Thailand

Transformer-based language models (LMs) pretrained on large text collections implicitly store a wealth of lexical semantic knowledge, but it is non-trivial to extract that knowledge effectively from their parameters. Inspired by prior work on semantic specialization of static word embedding (WE) models, we show that it is possible to expose and enrich lexical knowledge from the LMs, that is, to specialize them to serve as effective and universal &quot;decontextualized&quot; word encoders even when fed input words &quot;in isolation&quot; (i.e., without any context). Their transformation into such word encoders is achieved through a simple and efficient lexical fine-tuning procedure (termed LexFit) based on dual-encoder network structures. Further, we show that LexFit can yield effective word encoders even with limited lexical supervision and, via cross-lingual transfer, in different languages without any readily available external knowledge. Our evaluation over four established, structurally different lexical-level tasks in 8 languages indicates the superiority of LexFit-based WEs over standard static WEs (e.g., fastText) and WEs from vanilla LMs. Other extensive experiments and ablation studies further profile the LexFit framework, and indicate best practices and performance variations across LexFit variants, languages, and lexical tasks, also directly questioning the usefulness of traditional WE models in the era of large neural models.

ACL-IJCNLP 2021

LexFit: Lexical Fine-Tuning of Pretrained Language Models

lexical fine-tuning

lexical semantics

pretrained language models

**Welcome to ACL-IJCNLP 2021!**

The great event is jointly organized by the Association for Computational Linguistics (ACL) and Asian Federation of Natural Language Processing (AFNLP). 

As in previous years, the program of the conference includes a poster session, tutorials, workshops and demonstrations in addition to the main conference.


We were able to keep the registration fees similar to those charged for the virtual ACL 2020. The one fee allows attendance at the main conference and any/all tutorials and workshops. These fees would be $125 Regular Early and $175 Regular Late; $50 Student Early and $75 Student Late. Early registration closes at midnight July 11, 2021 (Eastern Daylight Time).

**Reminder:** It is ACL’s policy that at least one author of each accepted paper (including ACL Finding papers) must register for the conference.

**Reminder2:** Underline site will open closer to the event. If you already registered you will receive access detail

Registration is now open

The great event is jointly organized by the Association for Computational Linguistics (ACL) and Asian Federation of Natural Language Processing (AFNLP).

Transformer-based language models (LMs) pretrained on large text collections implicitly store a wealth of lexical semantic knowledge, but it is non-trivial to extract that knowledge effectively from their parameters. Inspired by prior work on semantic specialization of static word embedding (WE) models, we show that it is possible to expose and enrich lexical knowledge from the LMs, that is, to specialize them to serve as effective and universal "decontextualized" word encoders even when fed input words "in isolation" (i.e., without any context). Their transformation into such word encoders is achieved through a simple and efficient lexical fine-tuning procedure (termed LexFit) based on dual-encoder network structures. Further, we show that LexFit can yield effective word encoders even with limited lexical supervision and, via cross-lingual transfer, in different languages without any readily available external knowledge. Our evaluation over four established, structurally different lexical-level tasks in 8 languages indicates the superiority of LexFit-based WEs over standard static WEs (e.g., fastText) and WEs from vanilla LMs. Other extensive experiments and ablation studies further profile the LexFit framework, and indicate best practices and performance variations across LexFit variants, languages, and lexical tasks, also directly questioning the usefulness of traditional WE models in the era of large neural models.

poster

In computational psycholinguistics, various language models have been evaluated against human reading behavior (e.g., eye movement) to build human-like computational models. However, most previous efforts have focused almost exclusively on English, despite the recent trend towards linguistic universal within the general community. In order to fill the gap, this paper investigates whether the established results in computational psycholinguistics can be generalized across languages. Specifically, we re-examine an established generalization ---\textit{the lower perplexity a language model has, the more human-like the language model is}--- in Japanese with typologically different structures from English. Our experiments demonstrate that this established generalization exhibits a surprising lack of universality; namely, lower perplexity is not always human-like. Moreover, this discrepancy between English and Japanese is further explored from the perspective of (non-)uniform information density. Overall, our results suggest that a cross-lingual evaluation will be necessary to construct human-like computational models.

Lower Perplexity is Not Always Human-Like

Lately proposed Word Sense Disambiguation (WSD) systems have approached the estimated upper bound of the task on standard evaluation benchmarks. However, these systems typically implement the disambiguation of words in a document almost independently, underutilizing sense and word dependency in context. In this paper, we convert the nearly isolated decisions into interrelated ones by exposing senses in context when learning sense embeddings in a similarity-based Sense Aware Context Exploitation (SACE) architecture. Meanwhile, we enhance the context embedding learning with selected sentences from the same document, rather than utilizing only the sentence where each ambiguous word appears. Experiments on both English and multilingual WSD datasets have shown the effectiveness of our approach, surpassing previous state-of-the-art by large margins (3.7% and 1.2% respectively), especially on few-shot (14.3%) and zero-shot (35.9%) scenarios.

Word Sense Disambiguation: Towards Interactive Context Exploitation from Both Word and Sense Perspectives

The advent of contextual word embeddings — representations of words which incorporate semantic and syntactic information from their context—has led to tremendous improvements on a wide variety of NLP tasks. However, recent contextual models have prohibitively high computational cost in many use-cases and are often hard to interpret. In this work, we demonstrate that our proposed distillation method, which is a simple extension of CBOW-based training, allows to significantly improve computational efficiency of NLP applications, while outperforming the quality of existing static embeddings trained from scratch as well as those distilled from previously proposed methods. As a side-effect, our approach also allows a fair comparison of both contextual and static embeddings via standard lexical evaluation tasks.

Obtaining Better Static Word Embeddings Using Contextual Embedding Models

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.

Meta-Learning with Variational Semantic Memory for Word Sense Disambiguation

While neural language models can generate text with remarkable fluency and coherence, controlling for factual correctness in generation remains an open research question. This major discrepancy between the surface-level fluency and the content-level correctness of neural generation has motivated a new line of research that seeks automatic metrics for evaluating the factuality of machine text. In this paper, we introduce GO FIGURE, a meta-evaluation framework for evaluating factuality evaluation metrics. We propose five necessary and intuitive conditions to evaluate factuality metrics on diagnostic factuality data across three different summarization tasks. Our benchmark analysis on ten factuality metrics reveals that our meta-evaluation framework provides a robust and efficient evaluation that is extensible to multiple types of factual consistency and standard generation metrics, including QA metrics. It also reveals that while QA metrics generally improve over standard metrics that measure factuality across domains, performance is highly dependent on the way in which questions are generated.

GO FIGURE: A Meta Evaluation of Factuality in Summarization

Answering questions about why characters perform certain actions is central to understanding and reasoning about narratives. Despite recent progress in QA, it is not clear if existing models have the ability to answer “why” questions that may require commonsense knowledge external to the input narrative. In this work, we introduce TellMeWhy, a new crowd-sourced dataset that consists of more than 30k questions and free-form answers concerning why characters in short narratives perform the actions described. For a third of this dataset, the answers are not present within the narrative. Given the limitations of automated evaluation for this task, we also present a systematized human evaluation interface for this dataset. Our evaluation of state-of-the-art models show that they are far below human performance on answering such questions. They are especially worse on questions whose answers are external to the narrative, thus providing a challenge for future QA and narrative understanding research.

TellMeWhy: A Dataset for Answering Why-Questions in Narratives

A summary quality measure is judged by how well it correlates with quality scores produced by human annotators. A higher correlation with human scores is considered to be a decisive indicator of a better measure. In this work we present observations that cast doubt on this view. We also show a possibility of an alternative indicator for selecting the best measure from a family of measures, a criterion that does not rely on human scores.

Is Human Scoring the Best Criteria for Summary Evaluation?

Since the late 1990s, automatic text simplification (ATS) was promoted as a natural language processing (NLP) task with great potential to make texts more accessible to people with various reading or cognitive disabilities, and enable their better social inclusion. Large multidisciplinary projects showed promising steps in that direction. Since 2010, the field started attracting more attention but at the cost of major shifts in system architecture, target audience, and evaluation strategies. Somewhere along the way, the focus has shifted from ATS for social good towards building complex end-to-end neural architectures that are not aimed at any particular target population. This study presents the trajectory of ATS for social good, the main issues in current ATS trends, and the ways forward that could bring the field back to its initial goals.

Automatic Text Simplification for Social Good: Progress and Challenges

Open-domain neural dialogue models have achieved high performance in response ranking and evaluation tasks. These tasks are formulated as a binary classification of responses given in a dialogue context, and models generally learn to make predictions based on context-response content similarity. However, over-reliance on content similarity makes the models less sensitive to the presence of inconsistencies, incorrect time expressions and other factors important for response appropriateness and coherence. We propose approaches for automatically creating adversarial negative training data to help ranking and evaluation models learn features beyond content similarity. We propose mask-and-fill and keyword-guided approaches that generate negative examples for training more robust dialogue systems. These generated adversarial responses have high content similarity with the contexts but are either incoherent, inappropriate or not fluent. Our approaches are fully data-driven and can be easily incorporated in existing models and datasets. Experiments on classification, ranking and evaluation tasks across multiple datasets demonstrate that our approaches outperform strong baselines in providing informative negative examples for training dialogue systems.

Synthesizing Adversarial Negative Responses for Robust Response Ranking and Evaluation

We introduce a novel dataset of human judgments of machine-edited text and initial models of those perceptions. Six machine-editing methods ranging from character swapping to variational autoencoders are applied to collections of English-language social media text and scientific abstracts. The edits are judged in context for detectability and the extent to which they preserve the meaning of the original. Automated measures of semantic similarity and fluency are evaluated individually and combined to produce composite models of human perception. Both meaning preservation and detectability are predicted within 6% of the upper bound of human consensus labeling.

Downloads

Next from ACL-IJCNLP 2021

Lower Perplexity is Not Always Human-Like

Similar lecture

Analysis and Evaluation of Language Models for Word Sense Disambiguation