The objective of Geo-IA is to research, design and implement a Geo-Smart Artificial Intelligence technology platform for public and private business organizations. The GeoIA project presents a geolocation platform that integrates technological innovation to support a strategy for the creation of a Smart Territories. To do this, Text Mining, Machine Learning (including deep learning) and Natural Language Processing technologies are deployed. The functionality of the geolocation platform is to analyze, integrate, share data, visualize and represent territorial indicators, with the aim of facilitating the monitoring and fulfillment of territorial strategies. In short, GeoIA promotes interoperability between public administration bodies and also provides citizens with mechanisms to access information of interest, where the magnitude of the integrated and interrelated data permits. GeoIA also provides digital knowledge (tools, linked information, semantics, virtual assistants) for use by public administrations to enhance their decision making through greater knowledge of the environment and to improve services to citizens.
knowledge of the environment
With the above goal in mind, this project will apply a A smart society uses advances in information and com- discipline based on continuous evolution that is central munication technologies (ICT) to promote sustainable to Artificial Intelligence (AI), and grounded in three basic improvements that better the lives of citizens. A "smart" pillars: (i) data and text mining that will integrate and environment provides people with advanced solutions to extract information from diferent heterogeneous data solve problems or apply innovative technologies to create sources to transform it into an understandable structure eficient and adaptable services, connected cities and com- for further use, (ii) a prediction system, to make the best munities, informed, participative and satisfied citizens decisions of future actions in public administration, as and, above all, intelligent solutions for the provision of well as citizens’ behaviors for decision making based on services. Thus, in order to achieve these objectives and re- machine learning (ML) and, (iii) a simplified text generaspond to the needs of citizens, 21st-century public admin- tion system to prescribe citizens’ needs based on natural istration must deepen the digitization and improvement language processing (NLP). The starting hypotheses to of its services by promoting digital transformation as an address this project will be the following: organizational system or ecosystem in which all stakeholders (citizens, municipalities and companies) must develop at the same speed. However, the public sector adapts to change less rapidly than the private sector. It is therefore a worthwhile endeavour to provide digital knowledge (tools, linked information, semantics, virtual assistants) to public administration bodies so that they can improve services to citizens and thereby enhance public administration decision making through greater • (H1) Ontologies, the backbone technology that supports Linked Data and the semantic web, provide a means to overcome some of these challenges and thus help to obtain accurate profiles to produce satisfactory recommendations and personalized services. • (H2) An ontology-centric knowledge base allows the integration of data from heterogeneous sources and enables their analysis through advanced reasoning and inference processes. • (H3) Natural Language Processing, especially focused on entity-oriented machine learning, is essential to automate the construction and population of these knowledge bases, as well as to automatically generate simplified and synthesized text.
The main objective of the project is the research, design
and implementation of a GeoIntelligence Artificial
Intelligence (GeoIA) technology platform for public and
private business organizations. Specifically, the platform
will carry out the following activities:
• (H4) Knowledge-based Artificial GeoIntelligence language have been modeled through their
identificaapplied to public sector sources such as OpenData, tion, characterization, representation and exploitation.
GIS, Meteorology and Social Media data will be The participating team has also worked in projects
reessential to generate predictive and prescriptive lated to knowledge generation and bias in language
models to improve decision making processes in modeling, both technologies of great relevance for the
Public Administration. present project proposal. Specifically, they are the
REDES1 project (TIN2015-65136-C2-1-R,
TIN2015-65136-C22-R) and LIVING-LANG2 (RTI2018-094653-B-C22). The
goal of REDES was to go a step beyond the
representation of Digital Entities, enhancing the development of
technologies to automatically discover Digital Entities
from diferent heterogeneous sources to populate
semantic structures and link them to shared data. This process
involved defining Digital Entities for diferent domains
and processing heterogeneous information from the web,
• The design and development of a flexible, scalable the social web and the web of data. The digital entities
and robust technological architecture for data were then semantically enriched and spatiotemporally
integrating from diferent sectors of the public controlled and, finally, the generated information was
inadministration, such as Geographic Information tegrated into the digital entity model. REDES project was
Systems (GIS), cadastral data, census data, infras- the starting point to generate the following publications
tructure data, consumption data, and endless sec- in indexed journals by the research team also involved
tor data. in GeoIA: [
Figure 1 provides a summary of how the above projects have led to a number of Artificial Intelligence technologies, in particular Language Technologies, which have evolved incrementally.
Thus, it is essential for public and private sector organizations to develop a data and text integration model capable of defining and creating semantic networks of geolocalized digital entities, understanding digital entity as any concept that has associated information that characterizes it (a company, person, city, building, organization, etc.). These tools will allow the detection and extraction of semantic relationships between digital entities, obtaining the information from diferent types of sources (unstructured, structured and linked open data), as well as determining the quality, consistency and veracity of these relationships. In addition, from the previously created knowledge bases, machine learning techniques and algorithms will be adapted to work with spatio-temporal data to define GeoIA knowledge-based models.
