WordNet Domains (WND) is a lexical resource where synsets have been semi-automatically annotated with one or more domain labels from a set of 170 hierarchically organized domains. The uses of WND include the power to reduce the polysemy degree of the words, grouping those senses that belong to the same domain. This paper presents a novel automatic method to propagate domain information through WordNet. We compare both labellings (the original and the new one) allowing us to detect anomalies in the original WND labels. We also compare the quality of both resources (the original labelling and the new one) in a common Word Sense Disambiguation task. The results show that the new labelling clearly outperform the original one by a large margin.

SMS language presents special phenomena and important deviations from natural language. Every day, an impressive amount of chat messages, SMS messages, and e-mails are sent all over the world. This widespread use makes important the development of systems that normalize SMS language into natural language. However, typical machine translation approaches are difficult to adapt to SMS language because of many irregularities that are shown by this kind of language. This paper presents a new approach for SMS normalization that combines lexical and phonological translation techniques with disambiguation algorithms at two different levels: lexical and semantic. The method proposed does not depend on big annotated corpus, which is difficult to build and is applied in two different domains showing its easiness of adaptation across different languages and domains. The results obtained by the system outperform some of the existing methods of SMS normalization despite the fact that the Spanish language and the corpus created have some features that complicate the normalization task.

No abstract

<p>The thickness of the seismogenic crust (Ts) controls the location and magnitude of crustal earthquakes. Its upper limit is generally found near the surface and correlates to crustal seismicity onset depth (SOD) while its base correlates to the brittle-ductile transition in the crust and the seismicity cutoff depth (SCD) (Chiarabba & De Gori, 2016; Wu <em>et al., </em>2017 and Zuza & Cao, 2020). Thus, it is a proxy of the brittle crust thickness and limits how deep earthquake ruptures may propagate, influencing their magnitude. Furthermore, crust with a thin Ts is inherently weaker and may concentrate more earthquakes (Burov, 2010 and Zuza & Cao, 2020). Given these factors, knowledge of Ts can help constrain future earthquake’s locations and magnitudes, aiding in seismic hazard assessment and mitigation.</p> <p>Previous authors have used seismic data to calculate Ts in Italy, California and Taiwan considering the depth distribution of earthquakes (eg., Chiarabba & De Gori; 2016, Wu <em>et al., </em>2017 and Zuza & Cao, 2020). However, the Chilean case presents a special and complex scenario. Here, the Nazca plate subducts below the South American plate producing an abundance of subduction earthquakes. Comparatively, crustal seismicity is sparse which presents a challenge. Adding to the complexity of the problem, the geometry of subduction as well as crustal thickness change considerably in latitude and longitude.</p> <p>In this work, we present the first attempt at a Ts map of Chile. Following the methodologies of Chiarabba & De Gori, 2016; Wu <em>et al., </em>2017 and Zuza & Cao, 2020 we divided the study area (ie. the Chilean margin between 15º and 45ºS) into a grid of square cells superposed by 2/3 of their width and calculated the depth distribution of earthquakes in each cell. As no consensus on which depth percentile to use for SOD and SCD exists, we calculated the percentiles 1, 5 and 10 for SOD and 90, 95 and 99 for SCD. Ts was then calculated as the difference between SCD and SOD. We compared the different outcomes.</p> <p>Furthermore, we test a new methodology, relaying on cells of variable radius. Here, cell size changes according to earthquake density. We believe this approach is optimal for heterogenous catalogues, such as is the case in Chile.</p> <p>Our results indicate that Ts in Chile varies latitudinally and longitudinally. Longitudinally it is generally thin at or close to the subduction trench, becomes thicker towards the east, reaching a maximum thickness below the central valley and then becomes thinner once again towards the volcanic arc. Latitudinally, it varies with crustal thickness as well as with subduction geometry (ie. it is thicker above the flat slab region).</p>