Design And Implementation Of Hotel-Ordering Platform Search System

With the development of the society and the enhancement of the needs of people’s material and cultural enhancing, tourism as the emerging industry of tertiary industry has been rapidly developed. In recent years, a growing number of travel companies are involved in tourism e-commerce with the development of mobile Internet. Tourism e-commerce offers customers more convenient services. Meanwhile, the booming tourism industry accelerates the development of hotel industry, the Chinese hotel industrial scale has been increased by multiple times. The integration of computer networks and hotel industry has brought new vitality to hotel industry, the hotel online ordering becomes a new consumption pattern for customers. However, there are still a lot of problems in hotel online ordering:the hotel business offers a plenty of complicated hotel information which makes consumers dizzying under such a virtual environment. In order to facilitate the consumers to select the target from the massive information, it is inevitable to establish an exclusive search system for hotel ordering platform.The search system of hotel ordering platform has two parts:search online-service and search offline-service. In the process of design and implementation of search online-service, the work has been done by the author as follow:(1) Completed the design and implementation of Hotword module independently. Hotword module offers popular words of hotel and gives guide to the customers when they have no targets. In order to facilitate the operation of caller, the development of Hotword module is based on Thrift and only offers the interface of service to the caller. The resource of Hotword module is from log information combined with search history, it is updated by ETL file regularly to ensure the accuracy and real-time of service. Hotword module offers multiple categories of hot words to the customers and it will give different words with the change of destination.(2) Completed the design and implementation of Suggest module independently. There are two different applications of Suggest module:Destination-Change sub-module and Keyword-Suggest sub-module. Both of these two sub-modules are based on Thrift. Suggest module uses the structure of Trie tree to implement prefix-matching, and uses redundant storage of words to accomplish fuzzy-matching. This module supports these four input modes:pure Chinese, pure spelling, logogram of spelling and combination of Chinese with spelling. Considering the semantic diversity, Suggest module uses semantic rules to do synonymous and homonyms extension of words. According to these extensions, the module offers intelligent error correction to the customers.(3) Participated in the design and implementation of Basic Search module. In order to ensure the efficiency of index, this module uses the full-text search technology which based on Sphinx engine. For the purpose of accuracy of search, the Finite State Machine is used to establish query syntax tree, and the query syntax tree is optimized to get high performance. The search results are sorted by using location-based weights and TF-IDF these two methods.The search system of hotel ordering platform has been already used now, and the modules designed and implemented by the author are operating normally. These modules not only meet the demand well but also have good scalability to convenient the future maintenance.