China
Africa
Explore chapters and articles related to this topic
Multilingual and Multimodal Access
Published in Sarika Jain, Understanding Semantics-Based Decision Support, 2021
There are multiple multilingual repositories of ontologies available from which we can associate linguistic information into the ontology, or we can make our own. Following are some of those repositories:WordNet: WordNet is a lexical database and can be considered an electronic dictionary. It contains nouns, verbs, adverbs, and adjectives along with idioms and simplex words.DBpedia: DBpedia is an effort at extracting all the structured information of Wikipedia, representing it semantically in the form of RDF triples, and making it available on the semantic web.BabelNet: BabelNet is a very large, wide-coverage multilingual semantic network. It is a resource which is constructed automatically through integration of knowledge from WordNet and Wikipedia. Along with this machine learning, translations are used to enrich the repository with lexical multilingual information.YAGO: YAGO is a combination of Wikipedia, WordNet, GeoNames, and more.
EASIER System. Evaluating a Spanish Lexical Simplification Proposal with People with Cognitive Impairments
Published in International Journal of Human–Computer Interaction, 2022
Rodrigo Alarcon, Lourdes Moreno, Paloma Martínez, José A. Macías
CWI seeks to detect words that are unusual or complex for a given user, using methods ranging from simple lexicon-based methods (Alarcon et al., 2019) to the popular machine learning-based methods (Alarcon et al., 2021a), which have been used in various competitions that have emerged over the years (G. Paetzold & Specia, 2016b; Yimam et al., 2017). The next step, SG, receives the complex words in the previous step to generate possible replacements in all the contexts it may have. Research in this step can be grouped into linguistic database-based methods such as Babelnet (Navigli & Ponzetto, 2010) or Wordnet (Thomas & Anderson, 2012) to automatic methods that focus on the disadvantages of linguistic databases (Pavlick & Callison-Burch, 2016; Qiang et al., 2021), by looking for less expensive ways to generate replacements. The third step, SS, receives the list of candidates generated in the previous step, to select the best candidate taking into account complexity and context. Research in this step is varied, from explicit approaches supported by the word sense disambiguation (WSD) task (G. H. Paetzold & Specia, 2017), implicit approaches (Nunes et al., 2013) or those based on word semantics (Alarcon et al., 2021b; Saggion et al., 2015). Finally, the SR stage obtains the contextualized candidates in order to sort them according to the complexity of the words, taking into account the needs of the target user. Different methods at this stage have been presented, from the most popular frequency-based methods (Rello et al., 2013; Specia et al., 2012), metric-based (Glavaš & Štajner, 2015) or, like CWI, machine learning-based methods, which try to deal with the resource dependency suffered by frequency-based methods (Maddela & Xu, 2018; G. H. Paetzold & Specia, 2017).