eSPERTo’s Paraphrastic Knowledge Applied to Question-Answering and Summarization

Multi3Generation: Multitask, Multilingual, and Multimodal Language Generation

“…The project “DIALECT – Natural Language Understanding for non-standard languages and dialects”, led by Prof. Barbara Plank 15 , MC member of Multi3Generation also addresses the topic of processing languages underrepresented in the Internet. We can include here undergoing projects, such as “eSPERTo – System for Paraphrasing in Editing and Revision of Texts 9 , 10 16 , and “Paraphrasary” (Portuguese version “Parafrasário”) 11 , both led by the Action’s Chair, Dr. Anabela Barreiro, developed for Portuguese as part of a larger multilingual generation project involving paraphrasing and translation, for which several experimental results have been already published 12 – 15 .…”

Large Language Models and OpenLogos: An Educational Case Scenario

“…Several articles have been written in previous years illustrating the Model 33 , sharing linguistic resources 34 , and comparing translations by OpenLogos to those generated by Google Translate 35 , among other experiments published in random articles. Similarly, several articles have described the work done on paraphrasing over the years using initial OpenLogos resources 26 , 28 , 36 , 37 , among others, including recent experiments comparing paraphrases generated by the eSPERTo paraphrasing tool for Portuguese to paraphrases generated by ChatGPT 38 . By using OpenLogos and derived paraphrasing tools, students can effectively rewrite texts in a non-plagiarized manner.…”

One Book, Two Language Varieties

“…In recent years, editorial revision processes have benefited not only from linguistic reflections and orientations, but also from natural language processing tools, which can serve to develop corpora analyses, generation and summarisation of texts, translation, paraphrasing, among others. In an attempt to complement tools that aid in writing systems available in the market (mostly spelling and superficial grammar correction), [2] created a paraphrase generation system called eSPERTo, an acronym for Sistema de Parafraseamento para Edição e Revisão de Texto (System of Paraphrasing in Editing and Revision of Text [9], which aims to assist in various tasks, such as summarizing and answering questions ( [10]), language teaching, support for machine translation ( [2]; controlled technical writing, editing and proofreading, stylistic adaptation ( [12], and text conversion between language varieties ( [4]), among others. In the last case study, the use of the tool allowed us to analyze the paraphrastic units of a complete literary work in the EP and BP varieties of Portuguese, following some previously performed experimental works of adaptation between them (cf.…”

“…(8) EP -fico sempre fascinado a olhar para os seus gestos cuidados BP -sempre fico fascinado observando seus gestos cuidadosos EN -I always get fascinated when I look at his elegant/careful gestures Word Order -A common conversion strategy between the EP and the BP versions is the word order, which implies words from different parts of speech. Example (9) illustrates the change in the word order of the quantifier todos (all). However, examples in the corpus exemplify changes in the word order of adjectives (e.g., mundo enfadonho-enfadonho mundo (boring world)), adverbs (e.g., aqui me trouxe-me trouxe aqui (brought me here)), or subjects, either pronouns (e.g., Que poderia eu-Que eu poderia (What could I) or nouns Como se chama o senhor?…”