![]() A study by Radke et al. ( 2019) that was focused specifically on this task reported relatively poor classifier performance, with the best precision only being 0.63. While there is some previous work on the demanding task of extraction of spatial relations in specifically geo-spatial contexts (e.g., Khan et al., 2013 Zenasni et al., 2018 Zhang et al., 2009, 2011), little progress has been made on development of methods focused specifically on identifying geo-spatial senses of prepositions and distinguishing them from other-spatial senses. It is an integral component of processes of automated spatial relation extraction that detect the located object, the spatial relational term (spatial indicator), and the reference object (D'Souza & Ng, 2015 Kordjamshidi et al., 2011 Rahgooy et al., 2018). The task of determining the sense of prepositions (or other parts of speech) is sometimes referred to as spatial indicator classification (Kordjamshidi et al., 2017). The distinction that we draw here between generic spatial senses and geo-spatial senses is important, because many locative expressions describe non-geographic situations (e.g., in the so-called table-top space) that cannot normally be geo-referenced. Furthermore, when a preposition is used in a spatial sense, it is very often the case that the spatial sense is not actually geo-spatial. For example, in the locative expression “Otaki Gorge Road near Otaki Forks,” the preposition “near” indicates a spatial proximity relationship between the named places “Otaki Gorge Road” and “Otaki Forks.” In spatial relational expressions of this form, “Otaki Gorge Road” is regarded as the located object (or trajector, locatum or figure) while “Otaki Forks” is the reference location (also referred to as a landmark, relatum, or ground).Īutomated detection of the presence of locative expressions that describe the spatial relationship of one entity to another is a challenge however in that preposition terms commonly used to convey spatial relations, such as near, in, and at, can also be used in a non-spatial sense. Descriptions of locations typically include relationships between some entity or event and a geographic place, where the spatial relationship is very commonly expressed with a preposition (Herskovits, 1987), though parts of speech such as verbs can also be employed (Dittrich et al., 2015). ![]() Natural language texts contain a great deal of geo-spatial information which, if extracted and geo-referenced to locations on the Earth's surface, constitute a massive potential source of data that could be exploited in geographic information systems. ![]() We also conducted experiments to detect generic spatial sense, in which the best F1 score, of 0.95, was again obtained with XLNet. The best performance was obtained with the Bidirectional Encoder Representation from Transformer-based XLNet transformer model, with a best precision of 0.96 and an F1 score of 0.94 when evaluated on a corpus of natural language expressions that were annotated for this task. ![]() We conduct machine learning experiments that demonstrate the clear benefit for geo-spatial sense detection of using transformer model deep learning methods when compared with a variety of methods, that include Naive Bayes, support vector machine, and random forest classifiers with handcrafted linguistic features, and a bag of words approach with a meta-classifier that adds geo-spatial features. This work focuses on disambiguation of prepositions in natural language, with the goal of distinguishing whether a preposition is used in a specifically geo-spatial sense. ![]() Automatic detection of the use of prepositions in a spatial and in particular a geo-spatial sense that refers to geographic context is of interest in supporting automated methods for determining the actual geographic location referred to by locative expressions. Thus, in the locative expressions “New York in the United States” and “the house on the river” the prepositions “in” and “on,” respectively, serve to communicate the relationships in space between the subject and object of the preposition. Spatial relations in natural language are frequently expressed through prepositions. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |