As a contribution to the Semantic Web this project seeks to preserve and disseminate historical data, through the publication of Linked Data of Historical Figures of Loja, Ecuador. The publication is carried out following the principles of Linked Data through a life-cycle that includes a research process for the speci cation of reliable data sources, the modeling of information, the generation and publication of RDF data, the validation of the RDF generation and nally its exploitation through the development of a Web site, where the Linked Data becomes an invisible layer to the users with interest in the domain of the published data.
Various demographic, technological and economic changes have been driven by
the Web. Behind them is the fundamental desire of people to connect,
communicate, share and participate [
The Semantic Web is a Web where information has much more meaning
than in the current Web [
The commitment to share data and semantics through the Web [
As a solution to this problem, this work aims to disseminate Historical Figures data from the city of Loja, using Linked Data techniques. This will be of great value within the cultural and social scope of our city and country. It allows the internationalization of this information, with certain standards recommended by W3C (organization that rules the Web in general).
This work aims to preserve the historical memory of Loja Historical Figures
through Linked Data techniques. For which it is sought to reuse the existing
information of Loja Historical Figures from di erent sources, to increase the
veracity, quality and quantity of the data. Once the data has been collected, it
is intended to transform the data into standard formats recommended by the
W3C, such as RDF [
The organization of this work is based on sections that represent the di erent phases of the life-cycle proposed below.
Linked Data is a combination of techniques, tools and Web standards that
allow the World Wide Web to evolve from a document network to a knowledge
graph (KG) [
The techniques of Semantic Web and Linked Data (or "Data Web") are of great importance for initiatives related to the historical knowledge management e orts of a locality, since they allow to preserve knowledge, share data, both among related institutions and all kinds of external domains that can reuse this knowledge in new and di erent ways. Next, the process carried out as part of this work is described.
The data to be published belong to Loja Historical Figures. According to Larousse
[
The gures data are obtained mainly from physical sources of local governments such as the libraries of the Municipality, Prefecture and House of Culture of Loja. Several sources have been identi ed in order to validate their veracity. In addition, semantic digital sources are identi ed, where data is obtained in RDF that enrich the data obtained from physical sources. Figure 2 shows the data sources identi ed. 3.1
Next, the data is extracted from the identi ed sources, digitized and nally structured, obtaining a total of 138 gures. The data is of quality, since its accuracy increases as there are more data sources that complement and con rm the information already collected.
URIs are global and permanent identi ers [
With the foundations of [
The most important recommendation made by [
The rst activity to carry out the ontological modeling is the Requirements Speci cation, where a series of tasks is performed, whose output is the Ontology Requirements Speci cation Document (ORSD). The Ontology Requirements Speci cation Document shows the purpose, scope, possible users and uses, as well as Competence Questions of the ontology to be developed. The latter allow to validate the ontology through Sparql queries, once implemented in an ontology language such as OWL. 4.2
After making the speci cation, where the ORSD is obtained as output, a highlevel ontological modeling is performed, as a rst approach to entities and relations to be modeled in the ontology. Followed, the search for vocabularies that allow representing the entities and relationships of the high-level model. For this, the website "Linked Open Vocabularies" (LOV, see http://lov.okfn.org) is used. Among the reused vocabularies we have: Friend of a Friend (foaf), Dbpedia Ontlogy (dbo), Schema, Dublin Core Metadata Terms (dct), Bbccore and Skos. 4.3
iPers Ontology
In this section, the iPers (Illustrious Personages) ontology is presented after its
conceptualization has been captured in the High Level Model, and the
vocabularies to be reused have also been selected. For its elaboration, the principles
for the design of ontologies mentioned in [
Figure 3 shows an extract of the iPers ontology, which describes the classes, properties and relationships that represent the gure basic information. The core element of the iPers ontology is the Person class, reused from the foaf vocabulary that models each gure, the basic data such as rst and last names are also modeled using foaf.
