Approaches for the Clustering of Geographic Metadata and the Automatic Detection of Quasi-Spatial Dataset Series

Javier Lacasta,Francisco Javier Lopez-Pellicer,Rubén Béjar,Javier Zarazaga-Soria,Javier Nogueras-Iso

doi:10.3390/ijgi11020087

Javier Lacasta, Francisco Javier Lopez-Pellicer + Show 3 more

Open Access

https://doi.org/10.3390/ijgi11020087

Copy DOI

Abstract

The discrete representation of resources in geospatial catalogues affects their information retrieval performance. The performance could be improved by using automatically generated clusters of related resources, which we name quasi-spatial dataset series. This work evaluates whether a clustering process can create quasi-spatial dataset series using only textual information from metadata elements. We assess the combination of different kinds of text cleaning approaches, word and sentence-embeddings representations (Word2Vec, GloVe, FastText, ELMo, Sentence BERT, and Universal Sentence Encoder), and clustering techniques (K-Means, DBSCAN, OPTICS, and agglomerative clustering) for the task. The results demonstrate that combining word-embeddings representations with an agglomerative-based clustering creates better quasi-spatial dataset series than the other approaches. In addition, we have found that the ELMo representation with agglomerative clustering produces good results without any preprocessing step for text cleaning.

Highlights

Geospatial catalogues are discovery and access systems that use metadata as the target for querying geospatial resources [1]
This paper has shown how spatial fragmentation in geospatial catalogues can cause ineffectiveness in “concept at location” searches
We have summarised existing information retrieval (IR) problems and described the existing dissonance between the continuous nature of geospatial information and the digital library based structure of these metadata catalogues

Summary

Introduction

Geospatial catalogues are discovery and access systems that use metadata as the target for querying geospatial resources [1]. The objective of any catalogue storage system is to make the contained resources findable, accessible, interoperable, and reusable, which is commonly known as the FAIR principle [3]. Users expect that geospatial catalogues will return information based on their conceptual, spatial, and temporal relevance with respect to a query. This approach is natural, but it is known to be ineffective in the real world without improving the different geospatial catalogue components with intelligent metadata curation methods or the use of advanced search engines [5]. There are multiple works in the literature proposing search improvements in the fulfilment of FAIR principles through adding semantics and ontologies to the metadata, using new ranking algorithms, or boosting data storage [6,7,8,9]

Objectives

Results

Discussion

Conclusion