摘要：【中文】本申请实施例提供一种数据接口监控方法及装置，方法包括：将配置有预设日志注解的目标数据接口与预设代理对象建立关联关系；通过所述代理对象对所述目标数据接口进行日志信息采集，并持久化存储采集得到的日志信息；本申请能够在不侵入原有系统的系统上，通过自定义注解的方式来实现日志的采集，减少日志代码的重复输出的同时能降低日志采集与现有业务系统的耦合，减少了大量重复性工作，提高了工作效率。 【EN】The embodiment of the application provides a data interface monitoring method and a device, wherein the method comprises the following steps: establishing an association relation between a target data interface configured with a preset log annotation and a preset proxy object; carrying out log information acquisition on the target data interface through the proxy object, and persistently storing the acquired log information; according to the method and the system, the collection of the logs can be realized in a self-defined annotation mode on a system which does not invade the original system, the repeated output of log codes is reduced, meanwhile, the coupling between the log collection and the existing service system can be reduced, a large amount of repetitive work is reduced, and the work efficiency is improved.

摘要：【中文】本发明提出了一种基于低内存实现超大文件的排序方法，包括：步骤S1，采用split分片法进行逻辑划分，对来自多个数据源的数据文件进行切割；步骤S2，利用对比去重和按照数据大小的排序原则将切割后的数据块进行重新排序；步骤S3，将重新排序后的多个数据块进行重新合并为一个文件；步骤S4，将合并后的文件与预存的历史存储文档中的数据进行排序对比以及去重复过滤处理后，得到最终数据结果并返回给数据请求方。本发明实现利用较低内存对超大文件进行排序整合，在确保文本数据的准确性以及高效性的前提条件下，使得低内存计算机也可以处理超大文本。 【EN】The invention provides a method for realizing ordering of oversized files based on low memory, which comprises the following steps: step S1, performing logic division by adopting a split fragmentation method, and cutting data files from a plurality of data sources; step S2, the cut data blocks are reordered by using comparison duplication removal and the ordering principle according to the data size; step S3, recombining the reordered data blocks into a file; and step S4, performing sequencing comparison and de-duplication filtering processing on the combined file and the data in the pre-stored history storage document to obtain a final data result and returning the final data result to the data requester. The invention realizes the sequencing and integration of the oversized files by using the lower memory, and ensures that a low-memory computer can also process the oversized files under the precondition of ensuring the accuracy and the high efficiency of the text data.

摘要：【中文】本申请提供一种基于策略规则的语音识别方法及装置，方法包括：接收目标音频文件和预设策略集；判断所述目标音频文件的对话文本是否匹配所述预设策略集中各策略的关键词标签，若是，则获取对应的质检分数；根据所述质检分数和预设质检阈值，确定所述目标音频文件的质检结果；本申请能够个性化、可配置化的进行语音识别，并能提升语音识别的准确性。 【EN】The application provides a voice recognition method and a voice recognition device based on policy rules, wherein the method comprises the following steps: receiving a target audio file and a preset strategy set; judging whether the dialog text of the target audio file is matched with the keyword labels of all the strategies in the preset strategy set, if so, acquiring a corresponding quality inspection score; determining a quality inspection result of the target audio file according to the quality inspection score and a preset quality inspection threshold value; the method and the device can perform voice recognition in a personalized and configurable mode, and can improve accuracy of the voice recognition.

摘要：【中文】本申请实施例提供一种基于WebSocket的分布式服务器集群交互方法及装置，方法包括：与客户端建立WebSocket连接，并将对应的标识信息存储至Redis缓存；接收用户在所述客户端触发的工作指令，并向Redis消息队列发布Topic消息，以使服务器集群中的另一订阅了所述Redis消息队列的Topic消息的服务器能够根据所述Topic消息中的标识信息判断该用户对应的session是否存在于本地，并根据判断结果执行对应消息推送操作；本申请能够有效确保每个用户实时、无误的收到服务器推送的消息。 【EN】The embodiment of the application provides a distributed server cluster interaction method and device based on WebSocket, and the method comprises the following steps: establishing WebSocket connection with a client, and storing corresponding identification information to a Redis cache; receiving a work instruction triggered by a user at the client, and publishing a Topic message to a Redis message queue, so that another server in a server cluster, which subscribes to the Topic message of the Redis message queue, can judge whether a session corresponding to the user exists locally according to identification information in the Topic message, and execute a corresponding message pushing operation according to a judgment result; the method and the device can effectively ensure that each user receives the message pushed by the server in real time and without errors.

