Thailand

Pretrained multilingual models (PMMs) enable zero-shot learning via cross-lingual transfer, performing best for languages seen during pretraining. While methods exist to improve performance for unseen languages, they have almost exclusively been evaluated using amounts of raw text only available for a small fraction of the world&#39;s languages. In this paper, we evaluate the performance of existing methods to adapt PMMs to new languages using a resource available for close to 1600 languages: the New Testament. This is challenging for two reasons: (1) the small corpus size, and (2) the narrow domain. While performance drops for all approaches, we surprisingly still see gains of up to 17.69% accuracy for part-of-speech tagging and 6.29 F1 for NER on average over all languages as compared to XLM-R. Another unexpected finding is that continued pretraining, the simplest approach, performs best. Finally, we perform a case study to disentangle the effects of domain and size and to shed light on the influence of the finetuning source language.

ACL 2021

How to Adapt Your Pretrained Multilingual Model to 1600 Languages

language adaptation

unseen

cross-lingual

zero-shot

multilingual

**Welcome to ACL-IJCNLP 2021!**

The great event is jointly organized by the Association for Computational Linguistics (ACL) and Asian Federation of Natural Language Processing (AFNLP). 

This Joint Conference of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (ACL-IJCNLP 2021) was to be held in Bangkok, Thailand, during August 1-6, 2021 and is now virtual on the Underline platform.

As in previous years, the program of the conference includes a poster session, tutorials, workshops and demonstrations in addition to the main conference.


We were able to keep the registration fees similar to those charged for the virtual ACL 2020. The one fee allows attendance at the main conference and any/all tutorials and workshops. These fees would be $125 Regular Early and $175 Regular Late; $50 Student Early and $75 Student Late. Early registration closes at midnight July 11, 2021 (Eastern Daylight Time).

**Reminder:** It is ACL’s policy that at least one author of each accepted paper (including ACL Finding papers) must register for the conference.

**Reminder2:** Underline site will open closer to the event. If you already registered you will receive access detail

Registration is now open

Pretrained multilingual models (PMMs) enable zero-shot learning via cross-lingual transfer, performing best for languages seen during pretraining. While methods exist to improve performance for unseen languages, they have almost exclusively been evaluated using amounts of raw text only available for a small fraction of the world's languages. In this paper, we evaluate the performance of existing methods to adapt PMMs to new languages using a resource available for close to 1600 languages: the New Testament. This is challenging for two reasons: (1) the small corpus size, and (2) the narrow domain. While performance drops for all approaches, we surprisingly still see gains of up to 17.69% accuracy for part-of-speech tagging and 6.29 F1 for NER on average over all languages as compared to XLM-R. Another unexpected finding is that continued pretraining, the simplest approach, performs best. Finally, we perform a case study to disentangle the effects of domain and size and to shed light on the influence of the finetuning source language.

poster

While pre-training techniques are working very well in natural language processing, how to pre-train a decoder and effectively use it for neural machine translation (NMT) still remains a tricky issue. The main reason is that the cross-attention module between the encoder and decoder cannot be pre-trained, and the combined encoder-decoder model cannot work well in the fine-tuning stage because the inputs of the decoder cross-attention come from unknown encoder outputs. In this paper, we propose a better pre-training method for NMT by defining a semantic interface (SemFace) between the pre-trained encoder and the pre-trained decoder. Specifically, we propose two types of semantic interfaces, including CL-SemFace which regards cross-lingual embeddings as an interface, and VQ-SemFace which employs vector quantized embeddings to constrain the encoder outputs and decoder inputs in the same language-independent space. We conduct massive experiments on six supervised translation pairs and three unsupervised pairs. Experimental results demonstrate that our proposed SemFace can effectively connect the pre-trained encoder and decoder, and achieves significant improvement by 3.7 and 1.5 BLEU points on the two tasks respectively compared with previous pre-training-based NMT models.

