What Can I Do There? Extracting Place Functionality Based on Analysis of User-Generated Textual Contents

Abstract

Recently, due to the rapid increase of social networks, user-generated textual contents have grown significantly. Sense of place emerges when a place is functionally different from its surroundings and is therefore recognizable. This paper aims to extract place functionality by analyzing user-generated textual contents. Therefore, using different natural language processing and machine learning methods, we show how much place functionality can be represented through the whole text with lemmatized words and only action verbs in descrining place. The evaluation results demonstrate that using the whole text achieves better performance and among different classifiers, SVM leads to better accuracy and f1-score.