Dominican Republic

Previous work has shown that human evaluations in NLP are notoriously under-powered. Here, we argue that there are two common factors which make this problem even worse: NLP studies usually (a) treat ordinal data as interval data and (b) operate under high variance settings while the differences they are hoping to detect are often subtle. We demonstrate through simulation that ordinal mixed effects models are better able to detect small differences between models, especially in high variance settings common in evaluations of generated texts. We release tools for researchers to conduct their own power analysis and test their assumptions. We also make recommendations for improving statistical power.

EMNLP 2021

What happens if you treat ordinal ratings as interval data? Human evaluations in NLP are even more under-powered than you think

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

poster

Style is an integral part of natural language. However, evaluation methods for style measures are rare, often task-specific and usually do not control for content. We propose the modular, fine-grained and content-controlled similarity-based STyle EvaLuation framework (STEL) to test the performance of any model that can compare two sentences on style. We illustrate STEL with two general dimensions of style (formal/informal and simple/complex) as well as two specific characteristics of style (contrac'tion and numb3r substitution). We find that BERT-based methods outperform simple versions of commonly used style measures like 3-grams, punctuation frequency and LIWC-based approaches. We invite the addition of further tasks and task instances to STEL and hope to facilitate the improvement of style-sensitive measures.

Does It Capture STEL? A Modular, Similarity-based Linguistic Style Evaluation Framework

Evaluation metrics are a key ingredient for progress of text generation systems. In recent years, several BERT-based evaluation metrics have been proposed (including BERTScore, MoverScore, BLEURT, etc.) which correlate much better with human assessment of text generation quality than BLEU or ROUGE, invented two decades ago. However, little is known what these metrics, which are based on black-box language model representations, actually capture (it is typically assumed they model semantic similarity). In this work, we use a simple regression based global explainability technique to disentangle metric scores along linguistic factors, including semantics, syntax, morphology, and lexical overlap. We show that the different metrics capture all aspects to some degree, but that they are all substantially sensitive to lexical overlap, just like BLEU and ROUGE. This exposes limitations of these novelly proposed metrics, which we also highlight in an adversarial test scenario.

Global Explainability of BERT-Based Evaluation Metrics by Disentangling along Linguistic Factors

Previous works introduce important challenges for improving the generalization performance, i.e., the model trained on a cross-domain text-to-SQL dataset (e.g., Spider) does not generalize to a new external database. In this work, we demonstrate that the generalization performance could be poor even when both the NL questions and SQL queries follow a similar distribution to the training set. Specifically, we constructed Spider-DK, a challenging variant of the Spider development set, with the focus of evaluating the model understanding of domain knowledge. We show that the state-of-the-art models consistently fail in cases when specific domain knowledge is required for prediction, even if the domain knowledge is moderately mentioned in the training data, and the models accurately predict the corresponding training samples. Such discrepancy suggests that the models do not properly learn the domain knowledge in order to fit the training set, thus improving the model capability to capture the domain knowledge is an important direction towards achieving the cross-domain generalization for text-to-SQL applications.

Exploring Underexplored Limitations of Cross-Domain Text-to-SQL Generalization

Standard train-dev-test splits used to benchmark multiple models against each other are ubiquitously used in Natural Language Processing (NLP). In this setup, the train data is used for training the model, the development set for evaluating different versions of the proposed model(s) during development, and the test set to confirm the answers to the main research question(s). However, the introduction of neural networks in NLP has led to a different use of these standard splits; the development set is now often used for model selection during the training procedure. Because of this, comparing multiple versions of the same model during development leads to overestimation on the development data. As an effect, people have started to compare an increasing amount of models on the test data, leading to faster overfitting and ``expiration'' of our test sets. We propose to use a tune-set when developing neural network methods, which can be used for model picking so that comparing the different versions of a new model can safely be done on the development data.

We Need to Talk About train-dev-test Splits

Recent development in NLP shows a strong trend towards refining pre-trained models with a domain-specific dataset. This is especially the case for response generation where emotion plays an important role. However, existing empathetic datasets remain small, delaying research efforts in this area, for example, the development of emotion-aware chatbots. One main technical challenge has been the cost of manually annotating dialogues with the right emotion labels. In our presentation, we describe a semi-automated process followed in curating a large-scale silver standard dialogue dataset consisting of 1M dialogues annotated with fine-grained emotions, and empathetic response intents. We describe our novel data curation pipeline that starts with a small seed of manually annotated data and eventually scales it to a satisfactory size. We compare its quality against a state-of-the-art gold dataset using offline experiments and visual validation methods. The resultant procedure can be used to create similar datasets in the same domain as well as in other domains.

