【作者】 熊鹰；

【导师】 刘文予；

【作者基本信息】 华中科技大学 ， 电子与通信工程， 2016， 硕士

【摘要】 近年来,伴随着移动通信技术和智能手机终端的发展,大量移动应用不断地涌现,移动互联网产业迎来了前所未有的大发展。而在移动互联网的众多领域中,线上到线下(Online to Offline,O2O)产业是当前发展最引人注目的领域之一。目前我国各大中型城市中,各大热门餐厅、银行网点、市政大厅、医院等经常出现排队场景。在当前线下排队场景中,排队者通常需要在排队现场拿号,而且排队者为了防止过号,通常只能留在现场等待,这种传统排队模式无疑浪费了排队者大量等待的时间。考虑到上述传统线下排队场景中这些突出问题,本文提出通过设计和实现一款移动应用使用户能够在线上取号排队,线下入座消费,从而节省排队者线下排队时间,解决当前线下排队场景中排队者需在现场等待的问题。在对同类型应用的研究中,本文发现目前各大移动应用的评论中普遍存在着大量垃圾评论,这些垃圾评论对互联网环境和用户体验造成了巨大的负面影响。本文在深入研究前沿的垃圾信息过滤算法的基础上,与以往算法针对恶意用户行为特征进行研究不同,本文提出基于对商户评论量异常变化以及可疑商户中相似评论的检测来过滤垃圾评论。为了更好适应线上排队系统的需求,本文提出了融合评论内容和可疑商户评论特征的垃圾评论过滤的方法。实际应用中,为了加快垃圾评论过滤的处理速率,本文根据系统的整体架构提出如下优化方案:一方面,基于可疑商户评论特征的垃圾评论过滤在实际应用中往往需要大量消耗大量服务器资源,故在实现中对该模块的运行时段限定为服务器的空闲时段;另一方面,考虑到服务器在承担高并发的情况下,大量文本的分词和去停用词的任务给服务器将带来巨大压力,本文提出将分词和去停用词两大模块部署在客户端上,用来分担服务器在高并发下的压力。这样使得服务器与客户端能够协同运行,增强了整个系统的工作效率的同时,又保障了系统的稳定性。最后,本文详细介绍了对整个系统的设计与实现方案,并呈现系统设计与实现的结果。更多还原

【Abstract】 In recent years, with the rapid development of the mobile communication technology and the great improvement of smart mobile phones, the mobile internet ushered in unprecedented new opportunities for its development. O2O(Online to Offline) is one of the most noticeably developing area in the mobile internet.It is easily found that there existing a large amount of queuing in restaurants, shopping malls, municipal halls, and even at hospitals in most cities of China. It is undoubtedly a waste of the queuers’ time that the queuers usually take the number at the scene and stand in the field waiting for their turn in traditional queuing scenarios. Considering the disadvantages of the traditional queuing offline, we propose to design and implement an Android App that will help the queuers save their time, and in which users are able to take the number online and receive services offline.In the study for the similar type of Apps, it has been found that nearly all major mobile O2 O Apps exist a lot of spam comments, which caused a negative influence on the internet environment and user experience. On the basis of deep study of previous research, a spam comments filtering algorithm has been proposed by detecting the abnormal change of the comments amount and detecting similar comments of the suspected stores, which is different from the algorithm based on malicious users’ behavioral characteristics. Furthermore, for the purpose of practical use in the online queuing system, the method of fusion spam comments filtering based on the content of the spam comments and the characteristics of the suspected stores are applied in the online queuing system.From the perspective of the practical application in the online queuing system, the following optimization scheme that was based on the overall online queuing system architecture has been presented: On one hand, due to the spam filtering algorithm based on suspected stores’ characteristics may cause massive consuming of the server’s resources, this part of processing starts only when the server is idle. On the other hand, taking the server in the case of high concurrency into account, a large number of tasks of word segment, removal of stop words and sensitive words filtering will bring enormous pressure to the server. Therefore, the module of word segment, removal of stop words and sensitive words filtering are deployed on the client to share the server’s concurrent pressure. This not only allows the server and client a better co-operation and to enhance the efficiency of the entire system at the same time, but also to ensure the stability of the entire system.Finally, a detailed description of the entire system design and implementation and the final results achieved have been presented.更多还原