Malta

This paper investigates the appropriate responses that Conversational Agent systems (CAs) should employ when subjected to sexual harassment by users. Previous studies indicate that conventional CAs often respond neutrally or evade such requests. Enhancing the responsiveness of CAs to offensive speech is crucial, as users might carry over these interactions into their social interactions. To address this issue, we selected evaluators to compare a series of responses to sexual harassment from four commercial CAs (Amazon Alexa, Apple Siri, Google Home, and Microsoft Cortana) with alternative responses we realized based on insights from psychological and sociological studies. Focusing on CAs with a female voice, given their increased likelihood of encountering offensive language, we conducted two experiments involving 22 evaluators (11 females and 11 males). In the initial experiment, participants assessed the responses in a textual format, while the second experiment involved the evaluation of responses generated with a synthetic voice exhibiting three different intonations (angry, neutral, and assertive). Results from the first experiment revealed a general preference for the responses we formulated. For the most voted replies, female evaluators exhibited a tendency towards responses with an assertive intent, emphasizing the sexually harassing nature of the request. Conversely, male evaluators leaned towards a more neutral response, aligning with prior findings that highlight gender-based differences in the perception of sexual harassment. The second experiment underscored a preference for assertive responses. The study&#39;s outcomes highlight the need to develop new, educational responses from CAs to instances of sexual harassment, aiming to discourage harmful behavior.

EACL 2024

How should Conversational Agent systems respond to sexual harassment?

assertive responses.

conversational agent systems

sexual harassment

Join AAPT and Physics Educators worldwide in developing and strengthening your professional connections by attending the 2021 AAPT Virtual Summer Meeting. AAPT meetings are a great way to reflect on your teaching, recommit to best practices and re-energize your career with new research, products, and techniques.

Listen to internationally known plenary speakers, attend sessions, workshops, and special activities. Attending the virtual summer meeting is an excellent opportunity to share your expertise and learn from and network with leaders in the field.

Please log in with your AAPT account or register for the event if you haven't already.

If you are registered please note that you need to be logged in with your AAPT email / credentials in order to get access.

AAPT SM 21

This paper investigates the appropriate responses that Conversational Agent systems (CAs) should employ when subjected to sexual harassment by users. Previous studies indicate that conventional CAs often respond neutrally or evade such requests. Enhancing the responsiveness of CAs to offensive speech is crucial, as users might carry over these interactions into their social interactions. To address this issue, we selected evaluators to compare a series of responses to sexual harassment from four commercial CAs (Amazon Alexa, Apple Siri, Google Home, and Microsoft Cortana) with alternative responses we realized based on insights from psychological and sociological studies. Focusing on CAs with a female voice, given their increased likelihood of encountering offensive language, we conducted two experiments involving 22 evaluators (11 females and 11 males). In the initial experiment, participants assessed the responses in a textual format, while the second experiment involved the evaluation of responses generated with a synthetic voice exhibiting three different intonations (angry, neutral, and assertive). Results from the first experiment revealed a general preference for the responses we formulated. For the most voted replies, female evaluators exhibited a tendency towards responses with an assertive intent, emphasizing the sexually harassing nature of the request. Conversely, male evaluators leaned towards a more neutral response, aligning with prior findings that highlight gender-based differences in the perception of sexual harassment. The second experiment underscored a preference for assertive responses. The study's outcomes highlight the need to develop new, educational responses from CAs to instances of sexual harassment, aiming to discourage harmful behavior.

workshop paper

Welcome to the 18th 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 has selected the Malta community for the 18th EACL. Considering the importance of physical interaction among researchers, the conference will be held at the Hotel Radisson Blu, St. Julians, in Malta, from 17 to 22 of March, 2024. The conference will also feature the possibility to participate virtually. 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. 
#### [Conference Handbook](https://drive.google.com/file/d/13Wn38q0yev6U3RTgnf4Tn5T9GuAuZ-xW/view?usp=sharing) *downloadable*

To access the EACL 2024 event page you need to register [**here**](https://acl.swoogo.com/EACL2024). 

Please register!

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.

Non-referential functions of language such as setting group boundaries, identity construction and regulation of social proximity have rarely found place in the language technology creation process. Nevertheless, their importance has been postulated in literature. While multiple methods to include social information in large language models (LLM) cover group properties (gender, age, geographic relations, professional characteristics), a combination of group social characteristics and individual features of an agent (natural or artificial) play a role in social interaction but have not been studied in generated language. This article explores the orchestration of prompt engineering and retrieval-augmented generation techniques to linguistic features of social proximity and distance in language generated by an LLM. We use the immediacy/distance model from literature to analyse language generated by an LLM for different recipients. This study reveals that kinship terms are almost the only way of displaying immediacy in LLM-made conversations.