GeoIntelligence system that focuses on the extraction of knowledge from heterogeneous sources and is capable of integrating diferent types of data, i.e. sectors of public administration, such as Geographic Information Systems (GIS), cadastral data, census data, infrastructure data, consumption data, and a host of sector data. This type of system has a direct impact on eco-digital transition policies, in addition to being able to ofer a direct service to the population through the use of virtual assistants, e.g. chatbots.
1https://gplsi.dlsi.ua.es/proyectos/redes/ 2https://gplsi.dlsi.ua.es/proyectos/livinglang/
The data to be processed can be of diferent nature and format. It is therefore necessary to implement data cleansing, normalization, and transformation techniques in order to be able to enter data into the platform in a standardized way. Cleaning consists of correcting possible errors in the data, sometimes manual or often caused by trying diferent encoding conversions (i.e. UTF-8, EPSG 3, etc.). In cases where the data is not recognizable because it has undergone a significant alteration, it is eliminated.
Standardization consists of unifying characteristics and information with the same meaning, under uniform criFigure 2: GeoIA solution in the context of digital transition. teria. For example: Alicante is equal to ALC. Transformation consists of extracting the data from the source document, recognizing its original format, and placing
Figure 2 is a macro view of how GeoIA’s knowledge- it in a common format among all the data incorporated based solutions are presented to support the digital tran- into the platform, for example, GeoPandas4. Examples of sition towards optimizing operations in organizations by types of data sources that may be relevant for this project incorporating digital technologies. are weather data API (e.g. OpenWeatherMap5, GIS (e.g.
The GeoIA project, as an open and accessible data GeoNet6), census, city cadastre, air quality data (e.g. Breestorage and processing center, will allow a controlled zometer7, social media user comments (e.g. Twitter 8 and collaborative model between the diferent agents for the Google9, and other data sources such as the National achievement of the project objectives, based on obtaining Institute of Statistics10 (INE). knowledge through the integration of the data collected and subsequently analyzed. Addressing each of the diferent phases requires an expert consortium both in the techniques involved and in the experience for its application to real areas of society. For this reason, the consortium is formed by: Instituto Tecnológico de Informática (ITI), the multinational company GFT, the company 1MillionBot, the company Gente Comunicación and the University of Alicante. The phases into which the system is divided are presented in Figure 3. 3https://epsg.io/ 4https://geopandas.org 5https://openweathermap.org/ 6https://www.geonet.es/ 7https://www.breezometer.com/ 8https://twitter.com/home 9https://www.google.com/ 10https://www.ine.es/ 3.2. Integration 11https://www.elastic.co/ 12https://lucene.apache.org/ 13https://leafletjs.com/ 14http://python-visualization.github.io/folium/ 15https://skos.um.es/TR/rdf-sparql-query/ 16https://neo4j.com/developer/cypher/ institutions, topics, named entities, etc.) through the semantic links that interconnect the network.
The data incorporated into the platform lacks semantic information. Therefore, from a previously designed model, which is able to capture the semantics of the domain, these data are instantiated, as pieces of semantic information, and linked together. The linking process consists of identifying common characteristics for each instance, and these characteristics are converted into semantic relationships.
This project is fully committed to the principle of eficient and rational public management, balancing control with productivity, by incorporating ICTs to improve administrative processes through advanced digitization tools and the provision of digital services to citizens and busi3.3. Storage nesses. In this context, GeoIA impacts the following lines of action that are considered central to making smart The storage phase allows us to ensure that all data that cities a reality: digital public services (e-Government), has been instantiated and linked can persist as long-term paperless administration, inter-administrative cooperamemory, and also coexist with other data that has been tion and interoperability, rational management of ICT previously incorporated. To guarantee this operation we resources and promotion of technological innovation in opted for a non-SQL database, in which each instance be- public management. The specific actions included in the comes a document and each relation an index. The tech- project are the promotion of the use and monitoring of nology stack chosen for this platform was ElasticSearch11 the quality of the public administration’s electronic serfor textual information. It is a Lucene12-based search vices, the promotion of the integration and exchange of server that provides a full-text, distributed, multitenancy- data and documents between administrations, the implecapable search engine with a RESTful web interface and mentation of an electronic administration platform, etc. JSON documents. For integration queries GeoPandas, In this line, the implementation of Geo.IA (GeoArtificial and Leaflet 13 y Folium14 for map creation and visualiza- Intelligence) will promote the generation of an ecosystem tion. The entire stack is built on the Python programming that revolves around Data, facilitating the development, language. prototyping and deployment of data analysis techniques in any domain of the economy and society. This will 3.4. Visualization enable the incorporation of value from the exploitation of the data available in the project by leveraging data In this phase, a series of queries are defined based on to improve management and decision making. The efthe user needs indicated in the user interface. From fective integration of innovation and research eforts these queries, data and metadata are obtained to generate in the project through the formed consortium will be visualizations that interweave data of diferent nature, boosted adding significant value. Geo.IA, serving as the source and domain, incorporated in the platform. With cornerstone and vehicle, will act as the guiding thread for this ability to interlink data, it is possible to carry out seamlessly incorporating these advancements into the cross-cutting studies and recommendations to facilitate productive and social fabric of the Valencian Community. decision-making and to propose new strategies, whether Geo.IA bases its value proposition on channeling the political, economic, geographic, or other, and mixtures eforts of the Valencian Community in accelerating the of these areas. digitization of our public sectors. The emphasis is on
One of the most important aspects to take into account knowing the requirements of Valencian Community citiin this task is semantic exploration and recommendation. zens in their interactions with public administration bodGiven the existence of a semantic database, it is neces- ies through an AI platform to extract knowledge. This sary to develop mechanisms for the exploration of the involves knowledge of citizens in line with their geospasemantic network, supported by SPARQL15 or Cypher16 tial position and knowledge for the public administration queries, of document profiles and other digital entities. to drive strategic decision making. Summarizing, this These mechanisms will allow to retrieve not only docu- proposal aims to boost to the digitization and technologments through metadata filters, but also to make aggre- ical advancement of citizens and the Valencian Commugate queries to discover statistical trends, and to make nity’s public administration, enabling the analysis of the recommendations of profiles (e.g., documents, authors, data generated in the GeoIA system to provide evidenced based decision making.