摘要：【中文】本申请实施例提供一种基于深度学习的反欺诈模型处理方法及装置，方法包括：对数据进行初步清洗，并根据数据的数据类型，分别执行对应的数据预处理；将经过数据预处理后的不同类型的数据进行合并，并将经过合并后的数据输入预设神经网络中进行训练，得到反欺诈神经网络模型；本申请能够有效提升对欺诈用户和正常用户的识别准确度和识别效率。 【EN】The embodiment of the application provides an anti-fraud model processing method and device based on deep learning, wherein the method comprises the following steps: preliminarily cleaning the data, and respectively executing corresponding data preprocessing according to the data type of the data; merging different types of data subjected to data preprocessing, inputting the merged data into a preset neural network for training to obtain an anti-fraud neural network model; the method and the device can effectively improve the identification accuracy and the identification efficiency of the fraudulent user and the normal user.

摘要：【中文】本发明提供了一种提供IVR语音播报的方法及IVR系统，所述方法包括：IVR系统响应用户的呼入请求，与用户建立IVR通话连接；所述IVR系统获取所述用户的接入号码并识别接入号码的归属地，及与所述用户建立IVR通话连接的系统时间；所述IVR系统根据流程匹配规则，获得与所述用户匹配的播报信息；所述IVR系统向所述用户提供所述播报信息的语音播报。本发明支持动态及静态信息发布及语音交互，使企业信息发布终端发布信息即时生效至当前任务，提升信息时效，改善交互质量，增强用户体验。 【EN】The invention provides a method for providing IVR voice broadcast and an IVR system, wherein the method comprises the following steps: the IVR system responds to the call-in request of the user and establishes IVR call connection with the user; the IVR system acquires the access number of the user and identifies the attribution of the access number and the system time for establishing IVR call connection with the user; the IVR system obtains broadcast information matched with the user according to the flow matching rules; and the IVR system provides voice broadcast of the broadcast information for the user. The invention supports dynamic and static information release and voice interaction, enables the information released by the enterprise information release terminal to take effect to the current task immediately, improves the information timeliness, improves the interaction quality and enhances the user experience.

摘要：【中文】本发明提出了一种基于消息队列的反欺诈策略优化方法，包括：获取待处理的反欺诈策略，并发起冠军挑战者任务；对所述反欺诈策略采用以下优化模式之一进行优化处理：(1)灰跑模式；比例模式。本发明通过灰跑模式和比例模式，这两种优化模式逐步对目标策略进行测试优化，在完成进件的同时进行优化测试，根据生成进件测试报告判断优化情况。本发明可以极大程度上提高相关策略的优化测试效率，有助于缩短开发周期，节省人力物力消耗，帮助企业更好更快的完成业务需求。 【EN】The invention provides a message queue-based anti-fraud policy optimization method, which comprises the following steps: acquiring an anti-fraud strategy to be processed, and initiating a champion challenger task; optimizing the anti-fraud policy by adopting one of the following optimization modes: (1) a gray running mode; proportional mode. The invention gradually tests and optimizes the target strategy through the grey running mode and the proportional mode, performs optimization test while finishing entering a piece, and judges the optimization condition according to the generated piece entering test report. The invention can greatly improve the optimized test efficiency of the related strategies, is beneficial to shortening the development period, saves the consumption of manpower and material resources and helps enterprises to better and more quickly finish the business requirements.

