摘要：【中文】本发明公开了一种键值存储系统键优化方法、存储介质、电子设备及系统，涉及软件工程领域，应用于优化键值存储系统中表意键占用的资源，所述表意键为键值存储系统中与值对应的表意键，其包括，获取键值存储系统的运行时内存空闲率以及表意键。若内存空闲率小于预设的第一阈值，则设置表意键和中间键的对应关系表，并将该对应关系表存储于键值存储系统外部，并根据所述对应关系使用中间键替换键值存储系统内的表意键，所述中间键为长度小于表意键的字符串。若内存空闲率大于预设的第一阈值，且表意键的长度超过预设的第二阈值，则对表意键键进行压缩得到压缩键，使用压缩键替换键值存储系统中的表意键。 【EN】The invention discloses a key optimization method, a storage medium, electronic equipment and a key optimization system of a key value storage system, relates to the field of software engineering, and is applied to optimizing resources occupied by an ideographic key in the key value storage system. If the memory vacancy rate is smaller than a preset first threshold value, setting a corresponding relation table of the ideographic keys and the intermediate keys, storing the corresponding relation table outside the key value storage system, and replacing the ideographic keys in the key value storage system by using the intermediate keys according to the corresponding relation, wherein the intermediate keys are character strings with the length smaller than that of the ideographic keys. If the memory vacancy rate is larger than a preset first threshold value and the length of the ideogram exceeds a preset second threshold value, compressing the ideogram key to obtain a compressed key, and replacing the ideogram in the key value storage system with the compressed key.

摘要：【中文】本发明实施例公开了一种弹幕显示方法、装置及电子设备。本发明实施例方法包括：构建一个显示弹幕的管理队列；建立弹幕textView控件与弹幕的唯一标识符之间的对应关系；获取用户的弹幕显示设置指令；取出当前正在显示的弹幕textView控件对应的目标唯一标识符，找到对应的目标textView控件；计算弹幕平滑变化的时间dT；根据弹幕显示区域高度设置信息，计算在dT时间内运动的垂直距离和垂直速度Vy；根据dT和Vy为所述目标textView控件构建一个垂直动画；将预先获取的水平动画和垂直动画添加到组合动画的对象中；通过调用组合动画开启水平动画和垂直动画的执行，在视频中实现基于所述目标textView控件的弹幕显示。本发明实施例使得弹幕切换更加平滑，弹幕显示更加合理。 【EN】The embodiment of the invention discloses a bullet screen display method and device and electronic equipment. The method provided by the embodiment of the invention comprises the following steps: constructing a management queue for displaying the bullet screen; establishing a corresponding relation between the bullet screen textView control and the unique identifier of the bullet screen; acquiring a bullet screen display setting instruction of a user; taking out a target unique identifier corresponding to the currently displayed bullet screen textView control, and finding out the corresponding target textView control; calculating the time dT of the smooth change of the bullet screen; calculating the vertical distance and the vertical speed Vy of movement within the dT time according to the height setting information of the bullet screen display area; constructing a vertical animation for the target textView control according to the dT and the Vy; adding the horizontal animation and the vertical animation which are acquired in advance into an object of the combined animation; and starting execution of horizontal animation and vertical animation by calling the combined animation, and realizing bullet screen display based on the target textView control in the video. The embodiment of the invention ensures that the bullet screen switching is smoother and the bullet screen display is more reasonable.

摘要：【中文】本发明公开了一种批量注册用户识别方法、存储介质、电子设备及系统，涉及大数据风控领域，该方法包括以下步骤：获取用户账号集合；用户账号集合中，任一用户账号均与其他用户账号做相似度计算，获得相似度w；根据用户账号集合中的各用户账号以及各相似度w构建用户结构图谱；对用户结构图谱进行聚类处理，获得多个用户聚类M；根据各用户聚类M计算各待识别用户账号的注册嫌疑分值；当批量注册嫌疑分值超过预设的批量注册嫌疑阈值时，则表示对应的待识别用户账号存在批量注册的操作。本发明对用户账号进行识别，判断其是否存在批量注册行为，识别效率高，为工作人员对用户账号的管理提供帮助。 【EN】The invention discloses a method, a storage medium, electronic equipment and a system for identifying users registered in batch, which relate to the field of big data wind control, and the method comprises the following steps: acquiring a user account set; in the user account set, similarity calculation is carried out on any user account and other user accounts to obtain similarity w; constructing a user structure map according to each user account and each similarity w in the user account set; clustering the user structure atlas to obtain a plurality of user clusters M; calculating the registration suspicion score of each user account to be identified according to each user cluster M; and when the batch registration suspicion score exceeds a preset batch registration suspicion threshold, indicating that the batch registration operation exists for the corresponding user account to be identified. The invention identifies the user account, judges whether the batch registration behavior exists, has high identification efficiency and provides help for the staff to manage the user account.

摘要：【中文】本发明实施例公开了一种消息处理方法，应用于通信技术领域，包括：服务器接收第一客户端发送的全量消息，将全量消息写入总Kafka集群，按照主题分类，将总Kafka集群中的全量消息分别写入各子Kafka集群，得到按不同主题分类的各子Kafka集群，以使获取到第二客户端发送的数据获取请求时，读取与该数据获取请求对应的子Kafka集群中的用户行为数据，并发送该用户行为数据至第二客户端。本发明实施例还公开了一种消息处理方法、服务器、终端、系统及存储介质，通过将消息分别写入各子Kafka集群，读取消息时，从相应的子Kafka集群中进行读取，使总Kafka集群只负责消息的写入，在不丢失数据时效性同时，减缓总Kafka集群压力。 【EN】The embodiment of the invention discloses a message processing method, which is applied to the technical field of communication and comprises the following steps: the server receives the full message sent by the first client, writes the full message into the total Kafka cluster, classifies according to the theme, writes the full message in the total Kafka cluster into each sub Kafka cluster respectively, obtains each sub Kafka cluster classified according to different themes, so that when a data acquisition request sent by the second client is acquired, reads the user behavior data in the sub Kafka cluster corresponding to the data acquisition request, and sends the user behavior data to the second client. The embodiment of the invention also discloses a message processing method, a server, a terminal, a system and a storage medium, wherein the message is respectively written into each sub Kafka cluster, and when the message is read, the message is read from the corresponding sub Kafka cluster, so that the total Kafka cluster is only responsible for writing the message, and the timeliness of the data loss is avoided, and the pressure of the total Kafka cluster is relieved.

摘要：【中文】本发明公开了基于用户行为的用户聚类基于用户行为的用户聚类方法、存储介质、设备及系统，涉及互联网直播领域，该方法包括获取相同行为下的每个用户的M个行为参数，所述M为大于0的整数。将各个行为参数作为M维空间坐标值，计算任意两个用户之间的欧式距离作为行为差距值。若N个用户之间的行为差距值小于预设阈值，且N大于预设的密集度个数，则将所述N个用户归为一簇，其中，N为大于1的整数。本发明能够更好的将差异较小的用户分为一个组/簇。 【EN】The invention discloses a user clustering method, a storage medium, equipment and a system based on user behaviors, and relates to the field of Internet live broadcast. And taking each behavior parameter as an M-dimensional space coordinate value, and calculating the Euclidean distance between any two users as a behavior gap value. If the behavior gap value between N users is smaller than a preset threshold value and N is larger than the preset intensity number, the N users are classified into a cluster, wherein N is an integer larger than 1. The invention can better divide users with smaller difference into one group/cluster.

摘要：【中文】本申请实施例提供了一种基于直播平台的数据处理方法以及相关设备，用于改善现有技术中用户只能观看直播间中直播的游戏，并不能参与游戏，可以提高用户的体验。该方法包括：客户端获取目标直播间的目标游戏对应的游戏信息；客户端根据游戏信息从目标游戏对应的服务器获取目标游戏对应的游戏数据；客户端根据目标游戏对应的游戏数据将目标游戏的显示界面在目标直播间进行展示；客户端接收用户的第一操作指令；客户端将第一操作指令发送至服务器，以使得服务器获取第一操作指令对应的第一操作数据，并将第一操作数据发送至客户端；客户端将第一操作数据对应的目标游戏的操作在目标游戏的显示界面进行展示。 【EN】The embodiment of the application provides a data processing method based on a live broadcast platform and related equipment, which are used for improving the problem that in the prior art, a user can only watch live games in a live broadcast room and cannot participate in the games, and the user experience can be improved. The method comprises the following steps: the client acquires game information corresponding to a target game in a target live broadcast room; the client acquires game data corresponding to the target game from a server corresponding to the target game according to the game information; the client displays a display interface of the target game in a target live broadcast room according to game data corresponding to the target game; the method comprises the steps that a client receives a first operation instruction of a user; the client sends the first operation instruction to the server so that the server obtains first operation data corresponding to the first operation instruction and sends the first operation data to the client; and the client displays the operation of the target game corresponding to the first operation data on a display interface of the target game.

摘要：【中文】本发明实施例公开了一种弹幕显示方法、装置及电子设备。本发明实施例方法包括：定义一个List集合；构建一个Map变量，在Map变量的codeToTextView集合内部构建弹幕TextView控件与弹幕的唯一标识符之间的映射关系；获取用户的弹幕显示设置指令；根据指令从List集合中取出所有的当前正在显示的弹幕textView控件对应的目标唯一标识符，并通过目标唯一标识符找到对应的目标textView控件；获取目标textView控件当前在弹幕容器对象中的高度位置信息；根据弹幕显示区域高度设置信息和目标textView控件当前在弹幕容器对象中的高度位置信息，调整目标textView控件在弹幕容器对象中的高度位置信息；根据调整后的目标textView控件在弹幕容器对象中的高度位置信息显示视频中弹幕。本发明实施例有效的保证弹幕数据的准确性。 【EN】The embodiment of the invention discloses a bullet screen display method and device and electronic equipment. The method provided by the embodiment of the invention comprises the following steps: defining a List set; constructing a Map variable, and constructing a mapping relation between a barrage TextView control and a unique identifier of the barrage in a codeToTextView set of the Map variable; acquiring a bullet screen display setting instruction of a user; taking out target unique identifiers corresponding to all currently displayed bullet screen textView controls from the List set according to the instruction, and finding out the corresponding target textView controls according to the target unique identifiers; acquiring the current height position information of the target textView control in the bullet screen container object; adjusting the height position information of the target textView control in the bullet screen container object according to the height setting information of the bullet screen display area and the current height position information of the target textView control in the bullet screen container object; and displaying the bullet screen in the video according to the height position information of the adjusted target textView control in the bullet screen container object. The embodiment of the invention effectively ensures the accuracy of the bullet screen data.