SemFace: Pre-training Encoder and Decoder with a Semantic Interface for Neural Machine Translation

The discrepancy between maximum likelihood estimation (MLE) and task measures such as BLEU score has been studied before for autoregressive neural machine translation (NMT) and resulted in alternative training algorithms (Ranzato et al., 2016; Norouzi et al., 2016; Shen et al., 2016; Wu et al., 2018). However, MLE training remains the de facto approach for autoregressive NMT because of its computational efficiency and stability. 
Despite this mismatch between the training objective and task measure, we notice that the samples drawn from an MLE-based trained NMT support the desired distribution -- there are samples with much higher BLEU score comparing to the beam decoding output. To benefit from this observation, we train an energy-based model to mimic the behavior of the task measure (i.e., the energy-based model assigns lower energy to samples with higher BLEU score), which is resulted in a re-ranking algorithm based on the samples drawn from NMT: energy-based re-ranking (EBR). We use both marginal energy models (over target sentence) and joint energy models (over both source and target sentences). Our EBR with the joint energy model consistently improves the performance of the Transformer-based NMT: +3.7 BLEU points on IWSLT'14 German-English, +3.37 BELU points on Sinhala-English, +1.4 BLEU points on WMT'16 English-German tasks.

Energy-Based Reranking: Improving Neural Machine Translation Using Energy-Based Models

In recent years, we have seen a colossal effort in pre-training multilingual text encoders using large-scale corpora in many languages to facilitate cross-lingual transfer learning. However, due to typological differences across languages, the cross-lingual transfer is challenging. Nevertheless, language syntax, e.g., syntactic dependencies, can bridge the typological gap. Previous works have shown that pre-trained multilingual encoders, such as mBERT \cite{devlin-etal-2019-bert}, capture language syntax, helping cross-lingual transfer. This work shows that explicitly providing language syntax and training mBERT using an auxiliary objective to encode the universal dependency tree structure helps cross-lingual transfer. We perform rigorous experiments on four NLP tasks, including text classification, question answering, named entity recognition, and task-oriented semantic parsing. The experiment results show that syntax-augmented mBERT improves cross-lingual transfer on popular benchmarks, such as PAWS-X and MLQA, by 1.4 and 1.6 points on average across all languages. In the \emph{generalized} transfer setting, the performance boosted significantly, with 3.9 and 3.1 points on average in PAWS-X and MLQA.

Syntax-augmented Multilingual BERT for Cross-lingual Transfer

Scientific knowledge is evolving at an unprecedented rate of speed, with new concepts constantly being introduced from millions of academic articles published every month. In this paper, we introduce a self-supervised end-to-end system, SciConceptMiner, for the automatic capture of emerging scientific concepts from both independent knowledge sources (semi-structured data) and academic publications (unstructured documents). First, we adopt a BERT-based sequence labeling model to predict candidate concept phrases with self-supervision data. Then, we incorporate rich Web content for synonym detection and concept selection via a web search API. This two-stage approach achieves highly accurate
(94.7%) concept identification with more than 740K scientific concepts. These concepts are deployed in the Microsoft Academic production system and are the backbone for its semantic search capability.

SciConceptMiner: A system for large-scale scientific concept discovery

This paper announces version 1.0 of the Classical Language Toolkit (CLTK), an NLP framework for pre-modern languages. The vast majority of NLP, its algorithms and software, is created with assumptions particular to living languages, thus neglecting certain important characteristics of largely non-spoken historical languages. Further, scholars of pre-modern languages often have different goals than those of living-language researchers. To fill this void, the CLTK adapts ideas from several leading NLP frameworks to create a novel software architecture that satisfies the unique needs of pre-modern languages and their researchers. Its centerpiece is a modular processing pipeline that balances the competing demands of algorithmic diversity with pre-configured defaults. The CLTK currently provides pipelines, including models, for almost 20 languages.

The {C}lassical {L}anguage {T}oolkit: {A}n {NLP} Framework for Pre-Modern Languages

Novel neural architectures, training strategies, and the availability of large-scale corpora haven been the driving force behind recent progress in abstractive text summarization. However, due to the black-box nature of neural models, uninformative evaluation metrics, and scarce tooling for model and data analysis, the true performance and failure modes of summarization models remain largely unknown. To address this limitation, we introduce SummVis, an open-source tool for visualizing abstractive summaries that enables fine-grained analysis of the models, data, and evaluation metrics associated with text summarization. Through its lexical and semantic visualizations, the tools offers an easy entry point for in-depth model prediction exploration across important dimensions such as factual consistency or abstractiveness. The tool together with several pre-computed model outputs is available at https://summvis.com.

