Design And Implementation Of A Vertical Search Engine For Online Meal Ordering

With the rapid development of Internet technology, the Internet has already become an important platform for global information dissemination. As of 2015, the number of Internet users in China reached 668 million, and the number of websites reached 4.13 million, which covered almost all areas. In the tide of the Internet, the traditional catering industry is also facing a huge challenge. People’s eating habits is gradually developing towards the direction of home delivery anytime and anywhere as well. This trend is particularly obvious among students who are more and more relying on the Internet to get access to information. Although our country’s online ordering service is currently still in a fledgling and fumble period, and there is still some gap compared with foreign countries, its research prospect is quite cheeerful. How to allow users to quickly and accurately find out food information they need from mass data has become one of the pressing problems to be solved, which is the purpose of this thesis lies in.Based on the research on the web crawler strategy, an improved and multi-strategy fused web crawler is put forward. The current web crawler strategy can be divided into two kinds. One is based on the URL link structure, and the other is based on the page content. The former strategy only considers the relationship between the pages, without taking into account the relevance of the search topics, while the latter is just the opposite.This thesis, according to the two strategies, puts forward a multi-strategy fused web crawler, an improved version with higher precision.This thesis analyzes Lucene’s own scoring mechanism of search results and points out its deficiencies in the present system, then puts up an improved scoring mechanism.The new scoring mechanismm mainly considers some actual needs of the users, such as monthly sales, restaurant ratings, whether goods are sold out or not,etc.Besides, this improved scoring mechanism is calculated based on Lucene’s own rating results. If Lucene’s own results show large monthly sales, high restaurant rating and with goods in stock, then the new score will be correspondly higher as well, which also means the sorting of the final results will be more closer to the first place.The new mechanism not only considers the relevance of the pages, but also considers the specific parameters in the system, which is more in line with the users’ actual needs.Based on the foregoing analysis and improvement mentioned in this thesis, an online-ordering oriented vertical search engine is designed and implementated, and a series of functional tests are conducted. The test results show that the improved crawler model has higher precision, the improved scoring mechanism is more in line with the actual needs of the system and can better improve the user experience, and the achieved search engine can generate more professinoal search results.