acl_id string | title string | abstract string | conference_name string | conference_track string | year int64 | url string | contribution_types list | openreview_id string | openreview_cycle string | openreview_history list | article_content string |
|---|---|---|---|---|---|---|---|---|---|---|---|
2024.findings-eacl.68 | Understanding and Mitigating Spurious Correlations in Text Classification with Neighborhood Analysis | Recent research has revealed that machine learning models have a tendency to leverage spurious correlations that exist in the training set but may not hold true in general circumstances. For instance, a sentiment classifier may erroneously learn that the token “performances” is commonly associated with positive movie r... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.68.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | Pkt8doM0TV | October 2023 | [] | [{"1 Introduction": ["Disclaimer: This paper contains examples that may be considered profane or offensive. These examples by no means reflect the authors' view toward any groups or entities.", "Pre-trained language models (PLMs) such as BERT Devlin et al. (2019) and its derivative models have shown impressive performa... |
2024.eacl-long.73 | No Error Left Behind: Multilingual Grammatical Error Correction with Pre-trained Translation Models | Grammatical Error Correction (GEC) enhances language proficiency and promotes effective communication, but research has primarily centered around English. We propose a simple approach to multilingual and low-resource GEC by exploring the potential of multilingual machine translation (MT) models for error correction. We... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.73.pdf | [
"NLP engineering experiment",
"Approaches to low-resource settings",
"Approaches low compute settings-efficiency"
] | vchiWnuieL | October 2023 | [] | [{"1 Introduction": ["Grammatical Error Correction (GEC) systems are a vital link between expert language use and clear communication, enhancing writing skills and language learning. However, GEC research has primarily focused on the English language with much less coverage for other languages, resulting in English-ori... |
2024.eacl-long.13 | Language Models as Inductive Reasoners | Inductive reasoning is a core component of human intelligence. In the past research of inductive reasoning within computer science, formal language is used as representations of knowledge (facts and rules, more specifically). However, formal language can cause systematic problems for inductive reasoning such as disabil... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.13.pdf | [
"NLP engineering experiment",
"Data resources"
] | MbM-nT-YfN | October 2023 | [] | [{"1 Introduction": ["Inductive reasoning is to reach to a hypothesis (usually a rule that explains an aspect of the law of nature) based on pieces of evidence (usually observed facts of the world), where the observations can not provide conclusive support to the hypothesis [19]. It is ampliative, which means that the ... |
2024.eacl-long.178 | Large-Scale Label Interpretation Learning for Few-Shot Named Entity Recognition | Few-shot named entity recognition (NER) detects named entities within text using only a few annotated examples. One promising line of research is to leverage natural language descriptions of each entity type: the common label PER might, for example, be verbalized as “person entity.” In an initial label interpretation l... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.178.pdf | [
"Approaches to low-resource settings",
"Data analysis"
] | ep9cuBomIC | October 2023 | [] | [{"1 Introduction": ["Few-shot named entity recognition (NER) refers to identifying and classifying named entities within text by learning from a few annotated examples. A widely adopted strategy in few-shot NER employs transfer learning with pre-trained language models (PLMs) to interpret labels based on their semanti... |
2024.eacl-long.176 | Do Moral Judgment and Reasoning Capability of LLMs Change with Language? A Study using the Multilingual Defining Issues Test | This paper explores the moral judgment and moral reasoning abilities exhibited by Large Language Models (LLMs) across languages through the Defining Issues Test. It is a well known fact that moral judgment depends on the language in which the question is asked. We extend the work of beyond English, to 5 new languages (... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.176.pdf | [
"Model analysis & interpretability"
] | jmysF33NjI | October 2023 | [] | [{"1 Introduction": ["In a recent work, Tanmay et al. (2023) used the Defining Issues Test (DIT) [19], a psychological assessment tool based on Kohlberg's Cognitive Moral Development (CMD) [18], to evaluate the moral reasoning capabilities of large language models (LLMs) such as GPT-4, ChatGPT, Llama2Chat-70B and PaLM-... |
2024.acl-long.438 | PokeMQA: Programmable knowledge editing for Multi-hop Question Answering | Multi-hop question answering (MQA) is one of the challenging tasks to evaluate machine's comprehension and reasoning abilities, where large language models (LLMs) have widely achieved the human-comparable performance. Due to the dynamics of knowledge facts in real world, knowledge editing has been explored to update mo... | acl | long | 2,024 | https://aclanthology.org/2024.acl-long.438.pdf | [
"NLP engineering experiment",
"Approaches to low-resource settings"
] | OQpPiCRTNdz | October 2023 | [] | [{"1 Introduction": ["Multi-hop question answering (MQA) requires a sequence of interacted knowledge facts to reach the final answer. For instance, considering the two-hop question in Figure 1, it is necessary to deduce the intermediate answer Inter Miami through the fact \"Messi plays for Inter Miami\", and then deduc... |
2024.acl-long.607 | Draft & Verify: Lossless Large Language Model Acceleration via Self-Speculative Decoding | We present a novel inference scheme, self-speculative decoding, for accelerating Large Language Models (LLMs) without the need for an auxiliary model. This approach is characterized by a two-stage process: drafting and verification. The drafting stage generates draft tokens at a slightly lower quality but more quickly,... | acl | long | 2,024 | https://aclanthology.org/2024.acl-long.607.pdf | [
"NLP engineering experiment",
"Approaches low compute settings-efficiency"
] | WueVcpFqKv | December 2023 | [
{
"contribution_types": [
"NLP engineering experiment",
"Approaches low compute settings-efficiency"
],
"contribution_types_has_changed": false,
"cycle": "October 2023",
"id": "-GPD_kF1--"
}
] | [{"1 Introduction": ["Transformer-based Large Language Models (LLMs), such as GPT-3/4, PaLM, and LLaMA, have been widely adopted in various real-world applications (Bommasani et al., 2021; Liang et al., 2022; Brown et al., 2020; Min et al., 2022; Chan et al., 2022; Touvron et al., 2023). However, their inference costs ... |
2024.findings-eacl.122 | Parameter-Efficient Fine-Tuning: Is There An Optimal Subset of Parameters to Tune? | The ever-growing size of pretrained language models (PLM) presents a significant challenge for efficiently fine-tuning and deploying these models for diverse sets of tasks within memory-constrained environments.In light of this, recent research has illuminated the possibility of selectively updating only a small subset... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.122.pdf | [
"Approaches low compute settings-efficiency"
] | -X-A7GO_bd | October 2023 | [] | [{"1 Introduction": ["In recent years, the number of parameters used in language models has risen much faster than the memory available in GPUs (Lialin et al., 2023). This creates high memory requirements for fine-tuning such models on available hardware. Further, this creates high memory requirements when deploying a ... |
2024.naacl-long.431 | Naive Bayes-based Context Extension for Large Language Models | Large Language Models (LLMs) have shown promising in-context learning abilities. However, conventional In-Context Learning (ICL) approaches are often impeded by length limitations of transformer architecture, which pose challenges when attempting to effectively integrate supervision from a substantial number of demonst... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.431.pdf | [
"NLP engineering experiment"
] | j0pVr1fIKQR | December 2023 | [
{
"contribution_types": [
"NLP engineering experiment",
"Approaches low compute settings-efficiency",
"Theory"
],
"contribution_types_has_changed": true,
"cycle": "October 2023",
"id": "NI5_JdrE-G"
}
] | [{"1 Introduction": ["Large Language Models (LLMs) have demonstrated remarkable capabilities in in-context learning (ICL), a paradigm that enables them to excel in various unseen tasks based on task examples or instructions within their context (Han et al., 2021; Qiu et al., 2020). Unlike traditional fine-tuning method... |
2024.findings-eacl.127 | Sequence Shortening for Context-Aware Machine Translation | Context-aware Machine Translation aims to improve translations of sentences by incorporating surrounding sentences as context. Towards this task, two main architectures have been applied, namely single-encoder (based on concatenation) and multi-encoder models. In this study, we show that a special case of multi-encoder... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.127.pdf | [
"NLP engineering experiment"
] | pL0PGOZ91S | October 2023 | [] | [{"1 Introduction": ["Following the introduction of the Transformer model Vaswani et al. (2017), Sentence-level Machine Translation, where the task is to translate separate sentences, has seen great success in recent years Vaswani et al. (2017); Hassan et al. (2018); Costa-jussa et al. (2022); Tiedemann et al. (2022). ... |
2024.eacl-long.181 | CCPrefix: Counterfactual Contrastive Prefix-Tuning for Many-Class Classification | Recently, prefix-tuning was proposed to efficiently adapt pre-trained language models to a broad spectrum of natural language classification tasks. It leverages soft prefix as task-specific indicators and language verbalizers as categorical-label mentions to narrow the formulation gap from pre-training language models.... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.181.pdf | [
"NLP engineering experiment"
] | pGd4LVnA7k4 | October 2023 | [] | [{"1 Introduction": ["While the fine-tuning approach has been highly successful in the field of natural language processing, enabling the effective application of knowledge to specific tasks, a significant disparity still exists between the pre-training and fine-tuning stages. This disparity can impede the efficient tr... |
2024.eacl-long.37 | MAFIA: Multi-Adapter Fused Inclusive Language Models | Pretrained Language Models (PLMs) are widely used in NLP for various tasks. Recent studies have identified various biases that such models exhibit and have proposed methods to correct these biases. However, most of the works address a limited set of bias dimensions independently such as gender, race, or religion. Moreo... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.37.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Approaches to low-resource settings",
"Publicly available software and/or pre-trained models",
"Data resources",
"Data analysis"
] | DKFcSREpZAe | October 2023 | [] | [{"2 Methodology": ["In our study, we examine four primary bias dimensions: gender, race (ethnicity), religion, and profession. First, we discuss the method for generating counterfactual (CF) pairs in Section 2.1. Subsequently, we outline the approach to train debiasing adapters (DBAs) for each dimension in Section 2.2... |
2024.findings-acl.700 | Visualizing Dialogues: Enhancing Image Selection through Dialogue Understanding with Large Language Models | For dialogue systems, the utilization of multimodal dialogue responses, as opposed to relying solely on text-only responses, offers the capability to describe different concepts through various modalities. This enhances the effectiveness of communication and elevates the overall conversational experience. However, curr... | acl | findings | 2,024 | https://aclanthology.org/2024.findings-acl.700.pdf | [
"NLP engineering experiment",
"Publicly available software and/or pre-trained models"
] | V9f9mHBXA3w | October 2023 | [] | [{"1 Introduction": ["In recent years, the landscape of online conversations has undergone a significant transformation thanks to the proliferation of instant messaging tools. Unlike the past, when these exchanges were confined to text alone, today's conversations have evolved into a multimodal experience, incorporatin... |
2024.naacl-long.438 | SuperGLEBer: German Language Understanding Evaluation Benchmark | We assemble a broad Natural Language Understanding benchmark suite for the German language and consequently evaluate a wide array of existing German-capable models in order to create a better understanding of the current state of German LLMs. Our benchmark consists of 29 different tasks ranging over different types suc... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.438.pdf | [
"Model analysis & interpretability",
"Publicly available software and/or pre-trained models",
"Data resources"
] | sWRVWp6B0d | October 2023 | [] | [{"1 Introduction": ["Fueled by the release of ChatGPT [22], the development of very capable, large language models (LLMs) has been accelerating, which also results in the release of more and more powerful models capable of the German language [13, 14]. From an NLP point of view, German is a language that apart from sm... |
2024.naacl-long.472 | LegalDiscourse: Interpreting When Laws Apply and To Whom | While legal AI has made strides in recent years, it still struggles with basic legal concepts: _when_ does a law apply? _Who_ does it applies to? _What_ does it do? We take a _discourse_ approach to addressing these problems and introduce a novel taxonomy for span-and-relation parsing of legal texts. We create a datase... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.472.pdf | [
"Data resources"
] | 0tyxAQgFCR | October 2023 | [] | [{"1 Introduction": ["AI practitioners have long explored how to use automation to _interpret the law1_Mehl (1958). Recent advances in NLP and information retrieval have already enabled practical applications Dale (2019), such as legal question answering bots 2, contract generation3, and automatic motion-filing Gibbs (... |
2024.eacl-short.34 | Corpus-Steered Query Expansion with Large Language Models | Recent studies demonstrate that query expansions generated by large language models (LLMs) can considerably enhance information retrieval systems by generating hypothetical documents that answer the queries as expansions. However, challenges arise from misalignments between the expansions and the retrieval corpus, resu... | eacl | short | 2,024 | https://aclanthology.org/2024.eacl-short.34.pdf | [
"NLP engineering experiment"
] | n6V9An2Cse | October 2023 | [] | [{"1 Introduction": ["Query expansion enhances the effectiveness of information retrieval systems by incorporating additional texts into the original query, which are traditionally identified via pseudo-relevance feedback (Amati and Van Rijsbergen, 2002; Robertson, 1990) or by leveraging external lexical knowledge sour... |
2024.eacl-short.41 | Pre-Training Methods for Question Reranking | One interesting approach to Question Answering (QA) is to search for semantically similar questions, which have been answered before. This task is different from answer retrieval as it focuses on questions rather than only on the answers, therefore it requires different model training on different data.In this work, we... | eacl | short | 2,024 | https://aclanthology.org/2024.eacl-short.41.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | 97P3lRAeeP | October 2023 | [] | [{"1 Introduction": ["An effective approach for answering user questions is to find semantically identical questions, which have been previously answered. Although this method cannot be applied to completely new questions, it provides optimal solutions for applications such as Frequently Asked Questions (FAQs) Sakata e... |
2024.naacl-long.254 | Backdoor Attacks on Multilingual Machine Translation | While multilingual machine translation (MNMT) systems hold substantial promise, they also have security vulnerabilities. Our research highlights that MNMT systems can be susceptible to a particularly devious style of backdoor attack, whereby an attacker injects poisoned data into a low-resource language pair to cause m... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.254.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Approaches to low-resource settings"
] | eqlK2Hl8wY | October 2023 | [] | [{"1 Introduction": ["Recently, multilingual neural machine translation (MNMT) systems have shown significant advantages Fan et al. (2021); Costa-jussa et al. (2022), in particular in greatly enhancing the translation performance on low-resource languages. Since MNMT training is strongly dependent on multilingual corpo... |
2024.eacl-long.5 | Leak, Cheat, Repeat: Data Contamination and Evaluation Malpractices in Closed-Source LLMs | Natural Language Processing (NLP) research is increasingly focusing on the use of Large Language Models (LLMs), with some of the most popular ones being either fully or partially closed-source. The lack of access to model details, especially regarding training data, has repeatedly raised concerns about data contaminati... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.5.pdf | [
"Data analysis",
"Surveys"
] | exbPWKOyzF | October 2023 | [] | [{"1 Introduction": ["The recent emergence of large language models (LLMs), that show remarkable performance on a wide range of tasks, has led not only to a dramatic increase in their use in research but also to a growing number of companies joining the race for the biggest and most powerful models. In pursuing a compe... |
2024.eacl-long.168 | Counterfactual Reasoning with Knowledge Graph Embeddings | Knowledge graph embeddings (KGEs) were originally developed to infer true but missing facts in incomplete knowledge repositories.In this paper, we link knowledge graph completion and counterfactual reasoning via our new task CFKGR. We model the original world state as a knowledge graph, hypothetical scenarios as edges ... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.168.pdf | [
"Data resources"
] | ZxOTyY9NI8 | October 2023 | [] | [{"1 Introduction": ["Reasoning about hypothetical situations (counterfactual reasoning) and anticipating the effects of a change in the current state of the world is central to human cognition [1, 14], and has been identified as a key concept in game theory [1, 19] and agent-based systems [1, 20]. It has even been arg... |
2024.findings-naacl.138 | How Interpretable are Reasoning Explanations from Prompting Large Language Models? | Prompt Engineering has garnered significant attention for enhancing the performance of large language models across a multitude of tasks. Techniques such as the Chain-of-Thought not only bolster task performance but also delineate a clear trajectory of reasoning steps, offering a tangible form of explanation for the au... | naacl | findings | 2,024 | https://aclanthology.org/2024.findings-naacl.138.pdf | [
"Model analysis & interpretability"
] | mNpuRZvDmbXM | December 2023 | [
{
"contribution_types": [
"Model analysis & interpretability"
],
"contribution_types_has_changed": false,
"cycle": "October 2023",
"id": "cQ9XRPIdPm"
}
] | [{"1 Introduction": ["In recent trends, Large Language Models (LLM) have shown impressive performance across a diverse array of tasks, primarily through extensive scaling of model size Brown et al. (2020). Techniques such as instruct-tuning Wei et al. (2021) applied across diverse tasks have empowered LLMs to execute i... |
2024.eacl-long.27 | Semantic Sensitivities and Inconsistent Predictions: Measuring the Fragility of NLI Models | Recent studies of the emergent capabilities of transformer-based Natural Language Understanding (NLU) models have indicated that they have an understanding of lexical and compositional semantics. We provide evidence that suggests these claims should be taken with a grain of salt: we find that state-of-the-art Natural L... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.27.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | T0UslFrC9N | October 2023 | [] | [{"1 Introduction": ["Transformer-based [20] Language Models (LMs) have shown solid performance across various NLU tasks [29]. These advances have led to suggestions regarding the emergent capabilities of the models in terms of syntactic [21, 19, 20], logic [22, 23] and semantic [24, 25] understanding. However, we pres... |
2024.eacl-long.145 | Advancing Precise Outline-Conditioned Text Generation with Task Duality and Explicit Outline Control | Existing works on outline-conditioned text generation typically aim to generate text using provided outlines as rough sketches, such as keywords and phrases. However, these approaches make it challenging to control the quality of text generation and assess consistency between outlines and generated texts due to lack of... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.145.pdf | [
"NLP engineering experiment",
"Data resources"
] | LLwYUggO2I | October 2023 | [] | [{"3 Methods": ["Precise Outline-conditioned Generation", "Baseline ApproachOur proposed precise outline-conditioned generation task takes specific sentence-level outlines and prompts as input, and requires generating long texts that are fluent, coherent, and match the input information. Specifically, the writing promp... |
2024.findings-eacl.12 | Large Language Models for Psycholinguistic Plausibility Pretesting | In psycholinguistics, the creation of controlled materials is crucial to ensure that research outcomes are solely attributed to the intended manipulations and not influenced by extraneous factors. To achieve this, psycholinguists typically pretest linguistic materials, where a common pretest is to solicit plausibility ... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.12.pdf | [
"NLP engineering experiment"
] | pw9cjunl9P2 | October 2023 | [] | [{"1 Introduction": ["Psycholinguistic research explores humans' exceptional language comprehension abilities, aiming to uncover underlying mechanisms through experiments and cognitive modelling (Frazier, 1987; Lewis and Vasishth, 2005; Gibson, 2000; Levy, 2008; MacDonald et al., 1994; Futrell et al., 2020; Tabor and H... |
2024.naacl-long.321 | Contextual Label Projection for Cross-Lingual Structured Prediction | Label projection, which involves obtaining translated labels and texts jointly, is essential for leveraging machine translation to facilitate cross-lingual transfer in structured prediction tasks. Prior research exploring label projection often compromise translation accuracy by favoring simplified label translation or... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.321.pdf | [
"NLP engineering experiment",
"Approaches to low-resource settings"
] | GcR_skwCvg | October 2023 | [] | [{"1 Introduction": ["Cross-lingual transfer for structured prediction tasks such as named entity recognition, relation extraction, and event extraction, has gained considerable attention recently [19, 20, 21, 22, 23, 24, 25]. It generalizes models trained in source languages to applications on other target languages [... |
2024.findings-eacl.62 | Do Language Models Know When They're Hallucinating References? | State-of-the-art language models (LMs) are notoriously susceptible to generating hallucinated information. Such inaccurate outputs not only undermine the reliability of these models but also limit their use and raise serious concerns about misinformation and propaganda. In this work, we focus on hallucinated book and a... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.62.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Publicly available software and/or pre-trained models",
"Data resources",
"Data analysis"
] | WzlaztFjmWq | October 2023 | [] | [{"1 Introduction": ["Despite their unparalleled capabilities, recent large language models (LLMs) still exhibit a tendency to generate seemingly credible yet incorrect or unfounded information. This phenomenon is often referred to as the \"hallucination\" problem in the field of natural-language processing (NLP).1 As ... |
2024.findings-eacl.112 | Teaching Probabilistic Logical Reasoning to Transformers | In this paper, we evaluate the capability of transformer-based language models in making inferences over uncertain text that includes uncertain rules of reasoning. We cover both Pre-trained Language Models (PLMs) and generative Large Language Models (LLMs). Our evaluation results show that both generations of language ... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.112.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | CEIS8n0hgXQ | October 2023 | [] | [{"1 Introduction": ["Language models have demonstrated high performance across a wide range of Natural Language Processing (NLP) tasks Liu et al. (2019) which in the case of Large Language Models holds even in zero-shot setting Chen (2023). However, they struggle to reason over uncertain text involving logical probabi... |
2024.naacl-long.464 | David helps Goliath: Inference-Time Collaboration Between Small Specialized and Large General Diffusion LMs | Diffusion-based language models are emerging as a promising alternative to autoregressive LMs: they approach the competence of autoregressive LMs while offering nuanced controllability at inference time. While autoregressive LMs have benefited immensely from scaling and instruction-based learning, existing studies of d... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.464.pdf | [
"NLP engineering experiment",
"Publicly available software and/or pre-trained models"
] | IP17In3wXW5 | October 2023 | [] | [{"1 Introduction": ["Following the footsteps of diffusion-based generative models for continuously valued data such as images, audio, and video (Ho et al., 2020; Kong et al., 2021; Ho et al., 2022), recent works have attempted to replicate these successes on discrete text data (Austin et al., 2021; Li et al., 2022c; H... |
2024.findings-eacl.151 | Joint Inference of Retrieval and Generation for Passage Re-ranking | Passage retrieval is a crucial component of modern open-domain question answering (QA) systems, providing information for downstream QA components to generate accurate and transparent answers. In this study we focus on passage re-ranking, proposing a simple yet effective method, Joint Passage Re-ranking (JPR), that opt... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.151.pdf | [
"NLP engineering experiment",
"Publicly available software and/or pre-trained models"
] | t7Y9sibi3Ks | October 2023 | [] | [{"1 Introduction": ["Passage retrieval is a crucial component in open-domain question answering (QA) (Chen and Yih, 2020), a task that requires answering questions from a wide range of domains and could be applied in systems that fulfill user's information needs (Voorhees et al., 1999). Retrieval offers downstream QA ... |
2024.findings-acl.971 | EHR-SeqSQL : A Sequential Text-to-SQL Dataset For Interactively Exploring Electronic Health Records | In this paper, we introduce EHR-SeqSQL, a novel sequential text-to-SQL dataset for Electronic Health Record (EHR) databases. EHR-SeqSQL is designed to address critical yet underexplored aspects in text-to-SQL parsing: interactivity, compositionality, and efficiency. To the best of our knowledge, EHR-SeqSQL is not only ... | acl | findings | 2,024 | https://aclanthology.org/2024.findings-acl.971.pdf | [
"Data resources",
"Data analysis"
] | S_DVSWve0Ehy | October 2023 | [] | [{"1 Introduction": ["Text-to-SQL provides a practical opportunity for non-experts to explore databases, even without prior knowledge of the database operations. Electronic Health Records (EHRs) are large-scale relational databases (RDBs) storing vast and comprehensive patient data Johnson et al. (2016); Pollard et al.... |
2024.naacl-long.359 | ToXCL: A Unified Framework for Toxic Speech Detection and Explanation | The proliferation of online toxic speech is a pertinent problem posing threats to demographic groups. While explicit toxic speech contains offensive lexical signals, implicit one consists of coded or indirect language. Therefore, it is crucial for models not only to detect implicit toxic speech but also to explain its ... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.359.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | 83EGEoQFnuV | October 2023 | [] | [{"1 Introduction": ["Warning: This paper discusses and contains content that can be offensive or upsetting.", "While social media has dramatically expanded democratic participation in public discourse, they have also become a widely recognized platform for the dissemination of toxic speech (Mathew et al., 2021; ElSher... |
2024.acl-long.340 | FastFiD: Improve Inference Efficiency of Open Domain Question Answering via Sentence Selection | Open Domain Question Answering (ODQA) has been advancing rapidly in recent times, driven by significant developments in dense passage retrieval and pretrained language models. State-of-the-art models typically incorporate the FiD framework, which is composed by a neural retriever alongside an encoder-decoder neural rea... | acl | long | 2,024 | https://aclanthology.org/2024.acl-long.340.pdf | [
"NLP engineering experiment",
"Approaches low compute settings-efficiency"
] | yF06l6KkSc | October 2023 | [] | [{"1 Introduction": ["Open Domain Question Answering(ODQA) is a longstanding task in Natural Language Processing that involves generating an answer solely based on a given question. Recent advancements in this field have typically adopted the Retriever-Reader framework [2, 16, 17, 18], which breaks down the task into t... |
2024.naacl-long.463 | REPLUG: Retrieval-Augmented Black-Box Language Models | We introduce REPLUG, a retrieval-augmented language modeling framework that treats the language model (LM) as a black box and augments it with a tuneable retrieval model. Unlike prior retrieval-augmented LMs that train language models with special cross-attention mechanisms to encode the retrieved text, REPLUG simply p... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.463.pdf | [
"NLP engineering experiment"
] | 6z_yPCrdCA4 | October 2023 | [] | [{"1 Introduction": ["Large language models (LMs) such as GPT-3 Brown et al. (2020) and Codex Chen et al. (2021), have demonstrated impressive performance on a wide range of language tasks. These models are typically trained on very large datasets and store a substantial amount of world or domain knowledge implicitly i... |
2024.findings-eacl.38 | Bootstrap Your Own PLM: Boosting Semantic Features of PLMs for Unsuperivsed Contrastive Learning | This paper aims to investigate the possibility of exploiting original semantic features of PLMs (pre-trained language models) during contrastive learning in the context of SRL (sentence representation learning). In the context of feature modification, we identified a method called IFM (implicit feature modification), w... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.38.pdf | [
"NLP engineering experiment"
] | 8P2Db19cs2 | October 2023 | [] | [{"1 Introduction": ["Contrastive learning has been successfully adopted in the field of VRL by constructing contrastive pairs (drawing positive pairs and repelling negative pairs) based on the sufficient background of augmentation strategies (He et al., 2020; Chen et al., 2020). After that, SRL (sentence representatio... |
2024.eacl-short.13 | On the Benefits of Fine-Grained Loss Truncation: A Case Study on Factuality in Summarization | Text summarization and simplification are among the most widely used applications of AI. However, such models are often prone to hallucination, which can result from training models on unaligned data. One efficient approach to address this issue is Loss Truncation (Kang and Hashimoto, 2020), an approach to modify the s... | eacl | short | 2,024 | https://aclanthology.org/2024.eacl-short.13.pdf | [
"Model analysis & interpretability"
] | QFGsa3f-plp | October 2023 | [] | [{"1 Introduction": ["Text summarization and simplification are among the most widely used NLP applications. However, such models are prone to generating hallucinations (Cao et al., 2022; Zhao et al., 2020; Maynez et al., 2020; Tang et al., 2023); this may have harmful real-world impact and hinder the adoption of such ... |
2024.findings-eacl.37 | Simple Temperature Cool-down in Contrastive Framework for Unsupervised Sentence Representation Learning | In this paper, we proposes a simple, tricky method to improve sentence representation of unsupervised contrastive learning. Even though contrastive learning has achieved great performances in both visual representation learning (VRL) and sentence representation learning (SRL) fields, we focus on the fact that there is ... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.37.pdf | [
"NLP engineering experiment"
] | 7WNQvITV-YV | October 2023 | [] | [{"1 Introduction": ["One of the most important breakthroughs in unsupervised representation learning is the introduction of contrastive learning into the field of deep learning (Chen et al., 2020; He et al., 2020). In the past few years, a number of studies have sought to analyze the success of contrastive learning. F... |
2024.findings-eacl.73 | Relabeling Minimal Training Subset to Flip a Prediction |
When facing an unsatisfactory prediction from a machine learning model, users can be interested in investigating the underlying reasons and exploring the potential for reversing the outcome. We ask: To flip the prediction on a test point (x_{t}), how to identify the smallest training subset (\mathcal{S}{t}) that we ne... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.73.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Position papers"
] | qDXW08uuZRQ | October 2023 | [] | [{"1 Introduction": ["The interpretability of machine learning systems is a crucial research area as it aids in understanding model behavior, facilitating debugging, and enhancing performance (Adebayo et al., 2020; Han et al., 2020; Pezeshkpour et al., 2022; Teso et al., 2021; Marx et al., 2019). A common approach invo... |
2024.findings-eacl.118 | Improving Backchannel Prediction Leveraging Sequential and Attentive Context Awareness | Backchannels, which refer to short and often affirmative or empathetic responses from a listener during a conversation, play a crucial role in effective communication. In this paper, we introduce CABP(Context-Aware Backchannel Prediction), a sequential and attentive context approach aimed at enhancing backchannel predi... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.118.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | vJQr_n0mNf | October 2023 | [] | [{"1 Introduction": ["Backchanneling is a conversational technique that involves providing short responses, such as \"Wow\" or \"Uh-huh,\" to display attention and engagement with the speaker's utterances (Ruede et al., 2019). Poppe et al. (2010) has shown that timely backchanneling can enhance the speaker's storytelli... |
