Author: Jasmin Wellnitz
Tutor: Bruno Baruque Zanón
Department: Ingeniería Civil, Área de Lenguajes y Sistemas Informáticos
Universidad de Burgos
Telegram Chatbot: @touristrecommenderbot
The chatbot takes a bit to respond if not accessed for a while, this can be explained by the Platform as a service running behaviour.
This project deals with the development of a chatbot application that is able to recommend tourist points of interest customized to the user's preferences. The chatbot can be accessed using the instant messaging app Telegram.
The project's main focus lies on the development of a server-side architecture to the Telegram interface. In order to do so, a Java web application is set up that uses the natural language processing platform API.AI to parse the user input. The application is deployed to the platform-as-a-service Heroku.
The recommendations given to the user are based on the information the user shares with the chatbot and already existing data of similar users. The applied recommendation algorithm combines the two most important approaches of recommendation theory, content-based and collaborative filtering.
Furthermore, a geographic database is set up using PostGis, providing the needed tourist information the recommendations are based on. In the presented example of the application, geographic information of Barcelona, Spain, is used and therefore limiting the provided recommendations to this city.