Croatia

When describing actions, subtle changes in word choice can evoke very different associations with the involved entities. For instance, a company `\{\textbackslash{}it employing\} workers&#39; evokes a more positive connotation than the one `\{\textbackslash{}it exploiting\}&#39; them. This concept is called \{\textbackslash{}it connotation\}. This paper investigates whether pre-trained language models (PLMs) encode such subtle connotative information about \{\textbackslash{}it power differentials\} between involved entities. We design a probing framework for power connotation, building on\textasciitilde{}\textbackslash{}citet\{sap-etal-2017-connotation\}&#39;s operationalization of \{\textbackslash{}it connotation frames\}. We show that zero-shot prompting of PLMs leads to above chance prediction of power connotation, however fine-tuning PLMs using our framework drastically improves their accuracy. Using our fine-tuned models, we present a case study of \{\textbackslash{}it power dynamics\} in US news reporting on immigration, showing the potential of our framework as a tool for understanding subtle bias in the media.

EACL 2023

Probing Power by Prompting: Harnessing Pre-trained Language Models for Power Connotation Framing

power connotation framing

probing power by prompting

When describing actions, subtle changes in word choice can evoke very different associations with the involved entities. For instance, a company `\{\textbackslash{}it employing\} workers' evokes a more positive connotation than the one `\{\textbackslash{}it exploiting\}' them. This concept is called \{\textbackslash{}it connotation\}. This paper investigates whether pre-trained language models (PLMs) encode such subtle connotative information about \{\textbackslash{}it power differentials\} between involved entities. We design a probing framework for power connotation, building on\textasciitilde{}\textbackslash{}citet\{sap-etal-2017-connotation\}'s operationalization of \{\textbackslash{}it connotation frames\}. We show that zero-shot prompting of PLMs leads to above chance prediction of power connotation, however fine-tuning PLMs using our framework drastically improves their accuracy. Using our fine-tuned models, we present a case study of \{\textbackslash{}it power dynamics\} in US news reporting on immigration, showing the potential of our framework as a tool for understanding subtle bias in the media.

poster

### Welcome to the 17th Conference of the European Chapter of the Association for Computational Linguistics (EACL) 
Continuing its mission of expanding and involving the science community of all European countries, EACL had selected the Ukrainian community for the 16th EACL, which took place online due to the COVID pandemic. Unfortunately, the ongoing war made the organisation in Kyiv impossible. Considering the importance of physical interaction among researchers, especially after the restrictions imposed by the COVID pandemic, in addition to an online mode, the conference will be held in Dubrovnik, Croatia, from 2 to 6 of May, 2023. The original aim of strengthening the connection with the Ukrainian community will not change as our program will feature a dedicated session and a workshop to highlight work on Ukrainian language technologies. As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

You need to register for the conference in order to access this site. Please visit https://2023.eacl.org/registration for more information.

As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

FEVEROUS is a fact extraction and verification task that requires systems to extract evidence of both sentences and table cells from a Wikipedia dump, then predict the veracity of the given claim accordingly. Existing works extract evidence in the two formats separately, ignoring potential connections between them. In this paper, we propose a Unified Evidence Extraction model (UnifEE), which uses a mixed evidence graph to extract the evidence in both formats. With the carefully-designed unified evidence graph, UnifEE allows evidence interactions among all candidates in both formats at similar granularity. Experiments show that, with information aggregated from related evidence candidates in the fusion graph, UnifEE can make better decisions about which evidence should be kept, especially for claims requiring multi-hop reasoning or a combination of tables and texts. Thus it outperforms all previous evidence extraction methods and brings significant improvement in the subsequent claim verification step.

UnifEE: Unified Evidence Extraction for Fact Verification

Recognizing medical self-disclosure is important in many healthcare contexts, but it has been under-explored by the NLP community. We conduct a three-pronged investigation of this task. We (1) manually expand and refine the only existing medical self-disclosure corpus, resulting in a new, publicly available dataset of 3,919 social media posts with clinically validated labels and high compatibility with the existing task-specific protocol. We also (2) study the merits of pretraining task domain and text style by comparing Transformer-based models for this task, pretrained from general, medical, and social media sources. Our BERTweet condition outperforms the existing state of the art for this task by a relative F1 score increase of 16.73\%. Finally, we (3) compare data augmentation techniques for this task, to assess the extent to which medical self-disclosure data may be further synthetically expanded. We discover that this task poses many challenges for data augmentation techniques, and we provide an in-depth analysis of identified trends.