The data related to the birth dates, death dates and gender are modeled with the Schema vocabulary, in the same way as birthplace and death place, for which the Place class is reused and linked by object properties to Person. The Bbccore vocabulary is also used with the knownFor property to describe the reason why the gure is remarkably well-known and important.
On the other hand, the vocabulary of Dbpedia (dbo) models important data of the gure like: occupation, that denotes the activities to which it is dedicated; residence, represents the places where they have lived habitually; nationality, very important since it not only represents the country of birth, but also other countries of which the nationality is a carrier due to long stays or other reasons; and nally the property of data birthName, which allows to distinguish the birth name of the gure if it were di erent from the current one, for example the birth name of Matilde Hidalgo de Procel would be "Dei lia Matilde Ines Hidalgo Navarro".
The Skos vocabulary with its Concept class categorize the gures by their occupation and years of birth and death. Finally, the last class created with reference to personal data and that models the gure image (photograph or portrait) is Image of the foaf vocabulary, which uses the properties of Dublin Core (dct) to represent its metadata. 4.4
iPers Ontology Implementation The iPers ontology implementation is done through the Protege tool, through the OWL ontological language, de ned as standard by the W3C to develop ontologies. Protege is an open source ontology editor developed by Stanford University that allows the creation and editing of ontologies in a very intuitive graphical environment.
First, the IRI of the iPers ontology is entered. Followed by the general annotations that describe the ontology, and nally establish the vocabularies pre xes to be reused. Then, based on the ontological model, the classes and properties of data and objects are created, together with their annotations, descriptions and respective characteristics. For all the elements of the ontology, label and comment annotations are de ned, while in the case of properties, the domain and range are added, and also characteristics such as transitivity or symmetry.
Finally, Figure 4 shows the implementation result of the iPers ontology in Protege. After completing the iPers ontology implementation process in the tool for creating and editing ontologies, Protege, the model validation is carried out using two tools: { OntOlogy Pitfall Scanner! (OOPS!): Developed by the Ontology Engineering Group of the Universidad Politecnica de Madrid for the detection of failures in the design of ontologies, where it also recommends what actions to take to solve them. { RDF Validator: RDF validation service provided by the W3C for the detection of errors in the triples analyzed. 5
The present phase consists of taking the data from the data sources selected in the Speci cation Phase and transforming them into RDF according to the ontology resulting from the Modeling Phase. This section details the necessary tasks for a correct generation of data, such as: transformation, cleaning and linking of data. 5.1
The transformation task is the main one of this phase, whose objective is to
convert to RDF the data of the Historical Figures after they were collected
in the rst phase. There are several tools that automate this task, called RDF
converters. They are chosen depending on the format in which the data are found
and the volume of them. In this case, the OpenRe ne tool is used in its version
2.7, since it allows to work with unordered data and perform tasks such as:
cleaning, transformation and linking of data [
The procedure for the transformation is simple, rst the le that contains
the data in the tool is loaded. Next, the "RDF Skeleton" is edited, which is the
one that speci es how the RDF data is generated based on the data contained
in the le. The base URI for resources (Abox) is de ned in the skeleton, which
is speci ed in the design of Cool URIs [
After con guring the RDF Skeleton that transforms the data based on the iPers ontological model, the export is made to an RDF le. For which the RDF Re ne extension allows exporting in two serialization formats: RDF/XML and Turtle.
Finally, it is veri ed that the transformation process has been carried out correctly, since in every automated task a subsequent veri cation is necessary to search for possible faults and rectify them. 5.2
The objective of data enrichment is to link the data with other external data sets, following the fourth principle of Linked Data, which states that links to other URIs should be included to discover related resources. The external data sets with which links are established are: Dbpedia (see http://dbpedia.org), Dbpedia in Spanish (see http://es.dbpedia.org) and DBpedia-Latinoamerica (see http://es-la.dbpedia.org), in that order.
