Place is a central category in the human experience. Across cultures, individuals describe experiences, express opinions, narrate stories set in and about places. The web provides a large, dynamic corpus of documents describing places from a myriad of viewpoints. Emotions and their expression play an important role in these representations of places, making some places appear joyful and beautiful, and others scary, sad, or even disgusting. In this paper we propose to tap the corpus of place descriptions from the emotional viewpoint, aiming at the development of a framework to model, extract, and analyze emotions relative to places. As first steps in this direction, we focus on place classes, i.e. the types of places that are discussed, such as city, forest, and road. To identify such classes, we design the Place Vocabulary, a linked semantic resource that contains nouns in English that are used to identify natural and built places. Subsequently, we propose a natural language processing technique to extract a multi-dimensional model of place emotion, based on the vocabulary in WordNet-Affect. The technique is applied to a corpus of about 100,000 travel blog posts from travelblog.org, enabling the exploration of the emotional structure of place classes.