酒店订购平台搜索系统的设计与实现

发布时间：2018-08-16 18:38

【摘要】：随着社会的进步,人们对物质文化需求的提高,旅游业作为第三产业中的新兴产业得到了快速地发展。而近年来,移动互联网的发展使越来越多的企业涉足旅游电子商务,通过旅游电子商务这种模式为客户提供更加便捷的服务。同时,旅游业的蓬勃发展带动了酒店行业,中国酒店业的产业规模成倍数地加速发展。计算机网络和酒店业的融合为酒店业带来了新的生机,酒店网络化订购成为了酒店用户新的消费模式。然而,酒店网上订购这样的电子商务模式依然存在很多问题。在虚拟环境下,酒店商家提供的酒店信息和数量多而繁杂,让消费者目不暇接。为了能让酒店用户从海量数据中挑选出自己所需的内容,开发专属酒店订购平台的搜索系统是必然的趋势。酒店订购平台搜索系统分为酒店搜索在线服务和酒店搜索离线服务两大部分。在酒店搜索在线服务的设计开发中,作者完成了如下工作：(1)独立完成了热词服务模块的设计、开发和功能性测试工作。热词服务为酒店用户提供酒店相关的热门词,在用户无法抉择目标或缺乏目标时给予导向,由此来提高用户的购买率。热词服务模块采用Thrift框架进行开发,对外只提供客户端服务接口,极大地方便了使用方调用热词服务。为保证热词服务的准确性和实时性,日志信息在关联用户搜索记录后作为热词的词源,并使用ETL文件定时地更新后写入到MySql数据库做为热词候选。为最大程度地满足酒店用户的需求,热词服务提供了类别丰富的热门词汇,且会根据用户所选目的地的不同而展示不同目的地下的热词。(2)独立完成了智能提示服务模块的设计、开发和功能性测试工作。智能提示服务模块在搜索系统中有两处不同的应用,分为目的地切换子模块和关键词搜索提示子模块两部分,两个子模块均是独立的服务模块并采用Thrift框架开发。智能提示服务模块基于Trie树实现了前缀词匹配,并通过冗余存储实现了模糊匹配功能。该服务支持纯汉字、纯拼音、拼音简拼和汉字拼音混合四种输入方式并返回相应的汉字推荐结果。考虑到语义多样性,智能提示服务根据规则对词源进行了同音同义的扩展,实现了智能纠错功能。为使用户可以直观地比较自己的输入内容和推荐结果,对推荐结果的对应部分进行了高亮处理。(3)参与完成了基础检索模块的设计与部分子功能的开发工作。为保证索引建立的高效性,采用了基于Sphinx引擎的全文检索技术。为保证检索的准确性,使用了有限状态机构造查询语法树,增加了查询语法树优化部分来提高性能,并采用了基于地理位置权重和基于TF-IDF的方法对检索结果进行排序。目前,酒店订购平台搜索系统已进入上线服务状态,作者负责设计和开发的各个模块均运行正常,在满足业务需求的基础上,各模块拥有良好的扩展性,简便了后续的维护工作。
[Abstract]:With the progress of society and the improvement of people's demand for material and culture, tourism, as a new industry in the tertiary industry, has been developed rapidly. In recent years, with the development of mobile internet, more and more enterprises are involved in tourism electronic commerce, which provides more convenient service for customers. At the same time, the vigorous development of tourism has led to the hotel industry, China's hotel industry scale multiple accelerated development. The combination of computer network and hotel industry has brought new vitality to hotel industry, and hotel network ordering has become a new consumption mode for hotel users. However, there are still many problems in the e-commerce mode of hotel online ordering. In the virtual environment, hotel merchants provide a large number of hotel information and complex, so that consumers dizzying. In order to enable hotel users to select their own content from the massive data, it is an inevitable trend to develop a search system with exclusive hotel ordering platform. Hotel order platform search system is divided into two parts: hotel search online service and hotel offline search service. In the design and development of hotel search online service, the author has completed the following work: (1) the design, development and functional testing of hot word service module have been completed independently. Hot word service provides hotel users with hotel-related hot words and gives them guidance when they can not choose or lack goals, thereby increasing the purchase rate of the users. The hot word service module is developed by Thrift framework and only provides the client service interface, which greatly facilitates the user to invoke the hot word service. In order to ensure the accuracy and real time of hot words service, log information is used as the etymology of hot words after the associated users search the records, and the ETL files are regularly updated and written to MySql database as hot word candidates. In order to meet the needs of hotel users to the greatest extent, hot word service provides a variety of hot words and shows hot words in different destinations according to the user's choice. (2) the design of intelligent prompt service module is completed independently. Development and functional testing. The intelligent prompt service module has two different applications in the search system. It is divided into two parts: destination switch submodule and keyword search prompt module. The two submodules are independent service modules and developed with Thrift framework. Intelligent prompt service module realizes prefix word matching based on Trie tree and fuzzy matching function by redundant storage. The service supports four input modes: pure Chinese character, pure pinyin, simple spelling and pinyin, and returns the corresponding Chinese character recommendation results. Considering the semantic diversity, the intelligent prompt service extends the synonym of etymology according to the rules and realizes the function of intelligent error correction. In order to enable users to compare their input content and recommendation results intuitively, the corresponding parts of the recommended results are highlighted. (3) the design of the basic retrieval module and the development of some sub-functions are completed. In order to ensure the efficiency of index building, the full-text retrieval technology based on Sphinx engine is adopted. In order to ensure the accuracy of retrieval, the query syntax tree is built with finite state mechanism, and the optimization part of query syntax tree is added to improve the performance. The method based on geographical position weight and TF-IDF is used to sort the retrieval results. At present, the hotel order platform search system has entered the online service state, the author is responsible for the design and development of each module is running normally, on the basis of meeting business requirements, each module has good scalability, Easy to follow up maintenance work.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP311.52;TP391.3

【相似文献】