The linking process for the present project is considered a hybrid between automatic and manual linking. The automatic linking is done for all the resources of place type by means of the reconciliation service o ered by the RDF Re ne extension, where the Dbpedia Endpoint Sparql is established and the property with which the related resources are searched, in this case rdfs: label . On the other hand, the manual linking is done with resources related to people, occupations, professions, organizations, awards and genres of the works. 6
The objective of this phase is to publish the transformed data to RDF, by storing it in a triplestore for later consumption. There are many database managers that support RDF storage, each with its characteristics and limitations. However, it is decided to use Open Virtuoso given its free code feature, in addition to allowing data loading in a simple way, and consulting them through an Endpoint Sparql.
In addition, the Pubby tool is used, an application developed in Java that
allows adding an interface to the Endpoint Sparql of Open Virtuoso, making it
easier to access data and navigating through each of the resources stored in the
triplestore [
Next, the validation of the RDF generation is shown by Sparql queries based on the ontological model developed in the second phase.
Figure 5 and Figure 6 show the validation of the names, surnames, treatment title, gender and birth names of the Historical Figures through Sparql queries.
The consumption and visualization consists in the exploitation of the published data through applications with graphical interfaces, friendly to normal users, who have little or no knowledge in terms of Semantic Web technologies. In addition, being the historical data the topic of the data published in the present project, the outcome of this phase represents a cultural contribution for the community of Loja.
The application developed in this work is web-based and allows the visualization and search of both the Historical Figures and their works. It is developed through the Django web framework with the SPARQLWrapper library, which allows consulting the Open Virtuoso Sparql, obtaining the result in JSON format and converting it to more manageable formats for processing. Regarding the design of the application, the Bootstrap framework is used as a basis. Figure 7 shows an extract of the web application, where the RDF data consumed and visualized through user-friendly interfaces are appreciated.
In the domain of digital historical and cultural heritage, open and linked data is key to advance in knowledge management methods, actions that allow data to be open, allow reuse, and admit new uses of data. These values are common in other academic communities, including researchers dedicated to open science and reproducibility, as well as to the creators of knowledge graphs in certain domains.
Initiatives related to digital humanities can bene t from the experimentation of Linked Open Data, since recently published data sets have the potential to directly drive new research threads.
Once the present project has been completed, meeting the stated objectives, it is concluded that: { The publication of data, especially those of a historical nature or with a cultural focus, represents an arduous work of research and data collection, given that these are not generally structured, and have not even been digitized, so so researchers must go to physical sources. Despite this, it is interesting because it allows preserving and disseminating data of great cultural value.
Expressing this way, the true sense of technology, its social bond. { The reuse of ontological resources, such as vocabularies, is a very important practice in the creation of ontologies. It allows to optimize the e ort and time invested in this activity. In addition, the vocabularies, when reviewed and agreed upon, represent a standardization for the modeling of certain data domains, so using them adds quality to the developed ontology and facilitates interoperability with the data sets that use them. The iPers ontology models the relevant data of a person's life and can be reused to represent the information of Historical Figures, not only from Loja, but from any other place. { The generation of RDF data of historical gures allows a semantic description of the information, which leads to more precise searches of the data generated and its easy reuse in other areas where such information is required. { The reconciliation of data was a fundamental activity in the generation of RDF and data enrichment of historical gures. This activity is aligned to one of the principles of Linked Data: include links to other URIs to discover related resources. In this way, the enrichment of the information through the interoperability of the data is allowed. { The consumption and visualization of data was the nal phase of this work.
The development of a web application that shows the published data is important especially if the data pretend to be available to people who, despite their interest in the data domain, do not have knowledge in Semantic Web technologies. Thus, the linked data becomes an invisible layer for the end user.
After the objectives proposed for this project have been met, possible future works are presented that will improve or expand their development: Automation of the data integration process, through the development of a system that allows to unify the data and show alerts in case of data di erent that correspond to the same value. Given that currently the language of the data collected is only Spanish, the translation to other languages is proposed to increase its visibility, thus obtaining greater dissemination. Increase the information of historical gures from other places in the country. Thus becoming the present work, the origin of an initiative at the national level to preserve the Ecuadorian Culture.