United States

Warning: This presentation contains content that is offensive and may be upsetting.

Biased or toxic speech can be harmful to various demographic groups. Therefore, it is not only important for models to detect these speech, but to also output explanations of why a given text is toxic. Previous literature has mostly focused on classifying and detecting toxic speech, and existing efforts on explaining stereotypes in toxic speech mainly use standard text generation approaches, resulting in generic and repetitive explanations. Building on these prior works, we introduce a novel knowledge informed encoder-decoder framework to utilize multiple knowledge sources to generate implications of biased text. Experiments show that our knowledge informed models outperform prior state-of-the-art models significantly, and can generate detailed explanations of stereotypes in toxic speech compared to baselines, both quantitatively and qualitatively.

NAACL 2022

Explaining Toxic Text via Knowledge Enhanced Text Generation

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible. The experience for on-site participants would closely approximate a normal pre-COVID *ACL conference, with 5-6 thematically organized parallel sessions and live Q/A and interactive discussion immediately after the talks. Presentations by virtual participants will be equitably interleaved with those of on-site participants, projected on the auditorium screens as if on-site, and also followed immediately by live Q/A and interactive discussion at a time during reasonable waking hours for the virtual presenter. For all participants, on-site and virtual, who are unable to attend a session due to either time-zone issues or because they are participating in another session live, talk recordings and slides will be available online at a minimum after the live presentation (and in many cases before as well), and questions may be submitted in advance on session-specific discussion boards and answered live in session with the usual visual aids if desired.

<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/ceby/" seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>

Please Note: The EMNLP registration system is not currently connected to the underline site as we are still in the process of building out EMNLP 2021. You will receive access instructions from underline the week of November 1st. 

Access is given only to EMNLP upon registration, if you have not registered please do so [here](https://2021.emnlp.org/registration).

Registered attendees will receive access the week of November 1st.

EMNLP 2021

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible.

technical paper

## Welcome to NAACL 2022!
The Annual Meeting of the North American Chapter of the Association for Computational Linguistics (NAACL) is taking place July 10-15, 2022 as a hybrid event, in Seattle, WA and online. We are happy to welcome all of you to conference! 
 
The main conference program features oral presentations, in-person and virtual posters and demo sessions, a plenary session for our best paper presentations and awards, keynote presentations, a plenary panel on the place of linguistics and symbolic structures in NLP, and an Industrial Track panel on Careers in NLP.
 
Posters (including Findings of NAACL 2022) and demos are grouped by areas for both the in-person and the virtual sessions. For the virtual component, the talks will be on Zoom and the posters and the demos will be in GatherTown. The Student Research Workshop and will have an oral session and a poster session.
 
The program also features 6 Tutorials and 26 Workshops.
 
 We wish you a wonderful conference!
 
[The NAACL 2022 Organizing Committee](https://2022.naacl.org/committees/organization/)

2022 Annual Conference of the North American Chapter of the Association for Computational Linguistics

In modern interactive speech-based systems, speech is consumed and transcribed incrementally prior to having disfluencies removed. While this post-processing step is crucial for producing clean transcripts and high performance on downstream tasks (e.g. machine translation), most current state-of-the-art NLP models such as the Transformer operate non-incrementally, potentially causing unacceptable delays for the user. In this work we propose a streaming BERT-based sequence tagging model that, combined with a novel training objective, is capable of detecting disfluencies in real-time while balancing accuracy and latency. This is accomplished by training the model to decide whether to immediately output a prediction for the current input or to wait for further context, in essence learning to dynamically size the lookahead window. Our results demonstrate that our model produces comparably accurate predictions and does so sooner than our baselines, with lower flicker. Furthermore, the model attains state-of-the-art latency and stability scores when compared with recent work on incremental disfluency detection.

Teaching BERT to Wait: Balancing Accuracy and Latency for Streaming Disfluency Detection

We present EASE, a novel method for learn- ing sentence embeddings via contrastive learn- ing between sentences and their related enti- ties. The advantage of using entity supervision is twofold: (1) entities have been shown to be a strong indicator of text semantics and thus should provide rich training signals for sen- tence embeddings; (2) entities are defined in- dependently of languages and thus offer useful cross-lingual alignment supervision. We eval- uate EASE against other unsupervised mod- els both in monolingual and multilingual set- tings. We show that EASE exhibits compet- itive or better performance in English seman- tic textual similarity (STS) and short text clus- tering (STC) tasks and it significantly outper- forms baseline methods in multilingual set- tings on a variety of tasks. Our source code, pretrained models, and newly constructed mul- tilingual STC dataset are available at https: //github.com/studio-ousia/ease.

EASE: Entity-Aware Contrastive Learning of Sentence Embedding

Multihop reasoning remains an elusive goal as existing multihop benchmarks are known to be largely solvable via shortcuts. Can we create a question answering (QA) dataset that, by construction, \emph{requires} proper multihop reasoning? To this end, we introduce a bottom-up approach that systematically selects composable pairs of single-hop questions that are connected, i.e., where one reasoning step critically relies on information from another. This bottom-up methodology lets us explore a vast space of questions and add stringent filters as well as other mechanisms targeting connected reasoning. It provides fine-grained control over the construction process and the properties of the resulting $k$-hop questions. We use this methodology to create MuSiQue-Ans, a new multihop QA dataset with 25K 2-4 hop questions. Relative to existing datasets, MuSiQue-Ans is more difficult overall (3x increase in human-machine gap), and harder to cheat via disconnected reasoning (e.g., a single-hop model has a 30 point drop in F1). We further add unanswerable contrast questions to produce a more stringent dataset, MuSiQue-Full. We hope our datasets will help the NLP community develop models that perform genuine multihop reasoning.

