑项目成员中确定具有实际事实性关系的项目成员为不可信项目成员;determining, among the suspicious project members, a project member with an actual factual relationship as an untrusted project member based on the factual relationship;
确定所述不可信项目成员中提交具有同一聚类标识的所述案件审核图像的成员集合为可疑团伙。Determining a set of members among the untrustworthy project members who submitted the case review image with the same cluster ID as a suspicious gang.
可选的,所述获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集步骤执行之前,还包括:Optionally, before the step of obtaining the case review image set composed of the case review images submitted by the project members participating in the project is executed, it may further include:
获取所述项目成员的事实性关系;obtain de facto relationships of said project members;
基于所述事实性关系确定具有实际事实性关系的项目成员;determining project members with actual factual relationships based on the factual relationships;
相应的,所述获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集步骤中,获取所述具有实际事实性关系的项目成员的案件审核图像,组成所述案件审核图像集。Correspondingly, in the step of obtaining the case review image set composed of the case review images submitted by the project members participating in the project, the case review images of the project members with actual factual relationship are obtained to form the case review image set.
可选的,所述项目成员的事实性关系,包括下述至少一项:Optionally, the de facto relationship of the project members includes at least one of the following:
所述项目成员之间的资金转账关系、通信关系、通讯录关系、设备关系。Fund transfer relationship, communication relationship, address book relationship, and equipment relationship among project members.
可选的,所述深度学习模型,基于下述任意一种神经网络构建:Optionally, the deep learning model is constructed based on any of the following neural networks:
卷积神经网络、深度神经网络。Convolutional Neural Networks, Deep Neural Networks.
本申请一实施例还提供一种计算机可读存储介质,其存储有计算机指令,该指令被处理器执行时实现如下:An embodiment of the present application also provides a computer-readable storage medium, which stores computer instructions, and the instructions are implemented as follows when executed by a processor:
获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集;Obtain a case review image set composed of case review images submitted by project members participating in the project;
提取所述案件审核图像集中所述案件审核图像的特征向量;Extracting the feature vector of the case review image in the case review image set;
根据所述特征向量确定所述案件审核图像之间的图像相似度;determining the image similarity between the case review images according to the feature vector;
基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类;clustering the case review images based on the similarity weight determined by the image similarity;
根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员。Suspicious project members among the project members are determined according to the clustering results of the clustering.
可选的,所述提取所述案件审核图像集中所述案件审核图像的特征向量,包括:Optionally, the extracting the feature vector of the case review image in the case review image set includes:
将所述案件审核图像输入预先训练好的深度学习模型进行图像向量化处理,输出所述案件审核图像的特征向量。The case review image is input into a pre-trained deep learning model for image vectorization processing, and the feature vector of the case review image is output.
可选的,所述根据所述特征向量确定所述案件审核图像之间的图像相似度,包括:Optionally, the determining the image similarity between the case review images according to the feature vectors includes:
根据所述特征向量利用词频逆文本频率指数算法计算所述案件审核图像之间的图像相似度。The image similarity between the case review images is calculated by using the word frequency inverse text frequency index algorithm according to the feature vector.
可选的,所述根据所述特征向量确定所述案件审核图像之间的图像相似度,包括:Optionally, the determining the image similarity between the case review images according to the feature vectors includes:
基于所述案件审核图像的特征向量之间的向量距离,确定所述案件审核图像之间的图像相似度。An image similarity between the case review images is determined based on a vector distance between feature vectors of the case review images.