A Large-Scale Dataset for Empathetic Response Generation

**Anthology paper link:**

https://aclanthology.org/2021.emnlp-main.817/

**Abstract:**



Automatic Text Evaluation through the Lens of Wasserstein Barycenters

The design of widespread vision-and-language datasets and pre-trained encoders directly adopts, or draws inspiration from, the concepts and images of ImageNet. While one can hardly overestimate how much this benchmark contributed to progress in computer vision, it is mostly derived from lexical databases and image queries in English, resulting in source material with a North American or Western European bias. Therefore, we devise a new protocol to construct an ImageNet-style hierarchy representative of more languages and cultures. In particular, we let the selection of both concepts and images be entirely driven by native speakers, rather than scraping them automatically. Specifically, we focus on a typologically diverse set of languages, namely, Indonesian, Mandarin Chinese, Swahili, Tamil, and Turkish. On top of the concepts and images obtained through this new protocol, we create a multilingual dataset for {M}ulticultur{a}l {R}easoning over {V}ision and {L}anguage (MaRVL) by eliciting statements from native speaker annotators about pairs of images. The task consists of discriminating whether each grounded statement is true or false. We establish a series of baselines using state-of-the-art models and find that their cross-lingual transfer performance lags dramatically behind supervised performance in English. These results invite us to reassess the robustness and accuracy of current state-of-the-art models beyond a narrow domain, but also open up new exciting challenges for the development of truly multilingual and multicultural systems.

Visually Grounded Reasoning across Languages and Cultures

Dialogue summarization has drawn much attention recently. Especially in the customer service domain, agents could use dialogue summaries to help boost their works by quickly knowing customer’s issues and service progress. These applications require summaries to contain the perspective of a single speaker and have a clear topic flow structure, while neither are available in existing datasets. Therefore, in this paper, we introduce a novel Chinese dataset for Customer Service Dialogue Summarization (CSDS). CSDS improves the abstractive summaries in two aspects: (1) In addition to the overall summary for the whole dialogue, role-oriented summaries are also provided to acquire different speakers’ viewpoints. (2) All the summaries sum up each topic separately, thus containing the topic-level structure of the dialogue. We define tasks in CSDS as generating the overall summary and different role-oriented summaries for a given dialogue. Next, we compare various summarization methods on CSDS, and experiment results show that existing methods are prone to generate redundant and incoherent summaries. Besides, the performance becomes much worse when analyzing the performance on role-oriented summaries and topic structures. We hope that this study could benchmark Chinese dialogue summarization and benefit further studies.

CSDS: A Fine-Grained Chinese Dataset for Customer Service Dialogue Summarization

The Winograd Schema (WS) has been proposed as a test for measuring commonsense capabilities of models. Recently, pre-trained language model-based approaches have boosted performance on some WS benchmarks but the source of improvement is still not clear. This paper suggests that the apparent progress on WS may not necessarily reflect progress in commonsense reasoning. To support this claim, we first show that the current evaluation method of WS is sub-optimal and propose a modification that uses twin sentences for evaluation. We also propose two new baselines that indicate the existence of artifacts in WS benchmarks. We then develop a method for evaluating WS-like sentences in a zero-shot setting to account for the commonsense reasoning abilities acquired during the pretraining and observe that popular language models perform randomly in this setting when using our more strict evaluation. We conclude that the observed progress is mostly due to the use of supervision in training WS models, which is not likely to successfully support all the required commonsense reasoning skills and knowledge. The code and evaluation are available at: https://github.com/yanaiela/winograd_square_one

Back to Square One: Artifact Detection, Training and Commonsense Disentanglement in the Winograd Schema

“Michael Jordan published a new paper on machine learning”. Did they mean MJ the scientist or MJ the basketball player? The answer is obvious to humans, but for entity linking systems it remains complicated: when in doubt, they tend to output more common entities even if it means losing relevance. We define this phenomenon as entity overshadowing and explore it using ShadowLink, our new dataset made to bring overshadowed entities to light. The results are intriguing: modern state-of-the-art methods get lost and confused, while a good old model from 2011 achieves the best scores. Our presentation will discuss how and why this happens.

What happens if you treat ordinal ratings as interval data? Human evaluations in NLP are even more under-powered than you think

Downloads

Next from EMNLP 2021

Does It Capture STEL? A Modular, Similarity-based Linguistic Style Evaluation Framework

Similar lecture

XTREME-R: Towards More Challenging and Nuanced Multilingual Evaluation