Non-Referential Functions of Language in Social Agents: The Case of Social Proximity

Conversation systems facilitate interaction from many different types of users with distinct personalities and writing styles. In the context of multi-turn dialogue modeling, in this paper, we explore the effects of varying lengths of utterances on the quality of follow-up responses generated by conversation models. Using GPT-3 as the base model, multiple dialogue datasets, and several metrics, we comprehensively study this aspect of conversational models. In general, our analysis sheds light on the complex relationship between utterance lengths and the quality of follow-up responses generated by dialogue systems. Our empirical analysis suggests that, for certain types of conversations, the lengths of utterances can be reduced by up to 72% without any noticeable difference in the quality of follow-up responses.

Making a Long Story Short in Conversation Modeling

Large language models are increasingly deployed for high-stakes decision making, for example in financial and medical applications. In such applications, it is imperative that we be able to estimate our confidence in the answers output by a language model in order to assess risks. Although we can easily compute the probability assigned by a language model to the sequence of tokens that make up an answer, we cannot easily compute the probability of the answer itself, which could be phrased in numerous ways. While other works have engineered ways of assigning such probabilities to LLM outputs, a key problem remains: existing language models are poorly calibrated, often confident when they are wrong or unsure when they are correct. In this work, we devise a protocol called *calibration tuning* for finetuning LLMs to output calibrated probabilities. Calibration-tuned models demonstrate superior calibration performance compared to existing language models on a variety of question-answering tasks, including open-ended generation, without affecting accuracy. We further show that this ability transfers to new domains outside of the calibration-tuning train set.

Calibration-Tuning: Teaching Large Language Models to Know What They Don't Know

Large language models (LLMs) have the remarkable ability to solve new tasks with just a few examples, but they need access to the right tools. Retrieval Augmented Generation (RAG) addresses this problem by retrieving a list of relevant tools for a given task. However, RAG's tool retrieval step requires all the required information to be explicitly present in the query. This is a limitation, as semantic search, the widely adopted tool retrieval method, can fail when the query is incomplete or lacks context. To address this limitation, we propose Context Tuning for RAG, which employs a smart context retrieval system to fetch relevant information that improves both tool retrieval and plan generation. Our lightweight context retrieval model uses numerical, categorical, and habitual usage signals to retrieve and rank context items. Our empirical results demonstrate that context tuning significantly enhances semantic search, achieving a 3.5-fold and 1.5-fold improvement in Recall@K for context retrieval and tool retrieval tasks respectively, and resulting in an 11.6% increase in LLM-based planner accuracy. Additionally, we show that our proposed lightweight model using Reciprocal Rank Fusion (RRF) with LambdaMART outperforms GPT-4 based retrieval. Moreover, we observe context augmentation at plan generation, even after tool retrieval, reduces hallucination.

Context Tuning for Retrieval Augmented Generation

This work explores the effectiveness of employing Clinical BERT for Relation Extraction (RE) tasks in medical texts within an Active Learning (AL) framework. Our main objective is to optimize RE in medical texts through AL while examining the trade-offs between performance and computation time, comparing it with alternative methods like Random Forest and BiLSTM networks. Comparisons extend to feature engineering requirements, performance metrics, and considerations of annotation costs, including AL step times and annotation rates. The utilization of AL strategies aligns with our broader goal of enhancing the efficiency of relation classification models, particularly when dealing with the challenges of annotating complex medical texts in a Human-in-the-Loop (HITL) setting. The results indicate that uncertainty-based sampling achieves comparable performance with significantly fewer annotated samples across three categories of supervised learning methods, thereby reducing annotation costs for clinical and biomedical corpora. While Clinical BERT exhibits clear performance advantages across two different corpora, the trade-off involves longer computation times in interactive annotation processes. In real-world applications, where practical feasibility and timely results are crucial, optimizing this trade-off becomes imperative.