What Clued the AI Doctor In? On the Influence of Data Source and Quality for Transformer-Based Medical Self-Disclosure Detection

Text-based environments enable RL agents to learn to converse and perform interactive tasks through natural language. However, previous RL approaches applied to text-based environments show poor performance when evaluated on unseen games. 
This paper investigates the improvement of generalisation performance through the simple switch from a value-based update method to a policy-based one, within text-based environments. We show that by replacing commonly used value-based methods with REINFORCE with baseline, a far more general agent is produced. The policy-based agent is evaluated on Coin Collector and Question Answering with interactive text (QAit), two text-based environments designed to test zero-shot performance. We see substantial improvements on a variety of zero-shot evaluation experiments, including tripling accuracy on various QAit benchmark configurations. The results indicate that policy-based RL has significantly better generalisation capabilities than value-based methods within such text-based environments, suggesting that RL agents could be applied to more complex natural language environments.

Policy-based Reinforcement Learning for Generalisation in Interactive Text-based Environments

Unsupervised anomaly detection seeks to identify deviant data samples in a dataset without using labels and constitutes a challenging task, particularly when the majority class is heterogeneous. This paper addresses this topic for textual data and aims to determine whether a text sample is an outlier within a potentially multi-topic corpus. To this end, it is crucial to grasp the semantic aspects of words, particularly when dealing with short texts, since it is difficult to syntactically discriminate data samples based only on a few words. Thereby we make use of word embeddings to represent each sample by a dense vector, efficiently capturing the underlying semantics. Then, we rely on the Mixture Model approach to detect which samples deviate the most from the underlying distributions of the corpus. Experiments carried out on real datasets show the effectiveness of the proposed approach in comparison to state-of-the-art techniques both in terms of performance and time efficiency, especially when more than one topic is present in the corpus.

Unsupervised Anomaly Detection in Multi-Topic Short-Text Corpora

The Appraisal framework in linguistics defines the framework for fine-grained evaluations and opinions and has contributed to sentiment analysis and opinion mining. As developing appraisal-annotated resources requires tagging of several dimensions with distinct semantic taxonomies, it has been primarily conducted manually by human experts through expensive and time-consuming processes. In this paper, we study how to automatically identify and annotate text segments for appraisal. 
We formulate the problem as a sequence tagging problem and propose novel task and sentiment adapters based on language models for appraisal tagging. Our model, named Adaptive Appraisal (A\$\^{}2\$), achieves superior performance than baseline adapter-based models and other neural classification models, especially for cross-domain and cross-language settings. Source code for A\$\^{}2\$ is available at: https://github.com/ltian678/AA-code.git

Task and Sentiment Adaptation for Appraisal Tagging

Event scenarios are often complex and involve multiple event sequences connected through different entity participants. Exploring such complex scenarios requires an ability to branch through different sequences, something that is difficult to achieve with standard event language modeling. To address this, we propose a question-guided generation framework that models events in complex scenarios as answers to questions about participants. At any step in the generation process, the framework uses the previously-generated events as context, but generates the next event as an answer to one of three questions: what else a participant did, what else happened to a participant, or what else happened. The participants and the questions themselves can be sampled or be provided as input from a user, allowing for controllable exploration. Our empirical evaluation shows that this question-guided generation provides better coverage of participants, diverse events within a domain, comparable perplexities for modeling event sequences, and more effective control for interactive schema generation.

