
A multi-agent-based approach for address geocoding in poorly mapped areas through public company data
Author(s) -
Mamadou Kébé,
Roger Marcelin Faye,
Claude Lishou
Publication year - 2021
Publication title -
international journal of information technology and applied sciences
Language(s) - English
Resource type - Journals
ISSN - 2709-2208
DOI - 10.52502/ijitas.v3i1.14
Subject(s) - geocoding , computer science , context (archaeology) , process (computing) , negotiation , information retrieval , data science , data mining , geography , cartography , political science , archaeology , law , operating system
In this study, we present an original method that enhances geocoding systems in poorly mapped areas thanks to public company data and a multi-agent system. In contrast with industrialized countries, many developing countries lack formal postal address systems assignments and usage, making the operation of translating text-based addresses to absolute spatial coordinates, known as geocoding, a big challenge. We recreated a standard of address as it is perceived and used by local people, a kind of non-official national address standard since there is no official one in these areas. Then, we designed a multi-agent system in which agents are assigned different tasks of geocoding process and can perform negotiation to achieve a global objective: find the best possible match or approximation of a location-based on current knowledge. Verification of the usefulness of the proposed approach is made in comparison with Google Geocoding API which shows that the proposed approach has great potential to geocode addresses considering local context semantic issues.