MuSiQue: Multi-hop Questions via Single-hop Question Composition

The bi-encoder design of dense passage retriever (DPR) is a key factor to its success in open-domain question answering (QA), yet it is unclear how DPR's question encoder and passage encoder individually contributes to overall performance, which we refer to as the \textit{encoder attribution} problem.
The problem is important as it helps us identify the factors that affect individual encoders to further improve overall performance.
In this paper, we formulate our analysis under a probabilistic framework called {\it encoder marginalization}, where 
we quantify the contribution of a single encoder by marginalizing other variables.
First, we find that the passage encoder contributes more than the question encoder to in-domain retrieval accuracy.
Second, we demonstrate how to find the affecting factors for each encoder, where we train DPR with different amounts of data and use encoder marginalization to analyze the results.
We find that positive passage overlap and corpus coverage of training data have big impacts on the passage encoder, while the question encoder is mainly affected by training sample complexity under this setting.
Based on this framework, we can devise data-efficient training regimes:\ for example, we manage to train a passage encoder on SQuAD using 60\% less training data without loss of accuracy.

An Encoder Attribution Analysis for Dense Passage Retriever in Open Domain Question Answering

The widespread use of Artificial Intelligence (AI) in consequential domains, such as
healthcare and parole decision making systems, has drawn intense scrutiny on the
fairness of these methods. However, ensuring fairness is often insufficient as the
rationale for a contentious decision needs to be audited, understood, and defended.
We propose that the attention mechanism can be used to ensure fair outcomes while simultaneously providing feature attributions to account for how a decision was made. Toward this goal, we design an attention-based model that can be leveraged as an attribution framework. It can identify features responsible for both performance and fairness of the model through attention interventions and attention weight manipulation. Using this attribution framework, we then design a post-processing bias mitigation strategy and compare it with a suite of baselines. We demonstrate the versatility of our approach by conducting experiments on two distinct data types, tabular and textual.

Attributing Fair Decisions with Attention Interventions

In an effort to guarantee that machine learning model outputs conform with human moral values, recent work has begun exploring the possibility of explicitly training models to learn the difference between right and wrong. This is typically done in a bottom-up fashion, by exposing the model to different scenarios, annotated with human moral judgements. One question, however, is whether the trained models actually learn any consistent, higher-level ethical principles from these datasets – and if so, what? Here, we probe the Allen AI Delphi model with a set of standardized morality questionnaires, and find that, despite some inconsistencies, Delphi tends to mirror the moral principles associated with the demographic groups involved in the annotation process. We question whether this is desirable and discuss how we might move forward with this knowledge.

Does Moral Code Have a Moral Code? Probing Delphi’s Moral Philosophy

Manuel Montes-y-Gómez: NLP in Mexican Spanish: One of many stories

As large language models have started to be used to mediate, modify, and generate messages for users, how to enable this process to be more user-centric and trustworthy becomes increasingly important. Our recent work takes a closer look at the social aspect of language via two studies towards building more socially responsible language technologies. The first study analyzes users’ perceived trust in AI-mediated writing and how such perceptions change in accordance with social context, and the second one examines positive reframing by neutralizing a negative point of view and generating a more positive perspective without contradicting its original meaning.

Building Positive and Trustworthy Language Technologies

Artificial intelligence has emerged as a promising technique for helping with healthcare from various aspects such as disease diagnostics and prognostics, personalized treatment and management, therapeutics research and development, disease surveillance and public health policy making, etc. Because AI relies more on the data rather than human judgement, they hold the promise of suffering less from subjective bias and thus reduce healthcare disparities. At the same time, due to the various issue of health data and fragmented nature of our healthcare system, AI can introduce implicit additional algorithmic bias and worsen healthcare disparities. This makes AI a double-edged sword. In this talk, I will introduce existing research and examples from both aspects, and discuss important considerations on designing fair healthcare AI algorithms.

Towards Building Trustworthy Machine Learning Models in Medicine: Evaluation vs Explanation

I will talk about our research on extracting issues and evaluations for
the administrative aspects of local governments from the analysis of
citizens' voices on social media. Specifically, I will explain our
method of collecting and analyzing citizens' voices from Twitter on a
city-by-city basis and analyze how the actions of local governments in
Japan changed citizens' voices. As an applied case study, I will also
present our work on understanding the mood of citizens' interest for the
region and our analysis of citizens' opinions on childcare and
restaurant take-out services in the COVID-19 disaster.

Explaining Toxic Text via Knowledge Enhanced Text Generation

Downloads

Next from NAACL 2022

Teaching BERT to Wait: Balancing Accuracy and Latency for Streaming Disfluency Detection

Similar lecture

Style Pooling: Automatic Text Style Obfuscation for Improved Classification Fairness

Explaining Toxic Text via Knowledge Enhanced Text Generation

.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 NAACL 2022

Teaching BERT to Wait: Balancing Accuracy and Latency for Streaming Disfluency Detection

Similar lecture

Style Pooling: Automatic Text Style Obfuscation for Improved Classification Fairness

Downloads