Summary In this work, we survey a breadth of literature that has revealed the limitations of predominant practices for dataset collection and use in the field of machine learning. We cover studies that critically review the design and development of datasets with a focus on negative societal impacts and poor outcomes for system performance. We also cover approaches to filtering and augmenting data and modeling techniques aimed at mitigating the impact of bias in datasets. Finally, we discuss works that have studied data practices, cultures, and disciplinary norms and discuss implications for the legal, ethical, and functional challenges the field continues to face. Based on these findings, we advocate for the use of both qualitative and quantitative approaches to more carefully document and analyze datasets during the creation and usage phases.

Data and its (dis)contents: A survey of dataset development and use in machine learning research.

AI is undergoing a paradigm shift with the rise of models (e.g., BERT, DALL-E, GPT-3) that are trained on broad data at scale and are adaptable to a wide range of downstream tasks. We call these models foundation models to underscore their critically central yet incomplete character. This report provides a thorough account of the opportunities and risks of foundation models, ranging from their capabilities (e.g., language, vision, robotics, reasoning, human interaction) and technical principles(e.g., model architectures, training procedures, data, systems, security, evaluation, theory) to their applications (e.g., law, healthcare, education) and societal impact (e.g., inequity, misuse, economic and environmental impact, legal and ethical considerations). Though foundation models are based on standard deep learning and transfer learning, their scale results in new emergent capabilities,and their effectiveness across so many tasks incentivizes homogenization. Homogenization provides powerful leverage but demands caution, as the defects of the foundation model are inherited by all the adapted models downstream. Despite the impending widespread deployment of foundation models, we currently lack a clear understanding of how they work, when they fail, and what they are even capable of due to their emergent properties. To tackle these questions, we believe much of the critical research on foundation models will require deep interdisciplinary collaboration commensurate with their fundamentally sociotechnical nature.

On the Opportunities and Risks of Foundation Models.

We present Samanantar, the largest publicly available parallel corpora collection for Indic languages. The collection contains a total of 46.9 million sentence pairs between English and 11 Indic languages (from two language families). In particular, we compile 12.4 million sentence pairs from existing, publicly-available parallel corpora, and we additionally mine 34.6 million sentence pairs from the web, resulting in a 2.8X increase in publicly available sentence pairs. We mine the parallel sentences from the web by combining many corpora, tools, and methods. In particular, we use (a) web-crawled monolingual corpora, (b) document OCR for extracting sentences from scanned documents (c) multilingual representation models for aligning sentences, and (d) approximate nearest neighbor search for searching in a large collection of sentences. Human evaluation of samples from the newly mined corpora validate the high quality of the parallel sentences across 11 language pairs. Further, we extracted 82.7 million sentence pairs between all 55 Indic language pairs from the English-centric parallel corpus using English as the pivot language. We trained multilingual NMT models spanning all these languages on Samanantar and compared with other baselines and previously reported results on publicly available benchmarks. Our models outperform existing models on these benchmarks, establishing the utility of Samanantar. Our data (this https URL) and models (this https URL) will be available publicly and we hope they will help advance research in Indic NMT and multilingual NLP for Indic languages.

Samanantar: The Largest Publicly Available Parallel Corpora Collection for 11 Indic Languages.

Large language models have led to remarkable progress on many NLP tasks, and researchers are turning to ever-larger text corpora to train them. Some of the largest corpora available are made by scraping significant portions of the internet, and are frequently introduced with only minimal documentation. In this work we provide some of the first documentation for the Colossal Clean Crawled Corpus (C4; Raffel et al., 2020), a dataset created by applying a set of filters to a single snapshot of Common Crawl. We begin by investigating where the data came from, and find a significant amount of text from unexpected sources like patents and US military websites. Then we explore the content of the text itself, and find machine-generated text (e.g., from machine translation systems) and evaluation examples from other benchmark NLP datasets. To understand the impact of the filters applied to create this dataset, we evaluate the text that was removed, and show that blocklist filtering disproportionately removes text from and about minority individuals. Finally, we conclude with some recommendations for how to created and document web-scale datasets from a scrape of the internet.

Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus

Abstract We present a survey covering the state of the art in low-resource machine translation (MT) research. There are currently around 7,000 languages spoken in the world and almost all language pairs lack significant resources for training machine translation models. There has been increasing interest in research addressing the challenge of producing useful translation models when very little translated training data is available. We present a summary of this topical research field and provide a description of the techniques evaluated by researchers in several recent shared tasks in low-resource MT. 

Survey of Low-Resource Machine Translation

With the success of large-scale pre-training and multilingual modeling in Natural Language Processing (NLP), recent years have seen a proliferation of large, web-mined text datasets covering hundreds of languages. However, to date there has been no systematic analysis of the quality of these publicly available datasets, or whether the datasets actually contain content in the languages they claim to represent. In this work, we manually audit the quality of 205 language-specific corpora released with five major public datasets (CCAligned, ParaCrawl, WikiMatrix, OSCAR, mC4), and audit the correctness of language codes in a sixth (JW300). We find that lower-resource corpora have systematic issues: at least 15 corpora are completely erroneous, and a significant fraction contains less than 50% sentences of acceptable quality. Similarly, we find 82 corpora that are mislabeled or use nonstandard/ambiguous language codes. We demonstrate that these issues are easy to detect even for non-speakers of the languages in question, and supplement the human judgements with automatic analyses. Inspired by our analysis, we recommend techniques to evaluate and improve multilingual corpora and discuss the risks that come with low-quality data releases.

Monang Setyawan

Papers

Quality at a Glance: An Audit of Web-Crawled Multilingual Datasets