摘要：【中文】本发明公开了一种面向移动终端的去中心化目标检测模型训练方法及系统，其中，方法包括以下步骤：获取当前设备产生图片以及其中标注的目标信息；对当前设备产生图片以及其中标注的目标信息进行预处理，得到处理后的数据；利用处理后的数据训练SSD模型，并通过基于K‑L散度衡量模型间的差异性对SSD模型进行联合优化。该方法可以在保护数据隐私的前提下，充分利用用户数据及用户设备的计算能力，进行模型优化，降低目标检测模型的训练成本，并且获得与在集中式数据使用传统训练方式获得的模型具有同等精度。 【EN】The invention discloses a mobile terminal-oriented decentralized target detection model training method and a system, wherein the method comprises the following steps: acquiring a picture generated by current equipment and target information marked in the picture; preprocessing a picture generated by current equipment and target information marked in the picture to obtain processed data; and training the SSD model by using the processed data, and carrying out combined optimization on the SSD model by measuring the difference between the models based on the K-L divergence. The method can fully utilize the user data and the computing power of the user equipment to optimize the model on the premise of protecting the data privacy, reduce the training cost of the target detection model, and obtain the model with the same precision as that of the model obtained by using the traditional training mode on the centralized data.

摘要：【中文】本发明公开了一种基于Web和Kafka的分布式数据集成系统及方法，其中，系统包括：控制台模块，用于为用户提供控制台，使得用户以Web页面操作的形式进行ETL任务的创建和监控；管理服务模块，用于为控制台模块提供管理服务API；模式管理模块，用于管理数据源端的schema和目的地端的schema及其映射；数据抽取模块，用于将管理数据源端的数据抽取数据到消息队列；数据处理模块，用于对数据进行清洗和转换；数据加载模块，用于将数据从消息队列加载到目的地。该系统使得基于Kafka Connect创建ETL实例的过程操作更简单、管理更规范、配置更灵活，并且使得ETL程序耦合度低、容错性高，易于扩展和集成。 【EN】The invention discloses a distributed data integration system and a method based on Web and Kafka, wherein the system comprises: the control console module is used for providing a control console for a user so that the user can establish and monitor an ETL task in a Web page operation mode; the management service module is used for providing a management service API for the console module; the mode management module is used for managing the schema of the data source end, the schema of the destination end and mapping of the schema; the data extraction module is used for extracting data from the management data source end to the message queue; the data processing module is used for cleaning and converting data; and the data loading module is used for loading the data from the message queue to the destination. The system enables the process of creating the ETL instance based on the Kafka Connect to be simpler in operation, more standard in management and more flexible in configuration, and enables the ETL program to be low in coupling degree and high in fault tolerance and to be easy to expand and integrate.

摘要：【中文】本发明公开了一种高阶邻域混合的网络表示学习方法及装置，在原始图卷积层的基础上加入自注意力机制和级联聚集层，其中，方法包括以下步骤：运用自注意力机制将图的拉普拉斯矩阵变换成节点对图注意力矩阵，且训练权重参数学习不同的注意力系数；通过级联聚集层汇聚不同距离信息流，并将上一阶的输出用作下一阶的输入，以控制计算复杂度；确定嵌入向量输出到下游机器学习任务，或者输出分类结果。该方法可以实现真正意义上的端到端训练，有效地提高模型的训练速度，且提出的网络高低阶信息混合学习的思想具有领域可扩展性，简单易实现。 【EN】The invention discloses a high-order neighborhood mixed network representation learning method and a device, wherein a self-attention mechanism and a cascade aggregation layer are added on the basis of an original graph convolution layer, and the method comprises the following steps: transforming the Laplace matrix of the graph into a node map attention matrix by using a self-attention mechanism, and training weight parameters to learn different attention coefficients; converging information flows with different distances through a cascade aggregation layer, and using the output of the previous step as the input of the next step to control the calculation complexity; and determining whether the embedded vector is output to a downstream machine learning task or outputting a classification result. The method can realize end-to-end training in the true sense, effectively improves the training speed of the model, and the proposed idea of network high-low order information hybrid learning has field expandability and is simple and easy to realize.

摘要：【中文】本发明公开了一种基于深度学习的眼科预问诊方法与装置，其中，装置包括：基础数据模块，基础数据模块存储基础数据；算法引擎模块，用于采用基于深度学习的自然语言处理相应对话流程需求，以根据自然语言生成SQL语句；对话流程模块，用于根据SQL语句获取用户的问诊信息，并输出对应的眼科预问诊结果；平台能力模块，用于根据问诊信息和基础数据得到眼科预问诊结果，并将眼科预问诊结果发送至对话流程模块。该装置可以提高就诊前患者信息收集的效率和准确度，实现就诊前智能化收集患者信息，进而提升后期医生诊断效率。 【EN】The invention discloses an ophthalmic pre-interrogation method and device based on deep learning, wherein the device comprises: the basic data module stores basic data; the algorithm engine module is used for processing the corresponding conversation process requirements by adopting the natural language based on deep learning so as to generate SQL sentences according to the natural language; the dialogue flow module is used for acquiring inquiry information of the user according to the SQL statement and outputting a corresponding ophthalmic pre-inquiry result; and the platform capability module is used for obtaining an ophthalmic pre-inquiry result according to the inquiry information and the basic data and sending the ophthalmic pre-inquiry result to the conversation process module. The device can improve the efficiency and the degree of accuracy of patient information collection before seeing a doctor, realizes intellectuality collection patient information before seeing a doctor, and then promotes later stage doctor diagnostic efficiency.

摘要：【中文】本发明公开了一种眼科预问诊装置的多轮对话实现方法及装置，其中，方法包括以下步骤：依次采集患者的基本信息、现病史信息和既往史信息；根据基本信息、现病史信息、既往史信息和眼科医学知识库预测患者的疾病信息；采集患者的追问信息，并根据追问信息和用户口语表达表生成答疑信息，及生成信息清单。该方法采用了整体串行，部分并行的对话设计流程，从而更好的符合场景要求，大幅度提升了流程实现速度和减小了标注成本。 【EN】The invention discloses a multi-round conversation realization method and a multi-round conversation realization device of an ophthalmic pre-inquiry device, wherein the method comprises the following steps: sequentially collecting basic information, current medical history information and past history information of a patient; predicting disease information of the patient according to the basic information, the current medical history information, the past medical history information and the ophthalmologic medical knowledge base; collecting the question-following information of the patient, generating question-answering information according to the question-following information and the spoken language expression table of the user, and generating an information list. The method adopts the whole serial and partial parallel dialogue design flow, thereby better meeting the scene requirement, greatly improving the flow realization speed and reducing the marking cost.

摘要：【中文】本发明公开了一种基于JSON树的可视化API组合方法及系统，其中，方法包括以下步骤：在同步解释执行时，根据父节点的API执行结果确定要执行的孩子节点，并且根据预设的MVEL脚本机制进行http请求响应消息的格式转换，以将不同风格的接口统一封装；在异步解释执行时，引入RabbitMQ消息队列，并且引入Redis缓存以及多线程机制，以实现事件的订阅及发布机制。该方法可以将不同风格的接口快速统一封装，并有效提高整个系统的性能，简单易实现。 【EN】The invention discloses a JSON tree-based visual API combination method and system, wherein the method comprises the following steps: during synchronous interpretation execution, determining child nodes to be executed according to the API execution result of the father node, and performing format conversion of http request response messages according to a preset MVEL script mechanism so as to uniformly package interfaces of different styles; during asynchronous interpretation execution, a RabbitMQ message queue is introduced, and Redis caching and a multi-thread mechanism are introduced to realize a subscription and publication mechanism of events. The method can quickly and uniformly package the interfaces with different styles, effectively improves the performance of the whole system, and is simple and easy to implement.

摘要：【中文】本发明公开了一种基于多数据源的统一数据服务开放方法及系统，其中，系统包括：动态数据服务子系统，用于将Web服务配置页面传来的SQL语句进行解析，生成相关数据源查询任务、用户指定服务的输入数据、输出数据模型以及对应的Restful服务，并且当发生服务请求时，发送服务对应的多个数据源查询任务，以从大数据系统中获取目标数据，并将最终结果返回给调用者；多数据源查询引擎子系统，用于将动态数据服务子系统传来的多个数据源查询任务，封装成具体查询任务，通过数据源连接中间件向不同平台、数据源请求目标数据，实现和大数据系统底层组件的交互，且将最终结果进行统一封装返回并缓存。该系统具有高灵活性和高效性的优点，简单易实现。 【EN】The invention discloses a unified data service opening method and a system based on multiple data sources, wherein the system comprises the following steps: the dynamic data service subsystem is used for analyzing SQL sentences transmitted by the Web service configuration page, generating related data source query tasks, input data of a user-specified service, an output data model and a corresponding Restful service, and sending a plurality of data source query tasks corresponding to the service when a service request occurs so as to obtain target data from the big data system and return a final result to a caller; and the multi-data source query engine subsystem is used for encapsulating a plurality of data source query tasks transmitted by the dynamic data service subsystem into specific query tasks, requesting target data from different platforms and data sources through the data source connection middleware, realizing interaction with a bottom component of the big data system, and uniformly encapsulating and returning final results for caching. The system has the advantages of high flexibility and high efficiency, and is simple and easy to implement.

摘要：【中文】本发明公开了一种分布式机器学习可视化装置，包括：组件模块、机器学习工作模块、配置模块、日志模块和报告模块，其中，组件模块用于提供拖拽组件以及报告的可查看编辑组件；机器学习工作模块用于为机器学习提供工作区域，允许将拖拽组件拖拽进入本模块，并进行流程图式连接；配置模块用于提供组件配置内容，并根据当前配置动态更新；日志模块，用于提供当前运行状态；报告模块用于在生成报告时，提供当前工作区域内各节点的详情以及运行结果的可视化内容。该装置可以为非机器学习专业人员提供一个门槛低、可视化程度高的一个分布式机器学习平台，并可以有效应对海量数据以及高精度机器学习的问题，简单易实现。 【EN】The invention discloses a distributed machine learning visualization device, which comprises: the system comprises a component module, a machine learning work module, a configuration module, a log module and a report module, wherein the component module is used for providing a dragging component and a viewable editing component of a report; the machine learning working module is used for providing a working area for machine learning, allowing the dragging component to be dragged into the machine learning working module, and performing flow diagram connection; the configuration module is used for providing component configuration content and dynamically updating according to the current configuration; the log module is used for providing a current running state; and the report module is used for providing the details of each node in the current working area and the visualized content of the operation result when generating a report. The device can provide a distributed machine learning platform with low threshold and high visualization degree for non-machine learning professionals, can effectively deal with the problems of mass data and high-precision machine learning, and is simple and easy to realize.

摘要：【中文】本发明公开了一种基于人在回路的混合增强智能需求精准感知方法及系统，其中，系统包括：知识图谱子系统，用于存储科技资源子图谱与用户/企业信息子图谱，以根据用户需求确定不同实体之间的关系；对话子系统，用于以自然语言的方式和用户进行交互，收集用户需求，并将感知结果实时返回给用户；特征感知与推荐子系统，用于根据对话子系统和知识图谱子系统的客观数据、痕迹数据，整合相关数据生成感知结果，并生成推荐信息推荐至用户。该系统利用对话系统建立起人在回路，结合用户痕迹数据和科技资源数据进行用户需求充分挖掘与特征感知，实现用户需求精准感知，有效解决现有技术存在没有利用客观数据、无法深度感知用户需求等问题。 【EN】The invention discloses a human-in-loop-based method and a human-in-loop-based system for accurately sensing hybrid enhanced intelligence demands, wherein the system comprises: the knowledge graph subsystem is used for storing a scientific and technological resource sub-graph and a user/enterprise information sub-graph so as to determine the relation between different entities according to the user requirements; the dialogue subsystem is used for interacting with the user in a natural language mode, collecting user requirements and returning a perception result to the user in real time; and the characteristic perception and recommendation subsystem is used for integrating related data to generate a perception result and generating recommendation information to be recommended to a user according to the objective data and trace data of the conversation subsystem and the knowledge map subsystem. The system utilizes a dialogue system to establish a human-in-the-loop, fully excavates the user requirements and senses the characteristics by combining the user trace data and the scientific and technological resource data, realizes accurate sensing of the user requirements, and effectively solves the problems that objective data is not utilized, the user requirements cannot be deeply sensed in the prior art and the like.

摘要：【中文】本发明公开了一种基于共指消解的科技服务资源分类体系构建方法及装置，其中，方法包括以下步骤：采集至少一个分类体系；对至少一个分类体系中不满足第一预设条件的类目进行预处理；根据共指消解规则整合预处理后的至少一个分类体系，并根据整合后的至少一个分类体系构建最终分类体系。该方法应用共指消解思想来构建科技服务资源分类体系，整合了各大平台的优点，构建的分类体系也更加科学和专业、也更加完善和标准。 【EN】The invention discloses a scientific and technological service resource classification system construction method and device based on coreference resolution, wherein the method comprises the following steps: collecting at least one classification system; preprocessing categories which do not meet a first preset condition in at least one classification system; and integrating the at least one pretreated classification system according to the coreference resolution rule, and constructing a final classification system according to the at least one integrated classification system. The method applies the coreference resolution idea to construct a scientific and technological service resource classification system, integrates the advantages of each large platform, and the constructed classification system is more scientific and professional and more perfect and standard.