Data is the central element of a geographic information system (GIS) and its cost is often high because of the substantial investment that allows its production. However, these data are often restricted to a service or a category of users. This has highlighted the need to propose and optimize the means of enriching spatial information relevant to a larger number of users. In this chapter, a data enrichment approach that integrates recent advances in machine learning; more precisely, the use of deep learning to optimize the enrichment of GDBs is proposed, specifically, during the topic identification phase. The evaluation of the approach was completed showing its performance.
TopBackground
The enrichment of the GIS allows the acquisition of additional information essential for good decision-making. We talk about spatial enrichment and semantic enrichment. Concerning the spatial aspect and within the framework of the generalization process (Plazanet, 1996), for example, enrichment provides the GDBs with information in terms of structure of forms, knowledge relating to the order of operations and the appropriate algorithms. Another stream of work relates to the semantic (also called factual or descriptive) aspect of GDBs. In this category, we can cite Metacarta (Kornai, 2005) and GeoNode (Hyland et al., 1999).
The Metacarta project accomplished the enrichment with its Geographic Text Search (GTS) product. GTS allows the linking of text documents to geographic features located on the map to enrich the GDB data. MetaCarta GTS is offered as an extension to the ArcGIS geographic information system.
GeoNode (Geographic News On Demand Environment) exploits the information extraction technique to achieve enrichment, via the Alembic system. GeoNode allows the extraction of named entities and associated events to be visualized in a geospatial way. Moreover, ArcView GIS supports GeoNode.
Persus (Smith, 2002) focuses on historical documents relative to past events. Persus has incorporated tools allowing GIS to use the historical collection to bring out knowledge to enrich the GDB. The collection's processing consists of determining the terms, the toponyms, the dates, and estimating the co-occurrence of the dates and emplacements to determine eventual events. The TimeMap GIS explores this means of enrichment.
We can also mention the PIV 1 “Pyrénées Itinéraires Virtuels”) (Gaio et al., 2008) consists of managing a documentary collection of electronic versions of documents from the nineteenth century devoted to the Pyrenees and consisting of newspapers, novels, and travel stories. It is a stable fund (few deletions and modifications, regular insertions of new documents) and small (a few tens of thousands of documents).
Another tool that could be used for GDB enrichment purposes is the SpatialML1 language, developed by the Mitre Corporation. It is a markup language for annotating places and spatial relationships.
On the other hand, Artificial intelligence is increasingly being integrated into automatic language processing systems. Especially, Deep Learning (DL) which promises to extend the advances of Artificial Intelligence to another level.