Indonesia

Pre-trained language models trained on large-scale data have learned serious levels of social biases. 
Consequently, various methods have been proposed to debias pre-trained models.
Debiasing methods need to mitigate only discriminatory bias information from the pre-trained models, while retaining information that is useful for the downstream tasks.
In previous research, whether useful information is retained has been confirmed by the performance of downstream tasks in debiased pre-trained models.
On the other hand, it is not clear whether these benchmarks consist of data pertaining to social biases and are appropriate for investigating the impact of debiasing.
For example in gender-related social biases, data containing female words, male words, and stereotypical words are considered to be the most affected by debiasing.
If there is not much data containing these words in a benchmark dataset for a target task, there is the possibility of erroneously evaluating the effects of debiasing.
In this study, we compare the impact of debiasing on performance across multiple downstream tasks using a wide-range of benchmark datasets that containing female, male, and stereotypical words.
Experiments show that the effects of debiasing are consistently \emph{underestimated} across all tasks.
Moreover, the effects of debiasing could be reliably evaluated by separately considering instances containing female, male, and stereotypical words than all of the instances in a benchmark dataset.

IJCNLP-AACL 2023

The Impact of Debiasing on the Performance of Language Models in Downstream Tasks is Underestimated

technical paper

### Welcome to IJCNLP-AACL 2023!
I am honored and happy to welcome you to IJCNLP-AACL 2023! The conference hosts both the 13th
International Joint Conference on Natural Language Processing (IJCNLP) and the 3rd Conference of
the Asia-Pacific Chapter of the Association for Computational Linguistics (AACL). 
The joint conferences of IJCNLP and AACL have been organized by alternating leadership in the
Asia-Pacific region, in odd years by Asian Federation of Natural Language Processing (AFNLP) and
in even years by AACL. The conference is organized by AFNLP this year. 

Considering the current situation that we are not yet fully over the coronavirus pandemic, the
conference is offered in a hybrid mode, off-line in Nusa Dua, Bali, Indonesia, and on-line with the help
of Underline, during November 1-4, 2023. For those attending off-line, I hope that you will find time to
explore the exciting local area, which remains quite warm and pleasant at this time of the year. For
those attending on-line, I hope that the attendance is not much hampered, sometimes even enhanced
and comfortable, though the on-line arrangements have been made without ample time. 

Message from Jong C. Park, Korea Advanced Institute of Science and Technology (KAIST) 
IJCNLP-AACL 2023 General Chair 
###### [Conference Handbook](https://drive.google.com/file/d/1S2JSXa105xJxIkNBfhYmGH20Om6wARBx/view?usp=sharing)

In order to acces this event page, you need to register [here](http://www.ijcnlp-aacl2023.org/)

The 13th International Joint Conference on Natural Language Processing and the 3rd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics

Two distinct approaches have been proposed for relational triple extraction - pipeline and joint. Joint models, which capture interactions across triples, are the more recent development, and have been shown to outperform pipeline models for sentence-level extraction tasks. Document-level extraction is a more challenging setting where interactions across triples can be long-range, and individual triples can also span across sentences. Joint models have not been applied for document-level tasks so far. In this paper, we benchmark state-of-the-art pipeline and joint extraction models on sentence-level as well as document-level datasets. Our experiments show that while joint models outperform pipeline models significantly for sentence-level extraction, their performance drops sharply below that of pipeline models for the document-level dataset.

Do the Benefits of Joint Models for Relation Extraction Extend to Document-level Tasks?

Relation Classification (RC)---the task of identifying the relation between a pair of target entities---is a fundamental sub-task of information extraction. RC models built on top of entity information are prevalent, with different variants using entity information, especially entity type information, differently. However, RC models are often benchmarked on datasets that human annotators provide near-perfect entity information, and, state-of-the-art results are reported using gold entity type information. We believe there is a need to understand how the effectiveness of RC models is affected by the correctness of entity type information because in practice this information is provided by imperfect entity recognition models. Our results on six datasets across four domains show that although using gold entity type improves the effectiveness of RC models, incorrect entity types may cause large effectiveness drops on some (but not all) datasets. We propose using Pointwise Mutual Information (PMI) to identify datasets on which RC models may be negatively impacted by incorrect entity type information.

Rethinking the Role of Entity Type in Relation Classification

Despite decades of research
on authorship attribution (AA) 
and authorship verification (AV),
inconsistent dataset splits/filtering 
and mismatched evaluation methods 
make it difficult to assess the state of the art.
In this paper, 
we present a survey of the fields, 
resolve points of confusion,
introduce Valla that 
standardizes and benchmarks AA/AV datasets and metrics, 
provide a large-scale empirical evaluation,
and provide apples-to-apples comparisons 
between existing methods.
We evaluate eight promising methods on fifteen datasets
(including distribution shifted challenge sets) 
and introduce a new dataset 
based on texts archived by Project Gutenberg. 
Surprisingly, we find that a traditional \ngram-based model 
performs best on 5 (of 7) AA tasks,
achieving an average macro-accuracy of 76.50%
(compared to 66.71% for a BERT-based model).
However, on the two AA datasets with the greatest number of words per author, 
as well as on the AV datasets, 
BERT-based models perform best. 
While AV methods are easily applied to AA,
they are seldom included as baselines in AA papers. 
We show that through the application of hard-negative mining, 
AV methods are competitive alternatives to AA methods.
Valla and all experiment code can be found at
https://github.com/JacobTyo/Valla

Valla: Standardizing and Benchmarking Authorship Attribution and Verification Through Empirical Evaluation and Comparative Analysis

Media bias detection requires comprehensive integration of information derived from multiple news sources. Sentence-level political bias detection in news is no exception, and has proven to be a challenging task that requires an understanding of bias in consideration of the context. Inspired by the fact that humans exhibit varying degrees of writing styles, resulting in a diverse range of statements with different local and global contexts, previous work in media bias detection has proposed augmentation techniques to exploit this fact. Despite their success, we observe that these techniques introduce noise by over-generalizing bias context boundaries, which hinders performance. To alleviate this issue, we propose techniques to more carefully search for context using a bias-sensitive, target-aware approach for data augmentation. Comprehensive experiments on the well-known BASIL dataset show that when combined with pre-trained models such as BERT, our augmentation techniques lead to state-of-the-art results. Our approach outperforms significantly, obtaining an F1-score of 58.15 over state-of-the-art bias detection task.

Target-Aware Contextual Political Bias Detection in News

Democratizing access to natural language processing (NLP) technology is crucial, especially for underrepresented and extremely low-resource languages. Previous research has focused on developing labeled and unlabeled corpora for these languages through online scraping and document translation. While these methods have proven effective and cost-efficient, we have identified limitations in the resulting corpora, including a lack of lexical diversity and cultural relevance to local communities. To address this gap, we conduct a case study on Indonesian local languages. We compare the effectiveness of online scraping, human translation, and paragraph writing by native speakers in constructing datasets. Our findings demonstrate that datasets generated through paragraph writing by native speakers exhibit superior quality in terms of lexical diversity and cultural content. In addition, we present the \datasetname{} benchmark, encompassing 12 underrepresented and extremely low-resource languages spoken by millions of individuals in Indonesia. Our empirical experiment results using existing multilingual large language models emphasize the need to extend these models to more underrepresented languages.

NusaWrites: Constructing High-Quality Corpora for Underrepresented and Extremely Low-Resource Languages

Persuasive meme identification is a crucial task in automatically categorizing memes based on their persuasive nature. Memes, being highly influential in online communication, have the ability to shape individuals' attitudes, behaviors, and beliefs, both positively and negatively. They can be utilized to promote positive actions, challenge social norms, and raise awareness, but they can also perpetuate harmful ideologies, spread misinformation, stereotype, and manipulate emotions. In this paper, we are addressing this challenge by empirically investigating three novel tasks: (i) Task 1: Persuasive meme detection (ii) Task 2: Identification of the effectiveness of persuasive memes, and (iii) Task 3: Identification of persuasion techniques used in persuasive memes. To this end, we make the very first attempt to release a high-quality, large-scale dataset, Persuasive_meme since there is no publicly available such dataset for Hindi-English code-mixed (Hinglish) domain. We further developed several baseline unimodal and multimodal models for these tasks. Empirical evaluations, including both qualitative and quantitative analysis, on the Persuasive_meme dataset highlight the significance of multimodality in addressing these tasks effectively. Additionally, also we discuss the limitations of the current models and emphasize the need for further research to overcome these challenges.