SummVis: Interactive Visual Analysis of Models

Why do large pre-trained transformer-based models perform so well across a wide variety of NLP tasks? Recent research suggests the key may lie in multi-headed attention mechanism's ability to learn and represent linguistic information. Understanding how these models represent both syntactic and semantic knowledge is vital to investigate why they succeed and fail, what they have learned, and how they can improve. We present Dodrio, an open-source interactive visualization tool to help NLP researchers and practitioners analyze attention mechanisms in transformer-based models with linguistic knowledge. Dodrio tightly integrates an overview that summarizes the roles of different attention heads, and detailed views that help users compare attention weights with the syntactic structure and semantic information in the input text. To facilitate the visual comparison of attention weights and linguistic knowledge, Dodrio applies different graph visualization techniques to represent attention weights scalable to longer input text. Case studies highlight how Dodrio provides insights into understanding the attention mechanism in transformer-based models. Dodrio is available at https://poloclub.github.io/dodrio/.

Dodrio: Exploring Transformer Models with Interactive Visualization

Online conversations, just like offline ones, are susceptible to influence by bad actors. These users have the capacity to derail neutral or even prosocial discussions through adverse behavior. Moderators and users alike would benefit from more resilient online conversations, i.e., those that can survive the influx of antisocial behavior to which many conversations fall victim. In this paper, we examine the notion of conversational resilience: what makes a conversation more or less capable of withstanding an antisocial interruption? Understanding the factors that contribute to resilience in online conversations can help identify what behaviors we should encourage and better inform moderation practices. Working with 10.5M comments from five mainstream subreddits (over 2.9M subscribers each), we compiled more than 5.8M comment threads (i.e., conversations). Using 20K relevant conversations, we investigate the effects of different norm violations on the outcome of a conversation while accounting for community characteristics and the context in which the violation occurs. We find that norm violations that occur at the start of a conversation do not necessarily stop the flow of subsequent comments. Additionally, highly-toxic violations are more likely to occur in comment threads containing other toxic comments, both before and after the violation itself. Finally, more severe violations may not necessarily lead to higher rates of subsequent violations, while the comment author’s history of toxic behavior might.

Conversation-Level Resilience to Bad Actors in Reddit Communities

Amidst rising mental health needs in society, virtual agents are increasingly deployed in counselling. In order to give pertinent advice, counsellors must first gain an understanding of the issues at hand by eliciting sharing from the counsellee. It is thus important for the counsellor chatbot to encourage the user to open up and talk. One way to sustain the conversation flow is to acknowledge the counsellee's key points by restating them, or probing them further with questions. This paper applies models from two closely related NLP tasks --- summarization and question generation --- to restatement and question generation in the counselling context. We conducted experiments on a manually annotated dataset of Cantonese post-reply pairs on topics related to loneliness, academic anxiety and test anxiety. We obtained the best performance in both restatement and question generation by fine-tuning BertSum, a state-of-the-art summarization model, with the in-domain manual dataset augmented with a large-scale, automatically mined open-domain dataset.

Restatement and Question Generation for Counsellor Chatbot

The debate around climate change (CC)—its extent, its causes, and the necessary re- sponses—is intense and of global importance. Yet, in the natural language processing (NLP) community, this domain has so far received little attention. In contrast, it is of enor- mous prominence in various social science disciplines, and some of that work follows the ”text-as-data” paradigm, seeking to em- ploy quantitative methods for analyzing large amounts of CC-related text. Other research is qualitative in nature and studies details, nu- ances, actors, and motivations within CC dis- courses. Coming from both NLP and Po- litical Science, and reviewing key works in both disciplines, we discuss how social sci- ence approaches to CC debates can inform advances in text-mining/NLP, and how, in re- turn, NLP can support policy-makers and ac- tivists in making sense of large-scale and com- plex CC discourses across multiple genres, channels, topics, and communities. This is paramount for their ability to make rapid and meaningful impact on the discourse, and for shaping the necessary policy change.

How to Adapt Your Pretrained Multilingual Model to 1600 Languages

Downloads

Next from ACL 2021

SemFace: Pre-training Encoder and Decoder with a Semantic Interface for Neural Machine Translation

Similar lecture

Common Sense Beyond English: Evaluating and Improving Multilingual Language Models for Commonsense Reasoning

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES