Building a Corpus for the Underexplored Moroccan Dialect (CFMD) Through Audio Segmentations
Bibliographic record
Abstract
The advancement of artificial intelligence has deeply influenced numerous domains.One particular area that has experienced remarkable progress is natural language processing.This progress can be largely attributed to the widespread use and popularity of social media platforms.With the increasing use of social media, dialects have taken on a new importance, as the diversity of dialects has an important role to consider in the relevance of Natural Language Processing, as it allows a greater number of people to communicate using a pertinent and appropriate local context.As evidenced by the rise of Chatbots that allow people to interact with machines using their own native dialects.The significance of dialects, especially in the Arabic-speaking world, cannot be understated.Many Arabic dialects have been under-researched and not adequately addressed in natural language processing applications.Among these, the Moroccan dialect stands out, prompting researchers to focus their efforts on understanding and incorporating it into artificial intelligence technologies.To facilitate the development of Chatbots that can effectively understand and respond in Moroccan dialect, the availability of suitable datasets becomes vital.For this reason, we adopt a targeted strategy for creating datasets by exploiting the extensive resources offered by platforms such as YouTube, where audio content is highly diverse in terms of language.This involves classifying each audio according to its theme and dividing it into 30 second segments to simplify manual transcription into text.This meticulous process enabled us to accumulate and annotate a large volume of data.As a result, NLP models built on these extensive and comprehensive datasets can efficiently and accurately understand Moroccan dialect speech and text.With the aim to employ this dataset as training data for the future development of a Moroccan-dialect conversational Chatbot.The methodologies and techniques can be adapted and applied to other underexplored dialects, creating opportunities for further advancements in natural language processing in a global context.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.007 | 0.003 |
| Science and technology studies | 0.003 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.013 | 0.005 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".