可选的,所述基于所述图像相似度确定的相似度权重对所述案件审核图像进行聚类,包括:Optionally, the clustering of the case review images based on the similarity weight determined based on the image similarity includes:
构建连通图,将所述案件审核图像与所述连通图中的节点建立一一对应的关系,并将所述案件审核图像之间的图像相似度取对数值作为所述案件审核图像对应节点之间的边权重;Constructing a connected graph, establishing a one-to-one correspondence relationship between the case review image and the nodes in the connected graph, and taking the logarithmic value of the image similarity between the case review images as one of the corresponding nodes of the case review image The edge weight between;
将所述连通图输入聚类模型进行聚类,输出所述案件审核图像的聚类标识。The connectivity graph is input into a clustering model for clustering, and the cluster identification of the case review image is output.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员,包括:Optionally, the determining a suspicious project member among the project members according to the clustering result of the clustering includes:
确定具有相同聚类标识的所述案件审核图像为可疑图像;Determining that the case review images with the same cluster identification are suspicious images;
确定所述可疑图像的提交人为所述项目成员中的可疑项目成员。It is determined that the submitter of the suspicious image is a suspicious project member among the project members.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员步骤执行之后,包括:Optionally, after the step of determining suspicious project members among the project members according to the clustering result of the clustering is executed, the step includes:
确定所述可疑图像中具有同一聚类标识的所述案件审核图像的提交人的集合为可疑团伙。Determining that the set of submitters of the case review image with the same cluster identifier in the suspicious image is a suspicious gang.
可选的,所述根据所述聚类的聚类结果确定所述项目成员中的可疑项目成员步骤执行之后,还包括:Optionally, after the step of determining suspicious project members among the project members according to the clustering results of the clustering is performed, it further includes:
获取所述可疑项目成员的事实性关系;Obtain the factual relationship of the suspicious project members;
基于所述事实性关系在所述可疑项目成员中确定具有实际事实性关系的项目成员为不可信项目成员;determining, among the suspicious project members, a project member with an actual factual relationship as an untrusted project member based on the factual relationship;
确定所述不可信项目成员中提交具有同一聚类标识的所述案件审核图像的成员集合为可疑团伙。Determining a set of members among the untrustworthy project members who submitted the case review image with the same cluster ID as a suspicious gang.
可选的,所述获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集步骤执行之前,还包括:Optionally, before the step of obtaining the case review image set composed of the case review images submitted by the project members participating in the project is executed, it may further include:
获取所述项目成员的事实性关系;obtain de facto relationships of said project members;
基于所述事实性关系确定具有实际事实性关系的项目成员;determining project members with actual factual relationships based on the factual relationships;
相应的,所述获取参与项目的项目成员提交的案件审核图像组成的案件审核图像集步骤中,获取所述具有实际事实性关系的项目成员的案件审核图像,组成所述案件审核图像集。Correspondingly, in the step of obtaining the case review image set composed of the case review images submitted by the project members participating in the project, the case review images of the project members with actual factual relationship are obtained to form the case review image set.
可选的,所述项目成员的事实性关系,包括下述至少一项:Optionally, the de facto relationship of the project members includes at least one of the following:
所述项目成员之间的资金转账关系、通信关系、通讯录关系、设备关系。Fund transfer relationship, communication relationship, address book relationship, and equipment relationship among project members.
可选的,所述深度学习模型,基于下述任意一种神经网络构建:Optionally, the deep learning model is constructed based on any of the following neural networks:
卷积神经网络、深度神经网络。Convolutional Neural Networks, Deep Neural Networks.
上述为本实施例的一种计算机可读存储介质的示意性方案。需要说明的是,该存储介质的技术方案与上述的基于图像处理的可疑项目成员识别方法的技术方案属于同一构思,存储介质的技术方案未详细描述的细节内容,均可以参见上述基于图像处理的可疑项目成员识别方法的技术方案的描述。The foregoing is a schematic solution of a computer-readable storage medium in this embodiment. It should be noted that the technical solution of the storage medium and the technical solution of the above-mentioned method for identifying suspicious project members based on image processing belong to the same idea, and details not described in detail in the technical solution of the storage medium can be found in the above-mentioned A description of the technical solution of the suspicious project member identification method.
所述计算机指令包括计算机程序代码,所述计算机程序代码可以为源代码形式、对象代码形式、可执行文件或某些中间形式等。所述计算机可读介质可以包括:能够携带所述计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器(ROM,Read-Only Memory)、随机存取存储器(RAM,Random Access Memory)、电载波信号、电信信号以及软件分发介质等。需要说明的是,所述计算机可读介质包含的内容可以根据司法管辖区内立法和专利实践的要求进行适当的增减,例如在某些司法管辖区,根据立法和专利实践,计算机可读介质不包括电载波信号和电信信号。The computer instructions include computer program code, which may be in source code form, object code form, executable file or some intermediate form, and the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a removable hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM, Read-Only Memory) , Random Access Memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable Excludes electrical carrier signals and telecommunication signals.