Modeling Complex Event Scenarios via Simple Entity-focused Questions

Recent research has shown that language models exploit `artifacts' in benchmarks to solve tasks, rather than truly learning them, leading to inflated model performance. In pursuit of creating better benchmarks, we propose VAIDA, a novel benchmark creation paradigm for NLP, that focuses on guiding crowdworkers, an under-explored facet of addressing benchmark idiosyncrasies. VAIDA facilitates sample correction by providing realtime visual feedback and recommendations to improve sample quality. Our approach is domain, model, task, and metric agnostic, and constitutes a paradigm shift for robust, validated, and dynamic benchmark creation via human-and-metric-in-the-loop workflows. We evaluate via expert review and a user study with NASA TLX. We find that VAIDA decreases effort, frustration, mental, and temporal demands of crowdworkers and analysts, simultaneously increasing the performance of both user groups with a 45.8\% decrease in the level of artifacts in created samples. As a by product of our user study, we observe that created samples are adversarial across models, leading to decreases of 31.3\% (BERT), 22.5\% (RoBERTa), 14.98\% (GPT-3 fewshot) in performance.

Real-Time Visual Feedback to Guide Benchmark Creation: A Human-and-Metric-in-the-Loop Workflow

The semantic code search is to find code snippets from the collection of candidate code snippets with respect to a user query that describes functionality. Recent work on code search proposes data augmentation of queries for contrastive learning. This data augmentation approach modifies random words in queries. When a user web query for searching code snippet is too brief, the important word that represents the search intent of the query could be undesirably modified. A code snippet has informative components such as function name and documentation that describe its functionality. We propose to utilize these code components to identify important words and preserve them in the data augmentation step. We present KeyDAC (Keyword-based Data Augmentation for Contrastive learning) that identifies important words for code search from queries and code components based on term matching. KeyDAC augments query-code pairs while preserving keywords, and then leverages generated training instances for contrastive learning. We use KeyDAC to fine-tune various pre-trained language models and evaluate the performance of code search and code question answering via CoSQA and WebQueryTest. The experimental results confirm that KeyDAC substantially outperforms the current state-of-the-art performance, and achieves the new state-of-the-arts for both tasks.

Contrastive Learning with Keyword-based Data Augmentation for Code Search and Code Question Answering

Social commonsense contains many human biases due to social and cultural influence (Sap et al., 2020; Emelin et al., 2020). We focus on identifying cultural biases in data, specifically causal assumptions and commonsense implications, that strongly influence model decisions for a variety of tasks designed for social impact. This enables us to examine data for bias by making explicit the causal (if-then, inferential) relations in social commonsense knowledge used for decision making, furthering interpretable commonsense reasoning from a dataset perspective. We apply our methods on 2 social tasks: emotion detection and perceived value detection. We identify influential social commonsense knowledge to explain model behavior in the following ways. First, we augment large-scale language models with social knowledge and show improvements for the tasks, indicating the implicit assumptions a model requires to be successful on each dataset. Second, we identify influential events in the datasets by using social knowledge to cluster data and demonstrate the influence that these events have on model behavior via leave-K-out experiments. This allows us to gain a dataset-level understanding of the events and causal commonsense relationships that strongly influence predictions. We then analyze these relationships to detect influential cultural bias in each dataset. Finally, we use our influential event identification for detecting mislabeled examples and improve training and performance through their removal. We support our findings with manual analysis.

Social Commonsense for Explanation and Cultural Bias Discovery

Bias-measuring datasets play a critical role in detecting biased behavior of language models and in evaluating progress of bias mitigation methods. In this work, we focus on evaluating gender bias through coreference resolution, where previous datasets are either hand-crafted or fail to reliably measure an explicitly defined bias. To overcome these shortcomings, we propose a novel method to collect diverse, natural, and minimally distant text pairs via counterfactual generation, and construct Counter-GAP, an annotated dataset consisting of 4008 instances grouped into 1002 quadruples. We further identify a bias cancellation problem in previous group-level metrics on Counter-GAP, and propose to use the difference between inconsistency across genders and within genders to measure bias at a quadruple level. Our results show that four pre-trained language models are significantly more inconsistent across different
gender groups than within each group, and that a name-based counterfactual data augmentation method is more effective to mitigate such bias than an anonymization-based method.

Downloads

Next from EACL 2023

UnifEE: Unified Evidence Extraction for Fact Verification