EMNLP 2023 Papers

Generating concise summaries of news events is a challenging natural language processing task. While journalists often curate timelines to highlight key sub-events, newcomers to a news event face challenges in catching up on its historical context. In this paper, we address this need by introducing the task of background news summarization, which complements each timeline update with a background summary of relevant preceding events. We construct a dataset by merging existing timeline datasets and asking human annotators to write a background summary for each timestep of each news event. We establish strong baseline performance using state-of-the-art summarization systems and propose a query-focused variant to generate background summaries. To evaluate background summary quality, we present a question-answering-based evaluation metric, Background Utility Score (BUS), which measures the percentage of questions about a current event timestep that a background summary answers. Our experiments show the effectiveness of instruction fine-tuned systems such as Flan-T5, in addition to strong zero-shot performance using GPT-3.5.

We present Superlim, a multi-task NLP benchmark and analysis platform for evaluating Swedish language models, a counterpart to the English-language (Super)GLUE suite. We describe the dataset, the tasks, the leaderboard and report the baseline results yielded by a reference implementation. The tested models do not approach ceiling performance on any of the tasks, which suggests that Superlim is truly difficult, a desirable quality for a benchmark. We address methodological challenges, such as mitigating the Anglocentric bias when creating datasets for a less-resourced language; choosing the most appropriate measures; documenting the datasets and making the leaderboard convenient and transparent. We also highlight other potential usages of the dataset, such as, for instance, the evaluation of cross-lingual transfer learning.

We introduce ZeroSCROLLS, a zero-shot benchmark for natural language understanding over long texts, which contains only test and small validation sets, without training data. We adapt six tasks from the SCROLLS benchmark, and add four new datasets, including two novel information fusing tasks, such as aggregating the percentage of positive reviews. Using ZeroSCROLLS, we conduct a comprehensive evaluation of both open-source and closed large language models, finding that Claude outperforms ChatGPT, and that GPT-4 achieves the highest average score. However, there is still room for improvement on multiple open challenges in ZeroSCROLLS, such as aggregation tasks, where models struggle to pass the naive baseline. As the state of the art is a moving target, we invite researchers to evaluate their ideas on the live ZeroSCROLLS leaderboard.

Large language models (LLMs) have shown remarkable reasoning capabilities, particularly with Chain-of-Thought-style prompts. However, LLMs can still struggle with problems that are easy for humans, such as generating action plans for executing tasks or performing complex math or logical reasoning. This is due to LLMs' absence of an internal world model for predicting world states (e.g., environment status, variable values) and simulating long-term action outcomes of actions. This prevents LLMs from performing deliberate planning akin to human brains, which involves exploring alternative reasoning paths, anticipating future states and rewards, and iteratively refining existing reasoning steps. To overcome the limitations, we propose a new LLM reasoning framework, Reasoning via Planning (RAP). RAP repurposes the LLM as both a world model and a reasoning agent, and incorporates a principled planning algorithm (based on Monte Carlo Tree Search) for strategic exploration in the vast reasoning space. During reasoning, the LLM (as agent) incrementally builds a reasoning tree under the guidance of the LLM (as world model) and task-specific rewards, properly balancing exploration v.s. exploitation to achieve a high-reward reasoning path efficiently. We apply RAP to a variety of challenging reasoning problems, such as plan generation, math reasoning, and logical inference. Empirical results demonstrate the superiority of RAP over various strong baselines, including CoT and least-to-most prompting with self-consistency, e.g., RAP on LLaMA-33B surpasses CoT on GPT-4 with 33\% relative improvement in plan generation.

Self-training has proven to be an effective approach for cross-domain tasks, and in this study, we explore its application to cross-domain constituency parsing. Traditional self-training methods rely on limited and potentially low-quality raw corpora. To overcome this limitation, we propose enhancing self-training with the large language model (LLM) to generate domain-specific raw corpora iteratively. For the constituency parsing, we introduce grammar rules that guide the LLM in generating raw corpora and establish criteria for selecting pseudo instances. Our experimental results demonstrate that self-training for constituency parsing, equipped with an LLM, outperforms traditional methods regardless of the LLM's performance. Moreover, the combination of grammar rules and confidence criteria for pseudo-data selection yields the highest performance in the cross-domain constituency parsing.

