We present an approach for the development of Language Understanding systems from a Transduction point of view. We describe the use of two types of automatically inferred transducers as the appropriate models for the understanding phase in dialog systems.

As many as two-thirds of individuals with an Autism Spectrum Disorder (ASD) also have language impairments, which can range from mild limitations to complete non-verbal behavior. For such cases, there are several Augmentative and Alternative Communication (AAC) devices available. These are computer-designed tools in order to help people with ASD to palliate or overcome such limitations, at least partially. Some of the most popular AAC devices are based on pictograms, so that a pictogram is the graphical representation of a simple concept and sentences are composed by concatenating a number of such pictograms. Usually, these tools have to manage a vocabulary made up of hundreds of pictograms/concepts, with no or very poor knowledge of the language at semantic and pragmatic level. In this paper we present Pictogrammar, an AAC system which takes advantage of SUpO and PictOntology. SUpO (Simple Upper Ontology) is a formal semantic ontology which is made up of detailed knowledge of facts of everyday life such as simple words, with special interest in linguistic issues, allowing automated grammatical supervision. PictOntology is an ontology developed to manage sets of pictograms, linked to SUpO. Both ontologies make possible the development of tools which are able to take advantage of a formal semantics.

Today social networks play an important role, where people can share information related to health. This information can be used for public health monitoring tasks through the use of Natural Language Processing (NLP) techniques. Social Media Mining for Health Applications (SMM4H) provides tasks such as those described in this document to help manage information in the health domain. This document shows the first participation of the SINAI group in SMM4H. We study approaches based on machine learning and deep learning to extract adverse drug reaction mentions from highly informal texts in Twitter. The results obtained in the tasks are encouraging, we are close to the average of all participants and even above in some cases.

In this paper we describe a new named entity extraction system. Our work proposes a system for the identification and annotation of drug names in Spanish biomedical texts based on machine learning and deep learning models. Subsequently, a standardized code using Snomed is assigned to these drugs, for this purpose, Natural Language Processing tools and techniques have been used, and a dictionary of different sources of information has been built. The results are promising, we obtain 78% in F1 score on the first sub-track and in the second task we map with Snomed correctly 72% of the found entities.