2024.findings-eacl.125 | CODET: A Benchmark for Contrastive Dialectal Evaluation of Machine Translation | Neural machine translation (NMT) systems exhibit limited robustness in handling source-side linguistic variations. Their performance tends to degrade when faced with even slight deviations in language usage, such as different domains or variations introduced by second-language speakers. It is intuitive to extend this o... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.125.pdf | [
"Approaches to low-resource settings",
"Data resources",
"Data analysis"
] | DE9E3fLU3x | October 2023 | [] | [{"1 Introduction": ["Progress in natural language processing (NLP) and other varieties of human language technology throughout the 2010s has been undeniably swift. However, such advances are limited to a set of languages with largely available resources (Joshi et al., 2020; Blasi et al., 2022); they have focused solel... |
2024.findings-naacl.51 | DivTOD: Unleashing the Power of LLMs for Diversifying Task-Oriented Dialogue Representations | Language models pre-trained on general text have achieved impressive results in diverse fields. Yet, the distinct linguistic characteristics of task-oriented dialogues (TOD) compared to general text limit the practical utility of existing language models. Current task-oriented dialogue pre-training methods overlook the... | naacl | findings | 2,024 | https://aclanthology.org/2024.findings-naacl.51.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | 50vk4eo8Pi | October 2023 | [] | [{"1 Introduction": ["Many NLP applications frequently utilize pre-trained language models (PLMs) [4, 15], which are based on extensive general text corpora [16].These models are pre-trained in a self-supervised manner and then fine-tuned for supervised downstream tasks. The Pretrain and Finetune paradigm has significa... |
2024.findings-eacl.69 | On the Intractability to Synthesize Factual Inconsistencies in Summarization | Factual consistency detection has gotten raised attention in the task of abstractive summarization. Many existing works rely on synthetic training data, which may not accurately reflect or match the inconsistencies produced by summarization models. In this paper, we first systematically analyze the shortcomings of the ... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.69.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Publicly available software and/or pre-trained models",
"Data analysis"
] | bVpIOOgXKpS | October 2023 | [] | [{"1 Introduction": ["With the advancements in neural conditioned generation, abstractive summarization systems, which are dominantly based on neural networks, have achieved phenomenal performances. However, summaries generated so often contain content that is factually inconsistent with the source documents (Kryscinsk... |
2024.naacl-long.165 | ChatGPT as an Attack Tool: Stealthy Textual Backdoor Attack via Blackbox Generative Model Trigger | Textual backdoor attacks, characterized by subtle manipulations of input triggers and training dataset labels, pose significant threats to security-sensitive applications. The rise of advanced generative models, such as GPT-4, with their capacity for human-like rewriting, makes these attacks increasingly challenging to... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.165.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | 7B6iaJD6qwi | October 2023 | [] | [{"1 Introduction": ["Deep Learning models have achieved remarkable success in natural language processing (NLP) tasks Devlin et al. (2019); Lewis et al. (2020); Radford et al. (2019); Xue et al. (2020); Raffel et al. (2020); Brown et al. (2020); OpenAI (2023). However, these models are susceptible to _backdoor attacks... |
2024.findings-eacl.85 | Efficiently Aligned Cross-Lingual Transfer Learning for Conversational Tasks using Prompt-Tuning | Cross-lingual transfer of language models trained on high-resource languages like English has been widely studied for many NLP tasks, but focus on conversational tasks has been rather limited. This is partly due to the high cost of obtaining non-English conversational data, which results in limited coverage. In this wo... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.85.pdf | [
"Approaches to low-resource settings",
"Approaches low compute settings-efficiency",
"Data resources"
] | CODSR5H5wh8 | October 2023 | [] | [{"1 Introduction": ["It has long been known that NLP research and applications are concentrated on high-resource languages such as English, French, and Japanese. This limitation introduces bias and prevents people in minority language groups from accessing recent NLP technologies.", "Driven by advances in large-scale ... |
2024.findings-eacl.107 | Exploiting Class Probabilities for Black-box Sentence-level Attacks | Sentence-level attacks craft adversarial sentences that are synonymous with correctly-classified sentences but are misclassified by the text classifiers. Under the black-box setting, classifiers are only accessible through their feedback to queried inputs, which is predominately available in the form of class probabili... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.107.pdf | [
"Model analysis & interpretability"
] | 5FzUClAyLYg | October 2023 | [] | [{"1 Introduction": ["Despite the tremendous success of text classification models (Devlin et al., 2018; Liu et al., 2019), studies have exposed their susceptibility to adversarial examples, i.e., carefully crafted sentences with human-unrecognizable changes to the inputs that are misclassified by the classifiers (Zhan... |
2024.findings-eacl.59 | MEDs for PETs: Multilingual Euphemism Disambiguation for Potentially Euphemistic Terms | Euphemisms are found across the world's languages, making them a universal linguistic phenomenon. As such, euphemistic data may have useful properties for computational tasks across languages. In this study, we explore this premise by training a multilingual transformer model (XLM-RoBERTa) to disambiguate potentially e... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.59.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Approaches to low-resource settings"
] | c4I1_b3HsA | October 2023 | [] | [{"1 Introduction": ["Euphemisms are a linguistic device used to soften or neutralize language that may otherwise be harsh or awkward to state directly (e.g. \"between jobs\" instead of \"unemployed\", \"late\" instead of \"dead\", \"collateral damage\" instead of \"war-related civilian deaths\"). By acting as alternat... |
2024.naacl-long.416 | LeanReasoner: Boosting Complex Logical Reasoning with Lean | Large language models (LLMs) often struggle with complex logical reasoning due to logical inconsistencies and the inherent difficulty ofsuch reasoning. We use Lean, a theorem proving framework, to address these challenges. By formalizing logical reasoning problems intotheorems within Lean, we can solve them by proving ... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.416.pdf | [
"NLP engineering experiment",
"Approaches to low-resource settings",
"Data resources"
] | Rt6N1GIzp6 | December 2023 | [
{
"contribution_types": [
"NLP engineering experiment"
],
"contribution_types_has_changed": true,
"cycle": "October 2023",
"id": "maT9JAAz_u"
}
] | [{"1 Introduction": ["Logical reasoning, a bedrock of intelligence and a core capability of humans, has been a challenging issue for machine learning systems for a long time. LLMs, despite their impressive abilities to understand and generate natural language, often fall short when dealing with complex logical reasonin... |
2024.findings-eacl.111 | Low-Resource Counterspeech Generation for Indic Languages: The Case of Bengali and Hindi | With the rise of online abuse, the NLP community has begun investigating the use of neural architectures to generate counterspeech that can “counter” the vicious tone of such abusive speech and dilute/ameliorate their rippling effect over the social network. However, most of the efforts so far have been primarily focus... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.111.pdf | [
"Model analysis & interpretability",
"Approaches to low-resource settings",
"Data resources",
"Data analysis"
] | qiXdtnhqMY | October 2023 | [] | [{"1 Introduction": ["The rise of online hostility has become an ominous issue endangering the safety of targeted people and groups and the welfare of society as a whole (Statt, 2017; Vedeler et al., 2019; Johnson et al., 2019). Therefore, to mitigate the widespread use of such hateful content, social media platforms g... |
2024.eacl-long.105 | Learning to Retrieve In-Context Examples for Large Language Models | Large language models (LLMs) have demonstrated their ability to learn in-context, allowing them to perform various tasks based on a few input-output examples. However, the effectiveness of in-context learning is heavily reliant on the quality of the selected examples. In this paper, we propose a novel framework to iter... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.105.pdf | [
"NLP engineering experiment"
] | lB0Xfv3Q8D | October 2023 | [] | [{"1 Introduction": ["In-context learning (ICL) Brown et al. (2020) is an emerging learning paradigm that allows LLMs to perform tasks with few-shot examples, without requiring any updates to the model parameters. This approach stands in stark contrast to traditional machine learning, where models are typically trained... |
2024.naacl-long.410 | Parameter-Efficient Instruction Tuning of Large Language Models For Extreme Financial Numeral Labelling | We study the problem of automatically annotating relevant numerals (GAAP metrics) occurring in the financial documents with their corresponding XBRL tags. Different from prior works, we investigate the feasibility of solving this extreme classification problem using a generative paradigm through instruction tuning of L... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.410.pdf | [
"NLP engineering experiment"
] | LrFvyeb24- | October 2023 | [] | [{"1 Introduction": ["The U.S. Securities and Exchange Commission (SEC) mandates publicly traded companies to disclose periodic filings such as quarterly 10-Q & annual 10-K reports. These documents are important to finance professionals and investors who rely on SEC filings to make informed investment decisions. Each c... |
2024.findings-eacl.32 | Contextualization Distillation from Large Language Model for Knowledge Graph Completion | While textual information significantly enhances the performance of pre-trained language models (PLMs) in knowledge graph completion (KGC), the static and noisy nature of existing corpora collected from Wikipedia articles or synsets definitions often limits the potential of PLM-based KGC models. To surmount these chall... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.32.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Publicly available software and/or pre-trained models"
] | 1JidfThspS | October 2023 | [] | [{"1 Introduction": ["Knowledge graph completion (KGC) is a fundamental task in natural language processing (NLP), aiming at unveiling hidden insights within diverse knowledge graphs to explore novel knowledge patterns. Traditional KGC methods Nickel et al. (2011); Bordes et al. (2013) typically predict the missing par... |
