This repository is created for Chapter 5 of the thesis titled 'Towards Context-Aware Recommender Systems for Tourists'.
The chapter presents the last piece of the evolution of this thesis. This work proposes a new CARS algorithm to recommend tourist routes in a city, using dynamic contextual data sources from platforms like Google Places API and Weather Underground API. The study collected and filtered photos from popular locations in Europe discarded user-generated information, and created routes based on specific rules. We also scraped data from the Google Sightseeing website to gather popular places and contextual details, which were mapped along with the check-in information. Our experimental evaluation shows the difference between contextual data used to create routes.
You need Python 3.7 and higher to run the codes and you need the latest version of the following packages to run both codes and jupyter notebook. The list of this package is in the requirements.txt file. Please use pip to install it.
numpy,
pandas,
matplotlib,
sklearn,
jupyter,
fastread,
Files and their usage
codes -- Contains codes to run algorithms. main.py is the location where everything starts.
datasets -- Contains the latest version of the datasets for two cities
evaluation -- Contains experiment results
notebooks -- Contains notebook for data analysis and exploration.
If you have any questions, please get in touch with [email protected]
If this work is helpful for your research, please consider citing
Many thanks to my supervisors Prof Seth Bullock and Prof Jonathan Lawry for their invaluable input during this research.