Abstractive Text Summarization (ATS) models are commonly trained using large-scale data that is randomly shuffled. However, the impact of data selection and data ordering on ATS models remains a relatively unexplored research area, where a significant challenge lies in accurately assessing the learning difficulty of each training instance. This study introduces a Data Selection Curriculum (DSC) scoring system that incorporates both the difficulty of improving ATS model via an instance and the expected performance on this instance. By selectively excluding excessively simple and overly complex instances, the training efficiency can be optimized. Furthermore, curriculum learning is integrated to accelerate convergence and improve performance by gradually increasing the learning difficulty, inspired by human learners. Experimental results on the CNN/DailyMail dataset demonstrate that our approach surpasses potent baselines, utilizing a mere 20\% of the available instances.

Large multilingual pretrained language models (mPLMs) have become the de facto state of the art for cross-lingual transfer in NLP. However, their large-scale deployment to many languages, besides pretraining data scarcity, is also hindered by the increase in vocabulary size and limitations in their parameter budget. In order to boost the capacity of mPLMs to deal with low-resource and unseen languages, we explore the potential of leveraging transliteration on a massive scale. In particular, we explore the UROMAN transliteration tool, which provides mappings from UTF-8 to Latin characters for all the writing systems, enabling inexpensive romanization for virtually any language. We first focus on establishing how UROMAN compares against other language-specific and manually curated transliterators for adapting multilingual PLMs. We then study and compare a plethora of data- and parameter-efficient strategies for adapting the mPLMs to romanized and non-romanized corpora of 14 diverse low-resource languages. Our results reveal that UROMAN-based transliteration can offer strong performance for many languages, with particular gains achieved in the most challenging setups: on languages with unseen scripts and with limited training data without any vocabulary augmentation. Further analyses reveal that an improved tokenizer based on romanized data can even outperform non-transliteration-based methods in the majority of languages.

Continual Named Entity Recognition (CNER) is a burgeoning area, which involves updating an existing model by incorporating new entity types sequentially. Nevertheless, continual learning approaches are often severely afflicted by catastrophic forgetting. This issue is intensified in CNER due to the consolidation of old entity types from previous steps into the non-entity type at each step, leading to what is known as the semantic shift problem of the non-entity type. In this paper, we introduce a pooled feature distillation loss that skillfully navigates the trade-off between retaining knowledge of old entity types and acquiring new ones, thereby more effectively mitigating the problem of catastrophic forgetting. Additionally, we develop a confidence-based pseudo-labeling for the non-entity type, i.e., predicting entity types using the old model to handle the semantic shift of the non-entity type. Following the pseudo-labeling process, we suggest an adaptive re-weighting type-balanced learning strategy to handle the issue of biased type distribution. We carried out comprehensive experiments on ten CNER settings using three different datasets. The results illustrate that our method significantly outperforms prior state-of-the-art approaches, registering an average improvement of 6.3% and 8.0% in Micro and Macro F1 scores, respectively.

Differentiable Search Indices (DSIs) encode a corpus of documents in the parameters of a model and use the same model to map queries directly to relevant document identifiers. Despite the solid performance of DSI models, successfully deploying them in scenarios where document corpora change with time is an open problem. In this work, we introduce DSI++, a continual learning challenge for DSI with the goal of continuously indexing new documents while being able to answer queries related to both previously and newly indexed documents. Across different model scales and document identifier representations, we show that continual indexing of new documents leads to considerable forgetting of previously indexed documents. We also hypothesize and verify that the model experiences forgetting events during training, leading to unstable learning. To mitigate these issues, we investigate two approaches. The first focuses on modifying the training dynamics. Flatter minima implicitly alleviates forgetting, so we explicitly optimize for flatter loss basins and show that the model stably memorizes more documents (+12\%). Next, we introduce a parametric memory to generate pseudo-queries for documents and supplement them during incremental indexing to prevent forgetting for the retrieval task. Extensive experiments on a novel continual indexing benchmark based on Natural Questions demonstrate that our proposed solution mitigates the forgetting in DSI++ by a significant margin and improves the average Hits@10 by +21.1\% over competitive baselines.

It has been well documented that a reviewer’s opinion of the nativeness of expression in an academic paper affects the likelihood of it being accepted for publication. Previous works have also shone a light on the stress and anxiety authors who are non-native English speakers experience when attempting to publish in international venues. We explore how this might be a concern in the field of Natural Language Processing (NLP) through conducting a comprehensive statistical analysis of NLP paper abstracts, identifying how authors of different linguistic backgrounds differ in the lexical, morphological, syntactic and cohesive aspects of their writing. Through our analysis, we identify that there are a number of characteristics that are highly variable across the different corpora examined in this paper. This indicates potential for the presence of linguistic bias. Therefore, we outline a set of recommendations to publishers of academic journals and conferences regarding their guidelines and resources for prospective authors in order to help enhance inclusivity and fairness.

Editing model parameters directly in Transformers makes updating open-source transformer-based models possible without re-training. However, these editing methods have only been evaluated on statements about encyclopedic knowledge with a single correct answer. Commonsense knowledge with multiple correct answers, e.g., an apple can be green or red but not transparent, has not been studied but is as essential for enhancing transformers' reliability and usefulness. In this paper, we investigate whether commonsense judgments are causally associated with localized, editable parameters in Transformers, and we provide an affirmative answer. We find that directly applying the MEMIT editing algorithm results in sub-par performance and improve it for the commonsense domain by varying edit tokens and improving the layer selection strategy, i.e., $MEMIT_{CSK}$. GPT-2 Large and XL models edited using $MEMIT_{CSK}$ outperform best-fine-tuned baselines by 10.97% and 10.73% F1 scores on PEP3k and 20Q datasets. In addition, we propose a novel evaluation dataset, $PROBE\ SET$, that contains unaffected and affected neighborhoods, affected paraphrases, and affected reasoning challenges. $MEMIT_{CSK}$ performs well across the metrics while fine-tuning baselines show significant trade-offs between unaffected and affected metrics. These results suggest a compelling future direction for incorporating feedback about common sense into Transformers through direct model editing.

Controllable text generation (CTG) aims to generate text with desired attributes, and decoding-time-based methods have shown promising performance on this task. However, in this paper, we identify the phenomenon of Attribute Collapse for the first time. It causes the fluency of generated text to rapidly decrease when the control strength exceeds a critical value, rendering the text completely unusable. This limitation hinders the effectiveness of decoding methods in achieving high levels of controllability. To address this problem, we propose a novel lightweight decoding framework named Air-Decoding. Its main idea is reconstructing the attribute distributions to balance the weights between attribute words and non-attribute words to generate more fluent text. Specifically, we train prefixes by prefix-tuning to obtain attribute distributions. Then we design a novel attribute distribution reconstruction method to balance the obtained distributions and use the reconstructed distributions to guide language models for generation, effectively avoiding the issue of Attribute Collapse. Experiments on multiple CTG tasks prove that our method achieves a new state-of-the-art control performance.

Instruction-Following Language Models (IFLMs) are promising and versatile tools for solving many downstream, information-seeking tasks. Given their success, there is an urgent need to have a shared resource to determine whether existing and new IFLMs are prone to produce biased language interactions. In this paper, we propose Prompt Association Test (P-AT): a new resource for testing the presence of social biases in IFLMs. P-AT stems from WEAT (Caliskan et al., 2017) and generalizes the notion of measuring social biases to IFLMs. Basically, we cast WEAT word tests in promptized classification tasks, and we associate a metric - the bias score. Our resource consists of 2310 prompts. We then experimented with several families of IFLMs discovering gender and race biases in all the analyzed models. We expect P-AT to be an important tool for quantifying bias across different dimensions and, therefore, for encouraging the creation of fairer IFLMs before their distortions have consequences in the real world.

Open-ended Commonsense Reasoning is defined as solving a commonsense question without providing 1) a short list of answer candidates and 2) a pre-defined answer scope. Conventional ways of formulating the commonsense question into a question-answering form or utilizing external knowledge to learn retrieval-based methods are less applicable in the open-ended setting due to an inherent challenge. Without pre-defining an answer scope or a few candidates, open-ended commonsense reasoning entails predicting answers by searching over an extremely large searching space. Moreover, most questions require implicit multi-hop reasoning, which presents even more challenges to our problem. In this work, we leverage pre-trained language models to iteratively retrieve reasoning paths on the external knowledge base, which does not require task-specific supervision. The reasoning paths can help to identify the most precise answer to the commonsense question. We conduct experiments on two commonsense benchmark datasets. Compared to other approaches, our proposed method achieves better performance both quantitatively and qualitatively.

Transformers have become a key architecture in speech processing, but our understanding of how they build up representations of acoustic and linguistic structure is limited. In this study, we address this gap by investigating how measures of 'context-mixing' developed for text models can be adapted and applied to models of spoken language. We identify a linguistic phenomenon that is ideal for such a case study: homophony in French (e.g. livre vs livres), where a speech recognition model has to attend to syntactic cues such as determiners and pronouns in order to disambiguate spoken words with identical pronunciations and transcribe them while respecting grammatical agreement. We perform a series of controlled experiments and probing analyses on Transformer-based speech models. Our findings reveal that representations in encoder-only models effectively incorporate these cues to identify the correct transcription, whereas encoders in encoder-decoder models mainly relegate the task of capturing contextual dependencies to decoder modules.

Developing an efficient retriever to retrieve knowledge from a large-scale knowledge base (KB) is critical for task-oriented dialogue systems to effectively handle localized and specialized tasks. However, widely used generative models such as T5 and ChatGPT often struggle to differentiate subtle differences among the retrieved KB records when generating responses, resulting in suboptimal quality of generated responses. In this paper, we propose the application of maximal marginal likelihood to train a perceptive retriever by utilizing signals from response generation for supervision. In addition, our approach goes beyond considering solely retrieved entities and incorporates various meta knowledge to guide the generator, thus improving the utilization of knowledge. We evaluate our approach on three task-oriented dialogue datasets using T5 and ChatGPT as the backbone models. The results demonstrate that when combined with meta knowledge, the response generator can effectively leverage high-quality knowledge records from the retriever and enhance the quality of generated responses. The code of this work is available at https://github.com/shenwzh3/MK-TOD.

Although counterfactual reasoning is a fundamental aspect of intelligence, the lack of large-scale counterfactual open-domain question-answering (QA) benchmarks makes it difficult to evaluate and improve models on this ability. To address this void, we introduce the first such dataset, named IfQA, where each question is based on a counterfactual presupposition via an ``if'' clause. Such questions require models to go beyond retrieving direct factual knowledge from the Web: they must identify the right information to retrieve and reason about an imagined situation that may even go against the facts built into their parameters. The IfQA dataset contains 3,800 questions that were annotated by crowdworkers on relevant Wikipedia passages. Empirical analysis reveals that the IfQA dataset is highly challenging for existing open-domain QA methods, including supervised retrieve-then-read pipeline methods (F1 score 44.5), as well as recent few-shot approaches such as chain-of-thought prompting with ChatGPT (F1 score 57.2). We hope the unique challenges posed by IfQA will push open-domain QA research on both retrieval and reasoning fronts, while also helping endow counterfactual reasoning abilities to today's language understanding models.

We introduce DISSC, a novel, lightweight method that converts the rhythm, pitch contour and timbre of a recording to a target speaker in a textless manner. Unlike DISSC, most voice conversion (VC) methods focus primarily on timbre, and ignore people's unique speaking style (prosody). The proposed approach uses a pretrained, self-supervised model for encoding speech to discrete units, which makes it simple, effective, and fast to train. All conversion modules are only trained on reconstruction like tasks, thus suitable for any-to-many VC with no paired data. We introduce a suite of quantitative and qualitative evaluation metrics for this setup, and empirically demonstrate that DISSC significantly outperforms the evaluated baselines. Code and samples are available at https://pages.cs.huji.ac.il/adiyoss-lab/dissc/.

Although large language models (LLMs) are impressive in solving various tasks, they can quickly be outdated after deployment. Maintaining their up-to-date status is a pressing concern in the current era. This paper provides a comprehensive review of recent advances in aligning deployed LLMs with the ever-changing world knowledge. We categorize research works systemically and provide in-depth comparisons and discussions. We also discuss existing challenges and highlight future directions to facilitate research in this field.