需要说明的是,对于前述的各方法实施例,为了简便描述,故将其都表述为一系列的动作组合,但是本领域技术人员应该知悉,本申请并不受所描述的动作顺序的限制,因为依据本申请,某些步骤可以采用其它顺序或者同时进行。其次,本领域技术人员也应该知悉,说明书中所描述的实施例均属于优选实施例,所涉及的动作和模块并不一定都是本申请所必须的。It should be noted that, for the sake of simplicity of description, the aforementioned method embodiments are expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited by the described action sequence. Depending on the application, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification belong to preferred embodiments, and the actions and modules involved are not necessarily required by this application.
在上述实施例中,对各个实施例的描述都各有侧重,某个实施例中没有详述的部分,可以参见其它实施例的相关描述。In the above-mentioned embodiments, the descriptions of each embodiment have their own emphases, and for parts not described in detail in a certain embodiment, reference may be made to relevant descriptions of other embodiments.
以上公开的本申请优选实施例只是用于帮助阐述本申请。可选实施例并没有详尽叙述所有的细节,也不限制该发明仅为所述的具体实施方式。显然,根据本说明书的内容,可作很多的修改和变化。本说明书选取并具体描述这些实施例,是为了更好地解释本申请的原理和实际应用,从而使所属技术领域技术人员能很好地理解和利用本申请。本申请仅受权利要求书及其全部范围和等效物的限制。The preferred embodiments of the present application disclosed above are only used to help clarify the present application. Alternative embodiments are not exhaustive in all detail, nor are the inventions limited to specific implementations described. Obviously, many modifications and variations can be made based on the contents of this specification. This description selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present application, so that those skilled in the art can well understand and use the present application. This application is to be limited only by the claims, along with their full scope and equivalents.
Claims (14)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910238911.0A CN110163761B (en) | 2019-03-27 | 2019-03-27 | Suspicious item member identification method and device based on image processing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910238911.0A CN110163761B (en) | 2019-03-27 | 2019-03-27 | Suspicious item member identification method and device based on image processing |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110163761A CN110163761A (en) | 2019-08-23 |
CN110163761B true CN110163761B (en) | 2023-05-12 |
Family
ID=67638486
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910238911.0A Active CN110163761B (en) | 2019-03-27 | 2019-03-27 | Suspicious item member identification method and device based on image processing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110163761B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111325265B (en) * | 2020-02-17 | 2023-09-01 | 中国银联股份有限公司 | Detection method and device for tampered image |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107358596A (en) * | 2017-04-11 | 2017-11-17 | 阿里巴巴集团控股有限公司 | Method, device, electronic equipment and system for image-based vehicle damage assessment |
CN108573286A (en) * | 2018-05-10 | 2018-09-25 | 阿里巴巴集团控股有限公司 | A kind of data processing method, device, equipment and the server of Claims Resolution business |
CN108734479A (en) * | 2018-04-12 | 2018-11-02 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and the server of Insurance Fraud identification |
CN109271418A (en) * | 2018-08-14 | 2019-01-25 | 阿里巴巴集团控股有限公司 | Suspicious clique's recognition methods, device, equipment and computer readable storage medium |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8724904B2 (en) * | 2011-10-25 | 2014-05-13 | International Business Machines Corporation | Anomaly detection in images and videos |
US10628890B2 (en) * | 2017-02-23 | 2020-04-21 | International Business Machines Corporation | Visual analytics based vehicle insurance anti-fraud detection |
CN107403424B (en) * | 2017-04-11 | 2020-09-18 | 阿里巴巴集团控股有限公司 | Vehicle loss assessment method and device based on image and electronic equipment |
-
2019
- 2019-03-27 CN CN201910238911.0A patent/CN110163761B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107358596A (en) * | 2017-04-11 | 2017-11-17 | 阿里巴巴集团控股有限公司 | Method, device, electronic equipment and system for image-based vehicle damage assessment |
CN108734479A (en) * | 2018-04-12 | 2018-11-02 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and the server of Insurance Fraud identification |
CN108573286A (en) * | 2018-05-10 | 2018-09-25 | 阿里巴巴集团控股有限公司 | A kind of data processing method, device, equipment and the server of Claims Resolution business |
CN109271418A (en) * | 2018-08-14 | 2019-01-25 | 阿里巴巴集团控股有限公司 | Suspicious clique's recognition methods, device, equipment and computer readable storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110163761A (en) | 2019-08-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10031925B2 (en) | Method and system of using image recognition and geolocation signal analysis in the construction of a social media user identity graph | |
TWI705341B (en) | Feature relationship recommendation method and device, computing equipment and storage medium | |
CN116580257A (en) | Feature fusion model training and sample retrieval method, device and computer equipment | |
CN112231592B (en) | Graph-based network community discovery method, device, equipment and storage medium | |
WO2021179631A1 (en) | Convolutional neural network model compression method, apparatus and device, and storage medium | |
CN108399180B (en) | Knowledge graph construction method and device and server | |
WO2021169454A1 (en) | Graph feature processing method and device | |
CN111950889A (en) | A customer risk assessment method, device, readable storage medium and terminal device | |
CN110636445A (en) | WIFI-based indoor positioning method, device, equipment and medium | |
WO2023024408A1 (en) | Method for determining feature vector of user, and related device and medium | |
CN107368499B (en) | Client label modeling and recommending method and device | |
CN115470409A (en) | Interest recommendation method, device, electronic equipment, medium and program product | |
CN110163761B (en) | Suspicious item member identification method and device based on image processing | |
CN111275261A (en) | Resource flow prediction method and device | |
CN113487389A (en) | Information recommendation method and device | |
CN109472307A (en) | A method and apparatus for training an image classification model | |
CN118606218A (en) | Test case selection method and device | |
CN116935083B (en) | Image clustering method and device | |
CN113590721B (en) | Block chain address classification method and device | |
CN117688209A (en) | Node determination method and device | |
CN117217929A (en) | Registered object risk identification method, device, computer equipment and storage medium | |
JP2023544022A (en) | Group type identification methods, devices, computer equipment and computer programs | |
CN114579403A (en) | Alarm processing method and device | |
CN113344581A (en) | Service data processing method and device | |
CN111552846A (en) | Method and device for identifying suspicious relationship |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
TA01 | Transfer of patent application right | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20200929 Address after: 27 Hospital Road, George Town, Grand Cayman ky1-9008 Applicant after: Advanced innovation technology Co.,Ltd. Address before: A four-storey 847 mailbox in Grand Cayman Capital Building, British Cayman Islands Applicant before: Alibaba Group Holding Ltd. Effective date of registration: 20200929 Address after: 27 Hospital Road, George Town, Grand Cayman ky1-9008 Applicant after: Innovative advanced technology Co.,Ltd. Address before: 27 Hospital Road, George Town, Grand Cayman ky1-9008 Applicant before: Advanced innovation technology Co.,Ltd. |
|
TA01 | Transfer of patent application right | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20230404 Address after: 801-10, Section B, 8th floor, 556 Xixi Road, Xihu District, Hangzhou City, Zhejiang Province Applicant after: Ant financial (Hangzhou) Network Technology Co.,Ltd. Address before: 27 Hospital Road, George Town, Grand Cayman ky1-9008 Applicant before: Innovative advanced technology Co.,Ltd. |
|
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CP03 | Change of name, title or address | ||
CP03 | Change of name, title or address |
Address after: Room 514, 5th Floor, Building 2, No. 543-569 Xixi Road, Xihu District, Hangzhou City, Zhejiang Province, China 310023 Patentee after: Ant Intelligent (Hangzhou) Technology Co.,Ltd. Country or region after: China Address before: 801-10, Section B, 8th Floor, No. 556 Xixi Road, Xihu District, Hangzhou City, Zhejiang Province, 310063 Patentee before: Ant financial (Hangzhou) Network Technology Co.,Ltd. Country or region before: China |