27 oktober 2020
WordCrowd is a dynamic location-based service that analyzes and visualizes geolocated social media data. By spatially clustering the data, areas of interest and their descriptions can be extracted and compared on different geographical scales. When walking through the city, the application visualizes the nearest areas of interest and presents these in a word cloud. By aggregating the data based on the country of origin of the original poster, we discover differences and similarities in tourist interest between different countries.
Next, the approach was extended to also visualize linked open data sources. By extracting and localizing place names and persons of interest mentioned in Wikipedia articles and linking them with one another we can automatically build a basic LBS application for any given Wikipedia page. Currently, we are looking to incorporate more advanced NLP techniques in order to visualize and link any given textual data source (e.g. book or article) in an interactive LBS web application.