Large language models are known to produce output which sounds fluent and convincing, but is also often wrong, e.g. “unfaithful” with respect to a rationale as retrieved from a knowledge base. In this paper, we show that task-based systems which exhibit certain advanced linguistic dialog behaviors, such as lexical alignment (repeating what the user said), are in fact preferred and trusted more, whereas other phenomena, such as pronouns and ellipsis are dis-preferred. We use open-domain question answering systems as our test-bed for task based dialog generation and compare several open- and closed-book models. Our results highlight the danger of systems that appear to be trustworthy by parroting user input while providing an unfaithful response.

This work presents a two-stage approach for tackling low-resource AMR-to-text generation for Brazilian Portuguese. Our approach consists of (1) generating a masked surface realization in which some tokens are masked according to its Part-of-Speech class and (2) infilling the masked tokens according to the AMR graph and the previous masked surface realization. Results show a slight improvement over the baseline, mainly in BLEU (1.63) and METEOR (0.02) scores. Moreover, we evaluate the pipeline components separately, showing that the bottleneck of the pipeline is the masked surface realization. Finally, the human evaluation suggests that models still suffer from hallucinations, and some strategies to deal with the problems found are proposed.

We introduce GEM, a living benchmark for natural language Generation (NLG), its Evaluation, and Metrics. Measuring progress in NLG relies on a constantly evolving ecosystem of automated metrics, datasets, and human evaluation standards. Due to this moving target, new models often still evaluate on divergent anglo-centric corpora with well-established, but flawed, metrics. This disconnect makes it challenging to identify the limitations of current models and opportunities for progress. Addressing this limitation, GEM provides an environment in which models can easily be applied to a wide set of tasks and in which evaluation strategies can be tested. Regular updates to the benchmark will help NLG research become more multilingual and evolve the challenge alongside models. This paper serves as the description of the data for the 2021 shared task at the associated GEM Workshop.

This paper describes the submission by the NILC Computational Linguistics research group of the University of S ̃ao Paulo/Brazil to the English Track 2 (closed sub-track) at the Surface Realisation Shared Task 2020. The success of the current pre-trained models like BERT or GPT-2 in several tasks is well-known, however, this is not the case for data-to-text generation tasks and just recently some initiatives focused on it. This way, we explore how a pre-trained model (GPT-2) performs on the UD-to-text generation task. In general, the achieved results were poor, but there are some interesting ideas to explore. Among the learned lessons we may note that it is necessary to study strategies to represent UD inputs and to introduce structural knowledge into these pre-trained models.

This paper describes the submission by the NILC Computational Linguistics research group of the University of São Paulo/Brazil to the RDF-to-Text task for English at the WebNLG+ challenge. The success of the current pretrained models like BERT or GPT-2 in text-to-text generation tasks is well-known, however, its application/success on data-totext generation has not been well-studied and proven. This way, we explore how good a pretrained model, in particular BART, performs on the data-to-text generation task. The results obtained were worse than the baseline and other systems in almost all automatic measures. However, the human evaluation shows better results for our system. Besides, results suggest that BART may generate paraphrases of reference texts.

In hierarchical text classification, we perform a sequence of inference steps to predict the category of a document from top to bottom of a given class taxonomy. Most of the studies have focused on developing novels neural network architectures to deal with the hierarchical structure, but we prefer to look for efficient ways to strengthen a baseline model. We first define the task as a sequence-to-sequence problem. Afterwards, we propose an auxiliary synthetic task of bottom-up-classification. Then, from external dictionaries, we retrieve textual definitions for the classes of all the hierarchy’s layers, and map them into the word vector space. We use the class-definition embeddings as an additional input to condition the prediction of the next layer and in an adapted beam search. Whereas the modified search did not provide large gains, the combination of the auxiliary task and the additional input of class-definitions significantly enhance the classification accuracy. With our efficient approaches, we outperform previous studies, using a drastically reduced number of parameters, in two well-known English datasets.

Corpora curated by experts have sustained Natural Language Processing mainly in English, but the expensiveness of corpora creation is a barrier for the development in further languages. Thus, we propose a corpus generation strategy that only requires a machine translation system between English and the target language in both directions, where we filter the best translations by computing automatic translation metrics and the task performance score. By studying Reading Comprehension in Spanish and Word Sense Disambiguation in Portuguese, we identified that a more quality-oriented metric has high potential in the corpora selection without degrading the task performance. We conclude that it is possible to systematise the building of quality corpora using machine translation and automatic metrics, besides some prior effort to clean and process the data.

Abstract Meaning Representation (AMR) is a recent and prominent semantic representation with good acceptance and several applications in the Natural Language Processing area. For English, there is a large annotated corpus (with approximately 39K sentences) that supports the research with the representation. However, to the best of our knowledge, there is only one restricted corpus for Portuguese, which contains 1,527 sentences. In this context, this paper presents an effort to build a general purpose AMR-annotated corpus for Brazilian Portuguese by translating and adapting AMR English guidelines. Our results show that such approach is feasible, but there are some challenging phenomena to solve. More than this, efforts are necessary to increase the coverage of the corresponding lexical resource that supports the annotation.

This paper presents an exploratory study that aims to evaluate the usefulness of back-translation in Natural Language Generation (NLG) from semantic representations for non-English languages. Specifically, Abstract Meaning Representation and Brazilian Portuguese (BP) are chosen as semantic representation and language, respectively. Two methods (focused on Statistical and Neural Machine Translation) are evaluated on two datasets (one automatically generated and another one human-generated) to compare the performance in a real context. Also, several cuts according to quality measures are performed to evaluate the importance (or not) of the data quality in NLG. Results show that there are still many improvements to be made but this is a promising approach.

This paper presents a more recent literature review on Natural Language Generation. In particular, we highlight the efforts for Brazilian Portuguese in order to show the available resources and the existent approaches for this language. We also focus on the approaches for generation from semantic representations (emphasizing the Abstract Meaning Representation formalism) as well as their advantages and limitations, including possible future directions.

This paper describes the submission by the NILC Computational Linguistics research group of the University of São Paulo/Brazil to the Track 1 of the Surface Realization Shared Task (SRST Track 1). We present a neural-based method that works at the syntactic level to order the words (which we refer by NILC-SWORNEMO, standing for “Syntax-based Word ORdering using NEural MOdels”). Additionally, we apply a bottom-up approach to build the sentence and, using language-specific lexicons, we produce the proper word form of each lemma in the sentence. The results obtained by our method outperformed the average of the results for English, Portuguese and Spanish in the track.