摘要：【中文】本发明提出了一种基于雷达技术的银行卡拒件导流方法，包括：步骤S1，获取信用卡申请请求被银行拒绝的申请人信息；步骤S2，采用预设的雷达决策引擎对所述申请人的信息进行评分，获取相应的评分结果；步骤S3，将所述评分结果与预设标准进行比对，如果高于所述预设标准，则将对应的申请人信息打回银行，进行在此审查导流操作；如果高于所述预设标准，则向对应的申请人推荐与其评分结果匹配的金融渠道。本发明可以解决了现有存在的审查错误的问题。 【EN】The invention provides a bank card refusal part diversion method based on a radar technology, which comprises the following steps: step S1, obtaining the information of the applicant that the credit card application request is rejected by the bank; step S2, a preset radar decision engine is adopted to score the information of the applicant, and a corresponding scoring result is obtained; step S3, comparing the scoring result with a preset standard, if the scoring result is higher than the preset standard, returning the corresponding applicant information to the bank, and performing the examination and diversion operation; and if the preset standard is higher than the preset standard, recommending the financial channels matched with the grading results of the corresponding applicant. The invention can solve the problem of error examination in the prior art.

摘要：【中文】本申请公开了一种用于电话外呼的属地化线路部署、调度方法以及系统、服务器。该属地化线路部署、调度方法包括(1)部署的步骤建立属地化线路集，其中线路集包括：一个或多个线路的市级归属地；导入用户信息和服务信息，其中所述用户信息是指用户资料，所述服务信息是指电话外呼话术；(2)调度的步骤当接入用户电话，判别所述电话的归属地；对所述归属地执行属地化线路的外显匹配操作，并使用已匹配的属地化线路进行呼叫操作。本申请解决了频繁调配线路以及用户号码归属地与外显归属地不一致的技术问题。通过本申请可有效提高包括接通率、通话时长、投诉、客户满意度等各项指标。 【EN】The application discloses a method, a system and a server for deploying and scheduling a belonged line of a telephone outgoing call. The method for deploying and scheduling the home-based line comprises the following steps of (1) establishing a home-based line set, wherein the line set comprises the following steps: a city-level home of one or more lines; importing user information and service information, wherein the user information refers to user data, and the service information refers to external calling operation of a telephone; (2) in the dispatching step, when a user telephone is accessed, the attribution of the telephone is judged; and performing an explicit matching operation of the attribution line on the attribution, and performing a calling operation by using the matched attribution line. The method and the device solve the technical problems that lines are frequently allocated and the attribution place of the user number is inconsistent with the attribution place of the external display. By the method and the device, various indexes including call completing rate, call duration, complaints, customer satisfaction and the like can be effectively improved.

摘要：【中文】本发明提出了一种网关对用户请求进行统一拦截判断是否放行的方法，包括：步骤S1，当接收到用户的服务请求时，对用户进行登陆是否有效以及该服务请求是否合法进行拦截判断；步骤S2，根据预先设置的所有URL地址和过滤器应忽略的URL地址进行判断；步骤S3，判断URL地址是否为有效地址，并判断用户令牌token是否有效；步骤S4，如果用户不具备该地址访问权限，则过滤返回；如果通过，则对该服务请求进行相关用户信息的请求头的添加。本发明明采用网关对用户的请求进行统一的拦截并进行URL以及用户登陆是否有效等一系列的判断来决定是否将请求发送到指定的微服务中去，并根据请求路径路由到不同的微服务，保证尽量的简化开发。 【EN】The invention provides a method for a gateway to uniformly intercept and judge whether a user request is released, which comprises the following steps: step S1, when receiving the service request of the user, intercepting and judging whether the login of the user is effective and whether the service request is legal; step S2, according to all preset URL addresses and URL addresses that the filter should ignore; step S3, judging whether the URL address is an effective address and whether the user token is effective; step S4, if the user does not have the address access right, filtering and returning; if the service request passes, adding a request header of the related user information to the service request. The invention adopts the gateway to uniformly intercept the request of the user and judges whether the URL and the user login are effective or not to determine whether the request is sent to the specified micro-service or not, and routes the request to different micro-services according to the request path, thereby ensuring simplified development as much as possible.