Browsing by Author "Martins, Daniel Jorge Ribeiro Nunes"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
- Data extraction in e-commercePublication . Martins, Daniel Jorge Ribeiro Nunes; Cardoso, Pedro J. S.; Lam, RobertoEletronic commerce, know as e-commerce, is a system that consists in buying and selling produtcs/services over the internet. The internet is used by millions of people, making the management of the available information (e.g. competitor analysis market) a very difficult task for those operationg an e-commerce business. So that the managers can better position their companies against competitors, comes the need to create automatic mechanisms to extract information from various web sources (websites). The hotel business is a market where e-commerce is essential since the internet is their biggest selling point, either through sales channels or through their own websites. At the same time, these channels have important information, regarding the reputation of the hotel and their competitors, for instance in the form of guest comments. In this thesis a solution to some of those problems is presented, in which the main focus is the automatic extraction of information from sales channels, such as Booking. com. The extracted information is used to help the hoteliers in the analysis of the prices and opinions of hotel’s guests. That information will be extracted using web robots, able to analyze and interact with web pages, by simulating human behavior. This behavior simulation takes advantage of the navigation patterns present on most sales channels, so that users can easily follow the steps to the final purchase. Briefly describing the overall process, the web robot begins by filling the web site search form with a set of configurable parameters. For each hotel that met the search criteria the most relevant information is extracted, such as: prices, offers, comments and location of the hotel. The collected data is grouped and stored in an intermediate database. Once collected, the data is: (a) used by mathematical prediction models that analyze the prices of the hotels in recent years and generate a forecast of prices that hotels will practice in the future and, (b) used to check the hotel’s reputation taking into account the comments of the guests. This thesis presents a set of four papers resulting in past from the author’s work in project "SRM: Smart Revenue Management" financed by QREN I&DT, no. 38962, with promotor VISUALFORMA - Tecnologias de Informação, SA and co-promoter University of the Algarve.