Optimizing Relation Extraction in Medical Texts through Active Learning: A Comparative Analysis of Trade-offs

Large Language Models (LLMs) have taken the research field of Natural Language Processing by storm. Researchers are not only investigating their capabilities and possible applications, but also their weaknesses and how they may be exploited. This has resulted in various attacks and "jailbreaking" approaches that have gained large interest within the community. The vulnerability of LLMs to certain types of input may pose major risks regarding the real-world usage of LLMs in productive operations. We therefore investigate the relationship between a LLM's uncertainty and its vulnerability to jailbreaking attacks. To this end, we focus on a probabilistic point of view of uncertainty and employ a state-of-the art open-source LLM. We investigate an attack that is based on linguistic obfuscation. Our results indicate that the model is subject to a higher level of uncertainty when confronted with manipulated prompts that aim to evade security mechanisms. This study lays the foundation for future research into the link between model uncertainty and its vulnerability to jailbreaks.

Linguistic Obfuscation Attacks and Large Language Model Uncertainty

Automatically generated summaries can be evaluated along different dimensions, one being how faithfully the uncertainty from the source text is conveyed in the summary. We present a study on uncertainty alignment in automatic summarization, starting from a two-tier lexical and semantic categorization of linguistic expression of uncertainty, which we used to annotate source texts and automatically generate summaries. We collected a diverse dataset including news articles and personal blogs and generated summaries using GPT-4. Source texts and summaries were annotated based on our two-tier taxonomy using a markup language. The automatic annotation was refined and validated by subsequent iterations based on expert input. We propose a method to evaluate the fidelity of uncertainty transfer in text summarization. The method capitalizes on a small amount of expert annotations and on the capabilities of Large language models (LLMs) to evaluate how the uncertainty of the source text aligns with the uncertainty expressions in the summary.

Aligning Uncertainty: Leveraging LLMs to Analyze Uncertainty Transfer in Text Summarization

Sequence labeling is a core task in text understanding for IE/IR systems. Text generation models have increasingly become the go-to solution for such tasks (e.g., entity extraction and dialog slot filling). While most research has focused on the labeling accuracy, a key aspect -- of vital practical importance -- has slipped through the cracks: understanding model confidence. More specifically, we lack a principled understanding of how to reliably gauge the confidence of a model in its predictions for each labeled span. This paper aims to provide some empirical insights on estimating model confidence for generative sequence labeling. Most notably, we find that simply using the decoder's output probabilities \textbf{is not} the best in realizing well-calibrated confidence estimates. As verified over six public datasets of different tasks, we show that our proposed approach -- which leverages statistics from top-$k$ predictions by a beam search -- significantly reduces calibration errors of the predictions of a generative sequence labeling model.

How Does Beam Search improve Span-Level Confidence Estimation in Generative Sequence Labeling?

Learning from human feedback can improve models for text generation or passage ranking, aligning them better to a user's needs. Data is often collected by asking users to compare alternative outputs to a given input, which may require a large number of comparisons to learn a ranking function. The amount of comparisons needed can be reduced using Bayesian Optimisation (BO) to query the user about only the most promising candidate outputs. Previous applications of BO to text ranking relied on shallow surrogate models to learn ranking functions over candidate outputs, and were therefore unable to fine-tune rankers based on deep, pretrained language models. This paper leverages Bayesian deep learning (BDL) to adapt pretrained language models to highly specialised text ranking tasks, using BO to tune the model with a small number of pairwise preferences between candidate outputs. We apply our approach to community question answering (cQA) and extractive multi-document summarisation (MDS) with simulated noisy users, finding that our BDL approach significantly outperforms both a shallow Gaussian process model and traditional active learning with a standard deep neural network, while remaining robust to noise in the user feedback.

Efficiently Acquiring Human Feedback with Bayesian Deep Learning

The highest probability sequences of most neural language generation models tend to be degenerate in some way, a problem known as the inadequacy of the mode. While many approaches to tackling particular aspects of the problem exist, such as dealing with too short sequences or excessive repetitions, explanations of why it occurs in the first place are rarer and do not agree with each other. We believe none of the existing explanations paint a complete picture. In this position paper, we want to bring light to the incredible complexity of the modelling task and the problems that generalising to previously unseen contexts bring. We argue that our desire for models to generalise to contexts it has never observed before is exactly what leads to spread of probability mass and inadequate modes. While we do not claim that adequate modes are impossible, we argue that they are not to be expected either.

Downloads

Next from EACL 2024

Non-Referential Functions of Language in Social Agents: The Case of Social Proximity

Similar lecture

How to talk about Sexual Harassment

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from EACL 2024

Non-Referential Functions of Language in Social Agents: The Case of Social Proximity

Similar lecture

How to talk about Sexual Harassment

Downloads