Search

A critical challenge faced by supervised word sense disambiguation (WSD) is the lack of large annotated datasets with sufficient coverage of words in their diversity of senses. This inspired recent research on few-shot WSD using meta-learning. While such work has successfully applied meta-learning to learn new word senses from very few examples, its performance still lags behind its fully supervised counterpart. Aiming to further close this gap, we propose a model of semantic memory for WSD in a meta-learning setting. Semantic memory encapsulates prior experiences seen throughout the lifetime of the model, which aids better generalization in limited data settings. Our model is based on hierarchical variational inference and incorporates an adaptive memory update rule via a hypernetwork. We show our model advances the state of the art in few-shot WSD, supports effective learning in extremely data scarce (e.g. one-shot) scenarios and produces meaning prototypes that capture similar senses of distinct words., 15 pages, 5 figures

Conversational dialogue systems (CDSs) are hard to evaluate due to the complexity of natural language. Automatic evaluation of dialogues often shows insufficient correlation with human judgements. Human evaluation is reliable but labor-intensive. We introduce a human-machine collaborative framework, HMCEval, that can guarantee reliability of the evaluation outcomes with reduced human effort. HMCEval casts dialogue evaluation as a sample assignment problem, where we need to decide to assign a sample to a human or a machine for evaluation. HMCEval includes a model confidence estimation module to estimate the confidence of the predicted sample assignment, and a human effort estimation module to estimate the human effort should the sample be assigned to human evaluation, as well as a sample assignment execution module that finds the optimum assignment solution based on the estimated confidence and effort. We assess the performance of HMCEval on the task of evaluating malevolence in dialogues. The experimental results show that HMCEval achieves around 99% evaluation accuracy with half of the human effort spared, showing that HMCEval provides reliable evaluation outcomes while reducing human effort by a large amount.

We investigate the semantic knowledge of language models (LMs), focusing on (1) whether these LMs create categories of linguistic environments based on their semantic monotonicity properties, and (2) whether these categories play a similar role in LMs as in human language understanding, using negative polarity item licensing as a case study. We introduce a series of experiments consisting of probing with diagnostic classifiers (DCs), linguistic acceptability tasks, as well as a novel DC ranking method that tightly connects the probing results to the inner workings of the LM. By applying our experimental pipeline to LMs trained on various filtered corpora, we are able to gain stronger insights into the semantic generalizations that are acquired by these models., Comment: Published in ACL Findings 2021

Conversational Question Simplification (CQS) aims to simplify self-contained questions into conversational ones by incorporating some conversational characteristics, e.g., anaphora and ellipsis. Existing maximum likelihood estimation (MLE) based methods often get trapped in easily learned tokens as all tokens are treated equally during training. In this work, we introduce a Reinforcement Iterative Sequence Editing (RISE) framework that optimizes the minimum Levenshtein distance (MLD) through explicit editing actions. RISE is able to pay attention to tokens that are related to conversational characteristics. To train RISE, we devise an Iterative Reinforce Training (IRT) algorithm with a Dynamic Programming based Sampling (DPS) process to improve exploration. Experimental results on two benchmark datasets show that RISE significantly outperforms state-of-the-art methods and generalizes well on unseen data., Comment: 13 pages, 4 figures, Published in ACL 2021

Language pairs with limited amounts of parallel data, also known as low-resource languages, remain a challenge for neural machine translation. While the Transformer model has achieved significant improvements for many language pairs and has become the de facto mainstream architecture, its capability under low-resource conditions has not been fully investigated yet. Our experiments on different subsets of the IWSLT14 training data show that the effectiveness of Transformer under low-resource conditions is highly dependent on the hyper-parameter settings. Our experiments show that using an optimized Transformer for low-resource conditions improves the translation quality up to 7.3 BLEU points compared to using the Transformer default settings., To be published in COLING 2020

We present a novel, domain expert-controlled, replicable procedure for the construction of concept-modeling ground truths with the aim of evaluating the application of word embeddings. In particular, our method is designed to evaluate the application of word and paragraph embeddings in concept-focused textual domains, where a generic ontology does not provide enough information. We illustrate the procedure, and validate it by describing the construction of an expert ground truth, QuiNE-GT. QuiNE-GT is built to answer research questions concerning the concept of naturalized epistemology in QUINE, a 2-million-token, single-author, 20th-century English philosophy corpus of outstanding quality, cleaned up and enriched for the purpose. To the best of our ken, expert concept-modeling ground truths are extremely rare in current literature, nor has the theoretical methodology behind their construction ever been explicitly conceptualised and properly systematised. Expert-controlled concept-modeling ground truths are however essential to allow proper evaluation of word embeddings techniques, and increase their trustworthiness in specialised domains in which the detection of concepts through their expression in texts is important. We highlight challenges, requirements, and prospects for future work.

Cognitive and social traits of individuals are reflected in language use. Moreover, individuals who are prone to spread fake news online often share common traits. Building on these ideas, we introduce a model that creates representations of individuals on social media based only on the language they produce, and use them to detect fake news. We show that language-based user representations are beneficial for this task. We also present an extended analysis of the language of fake news spreaders, showing that its main features are mostly domain independent and consistent across two English datasets. Finally, we exploit the relation between language use and connections in the social graph to assess the presence of the Echo Chamber effect in our data., 9 pages, accepted at COLING 2020

We have investigated the interaction of fibroblast growth factor (FGF) with megakaryocytopoiesis. Acidic FGF (aFGF) stimulated the proliferation of murine megakaryocytes and human erythroleukemia (HEL) cells in a concentration-dependent manner. The concentrations of aFGF required to elicit half-maximum and maximum effects were similar for HEL and megakaryocytic colony formation. The effect of aFGF was comparable to that of basic FGF (bFGF) in both cell types. The effect of both FGFs was found to be synergistic with interleukin-3 (IL-3), and was abrogated by a monoclonal anti-IL-6 antibody. A specific cell surface receptor complex of approximately 120 Kd was detected for FGF by crosslinking experiments on HEL cells and total bone marrow (BM) cells. Single-cell autoradiography of megakaryocytes in BM smears and BM cultures showed binding sites for 125I-aFGF. Northern blot analysis of messenger RNA (mRNA) from total BM and HEL cells showed a 4.4-kb mRNA specific for FGF receptors type 1 (flg) and type 2 (bek). This was confirmed by polymerase chain reaction, which also showed the presence of FGF receptor mRNA in megakaryocytic-like cells, normal megakaryocytes, and platelets. Together, these results indicate that FGF is involved in megakaryocytopoiesis and suggest that this interaction may be mediated via FGF receptor type 1 and type 2 located on the megakaryocytic lineage or on accessory cells responsible for the release of megakaryocytic growth-promoting activities.

Licorice has been used as medicine and as sweetening agent in food products. The major water-soluble constituent of licorice is glycyrrhizin (GL), an oleanane triterpenoide, which is known to be partly hydrolyzed by glucuronidase to its aglycone glycyrrhetinic acid (GA) which exists in 18 alpha (alpha-GA) and 18 beta (beta-GA) stereoisomeric forms. In this study alpha-GA and beta-GA were found to inhibit the mutagenicity of benzo[a]pyrene (B[a]P), 2-aminofluorene and aflatoxin B1 in Salmonella typhimurium TA98 and TA100. beta-GA was more effective than alpha-GA as an antimutagen. In the two-stage skin tumorigenesis protocol using 7,12-dimethylbenz[a]anthracene (DMBA) as the tumor initiating agent followed by twice weekly applications of 12-O-tetradecanoylphorbol-13-acetate as tumor promoter, pretreatment of SENCAR mice with alpha-GA or beta-GA resulted in significant protection against tumor initiation as well as tumor promotion. As an anti-tumor initiating agent, beta-GA was found to be more effective than alpha-GA. Similarly, topical application of beta-GA was found to be more effective than alpha-GA in inhibiting the binding of both [3H]B[a]P and [3H]DMBA to epidermal DNA. However, as an anti-tumor promoter, alpha-GA and beta-GA showed comparable effects. Our results suggest that both alpha-GA and beta-GA possess substantial anti-skin tumor initiating and anti-skin tumor promoting activities.