[3] Y. Gutiérrez, S. Vázquez, A. Montoyo, Spreading semantic information by word sense disambiguation, GeoIA will provide solutions to integrate data from Geo- Knowledge-Based Systems 132 (2017) 47–61. graphic Information Systems (GIS), public sector organi- [4] Y. Gutiérrez, S. Vázquez, A. Montoyo, A semanzations, census, land registry, cadastre, national statistics tic framework for textual data enrichment, Exinstitute, weather forecast providers, social media, etc. pert Syst. Appl. 57 (2016) 248–269. URL: https:// These data will be integrated and transformed into Digi- doi.org/10.1016/j.eswa.2016.03.048. doi:10.1016/J. tal Entities (DE) to compose a Semantic Lake of Digital ESWA.2016.03.048.
Entities in which real entities coexist and interrelate with [5] B. Navarro-Colorado, E. Saquete, Cross-document external information, DbPedia for example. This will con- event ordering through temporal, lexical and distritribute to generate an ecosystem that can be exploited butional knowledge, Knowl. Based Syst. 110 (2016) by public or private organizations and third party Apps 244–254. URL: https://doi.org/10.1016/j.knosys.2016. oriented to citizen services. In addition, this allows anal- 07.032. doi:10.1016/J.KNOSYS.2016.07.032. ysis, intelligent visualization of data for decision making [6] S. Estevez-Velarde, Y. Gutiérrez, Y. Almeida-Cruz, at diferent scales. The models will improve traditional A. Montoyo, General-purpose hierarchical optimisageographic analytics, going from knowing where and tion of machine learning pipelines with grammatical when things happen to knowing why they happen in evolution, Inf. Sci. 543 (2021) 58–71. URL: https: those places. If we add to this the information coming //doi.org/10.1016/j.ins.2020.07.035. doi:10.1016/J. from public administration agencies (land registry, cen- INS.2020.07.035. sus, environmental information, etc.), a whole ecosystem [7] A. Piad-Morfis, Y. Gutiérrez, Y. Almeida-Cruz, is created, enriched by the use of knowledge that can be R. Muñoz, A computational ecosystem to exploited by public administration agencies or private support ehealth knowledge discovery technolocompanies and by third party Apps oriented to citizen gies in spanish, J. Biomed. Informatics 109 services. (2020) 103517. URL: https://doi.org/10.1016/j.jbi.2020. 103517. doi:10.1016/J.JBI.2020.103517.
Acknowledgments [8] S. Estevez-Velarde, Y. Gutiérrez, A. Montoyo, Y. Almeida-Cruz, Automl strategy based on grammatical evolution: A case study about knowledge discovery from text, in: A. Korhonen, D. R. Traum, L. Màrquez (Eds.), Proceedings of the 57th Conference of the Association for Computational Linguistics, ACL 2019, Florence, Italy, July 28- August 2, 2019, Volume 1: Long Papers, Association for Computational Linguistics, 2019, pp. 4356–4365.
URL: https://doi.org/10.18653/v1/p19-1428. doi:10.
18653/V1/P19-1428. [9] A. Piad-Morfis, Y. Gutiérrez, R. Muñoz, A corpus to support ehealth knowledge discovery technologies, J. Biomed. Informatics 94 (2019). URL: https://doi. org/10.1016/j.jbi.2019.103172. doi:10.1016/J.JBI. 2019.103172.
Innovation(AVI) and the European Regional Development Fund(ERDF) through the project "GeoIA: Artificial GeoIntelligence platform to solve citizens problems and facilitate strategic decision making in public administrations" (INNEST/2023/11), partially funded by the Generalitat Valenciana (Conselleria d’Educació, Investigació, Cultura i Esport) through the project NL4DISMIS: TLHs for an Equal and Accessible Inclusive Society (CIPROM/2021/021). Moreover, it was backed by the work of two COST Actions: CA19134 - “Distributed Knowledge Graphs” and CA19142 - “Leading Platform for European Citizens, Industries, Academia, and Policymakers in Media Accessibility”.