2024.eacl-long.119 | Investigating Agency of LLMs in Human-AI Collaboration Tasks | Agency, the capacity to proactively shape events, is central to how humans interact and collaborate. While LLMs are being developed to simulate human behavior and serve as human-like agents, little attention has been given to the Agency that these models should possess in order to proactively manage the direction of in... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.119.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Data resources",
"Data analysis"
] | yGMPClJoYP | October 2023 | [] | [{"1 Introduction": ["To be an agent is to intentionally cause events to occur through one's own actions. Humans operate with Agency to proactively plan their activities, direct their interaction and collaboration with other humans, and achieve their outcomes and goals [1].", "AI researchers have long strived to develo... |
2024.eacl-long.175 | Gradient-Based Language Model Red Teaming | Red teaming is a common strategy for identifying weaknesses in generative language models (LMs) by producing adversarial prompts that trigger models to generate unsafe responses. Red teaming is instrumental for both model alignment and evaluation, but is labor-intensive and difficult to scale when done by humans. In th... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.175.pdf | [
"NLP engineering experiment"
] | SL3ZqaKwkE | October 2023 | [] | [{"1 Introduction": ["Generative transformer-based language models (LMs) have achieved state-of-the-art results across many tasks, including in high-stakes domains such as medicine and education (Anil et al., 2023; OpenAI, 2023; Singhal et al., 2023; Touvron et al., 2023). These general-purpose models have an enormous ... |
2024.eacl-long.7 | GEAR: Augmenting Language Models with Generalizable and Efficient Tool Resolution | Augmenting large language models (LLM) to use external tools enhances their performance across a variety of tasks. However, prior works over-rely on task-specific demonstration of tool use that limits their generalizability and computational cost due to making many calls to large-scale LLMs. We introduce GEAR, a comput... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.7.pdf | [
"NLP engineering experiment",
"Approaches to low-resource settings",
"Approaches low compute settings-efficiency"
] | SnH8saq8EqY | October 2023 | [] | [{"1 Introduction": ["Recently there has been a surge in research on Augmented Language Model [13], which aims to enable models interface existing \"tools\" for various purposes, such as accessing the latest information [11], interacting with third-party services [11], performing precise calculations [14], or reasoning... |
2024.findings-eacl.64 | CEO: Corpus-based Open-Domain Event Ontology Induction | Existing event-centric NLP models often only apply to the pre-defined ontology, which significantly restricts their generalization capabilities.This paper presents CEO, a novel Corpus-based Event Ontology induction model to relax the restriction imposed by pre-defined event ontologies. Without direct supervision, CEO l... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.64.pdf | [
"NLP engineering experiment",
"Publicly available software and/or pre-trained models"
] | jdOVhXFJqKZ | October 2023 | [] | [{"1 Introduction": ["Extracting and understanding real-world events described in the text are crucial information extraction tasks that lay the foundations for downstream NLP applications Chen et al. (2021); Zhang et al. (2020); Fung et al. (2021). However, existing event-related studies are mostly restricted by the p... |
2024.naacl-long.209 | Analyzing the Role of Semantic Representations in the Era of Large Language Models | Traditionally, natural language processing (NLP) models often use a rich set of features created by linguistic expertise, such as semantic representations. However, in the era of large language models (LLMs), more and more tasks are turned into generic, end-to-end sequence generation problems. In this paper, we investi... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.209.pdf | [
"Model analysis & interpretability"
] | cgrxIjl9hq | October 2023 | [] | [{"1 Introduction": ["Formal representations of linguistic structure and meaning have long held an important role in the construction and evaluation of NLP systems. Semantic representations such as Abstract Meaning Representation (AMR; Banarescu et al., 2013) are designed to distill the semantic information of text to ... |
2024.eacl-long.142 | Investigating Content Planning for Navigating Trade-offs in Knowledge-Grounded Dialogue | Knowledge-grounded dialogue generation is a challenging task because it requires satisfying two fundamental, yet often competing constraints: being responsive in a manner that is specific to what the conversation partner has said while also being attributable to an underlying source document. In this work, we bring thi... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.142.pdf | [
"Model analysis & interpretability"
] | gj4gD3LoaX | October 2023 | [] | [{"1 Introduction": ["A knowledge-grounded dialogue system that aims to address a user's information needs must meet two fundamental requirements. First, the knowledge shared by the system must be credible. A common formulation for this constraint is that the system must share information that is faithful or attributab... |
2024.eacl-long.97 | Discovering and Articulating Frames of Communication from Social Media Using Chain-of-Thought Reasoning | Frames of Communication (FoCs) are ubiquitous in social media discourse. They define what counts as a problem, diagnose what is causing the problem, elicit moral judgments and imply remedies for resolving the problem. Most research on automatic frame detection involved the recognition of the problems addressed by frame... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.97.pdf | [
"Position papers"
] | abndH0SqGM | October 2023 | [] | [{"1 Introduction": ["The way in which we interpret information depends on how the information is framed (Entman, 2003; Reese et al., 2001; Scheufele, 2004; Chong and Druckman, 2012; Bolsen et al., 2014). For instance, if information about vaccines is framed to build our confidence in them, we can become vaccine enthus... |
2024.findings-naacl.260 | A Study on Scaling Up Multilingual News Framing Analysis | Media framing is the study of strategically selecting and presenting specific aspects of political issues to shape public opinion. Despite its relevance to almost all societies around the world, research has been limited due to the lack of available datasets and other resources. This study explores the possibility of d... | naacl | findings | 2,024 | https://aclanthology.org/2024.findings-naacl.260.pdf | [
"NLP engineering experiment",
"Approaches to low-resource settings",
"Data resources"
] | D29_hjgwXs | December 2023 | [
{
"contribution_types": [
"NLP engineering experiment",
"Approaches to low-resource settings",
"Data resources"
],
"contribution_types_has_changed": false,
"cycle": "October 2023",
"id": "kVffND8f7U"
}
] | [{"1 Introduction": ["News framing refers to the power of the news media to define and interpret events, issues, and policies by emphasizing certain aspects while downplaying or excluding others. According to Entman (1993), it can \"make a piece of information more noticeable, meaningful, or memorable to audiences\". I... |
2024.naacl-long.364 | PlanRAG: A Plan-then-Retrieval Augmented Generation for Generative Large Language Models as Decision Makers | In this paper, we conduct a study to utilize LLMs as a solution for decision making that requires complex data analysis. We define **Decision QA** as the task of answering the best decision, $d_{best}$, for a decision-making question $Q$, business rules $R$ and a database $D$. Since there is no benchmark that can exami... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.364.pdf | [
"NLP engineering experiment",
"Data resources"
] | 4sajV6NEnWE | October 2023 | [] | [{"1 Introduction": ["In many business situations, decision making plays a crucial role for the success of organizations (Kasie et al., 2017; Gupta et al., 2002). Here, decision making involves analyzing data, ultimately leading to the selection of the most suitable alternative to achieve a specific goal (Provost and F... |
2024.findings-eacl.133 | Unraveling the Dynamics of Semi-Supervised Hate Speech Detection: The Impact of Unlabeled Data Characteristics and Pseudo-Labeling Strategies | Despite advances in machine learning based hate speech detection, the need for larges amounts of labeled training data for state-of-the-art approaches remains a challenge for their application. Semi-supervised learning addresses this problem by leveraging unlabeled data and thus reducing the amount of annotated data re... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.133.pdf | [
"NLP engineering experiment",
"Approaches to low-resource settings"
] | jTd1Q2-bMt | October 2023 | [] | [{"1 Introduction": ["Topic shifts in online hate speech arising from changing social media trends or news poses a challenge for hate speech detection systems Florio et al. (2020). In order to keep the pace and follow such dynamic changes developers of such systems need to adapt their models to the continuously changin... |
2024.eacl-long.26 | Fréchet Distance for Offline Evaluation of Information Retrieval Systems with Sparse Labels | The rapid advancement of natural language processing, information retrieval (IR), computer vision, and other technologies has presented significant challenges in evaluating the performance of these systems. One of the main challenges is the scarcity of human-labeled data, which hinders the fair and accurate assessment ... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.26.pdf | [
"Approaches to low-resource settings",
"Theory"
] | g-q3nMipSa | October 2023 | [] | [{"1 Introduction": ["With the rapid advancement of technologies in fields such as natural language processing, natural language generation, computer vision, and information retrieval (IR), evaluating the performance of these systems is becoming increasingly challenging [13, 14, 15, 16]. We must develop new metrics, be... |
2024.findings-eacl.53 | Consistent Joint Decision-Making with Heterogeneous Learning Models | This paper introduces a novel decision-making framework that promotes consistency among decisions made by diverse models while utilizing external knowledge. Leveraging the Integer Linear Programming(ILP) framework, we map predictions from various models into globally normalized and comparable values by incorporating in... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.53.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Approaches to low-resource settings"
] | 8LVCb2HPh6 | October 2023 | [] | [{"1 Introduction": ["The rapid advance of AI has led to the widespread use of neural networks in tackling complex tasks that involve multiple output decisions, which may be derived from various models (Liu et al., 2022; Wang et al., 2022). However, in many cases, these decisions are interrelated and must conform to sp... |
2024.eacl-long.150 | Effective Controllable Bias Mitigation for Classification and Retrieval using Gate Adapters | Bias mitigation of Language Models has been the topic of many studies with a recent focus on learning separate modules like adapters for on-demand debiasing. Besides optimizing for a modularized debiased model, it is often critical in practice to control the degree of bias reduction at inference time, e.g., in order to... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.150.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Publicly available software and/or pre-trained models"
