Guadalupe Aguado-de-Cea

Guidelines for multilingual linked data

In this article, we argue that there is a growing number of linked datasets in different natural languages, and that there is a need for guidelines and mechanisms to ensure the quality and organic growth of this emerging multilingual data network. However, we have little knowledge regarding the actual state of this data network, its current practices, and the open challenges that it poses. Questions regarding the distribution of natural languages, the links that are established across data in different languages, or how linguistic features are represented, remain mostly unanswered. Addressing these and other language-related issues can help to identify existing problems, propose new mechanisms and guidelines or adapt the ones in use for publishing linked data including language-related features, and, ultimately, provide metrics to evaluate quality aspects. In this article we review, discuss, and extend current guidelines for publishing linked data by focusing on those methods, techniques and tools that can help RDF publishers to cope with language barriers. Whenever possible, we will illustrate and discuss each of these guidelines, methods, and tools on the basis of practical examples that we have encountered in the publication of the datos.bne.es dataset.

Applying the OntoLex Model to a Multilingual Terminological Resource

Terminesp is a multilingual terminological resource with terms from a range of specialized domains. Along with definitions, notes, scientific denominations and provenance information, it includes translations from Spanish into a variety of languages. A linked data resource with these features would represent a potentially relevant source of knowledge for NLP-based applications. In this contribution we show that Terminesp constitutes an appropriate validating test bench for OntoLex and its vartrans module, a newly developed model which evolves the lemon model to represent the lexicon-ontology interface. We present a first showcase of this module to account for variation across entries, while highlighting the modeling problems we encountered in this effort. Furthermore, we extend the resource with part-of-speech and syntactic information which was not explicitly declared in the original data with the aim of exploring its future use in NLP applications.

Publishing Linked Data on the Web: The Multilingual Dimension

Linked Data technologies and methods are enabling the creation of a data network where pieces of data are interconnected on the Web using machine-readable formats such as Resource Description Framework (RDF). This paradigm offers great opportunities to connect and make available knowledge in different languages. However, in order to make this vision a reality, there is a need for guidelines, techniques, and methods that allow publishers of data to overcome language and technological barriers. In this chapter, we review existing methodologies from the point of view of multilingualism and propose a series of guidelines to help publishers when publishing Linked Data in several languages.

Role-based model for Named Entity Recognition

Named Entity Recognition (NER) poses new challenges in real-world documents in which there are entities with different roles according to their purpose or meaning. Retrieving all the possible entities in scenarios in which only a subset of them based on their role is needed, produces noise on the overall precision. This work proposes a NER model that relies on role classification models that support recognizing entities with a specific role. The proposed model has been implemented in two use cases using Spanish drug Summary of Product Characteristics: identification of therapeutic indications and identification of adverse reactions. The results show how precision is increased using a NER model that is oriented towards a specific role and discards entities out of scope.

Expanding SNOMED-CT through Spanish Drug Summaries of Product Characteristics

Terminologies in the biomedical field are one of the main resources used in the clinical practice. Keeping them up-to-date to meet real-world use cases is a critical operation that even in the case of well maintained terminologies such as SNOMED-CT involves much effort from domain experts. Pharmacological products or drugs are constantly being approved and made available in the market and their clinical information should be also updated in terminologies. Each new drug is provided with its Summary of Product Characteristics (SPC), a document in natural language that contains its essential information. This paper proposes a method for populating the Spanish extension of SNOMED-CT with drug names using SPCs and representing their clinical data sections in the terminology. More precisely, the method has been applied to the therapeutic indication and the adverse reaction sections, in which disease names are recognized as named entities in the document and mapped to the terminology. The relations between the drug name and the mapped entities are also represented in the terminology based on the specific roles that they have in the document.