GPT-2 has been frequently adapted in story generation models as it provides powerful generative capability. However, it still fails to generate consistent stories and lacks diversity. Current story generation models leverage additional information such as plots or commonsense into GPT-2 to guide the generation process. These approaches focus on improving generation quality of stories while our work look at both quality and diversity. We explore combining BERT and GPT-2 to build a variational autoencoder (VAE), and extend it by adding additional objectives to learn global features such as story topic and discourse relations. Our evaluations show our enhanced VAE can provide better quality and diversity trade off, generate less repetitive story content and learn a more informative latent variable.
Most rumour detection models for social media are designed for one specific language (mostly English). There are over 40 languages on Twitter and most languages lack annotated resources to build rumour detection models. In this paper we propose a zero-shot cross-lingual transfer learning framework that can adapt a rumour detection model trained for a source language to another target language. Our framework utilises pretrained multilingual language models (e.g.\ multilingual BERT) and a self-training loop to iteratively bootstrap the creation of ''silver labels'' in the target language to adapt the model from the source language to the target language. We evaluate our methodology on English and Chinese rumour datasets and demonstrate that our model substantially outperforms competitive benchmarks in both source and target language rumour detection.
We present IndoBERTweet, the first large-scale pretrained model for Indonesian Twitter that is trained by extending a monolingually-trained Indonesian BERT model with additive domain-specific vocabulary. We focus in particular on efficient model adaptation under vocabulary mismatch, and benchmark different ways of initializing the BERT embedding layer for new word types. We find that initializing with the average BERT subword embedding makes pretraining five times faster, and is more effective than proposed methods for vocabulary adaptation in terms of extrinsic evaluation over seven Twitter-based datasets.
There is unison is the scientific community about human induced climate change. Despite this, we see the web awash with claims around climate change scepticism, thus driving the need for fact checking them but at the same time providing an explanation and justification for the fact check. Scientists and experts have been trying to address it by providing manually written feedback for these claims. In this paper, we try to aid them by automating generating explanation for a predicted veracity label for a claim by deploying the approach used in open domain question answering of a fusion in decoder augmented with retrieved supporting passages from an external knowledge. We experiment with different knowledge sources, retrievers, retriever depths and demonstrate that even a small number of high quality manually written explanations can help us in generating good explanations.
While automatic summarization evaluation methods developed for English are routinely applied to other languages, this is the first attempt to systematically quantify their panlinguistic efficacy. We take a summarization corpus for eight different languages, and manually annotate generated summaries for focus (precision) and coverage (recall). Based on this, we evaluate 19 summarization evaluation metrics, and find that using multilingual BERT within BERTScore performs well across all languages, at a level above that for English.
The COVID-19 pandemic has driven ever-greater demand for tools which enable efficient exploration of biomedical literature. Although semi-structured information resulting from concept recognition and detection of the defining elements of clinical trials (e.g. PICO criteria) has been commonly used to support literature search, the contributions of this abstraction remain poorly understood, especially in relation to text-based retrieval. In this study, we compare the results retrieved by a standard search engine with those filtered using clinically-relevant concepts and their relations. With analysis based on the annotations from the TREC-COVID shared task, we obtain quantitative as well as qualitative insights into characteristics of relational and concept-based literature exploration. Most importantly, we find that the relational concept selection filters the original retrieved collection in a way that decreases the proportion of unjudged documents and increases the precision, which means that the user is likely to be exposed to a larger number of relevant documents.
Existing work on probing of pretrained language models (LMs) has predominantly focused on sentence-level syntactic tasks. In this paper, we introduce document-level discourse probing to evaluate the ability of pretrained LMs to capture document-level relations. We experiment with 7 pretrained LMs, 4 languages, and 7 discourse probing tasks, and find BART to be overall the best model at capturing discourse -- but only in its encoder, with BERT performing surprisingly well as the baseline model. Across the different models, there are substantial differences in which layers best capture discourse information, and large disparities between models.
We introduce a grey-box adversarial attack and defence framework for sentiment classification. We address the issues of differentiability, label preservation and input reconstruction for adversarial attack and defence in one unified framework. Our results show that once trained, the attacking model is capable of generating high-quality adversarial examples substantially faster (one order of magnitude less in time) than state-of-the-art attacking methods. These examples also preserve the original sentiment according to human evaluation. Additionally, our framework produces an improved classifier that is robust in defending against multiple adversarial attacking methods. Code is available at: https://github.com/ibm-aur-nlp/adv-def-text-dist.
We introduce a top-down approach to discourse parsing that is conceptually simpler than its predecessors (Kobayashi et al., 2020; Zhang et al., 2020). By framing the task as a sequence labelling problem where the goal is to iteratively segment a document into individual discourse units, we are able to eliminate the decoder and reduce the search space for splitting points. We explore both traditional recurrent models and modern pre-trained transformer models for the task, and additionally introduce a novel dynamic oracle for top-down parsing. Based on the Full metric, our proposed LSTM model sets a new state-of-the-art for RST parsing.
In this paper, we propose FFCI, a framework for automatic summarization evaluation that comprises four elements: Faithfulness, Focus, Coverage, and Inter-sentential coherence. We design FFCI by comprehensively studying traditional evaluation metrics and model-based evaluations, including question answering (QA) approaches, STS, next-sentence prediction (NSP), and scores from 19 pre-trained language models. Our study reveals three key findings: (1) calculating BertSCORE between the summary and article sentences yields a higher correlation score than recently-proposed QA-based evaluation methods for faithfulness evaluation; (2) GPT2Score has the best Pearson's correlation for focus and coverage; and (3) a simple NSP model is effective at evaluating inter-sentential coherence.