Anass El Haddadi

Named Entity Recognition (NER) is a fundamental Natural Language Processing task, yet low-resource languages such as Tarifit Berber remain critically underexplored. This study introduces the first standardized and reproducible benchmark for Tarifit NER, addressing the lack of publicly available annotated corpora and unified evaluation protocols. A manually annotated corpus was developed using the IOB2 scheme, covering nine entity categories, and subjected to rigorous quality-assurance procedures, including data-leakage detection and out-of-vocabulary (OOV) analysis. Ten representative architectures were evaluated under an identical protocol, spanning traditional statistical models (CRF), neural sequence-labeling models (BiLSTM-CRF), multilingual Transformer models (mBERT, XLM-RoBERTa, RemBERT, LUKE), Arabic-oriented pretrained models (AraBERT, MARBERT, DarijaBERT), and the span-based GLiNER approach. Evaluation combined entity- and token-level metrics, category-wise analysis, lexical robustness across known, rare, and OOV entities, and bootstrap-based statistical reliability analysis. Results show that Transformer models consistently outperform traditional statistical and neural sequence-labeling models, with mBERT achieving the highest entity-level F1 score (95.76%), followed by MARBERT, DarijaBERT, and RemBERT. Location and Date were the most challenging categories, while GLiNER performed substantially worse under strict exact-match evaluation. These findings establish a reliable benchmark for Tarifit NER and provide a reusable foundation for future research in Amazigh and multilingual NLP.

This overview traced the evolution of data management, transitioning from traditional ETL processes to addressing contemporary challenges in Big Data, with a particular emphasis on data ingestion and schema matching. It explored the classification of data ingestion into batch, real-time, and hybrid processing, underscoring the challenges associated with data quality and heterogeneity. Central to the discussion was the role of schema mapping in data alignment, proving indispensable for linking diverse data sources. Recent advancements, notably the adoption of machine learning techniques, were significantly reshaping the landscape. The paper also addressed current challenges, including the integration of new technologies and the necessity for effective schema matching solutions, highlighting the continuously evolving nature of schema matching in the context of Big Data

Sentiment analysis, a crucial component of natural language processing (NLP), involves the classification of subjective information by extracting emotional content from textual data. This technique plays a significant role in the movie industry by analyzing public opinions about films. The present research addresses a gap in the literature by conducting a comparative analysis of various machine learning algorithms for sentiment analysis in film reviews, utilizing a dataset from Kaggle comprising 50,000 reviews. Classifiers such as Logistic Regression, Multinomial Naive Bayes, Linear Support Vector Classification (LinearSVC), and Gradient Boosting were employed to categorize the reviews into positive and negative sentiments. The emphasis was placed on specifying and comparing these classifiers in the context of film review sentiment analysis, highlighting their respective advantages and disadvantages. The dataset underwent thorough preprocessing, including data cleaning and the application of stemming techniques to enhance processing efficiency. The performance of the classifiers was rigorously evaluated using metrics such as accuracy, precision, recall, and F1-score. Among the classifiers, LinearSVC demonstrated the highest accuracy at 90.98%. This comprehensive evaluation not only identified the most effective classifier but also elucidated the contextual efficiencies of various algorithms. The findings indicate that LinearSVC excels at accurately classifying sentiments in film reviews, thereby offering new insights into public opinions on films. Furthermore, the extended comparison provides a step-by-step guide for selecting the most suitable classifier based on dataset characteristics and context, contributing valuable knowledge to the existing literature on the impact of different machine learning approaches on sentiment analysis outcomes in the movie industry.

The task of interpreting multi-variable time series data, while also forecasting outcomes accurately, is an ongoing challenge within the machine learning domain. This study presents an advanced method of utilizing Long Short-Term Memory (LSTM) recurrent neural networks in the analysis of such data, with specific attention to both target and exogenous variables. The novel approach aims to extract hidden states that are unique to individual variables, thereby capturing the distinctive dynamics inherent in multi-variable time series and allowing the elucidation of each variable's contribution to predictive outcomes. A pioneering mixture attention mechanism is introduced, which, by leveraging the aforementioned variable-specific hidden states, characterizes the generative process of the target variable. The study further enhances this methodology by formulating associated training techniques that permit concurrent learning of network parameters, variable interactions, and temporal significance with respect to the target prediction. The effectiveness of this approach is empirically validated through rigorous experimentation on three real-world datasets, including the 2022 closing prices of three major stocks-Apple (AAPL), Amazon (AMZN), and Microsoft (MSFT). The results demonstrated superior predictive performance, attributable to the successful encapsulation of the diverse dynamics of different variables. Furthermore, the study provides a comprehensive evaluation of the interpretability outcomes, both qualitatively and quantitatively. The presented framework thus holds substantial promise as a comprehensive solution that not only enhances prediction accuracy but also aids in the extraction of valuable insights from complex multi-variable datasets.