Spatial relation extraction is the process of identifying geographic entities from text and determining their corresponding spatial relations. Traditional spatial relation extraction mainly uses rule-based pattern matching, supervised learning-based or unsupervised learning-based methods. However, these methods suffer from poor time-sensitive, high labor cost and high dependence on large-scale data. With the development of pre-trained language models greatly alleviating the shortcomings of traditional methods, supervised learning methods incorporating pre-trained language models have become the mainstream relation extraction methods. Pipeline extraction and joint extraction, as the two most dominant ideas of relation extraction, both have obtained good performance on different datasets, and whether to share the contextual information of entities and relations is the main differences between the two ideas. In this paper, we compare the performance of two ideas oriented to spatial relation extraction based on Chinese corpus data in the field of geography and verify which method based on pre-trained language models is more suitable for Chinese spatial relation extraction. We fine-tuned the hyperparameters of the two models to optimize the extraction accuracy before the comparison experiments. The results of the comparison experiments show that pipeline extraction performs better than joint extraction of spatial relation extraction for Chinese text data with sentence granularity, because different tasks have different focus on contextual information, and it is difficult to take account into the needs of both tasks by sharing contextual information. In addition, we further compare the performance of the two models with the rule-based template approach in extracting topological, directional and distance relations, summarize the shortcomings of this experiment and provide an outlook for future work.

In Geographical Information Systems, geo-coding is used for the task of mapping from implicitly geo-referenced data to explicitly geo-referenced coordinates. At present, an enormous amount of implicitly geo-referenced information is hidden in unstructured text, e.g., Wikipedia, social data and news. Toponym recognition is the foundation of mining this useful geo-referenced information by identifying words as toponyms in text. In this paper, we propose an adapted toponym recognition approach based on deep belief network (DBN) by exploring two key issues: word representation and model interpretation. A Skip-Gram model is used in the word representation process to represent words with contextual information that are ignored by current word representation models. We then determine the core hyper-parameters of the DBN model by illustrating the relationship between the performance and the hyper-parameters, e.g., vector dimensionality, DBN structures and probability thresholds. The experiments evaluate the performance of the Skip-Gram model implemented by the Word2Vec open-source tool, determine stable hyper-parameters and compare our approach with a conditional random field (CRF) based approach. The experimental results show that the DBN model outperforms the CRF model with smaller corpus. When the corpus size is large enough, their statistical metrics become approaching. However, their recognition results express differences and complementarity on different kinds of toponyms. More importantly, combining their results can directly improve the performance of toponym recognition relative to their individual performances. It seems that the scale of the corpus has an obvious effect on the performance of toponym recognition. Generally, there is no adequate tagged corpus on specific toponym recognition tasks, especially in the era of Big Data. In conclusion, we believe that the DBN-based approach is a promising and powerful method to extract geo-referenced information from text in the future.

Typhoon events can cause serious environmental damage and economic losses. Understanding the development of typhoon events will provide valuable knowledge for disaster prevention and mitigation. In the age of big data, the sharp contrast between the sudden increase of mass information and the lack of a knowledge appreciation mechanism appears. There is an urgent need to promote the transformation of information services to knowledge services in the field of hazard management. Knowledge representation, as a strategy for symbolizing and formalizing knowledge, affects knowledge acquisition, storage, management, and application, and is the basis and prerequisite for the implementation of knowledge services. Based on the evolution law of typhoon events and human cognitive habits, a formal knowledge representation method for typhoon events (TKRM) is proposed in this paper. First, by analyzing the evolution characteristics of typhoon events, the TKRM framework with three layers consisting of “event–process–state” was constructed, which was used to describe the knowledge composition and relationship of the different granularity of typhoon events. Second, the formal representation of the TKRM framework was formed by using a finite state machine (FSM) as a reference, taking time and location as the basic conditions, and extending the hierarchical and parallel representation mechanism. Finally, the rationality and practical value of the TKRM were verified using a case study.

Naïve Geography, intelligent geographical information systems (GIS), and spatial data mining especially from social media all rely on natural-language spatial relations (NLSR) terms to incorporate commonsense spatial knowledge into conventional GIS and to enhance the semantic interoperability of spatial information in social media data. Yet, the inherent fuzziness of NLSR terms makes them challenging to interpret. This study proposes to interpret the fuzzy semantics of NLSR terms using the fuzzy random forest (FRF) algorithm. Based on a large number of fuzzy samples acquired by transforming a set of crisp samples with the random forest algorithm, two FRF models with different membership assembling strategies are trained to obtain the fuzzy interpretation of three line-region geometric representations using 69 NLSR terms. Experimental results demonstrate that the two FRF models achieve good accuracy in interpreting line-region geometric representations using fuzzy NLSR terms. In addition, fuzzy classification of FRF can interpret the fuzzy semantics of NLSR terms more fully than their crisp counterparts.

Natural language spatial relations have drawn much attention in recent years for bridging the gap between natural language and geographical information systems. Annotating spatial relations in corpora is the precondition for resolution of geographical information in text. This paper proposes an annotation scheme for marking expressions of spatial relations in natural language. This scheme is based on a hierarchical classification of geographical feature types and spatial relations. An annotation approach is illustrated with examples of Chinese text, and finally the natural language processing tool GATE is used to test the proposed scheme. This work can be useful for disambiguation of natural language spatial relations, integration of natural language with GIS services and spatial reasoning.