] | izWHHWreyX | October 2023 | [] | [{"1 Introduction": ["The purpose of this paper is to design a modularized debiased model to enhance the performance of the debiased version at inference time, e.g.", "bility by introducing Controllable Gate Adapter (ConGater). The proposed module is based on a novel gating mechanism, that learns to reduce protected at... |
2024.findings-eacl.129 | Non-Exchangeable Conformal Language Generation with Nearest Neighbors | Quantifying uncertainty in automatically generated text is important for letting humans check potential hallucinations and making systems more reliable. Conformal prediction is an attractive framework to provide predictions imbued with statistical guarantees, however, its application to text generation is challenging s... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.129.pdf | [
"NLP engineering experiment"
] | nRYJu0EuuCI | October 2023 | [] | [{"1 Introduction": ["Natural language generation (NLG) is a multi-faceted field spanning applications such as machine translation (MT), language modeling (LM), summarization, question answering and dialogue generation. Owing to the recent success of large language models (LLMs) such as GPT-4 (OpenAI, 2023), BLOOM (Sca... |
2024.findings-eacl.46 | Prosody in Cascade and Direct Speech-to-Text Translation: a case study on Korean Wh-Phrases | Speech-to-Text Translation (S2TT) has typically been addressed with cascade systems, where speech recognition systems generate a transcription that is subsequently passed to a translation model. While there has been a growing interest in developing direct speech translation systems to avoid propagating errors and losin... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.46.pdf | [
"Model analysis & interpretability",
"Data analysis"
] | Psa0F8ES2P | October 2023 | [] | [{"1 Introduction": ["Speech-to-Text Translation (S2TT) is the task of automatically generating a text translation in a target language given an input speech signal. Traditionally, S2TT has been achieved by concatenating two systems: one in charge of generating an intermediate transcription of the source speech signal ... |
2024.eacl-long.68 | HumBEL: A Human-in-the-Loop Approach for Evaluating Demographic Factors of Language Models in Human-Machine Conversations | While demographic factors like age and gender change the way people talk, and in particular, the way people talk to machines, there is little investigation into how large pre-trained language models (LMs) can adapt to these changes. To remedy this gap, we consider how demographic factors in LM language skills can be me... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.68.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Data resources"
] | KwJgFpVDVGd | October 2023 | [] | [{"1 Introduction": ["Demographic factors like age and gender impact the words we use [20, 1] and, more broadly, the way we interact and communicate with each other [4]. Moreover, these same factors carry over influence into our conversations with machines. Age group, in particular, impacts the way we converse with hou... |
2024.findings-eacl.135 | Can Large Language Models Understand Context? | Understanding context is key to understanding human language, an ability which Large Language Models (LLMs) have been increasingly seen to demonstrate to an impressive extent. However, though the evaluation of LLMs encompasses various domains within the realm of Natural Language Processing, limited attention has been p... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.135.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Data analysis"
] | f6CP5G3oDZ | October 2023 | [] | [{"1 Introduction": ["Discourse understanding, as one of the fundamental problems in NLP, focuses on modeling linguistic features and structures that go beyond individual sentences [10]. Understanding discourse requires resolving the relations between words/phrases (coreference resolution) and discourse units (discours... |
2024.findings-eacl.81 | Exploring efficient zero-shot synthetic dataset generation for Information Retrieval | The broad integration of neural retrieval models into Information Retrieval (IR) systems is significantly impeded by the high cost and laborious process associated with the manual labelling of training data. Similarly, synthetic training data generation, a potential workaround, often requires expensive computational re... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.81.pdf | [
"NLP engineering experiment",
"Approaches low compute settings-efficiency"
] | MG8BOJzMOv | October 2023 | [] | [{"1 Introduction": ["Deep Learning is at the heart of many current breakthroughs in AI in a wide range of fields. Typically, such progress is attributed to better computational capabilities, superior algorithms, and a larger corpus of high-quality training data. Particularly in the Information Retrieval (IR) field, si... |
2024.eacl-long.116 | It is not True that Transformers are Inductive Learners: Probing NLI Models with External Negation | NLI tasks necessitate a substantial degree of logical reasoning; as such, the remarkable performance of SoTA transformers on these tasks may lead us to believe that those models have learned to reason logically. The results presented in this paper demonstrate that (i) models fine-tuned on NLI datasets learn to treat ex... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.116.pdf | [
"Model analysis & interpretability",
"Theory"
] | Tv3RmfrTsR | October 2023 | [] | [{"1 Introduction": ["Natural language inference (NLI) tasks require detecting inferential relations between pairs of sentences [13]. For NLI datasets such as MultiNLI (MNLI; [20]) and Stanford NLI (SNLI; [21]), the task proceeds as follows: given a pair of sentences ((P,H)), an NLI model must determine whether the pre... |
2024.emnlp-main.1097 | Evaluating Diversity in Automatic Poetry Generation | Natural Language Generation (NLG), and more generally generative AI, are among the currently most impactful research fields. Creative NLG, such as automatic poetry generation, is a fascinating niche in this area. While most previous research has focused on forms of the Turing test when evaluating automatic poetry gener... | emnlp | main | 2,024 | https://aclanthology.org/2024.emnlp-main.1097.pdf | [
"Model analysis & interpretability",
"Data analysis"
] | wo7FcGvM8r | June 2024 | [
{
"contribution_types": [
"Model analysis & interpretability"
],
"contribution_types_has_changed": true,
"cycle": "October 2023",
"id": "3mmdJ3pRyM"
}
] | [{"1 Introduction": ["A key aspect of creative language generation is the ability to create new, original and interesting text, cf. Colton et al. (2012); Gatt and Krahmer (2018); Yi et al. (2020); Elgammal et al. (2017). To date, extremely little attention has been given to the evaluation of originality and creativity ... |
2024.findings-eacl.120 | Analyzing the Role of Part-of-Speech in Code-Switching: A Corpus-Based Study | Code-switching (CS) is a common linguistic phenomenon wherein speakers fluidly transition between languages in conversation. While the cognitive processes driving CS remain a complex domain, earlier investigations have shed light on its multifaceted triggers. This study delves into the influence of Part-of-Speech (POS)... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.120.pdf | [
"Data analysis"
] | 482C4VZvf5 | October 2023 | [] | [{"1 Introduction": ["Code-switching (CS), the integration of two languages within a single utterance, is pervasive across diverse language pairs. This phenomenon presents the flexibility and adaptability of individuals in their language use and therefore serves as a testing ground for research into the cognitive mecha... |
2024.eacl-long.21 | MultiMUC: Multilingual Template Filling on MUC-4 | We introduce MultiMUC, the first multilingual parallel corpus for template filling, comprising translations of the classic MUC-4 template filling benchmark into five languages: Arabic, Chinese, Farsi, Korean, and Russian. We obtain automatic translations from a strong multilingual machine translation system and manuall... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.21.pdf | [
"NLP engineering experiment",
"Data resources"
] | Idd4xM5BK0 | October 2023 | [] | [{"1 Introduction": ["The Message Understanding Conferences (MUCs) were a series of U.S. government-sponsored competitions that ran from the late 1980s through the late 1990s whose aim was to promote the development of systems for extracting complex relations from text, and which have been credited with inaugurating th... |
2024.eacl-long.83 | M4: Multi-generator, Multi-domain, and Multi-lingual Black-Box Machine-Generated Text Detection | Large language models (LLMs) have demonstrated remarkable capability to generate fluent responses to a wide variety of user queries. However, this has also raised concerns about the potential misuse of such texts in journalism, education, and academia. In this study, we strive to create automated systems that can detec... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.83.pdf | [
"NLP engineering experiment",
"Data resources",
"Data analysis"
] | ZTF2mX-Mpwh | October 2023 | [] | [{"1 Introduction": ["Large language models (LLMs) are becoming mainstream and easily accessible, ushering in an explosion of machine-generated content over various channels, such as news, social media, question-answering forums, educational, and even academic contexts. Recently introduced LLMs, such as ChatGPT, GPT-4,... |
2024.eacl-long.163 | Presentations by the Humans and For the Humans: Harnessing LLMs for Generating Persona-Aware Slides from Documents | Scientific papers and slides are two different representations of the same underlying information, but both require substantial work to prepare. While there had been prior efforts on automating document-to-slides generation, there is still a pressing need of customizing the presentation of content aligning with the per... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.163.pdf | [
"Model analysis & interpretability",
"Data resources"
] | jkay-UKF_1 | October 2023 | [] | [{"1 Presentations are Everywhere...How can we make them customized to end user needs?": ["From business to education to research, presentations are everywhere Zheng et al. (2022); Bhattacharyya (2014); Tarkhova et al. (2020). A recent 2023 survey reveals that 20.3 million people in the UK have used Powerpoint and over... |
2024.eacl-short.31 | A Comparative Analysis of Conversational Large Language Models in Knowledge-Based Text Generation | Generating natural language text from graph-structured data is essential for conversational information seeking. Semantic triples derived from knowledge graphs can serve as a valuable source for grounding responses from conversational agents by providing a factual basis for the information they communicate. This is esp... | eacl | short | 2,024 | https://aclanthology.org/2024.eacl-short.31.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | Eo-fHE5vU5 | October 2023 | [] | [{"1 Introduction": ["Accessing structured information through natural language interfaces has garnered significant research interest in natural language processing (NLP) Aliannejadi et al. (2021); Radlinski and Craswell (2017). For instance, the emerging information retrieval paradigm of conversational search frames i... |
2024.eacl-long.35 | NNOSE: Nearest Neighbor Occupational Skill Extraction | The labor market is changing rapidly, prompting increased interest in the automatic extraction of occupational skills from text. With the advent of English benchmark job description datasets, there is a need for systems that handle their diversity well. We tackle the complexity in occupational skill datasets tasks—comb... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.35.pdf | [
"NLP engineering experiment",
"Publicly available software and/or pre-trained models",
"Data analysis"
] | L1nTNqQS_J | October 2023 | [] | [{"2 Nearest Neighbor Skill Extraction": ["Skill Extraction.The task of SE is formulated as a sequence labeling problem. We define a set of job description sentences (\\mathcal{X}), where each (d\\in\\mathcal{X}) represents a set of sequences with the (j^{\\text{th}}) input sequence (\\mathcal{X}{d}^{j}={x{1},x_{2},...... |
2024.eacl-long.132 | Exploring Data Augmentation in Neural DRS-to-Text Generation | Neural networks are notoriously data-hungry. This represents an issue in cases where data are scarce such as in low-resource languages. Data augmentation is a technique commonly used in computer vision to provide neural networks with more data and increase their generalization power. When dealing with data augmentation... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.132.pdf | [
"NLP engineering experiment"
] | Mo4R53Cj6q62 | October 2023 | [] | [{"1 Introduction": ["Data augmentation is a systematic way of increasing data examples by altering the original data with controlled variations Feng et al. (2021). It is a prevalent technique in computer vision (CV) for increasing dataset size by introducing slightly different and contextually similar examples Yang et... |
2024.findings-eacl.27 | Re3val: Reinforced and Reranked Generative Retrieval | Generative retrieval models encode pointers to information in a corpus as an index within the model's parameters. These models serve as part of a larger pipeline, where retrieved information conditions generation for knowledge-intensive NLP tasks. However, we identify two limitations: the generative retrieval does not ... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.27.pdf | [
"Publicly available software and/or pre-trained models"
] | 1eRFOi-tnN | October 2023 | [] | [{"1 Introduction": ["The primary objective of retrieval models is to enhance the accuracy of answers by selecting the most relevant documents retrieved for a given query, ensuring models have sufficient information to help the downstream reasoning process. For instance, DRQA [20] introduces a \"retrieve and read\" pip... |
2024.acl-long.627 | Cross-Lingual Knowledge Editing in Large Language Models | Knowledge editing aims to change language models' performance on several special cases (i.e., editing scope) by infusing the corresponding expected knowledge into them. With the recent advancements in large language models (LLMs), knowledge editing has been shown as a promising technique to adapt LLMs to new knowledge ... | acl | long | 2,024 | https://aclanthology.org/2024.acl-long.627.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Reproduction study",
"Data resources"
] | lCtaGCFWMd | October 2023 | [] | [{"1 Introduction": ["The goal of knowledge editing is to adjust language models' behaviors within an expected scope (i.e., editing scope) and retain out-of-scope model performance ideally (Yao et al., 2023). Along with the dynamic changes in the world, knowledge editing could help models forget outdated knowledge and ... |
2024.eacl-long.149 | Ask, Assess, and Refine: Rectifying Factual Consistency and Hallucination in LLMs with Metric-Guided Feedback Learning | Recent advancements in Large Language Models (LLMs) have heralded unprecedented capabilities in information-seeking and text generation, as evidenced by applications like Bing Chat and perplexity.ai. Despite these strides, challenges on hallucination and factual inconsistency continue to impede their wider real-world a... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.149.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Publicly available software and/or pre-trained models",
"Theory"
] | mGyikIQFoH | October 2023 | [] | [{"1 Introduction": ["Recent pioneering works on Large Language Models (LLMs) have facilitated for information seeking and text generation, thereby showcasing the various real-world applications such as Bing Chat1 and perplexity.ai2. However, despite of significant advancements with a combination of supervised fine-tun... |
2024.findings-eacl.148 | ICE-Score: Instructing Large Language Models to Evaluate Code | Recent advancements in the field of natural language generation have facilitated the use of large language models to assess the quality of generated text. Although these models have shown promising results in tasks such as machine translation and summarization, their applicability in code intelligence tasks remains lim... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.148.pdf | [
"NLP engineering experiment",
"Publicly available software and/or pre-trained models",
"Data resources"
] | RoGZaCsGUW | October 2023 | [] | [{"1 Introduction": ["Natural language generation (NLG) systems have seen significant progress in developing large language models (LLMs). These models have shown great promise in generating high-quality and diverse texts that can be difficult to distinguish from human-written texts (Ouyang et al., 2022). However, eval... |
2024.eacl-short.36 | Sentence Representations via Gaussian Embedding | Recent progress in sentence embedding, which represents a sentence's meaning as a point in a vector space, has achieved high performance on several tasks such as the semantic textual similarity (STS) task.However, a sentence representation cannot adequately express the diverse information that sentences contain: for ex... | eacl | short | 2,024 | https://aclanthology.org/2024.eacl-short.36.pdf | [
"NLP engineering experiment",
"Publicly available software and/or pre-trained models"
] | ydUPOmNkrM0 | October 2023 | [] | [{"1 Introduction": ["Sentence embeddings are representations to describe a sentence's meaning and are widely used in natural language tasks such as document classification [13], sentence retrieval [23], and question answering [13]. In recent years, machine-learning-based sentence embedding methods with pre-trained lan... |
2024.findings-eacl.123 | Contextualized Topic Coherence Metrics | This article proposes a new family of LLM-based topic coherence metrics called Contextualized Topic Coherence (CTC) and inspired by standard human topic evaluation methods. CTC metrics simulate human-centered coherence evaluation while maintaining the efficiency of other automated methods. We compare the performance of... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.123.pdf | [
"Model analysis & interpretability"
] | dI4sICTbQC | October 2023 | [] | [{"1 Introduction": ["Topic models are a family of text-mining algorithms that identify themes in a large corpus of text data [1]. These models [13] are widely used for exploratory data analysis with the aim of organizing, understanding, and summarizing large amounts of text data [1]. Numerous techniques, algorithms, a... |
2024.eacl-long.117 | Polarized Opinion Detection Improves the Detection of Toxic Language | Distance from unimodality (DFU) has been found to correlate well with human judgment for the assessment of polarized opinions. However, its un-normalized nature makes it less intuitive and somewhat difficult to exploit in machine learning (e.g., as a supervised signal). In this work a normalized version of this measure... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.117.pdf | [
"NLP engineering experiment",
"Theory"
] | DKNaMP33ZL | October 2023 | [] | [{"1 Introduction": ["Annotations for subjective tasks are often aggregated, to form ground truth labels and allow supervised learning algorithms to be trained for these tasks. Given a text, for example, annotations are averaged to yield binary labels reflecting whether the text is misogynous or not (Kirk et al., 2023)... |
2024.eacl-long.12 | Few-Shot Data Synthesis for Open Domain Multi-Hop Question Answering | Few-shot learning for open domain multi-hop question answering typically relies on the in-context learning capability of large language models (LLMs). While powerful, these LLMs usually contain tens or hundreds of billions of parameters, making them rather inefficient at inference time. To improve performance of smalle... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.12.pdf | [
"NLP engineering experiment"
] | R-xmaDD6PR | October 2023 | [] | [{"1 Introduction": ["Few-shot learning for open domain multi-hop question answering seeks to answer complex questions by iteratively retrieving relevant information with a handful of human-annotated question answer pairs. It has become increasingly popular for evaluating the abilities of grounding to factual and up-to... |
2024.naacl-long.167 | ExpertQA: Expert-Curated Questions and Attributed Answers | As language models are adopted by a more sophisticated and diverse set of users, the importance of guaranteeing that they provide factually correct information supported by verifiable sources is critical across fields of study. This is especially the case for high-stakes fields, such as medicine and law, where the risk... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.167.pdf | [
"Model analysis & interpretability",
"Data resources"
] | RVe32S5r0y | December 2023 | [
{
"contribution_types": [
"Model analysis & interpretability",
"Data resources"
],
"contribution_types_has_changed": false,
"cycle": "October 2023",
"id": "irxDvAw7np"
}
] | [{"1 Introduction": ["As the influence of large language models (LLMs) grows beyond the computer science community, experts from various fields are rapidly adapting LLMs for assistance in information-seeking scenarios. For example, medical professionals are using these systems for performing differential diagnosis (Lee... |
2024.naacl-long.451 | Improving Machine Translation with Human Feedback: An Exploration of Quality Estimation as a Reward Model | Insufficient modeling of human preferences within the reward model is a major obstacle for leveraging human feedback to improve translation quality. Fortunately, quality estimation (QE), which predicts the quality of a given translation without reference, has achieved impressive alignment with human evaluations in the ... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.451.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Approaches to low-resource settings"
] | 1bFmdGHHfA | October 2023 | [] | [{"1 Introduction": ["Human feedback has greatly contributed to recent advances in large language models (LLMs), aligning model behavior with human preferences and thereby enhancing the helpfulness and harmlessness of LLMs (Dong et al., 2023; Yuan et al., 2023; Zhao et al., 2023; Rafailov et al., 2023). The common prac... |
2024.findings-eacl.60 | PromptExplainer: Explaining Language Models through Prompt-based Learning | Pretrained language models have become workhorses for various natural language processing (NLP) tasks, sparking a growing demand for enhanced interpretability and transparency. However, prevailing explanation methods, such as attention-based and gradient-based strategies, largely rely on linear approximations, potentia... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.60.pdf | [
"Model analysis & interpretability"
] | l8fM3WoWhF | October 2023 | [] | [{"1 Introduction": ["Recently, pretrained language models Devlin et al. (2019); Liu et al. (2019); OpenAI (2022); Touvron et al. (2023) have achieved remarkable success across a wide range of NLP tasks, such as text classification, question answering and machine translation. However, the inherent complexity of these m... |
2024.acl-long.147 | CopyNE: Better Contextual ASR by Copying Named Entities | End-to-end automatic speech recognition (ASR) systems have made significant progress in general scenarios. However, it remains challenging to transcribe contextual named entities (NEs) in the contextual ASR scenario. Previous approaches have attempted to address this by utilizing the NE dictionary. These approaches tre... | acl | long | 2,024 | https://aclanthology.org/2024.acl-long.147.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment"
] | peds9GhKeI | October 2023 | [] | [{"1 Introduction": ["End-to-end automatic speech recognition (ASR) systems have achieved impressive performance in general scenarios (Chan et al., 2016; Rao et al., 2017; Gulati et al., 2020; Boulianne, 2022). However, in the contextual ASR scenario where speech often contains numerous contextual entities, it remains ... |
2024.naacl-long.11 | Measuring and Improving Chain-of-Thought Reasoning in Vision-Language Models | Vision-language models (VLMs) have recently demonstrated strong efficacy as visual assistants that can parse natural queries about the visual content and generate human-like outputs. In this work, we explore the ability of these models to demonstrate human-like reasoning based on the perceived information. To address a... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.11.pdf | [
"Model analysis & interpretability",
"Data resources"
] | DvmMsdj7Uv | October 2023 | [] | [{"1 Introduction": ["Vision-language models (VLMs) exhibit competence at generating human-like responses by leveraging multimodal instructional data and large language models (LLMs) [14, 15, 16, 17]. A key direction in improving such VLMs is to enable grounded and consistent visual reasoning. We thus take a critical l... |
2024.naacl-long.296 | CASA: Causality-driven Argument Sufficiency Assessment | The argument sufficiency assessment task aims to determine if the premises of a given argument support its conclusion.To tackle this task, existing works often train a classifier on data annotated by humans. However, annotating data is laborious, and annotations are often inconsistent due to subjective criteria. Motiva... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.296.pdf | [
"NLP engineering experiment",
"Approaches to low-resource settings"
] | ix0H2mOc7y | October 2023 | [] | [{"1 Introduction": ["Argumentation is an integral part of our daily verbal communication (Fogelin and Sinnott-Armstrong, 2005; Stab and Gurevych, 2017a). An argument is a series of statements consisting of premises and a conclusion. Take the argument shown in Figure 1 as an example: You shouldn't trust Donald's views ... |
2024.naacl-long.15 | Unleashing the Emergent Cognitive Synergy in Large Language Models: A Task-Solving Agent through Multi-Persona Self-Collaboration | Human intelligence thrives on cognitive synergy, where collaboration among different minds yield superior outcomes compared to isolated individuals. In this work, we propose Solo Performance Prompting (SPP), which transforms a single LLM into a cognitive synergist by engaging in multi-turn self-collaboration with multi... | naacl | long | 2,024 | https://aclanthology.org/2024.naacl-long.15.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Approaches to low-resource settings"
] | 9EaiOXnkpf | October 2023 | [] | [{"1 Introduction": ["Although large language models (LLMs) have demonstrated impressive performance as general task-solving agents, they still encounter challenges Qin et al. (2023); Bang et al. (2023); OpenAI (2023); Bubeck et al. (2023) in various knowledge-intensive and reasoning-intensive tasks due to factual hall... |
2024.eacl-short.29 | Less is More for Long Document Summary Evaluation by LLMs | Large Language Models (LLMs) have shown promising performance in summary evaluation tasks, yet they face challenges such as high computational costs and the Lost-in-the-Middle problem where important information in the middle of long documents is often overlooked. To address these issues, this paper introduces a novel ... | eacl | short | 2,024 | https://aclanthology.org/2024.eacl-short.29.pdf | [
"NLP engineering experiment",
"Approaches low compute settings-efficiency"
] | 8-LyAk7xRJ | October 2023 | [] | [{"1 Introduction": ["The evaluation of text generation plays a crucial role in the development of high-quality text generation systems Celikyilmaz et al. (2020). However, the alignment of automatic evaluation metrics with human judgment remains a challenging task Bhandari et al. (2020); Fabbri et al. (2021). Recently,... |
2024.eacl-long.62 | Where Do We Go From Here? Multi-scale Allocentric Relational Inferencefrom Natural Spatial Descriptions | The concept of acquired spatial knowledge is crucial in spatial cognitive research, particularly when it comes to communicating routes. However, NLP navigation studies often overlook the impact of acquired knowledge on textual descriptions. Current navigation studies concentrate on egocentric local descriptions (e.g., ... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.62.pdf | [
"Publicly available software and/or pre-trained models",
"Data resources",
"Data analysis"
] | WMAGobNJT- | October 2023 | [] | [{"2 The RVS Task and Environment": ["In this work we address the task of following geospatial instructions given in colloquial language based on a dense urban map. The input to the RVS task is as follows: (i) a map with rich details, given as a knowledge graph; (ii) an explicit starting point, given in coordinates (la... |
2024.findings-eacl.96 | Aspect-based Key Point Analysis for Quantitative Summarization of Reviews | Key Point Analysis (KPA) is originally for summarizing arguments, where short sentences containing salient viewpoints are extracted as key points (KPs) and quantified for their prevalence as salience scores. Recently, KPA was applied to summarize reviews, but the study still relies on sentence-based KP extraction and m... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.96.pdf | [
"Data analysis"
] | u-5VifzRrA | October 2023 | [] | [{"1 Introduction": ["Summarization of user reviews on the online marketplace has become essential both for businesses to improve their product and service qualities and for customers to make purchasing decisions. Although the star ratings aggregated from customer reviews are widely used to measure quality of service f... |
2024.eacl-long.106 | EnCore: Fine-Grained Entity Typing by Pre-Training Entity Encoders on Coreference Chains | Entity typing is the task of assigning semantic types to the entities that are mentioned in a text. In the case of fine-grained entity typing (FET), a large set of candidate type labels is considered. Since obtaining sufficient amounts of manual annotations is then prohibitively expensive, FET models are typically trai... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.106.pdf | [
"NLP engineering experiment"
] | LPxEHRz8zMg | October 2023 | [] | [{"1 Introduction": ["Entity typing is a fundamental task in Natural Language Processing (NLP), with important applications to entity linking [14] and relation extraction [20, 15], among others. In recent years, the main focus has been on fine-grained entity typing [13, 16], where around 100 different entity types are ... |
2024.acl-long.472 | Faithful Chart Summarization with ChaTS-Pi | Chart-to-summary generation can help explore data, communicate insights, and help the visually impaired people. Multi-modal generative models have been used to produce fluent summaries, but they can suffer from factual and perceptual errors. In this work we present CHATS-CRITIC, a reference-free chart summarization met... | acl | long | 2,024 | https://aclanthology.org/2024.acl-long.472.pdf | [
"Model analysis & interpretability"
] | mbanOyQszUB | February 2024 | [
{
"contribution_types": [
"Model analysis & interpretability"
],
"contribution_types_has_changed": true,
"cycle": "October 2023",
"id": "LJxSqcF-sdG"
},
{
"contribution_types": [
"Model analysis & interpretability",
"NLP engineering experiment"
],
"contribution_... | [{"1 Introduction": ["Chart summarization requires faithfully extracting quantitative data and describing them using natural language. Recent natural language generation (NLG) studies have explored different flavors of chart-to-summary generation tasks including caption generation for scientific figures Hsu et al. (202... |
2024.eacl-long.57 | LaMini-LM: A Diverse Herd of Distilled Models from Large-Scale Instructions | Large language models (LLMs) with instruction fine-tuning demonstrate superior generative capabilities. However, these models are resource-intensive. To alleviate this issue, we explore distilling knowledge from instruction-tuned LLMs into much smaller ones. While other similar works have been done, they are often cond... | eacl | long | 2,024 | https://aclanthology.org/2024.eacl-long.57.pdf | [
"NLP engineering experiment",
"Approaches low compute settings-efficiency",
"Data resources"
] | cTv-SYLadz3 | October 2023 | [] | [{"1 Introduction": ["Large language models (LLMs) with instruction tuning have demonstrated remarkable capabilities in generating high-quality outputs for a diverse set of applications (Ouyang et al., 2022; Wei et al., 2022; Sanh et al., 2022; Chung et al., 2022; OpenAI, 2023). These models typically consist of billio... |
2024.findings-eacl.126 | QAEVENT: Event Extraction as Question-Answer Pairs Generation | We propose a novel representation of document-level events as question and answer pairs (QAEVENT). Under this paradigm: (1) questions themselves can define argument roles without the need for predefined schemas, which will cover a comprehensive list of event arguments from the document; (2) it allows for more scalable ... | eacl | findings | 2,024 | https://aclanthology.org/2024.findings-eacl.126.pdf | [
"Model analysis & interpretability",
"NLP engineering experiment",
"Data resources"
] | 2lw6U0z-SPo | October 2023 | [] | [{"1 Introduction": ["Event extraction (EE) is a challenging yet important task in information extraction research (Sundheim, 1992). The task aims at extracting event information from unstructured texts into a structured form, which mostly describes attributes such as \"who\", \"when\", \"where\", and \"what\" of real-... |
2024.naacl-short.50 | Lifelong Event Detection with Embedding Space Separation and Compaction | To mitigate forgetting, existing lifelong event detection methods typically maintain a memory module and replay the stored memory data during the learning of a new task. However, the simple combination of memory data and new-task samples can still result in substantial forgetting of previously acquired knowledge, which... | naacl | short | 2,024 | https://aclanthology.org/2024.naacl-short.50.pdf | [
"NLP engineering experiment"
] | OTBHoflCXxXF | December 2023 | [
{
"contribution_types": [
"NLP engineering experiment"
],
"contribution_types_has_changed": false,
"cycle": "October 2023",
"id": "QL69qAZgTnx"
}
] | [{"1 Introduction": ["Event detection (ED) aims to detect the event type of trigger words in a given sentence, e.g., extracting the event type injure from the trigger word scaled in text \"He was scaled by hot water\". Traditional ED methods typically consider a fixed pre-defined set of event types [4, 16, 17, 18]. How... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.