The Persuasive Memescape: Understanding Effectiveness and Societal Implications of Internet Memes

Conventional text style transfer approaches focus on sentence-level style transfer without considering contextual information, and the style is described with attributes (e.g., formality). When applying style transfer in conversations such as task-oriented dialogues, existing approaches suffer from these limitations as context can play an important role and the style attributes are often difficult to define in conversations. In this paper, we introduce conversation style transfer as a few-shot learning problem, where the model learns to perform style transfer by observing only a few example dialogues in the target style. We propose a novel in-context learning approach to solve the task with style-free dialogues as a pivot. Human evaluation shows that by incorporating multi-turn context, the model is able to match the target style while having better appropriateness and semantic correctness compared to utterance/sentence-level style transfer. Additionally, we show that conversation style transfer can also benefit downstream tasks. For example, in multi-domain intent classification tasks, the F1 scores improve after transferring the style of training data to match the style of the test data.

Conversation Style Transfer using Few-Shot Learning

Previous studies exploring the human-like capabilities of machine-learning models have primarily focused on pure language models. 
Limited attention has been given to investigating whether models exhibit human-like behavior when performing tasks that require the integration of visual and language information.
In this study, we investigate the impact of tags of semantic, phonological, and bilingual features on the visual question answering task performance of an unsupervised model. 
Our findings reveal its similarities with the influence of distractors in the picture-naming task (known as the picture-word-interference paradigm) observed in human experiments: 

1) Semantically-related tags have a more negative effect on task performance compared to unrelated tags, indicating a more robust competition between visual and tag information which are semantically closer to each other when generating an answer. 
2) Even presenting a partial section (wordpiece) of the originally detected tag significantly improves task performance, with the portion that plays a lesser role in determining the overall meaning of the original tag leading to a more pronounced improvement. 
3) Tags in two languages that refer to the same meaning exhibit a symmetrical-like effect on performance in balanced bilingual models.

Datasets and code of this project are released at https://github.com/NLPbelllabs/PWI.

Human-Like Distractor Response in Vision-Language Model

Most work on verbalising Knowledge-Graphs (KG) has focused on high-resource languages such as English, Russian, Czech or Arabic. In this paper, we focus on KG-to-Text generation where the output text is in Breton, Irish or Welsh. To overcome the small size of the parallel training data, we combine the strengths of a multilingual language model with denoising fine-tuning on monolingual data and Soft Prompt fine-tuning on a small quantity of KG/text data. We furthermore structure the soft prompt into multiple sub-prompts designed to capture the similarities and differences between English, Knowledge graphs and the three target languages. Our experiments show that our approach outperforms strong baselines and that all sub-prompts contribute to performance.

Phylogeny-Inspired Soft Prompts For Data-to-Text Generation in Low-Resource Languages

Prior work has shown that the ordering in which concepts are shown to a commonsense generator plays an important role, affecting the quality of the generated sentence. However, it remains a challenge to determine the optimal ordering of a given set of concepts such that a natural sentence covering all the concepts could be generated from a pretrained generator. To understand the relationship between the ordering of the input concepts and the quality of the generated sentences, we conduct a systematic study considering multiple language models (LMs) and concept ordering strategies. We find that BART-large model consistently outperforms all other LMs considered in this study when fine-tuned using the ordering of concepts as they appear in CommonGen training data as measured using multiple evaluation metrics. Moreover, the larger GPT3-based large language models (LLMs) variants do not necessarily outperform much smaller LMs on this task, even when fine-tuned on task-specific training data. Interestingly, human annotators significantly reorder input concept sets when manually writing sentences covering those concepts, and this ordering provides the best sentence generations independently of the LM used for the generation, outperforming a probabilistic concept ordering baseline.

Premium content

The Impact of Debiasing on the Performance of Language Models in Downstream Tasks is Underestimated

Downloads

Next from IJCNLP-AACL 2023

Do the Benefits of Joint Models for Relation Extraction Extend to Document-level Tasks?

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES