NAACL 2022 Affinity Workshop

Pre-recorded to enable our members to celebrate Eid Al Adha

Mailing List

Facebook

Twitter

Why pre-recorded? Our statement on D&I and conference timings:

After receiving a few requests for different arrangements from our participants, we have decided to postpone our NAACL workshop as the first days of the conference are Eid days, and it would not be ideal to organize a virtual social on the last day of the conference.

Our affinity group is about D&I and we do not want our participants to sacrifice family time to attend an event at a conference.

The talks will be available on our website, and we will make announcements regarding our NeurIPS workshop soon.

In the meantime, Eid Mubarak to those celebrating. Hopefully, future conference organizers can take this into account.

Program

We are very excited to host talks on three North African vernaculars: Tamazight, Tunisian Arabic, and Moroccan Darija! We are also honored to host speakers who are all NLP researchers and engineers based in North Africa.
Our invited speaker is Dr. Fadoua Ataa Allah, Director of Research at Morocco's Royal Institute of Amazigh Culture (IRCAM). Her talk summarizes the contributions of IRCAM for the digitalization of Standard Moroccan Tamazight, Morocco's indigenous language.
Our speaker program is as follows:

Abir Messaoudi and Hatem Haddad present their work on building a "Tunisian Dialectal Speech Recognition Model"
Abderrahmane Issam and Khalil Mrini present their work on "Goud.ma: A News Article Dataset for Summarization in Moroccan Darija"

Invited Speaker: Dr. Fadoua Ataa Allah
Director of Research at the Royal Institute of Amazigh Culture (IRCAM), Morocco

Dr. Fadoua Ataa Allah is currently a Director of Research at the Computer Science Studies, Information Systems, and Communication Center (CEISIC), in the Royal Institute of Amazigh Culture (IRCAM). She was Director of the CEISIC between 2016 and 2020, and Coordinator of the Information Processing and Communication Unit of Studies and Research, at the same center, between 2012 – 2015. Her areas of interest are Natural Language Processing, E-learning, and Information Retrieval.

She was educated at the Faculty of Science, University of Mohamed V-Agdal, in Rabat; where she was awarded the B.S. degree in Applied Mathematical Sciences in 2001, the M.S. and Ph.D. degree in Computer Sciences and Telecommunications respectively in 2003 and 2008. In 2005, she received a Fulbright grant to pursue her research at the University of Michigan-Dearborn.

Fadoua Ataa Allah has advised and co-advised more than 20 Master and Ph.D. theses. She is an editor of 6 conference proceedings, and a contributor to more than 80 papers in national and international conferences and journals. She has a Patent for a Braille Digital converter.

"Amazigh language: From the stele to the digital world"

Abstract:
Driven by the permanent evolution of technology and the constant digitization in which societies find themselves, human language technologies have undergone major progress. However, the deployment of these technologies for poorly endowed languages represents a great challenge, especially that are fully confronted with the lack of linguistic resources. Like many other languages, Amazigh is also affected by this gap.
In this talk, Fadoua Ataa Allah will present the Royal Institute of Amazigh Culture (IRCAM) strategy, in general, to preserve the Amazigh language and culture. Furthermore, she will focus on the IRCAM digitization vision that has tried to meet not only the citizen expression and communication needs but also those of institutions. This vision consists in digitalizing the Amazigh language, from a simple display on screens to the development of specialized tools. In this context, the Institute amassed many notable achievements, ensuring the Amazigh language to live in the "information society". Nevertheless, the greatest challenge is to maintain this success.

"Tunisian Dialectal Speech Recognition Model"

Abir Messaoudi and Hatem Haddad

iCompass, Tunisia

Abstract:
The Tunisian dialect (TD) is considered as an under-resourced language due to the lack of available data. Code-switching is the main characteristic of the Tunisian dialect, where Tunisians alternate between two or more than a language: between Modern Standard Arabic (MSA) and the local Tunisian dialect which is influenced mainly by French and English. The interest in building Automatic Speech Recognition (ASR) systems has increased, since most Tunisians tend to use voice performing voice search queries, sending voice messages and interacting with voice assistants on a daily basis.
Click here for the poster.

"Goud.ma: A News Article Dataset for Summarization in Moroccan Darija"

Abderrahmane Issam and Khalil Mrini

Archipel Cognitive (Morocco) and Meta (USA)

Abstract:
Moroccan Darija is a vernacular spoken by over 30 million people primarily in Morocco. Despite a high number of speakers, it remains a low-resource language. In this paper, we introduce GOUD.MA: a dataset of over 158k news articles for automatic summarization in code-switched Moroccan Darija. We analyze the dataset and find that it requires a high level of abstractive reasoning. We fine-tune the Arabic-language BERT (AraBERT), and the language models for the Moroccan (DarijaBERT), and Algerian (DziriBERT) national vernaculars for summarization on GOUD.MA. The results show that GOUD.MA is a challenging summarization benchmark dataset. We release our dataset publicly in an effort to encourage the diversity of evaluation tasks to improve language modeling in Moroccan Darija.
This paper was first accepted and published at the AfricaNLP workshop at ICLR 2022. It is available here.
Click here for the poster.

The Organizers

Alphabetically Ordered by First Name