摘要：【中文】本发明实施例公开了一种弹幕显示方法、装置及电子设备。本发明实施例方法包括：构建一个显示弹幕的管理队列；建立弹幕textView控件与弹幕的唯一标识符之间的对应关系；通过管理队列添加或删除textView控件对应的唯一标识符；获取用户的弹幕显示设置指令；根据指令从管理队列中取出所有的当前正在显示的弹幕textView控件对应的目标唯一标识符，并通过目标唯一标识符找到对应的目标textView控件；获取目标textView控件当前在弹幕容器对象中的高度位置信息；通过调整目标textView控件在所述弹幕容器对象中的高度位置信息来显示视频中弹幕。本发明实施例中实现弹幕快速切换生效，能够有效的解决切换速度慢的问题，弹幕显示切换效率高。 【EN】The embodiment of the invention discloses a bullet screen display method and device and electronic equipment. The method provided by the embodiment of the invention comprises the following steps: constructing a management queue for displaying the bullet screen; establishing a corresponding relation between the bullet screen textView control and the unique identifier of the bullet screen; adding or deleting a unique identifier corresponding to the textView control through a management queue; acquiring a bullet screen display setting instruction of a user; taking out all target unique identifiers corresponding to currently displayed barrage textView controls from the management queue according to the instruction, and finding out the corresponding target textView controls according to the target unique identifiers; acquiring the current height position information of the target textView control in the bullet screen container object; and displaying the bullet screen in the video by adjusting the height position information of the target textView control in the bullet screen container object. According to the embodiment of the invention, the bullet screen is rapidly switched to take effect, the problem of low switching speed can be effectively solved, and the bullet screen display switching efficiency is high.

摘要：【中文】本发明公开了一种实现长腿特效的方法、存储介质、设备及系统，涉及美颜特效领域，包括基于人体关键点检测算法，定位出图像中人体的人脸区域的顶端中心坐标和底端中心坐标；根据定位出的人脸区域位置坐标，计算出人脸长度L；基于人脸长度和人脸区域下端至人体腰部区域中点距离的比例，计算出人体腰部中心点坐标；对人体腰部中心点坐标的以下区域进行拉伸。本发明基于人脸长度，并根据相关比例，然后精确计算出人体腰部中心点坐标。 【EN】The invention discloses a method, a storage medium, equipment and a system for realizing a long leg special effect, which relate to the field of beauty special effects and comprise the steps of positioning the top center coordinate and the bottom center coordinate of a human face area of a human body in an image based on a human body key point detection algorithm; calculating the length L of the face according to the position coordinates of the positioned face area; calculating the coordinate of the center point of the waist of the human body based on the length of the human face and the proportion of the distance from the lower end of the human face region to the center point of the waist region of the human body; stretching the area below the center point coordinates of the waist of the human body. According to the invention, based on the length of the human face and the related proportion, the coordinates of the center point of the waist of the human body are accurately calculated.

摘要：【中文】本发明公开了一种基于人脸相似度的主播推荐方法、存储介质、设备及系统，涉及互联网直播应用领域，该方法包括获取含有人脸的照片；基于人脸检测算法，检测出获取的照片中的人脸区域，然后基于特征提取算法，提取出人脸区域全连接层的4096维特征；对提取出的每个特征的特征值进行归一化处理计算，然后得到人脸区域对应归一化处理后特征值的特征向量y；进行相似度计算，根据计算出的相识度θ值，按照相识度高低顺序排序选取设定个数的主播，并进一步筛选后推荐给用户。本发明在数据库的所有主播中，匹配出与该人像相似度较高的其它主播，按相似度高低进行排名，然后推荐给观众，满足用户自己个性化定制，自由度较高。 【EN】The invention discloses a anchor recommendation method, a storage medium, equipment and a system based on face similarity, and relates to the field of Internet live broadcast application, wherein the method comprises the steps of obtaining a picture containing a face; detecting a face region in the acquired picture based on a face detection algorithm, and extracting 4096-dimensional features of a face region full-connection layer based on a feature extraction algorithm; carrying out normalization processing calculation on the extracted feature value of each feature, and then obtaining a feature vector y of the feature value of the face region after corresponding normalization processing; and performing similarity calculation, sorting and selecting a set number of anchor according to the calculated acquaintance degree theta value and the acquaintance degree high-low sequence, and recommending the anchor to a user after further screening. According to the invention, other anchor broadcasters with higher similarity to the portrait are matched from all anchor broadcasters in the database, ranking is carried out according to the similarity, and then the ranking is recommended to the audience, so that the personalized customization of the user is met, and the degree of freedom is higher.