José Angel Daza

Even though SRL is researched for many languages, major improvements have mostly been obtained for English, for which more resources are available. In fact, existing multilingual SRL datasets contain disparate annotation styles or come from different domains, hampering generalization in multilingual learning. In this work we propose a method to automatically construct an SRL corpus that is parallel in four languages: English, French, German, Spanish, with unified predicate and role annotations… Meer weergeven

Even though SRL is researched for many languages, major improvements have mostly been obtained for English, for which more resources are available. In fact, existing multilingual SRL datasets contain disparate annotation styles or come from different domains, hampering generalization in multilingual learning. In this work we propose a method to automatically construct an SRL corpus that is parallel in four languages: English, French, German, Spanish, with unified predicate and role annotations that are fully comparable across languages. We apply high-quality machine translation to the English CoNLL-09 dataset and use multilingual BERT to project its high-quality annotations to the target languages. We include human-validated test sets that we use to measure the projection quality, and show that projection is denser and more precise than a strong baseline. Finally, we train different SOTA models on our novel corpus for mono- and multilingual SRL, showing that the multi-lingual annotations improve performance especially for the weaker languages. Minder weergeven

We explore a novel approach for Semantic Role Labeling (SRL) by casting it as a sequence-to-sequence process. We employ an attention-based model enriched with a copying mechanism to ensure faithful regeneration of the input sequence, while enabling interleaved generation of argument role labels. Here, we apply this model in a monolingual setting, performing PropBank SRL on English language data. The constrained sequence generation set-up enforced with the copying mechanism allows us to analyze… Meer weergeven

We explore a novel approach for Semantic Role Labeling (SRL) by casting it as a sequence-to-sequence process. We employ an attention-based model enriched with a copying mechanism to ensure faithful regeneration of the input sequence, while enabling interleaved generation of argument role labels. Here, we apply this model in a monolingual setting, performing PropBank SRL on English language data. The constrained sequence generation set-up enforced with the copying mechanism allows us to analyze the performance and special properties of the model on manually labeled data and benchmarking against state-of-the-art sequence labeling models. We show that our model is able to solve the SRL argument labeling task on English data, yet further structural decoding constraints will need to be added to make the model truly competitive. Our work represents a first step towards more advanced, generative SRL labeling setups. Minder weergeven

Are mind and machine capable of solving the same tasks? Creativity is one of the arguments that some philosophers and psychologists use as a proof of what computers cannot achieve; however, these arguments might be based on a misconception of what both intelligence and creativity mean. This book provides arguments supporting that creativity, as storytelling, can be emulated through computer programs. The assumption of creativity presents a major problem: Complexity. Even if we consider… Meer weergeven

Are mind and machine capable of solving the same tasks? Creativity is one of the arguments that some philosophers and psychologists use as a proof of what computers cannot achieve; however, these arguments might be based on a misconception of what both intelligence and creativity mean. This book provides arguments supporting that creativity, as storytelling, can be emulated through computer programs. The assumption of creativity presents a major problem: Complexity. Even if we consider creativity just as a product of novel ways of achieving a goal, the number of combinations found when dealing with the ‘real world’ is astronomically huge. We can recall The Library of Babel (Borges, 1944), a library that contains any possible book that could be written in the history of humanity. This metaphor reveals the combinatory problem that emerges if a brute force algorithm is designed to generate texts. According to our hypothesis, our proposal is a heuristic that uses simple syntactic and semantic properties found in a text corpus in order to generate novel and coherent fiction texts based on what has been already written. Minder weergeven

Most of the work dealing with automatic story production is based on a generic architecture for text generation; however, the resulting stories still lack a style that can be called literary. We believe that in order to generate automatically stories that could be compared with those by human authors, a specific methodology for fiction text generation should be defined. We also believe that it is essential for a story to convey the effect of originality to the person who is reading it. Our… Meer weergeven

Most of the work dealing with automatic story production is based on a generic architecture for text generation; however, the resulting stories still lack a style that can be called literary. We believe that in order to generate automatically stories that could be compared with those by human authors, a specific methodology for fiction text generation should be defined. We also believe that it is essential for a story to convey the effect of originality to the person who is reading it. Our methodology proposes corpus-based generation of stories that could be called creative and also have a style similar to human fiction texts. We also show how these stories have plausible syntax and coherence, and are perceived as interesting by human evaluators. Minder weergeven