Detecting Points of Interest
Introduction
1.1 Information Retrieval
Information retrieval research involves techniques from machine learning and other theoretical models, together with extensive experimentation to develop more accurate, fast and advanced information retrieval and search techniques for a variety of applications.
This course is given by Eduard Hoenkamp.
1.2 Introduction
This tool was implemented as an add-on for the Pacmaas project in order to find hiding places.
This project conciders a game which can be compared to the computer game called Pacman, but with the difference that it is played live in the inner city center of Maastricht. The players are walking in the city and use PDAs to get the needed game information (such as the position of the opponents, the locaton of the pills, ...) from a central server. Moreover, advise is generated for every player participating in the game. These advises contain information about what direction to walk in order to win the game. Detecting POIs allows pacman to improve his winning chances since locations are searched where he can hide when ghosts come too close.
1.3 The goal
The aim of this project is to create an engine which looks up GPS coordinations for points of interest (POIs) given a certain description query, which is provided by the user (and usually tends to be very vague or abstract). For example, a tourist might be interested in places that have to do with art in the city of Maastricht. Since he is not familiar with the city, he could try to google a lot of related terms (eg musea, graffiti art, workshops,... ) in the hope to find such places. It is clear that such a process can consume a lot of time and that it would be nice if the user could be visually presented some POIs on a map of Maastricht by just defining the term art. The engine itself then defines what art actually involves and what locations related to art in Maastricht can be really interesting to visit. This process is a typical implementation of information retrieval.