In recent years, we have witnessed a rise in fake news, i.e., provably false pieces of information created with the intention deception. The dissemination this type news poses serious threat to cohesion and social well-being, since it fosters political polarization distrust people respect their leaders. huge amount that is disseminated through media makes manual verification unfeasible, which has promoted design implementation automatic systems for detection. creators use various stylistic...

We introduce a method to reduce constituent parsing sequence labeling. For each word wt, it generates label that encodes: (1) the number of ancestors in tree words wt and wt+1 have common, (2) nonterminal symbol at lowest common ancestor. first prove proposed encoding function is injective for any without unary branches. In practice, approach made extensible all constituency trees by collapsing then use PTB CTB treebanks as testbeds propose set fast baselines. achieve 90% F-score on test...

Michalina Strzyz, David Vilares, Carlos Gómez-Rodríguez. Proceedings of the 2019 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). 2019.

It is often stated that human languages, as other biological systems, are shaped by cost-cutting pressures but, to what extent? Attempts quantify the degree of optimality languages means an score have been scarce and focused mostly on English. Here we recast problem word order a sentence optimization spatial network where vertices words, arcs indicate syntactic dependencies, space defined linear words in sentence. We introduce cognitive pressure reduce distance between linked The analysis...

We address the problem of performing polarity classification on Twitter over different languages, focusing English and Spanish, comparing three techniques: (1) a monolingual model which knows language in opinion is written, (2) that acts based decision provided by identification tool (3) multilingual trained dataset does not need any recognition step.Results show models are even able to outperform some sets.We introduce first code-switching corpus with sentiment labels, showing robustness approach.

Abstract We describe an opinion mining system which classifies the polarity of Spanish texts. propose NLP approach that undertakes pre-processing, tokenisation and POS tagging texts to then obtain syntactic structure sentences by means a dependency parser. This is used address three most significant linguistic constructions for purpose in question: intensification, subordinate adversative clauses negation. also semi-automatic domain adaptation method improve accuracy our specific application...

Millions of micro texts are published every day on T witter. Identifying the sentiment present in them can be helpful for measuring frame mind public, their satisfaction with respect to a product, or support social event. In this context, polarity classification is subfield analysis focused determining whether content text objective subjective, and latter case, if it conveys positive negative opinion. Most detection techniques tend take into account individual terms even some degree...

Daniel Fernández-González, Carlos Gómez-Rodríguez. Proceedings of the 2019 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). 2019.

We evaluate a range of recent LLMs on English creative writing, challenging and complex task that requires imagination, coherence, style. use difficult, open-ended scenario chosen to avoid training data reuse: an epic narration single combat between Ignatius J. Reilly, the protagonist Pulitzer Prize-winning novel A Confederacy Dunces (1980), pterodactyl, prehistoric flying reptile. ask several humans write such story conduct human evalution involving various criteria as fluency, originality,...

<title>Abstract</title> We conduct a quantitative analysis contrasting human-written English news text with comparable large language model (LLM) output from six different LLMs that cover three families and four sizes in total. Our spans several measurable linguistic dimensions, including morphological, syntactic, psychometric, sociolinguistic aspects. The results reveal various differences between human AI-generated texts. Human texts exhibit more scattered sentence length distributions,...

En el contexto de las industrias del petróleo y gas en tabasco, Mexico,en lo relativo al Empowerrment, se hace relevante determinar la relación entre organizaciones que son algo más simple suma partes un sistema actividades formadas por dos o personas –capital humano–, donde es esencial cooperación ellas para puedan existir, existe estrategia empresarial Empowerrment (en español: empoderamiento), consiste delegar autoridad responsabilidad a los individuos., dar poder, tareas “líder “a...

Transition-based parsing is a widely used approach for dependency that combines high efficiency with expressive feature models. Many different transition systems have been proposed, often formalized in slightly frameworks. In this article, we show large number of the known projective can be viewed as variants same stack-based system small set elementary transitions composed into complex and restricted ways. We call these divisible prove theoretical results about their expressivity...

We introduce an approach to train lexicalized parsers using bilingual corpora obtained by merging harmonized treebanks of different languages, producing that can analyze sentences in either the learned or even mix both. test on Universal Dependency Treebanks, training with MaltParser and MaltOptimizer. The results show these are more than competitive, as most combinations not only preserve accuracy, but some achieve significant improvements over corresponding monolingual parsers. Preliminary...

We present HEAD-QA, a multi-choice question answering testbed to encourage research on complex reasoning. The questions come from exams access specialized position in the Spanish healthcare system, and are challenging even for highly humans. then consider monolingual (Spanish) cross-lingual (to English) experiments with information retrieval neural techniques. show that: (i) HEAD-QA challenges current methods, (ii) results lag well behind human performance, demonstrating its usefulness as...

Recent analyses suggest that encoders pretrained for language modeling capture certain morpho-syntactic structure. However, probing frameworks word vectors still do not report results on standard setups such as constituent and dependency parsing. This paper addresses this problem does full parsing (on English) relying only pretraining architectures – no decoding. We first cast sequence tagging. then use a single feed-forward layer to directly map labels encode linearized tree. is used to:...

We introduce dependency parsing schemata, a formal framework based on Sikkel's schemata for constituency parsers, which can be used to describe, analyze, and compare algorithms. use this describe several well-known projective non-projective build correctness proofs, establish relationships between them. then the define new polynomial-time algorithms various mildly formalisms, including well-nested structures with their gap degree bounded by constant k in time O(n 5+2k ), class that includes...

The introduction of dynamic oracles has considerably improved the accuracy greedy transition-based dependency parsers, without sacrificing parsing efficiency.However, this enhancement is limited to projective parsing, and have not yet been implemented for parsers supporting non-projectivity.In paper we introduce first such oracle, a non-projective parser based on Attardi's parser.We show that training with oracle improves over conventional (static) wide range datasets.