Intelligent Hotel Search

Abstract

The travel industry is rapidly moving towards an online market, but still hasn t been able to oust the travel agents role. Travel agents are able to make better suggestions to consumers based on their expertise and experiences. However, in all the current online reservation systems, the user has to enter basic information such as city, date range, and star rating etc. Thus, bulk of the decision making, which is the harder part of the whole online reservation process, is still done by a human, while the computer will only do a simple relational database search. The objective of this thesis is to overcome this limitation and simplify the process to the user by creating an intelligent software travel agent. This document will discuss the design and implementation of the proposed method. The focus of this work is narrowed down to a hotel reservations and the corresponding ontology will only be consisted of Sri Lankan entities and terminologies. While this project touches two broad research areas text processing and ontology reasoning the latter will be the main focus of this work. Instead of doing simple searches, this system is expected to consult its users on their travelling needs, using semantic web techniques. Travel ontology will be created to be compatible with Sri Lankan terminologies. First, the user enters plain text (e.g. Back packers Beach hotels) that will be process by text parsers. These will be forwarded to the ontology that will return hotels, destinations, and keywords for the search. Relational database for hotels and destinations will also be maintained, which will be used to enrich the ontology search results. The output will be the list of hotels that matches user criteria. In evaluating, the main target is to validate the system against actual user inputs. The user queries were collected from questionnaires, prototype logs and Google Adverts . The system exhibits more than seventy percent success rate to actual user searches. From the technologies perspective, Stanford Dependency Parser and the WordsNet have been used for the text processing. Ontocat and OwlApi (With Fact++ reasoner) have been used for ontology searching and reasoning respectively. Mysql database is also maintained to get more information and contents.