| Titre : | Une Approche Big Data Social basée sur les Techniques de l'IA. |
| Auteurs : | Wafa Saadi, Auteur |
| Type de document : | Thése doctorat |
| Editeur : | Laallam, 2026 |
| Format : | 1 vol. (281 p.) / couv. ill. en coul / 30 |
| Langues: | Anglais |
| Langues originales: | Anglais |
| Résumé : |
The extensive use of social media platforms has contributed to the continuous creation ofmassive amounts of user-generated textual data. This expanding volume of digital content represents a valuable resource for both scientific research and practical applications,particularly in the domains of mental health assessment, large-scale analysis of public opinion,and computational investigation of emotional expression. Consequently, text-based emotion detection has become a prominent research domain at the intersection of affective computing,natural language processing, and computational social science. Despite recent advances, emotion detection in noisy, multilingual corpora remains a complex task, particularly for lowresource languages and dialects where annotated data and linguistic tools are limited. The success of emotion detection systems based on machine learning and deep learning is largelycontingent upon the widespread availability, high-quality annotated data. This difficulty becomes even more pronounced in the context of low-resource languages and dialectal variants,particularly the Algerian dialect, where manual annotation processes are both costly and timeintensive.To address this limitation, this thesis introduces an original framework for text-basedemotion detection that mitigates the lack of annotated resources by adopting an unsupervised learning approach. This thesis, proposes a framework based on an unsupervised ensemble clustering techniques, which combine multiple clustering algorithms to exploit their complementary strengths and mitigate the biases and instability of individual methods. Themain contribution of this thesis is the design of an ensemble clustering-based approach for the automatic generation of emotion-labeled corpora from unstructured social media text (Twitterand YouTube). By combining multiple clustering algorithms and exploiting their diversity through a consensus mechanism, the proposed method enhances clustering stability, robustness, and semantic coherence. The resulting clusters are subsequently mapped to discrete emotion categories defined by Ekman’s emotion model, consequently allowing the construction of highquality labeled datasets without human assistance. The resulting labeled datasets are intendedfor direct use in subsequent supervised learning tasks. Experimental results demonstrate that ensemble clustering significantly superior performance to individual clustering techniques measured by internal evaluation metrics and label consistency. Furthermore, the automatically generated corpora prove effective for training supervised emotion classification models, confirming the practical relevance and scalability of the proposed approach. Overall, this thesis contributes to the enhancement of affective computing by offering a language-adaptive,scalable, and resource-efficient solution for emotion detection in Big Data environments, withparticular relevance to low-resource and dialect-rich linguistic contexts. |
| Sommaire : |
Contents LIST OF FIGURES ..... IV LIST OF TABLES .......... V LIST OF PUBLICATIONS .................................................................. VI CHAPTER 1: INTRODUCTION ................................................................................................................... 1 1.1. CONTEXT....... 1 1.2. PROBLEM STATEMENT AND MOTIVATION.................................................................................. 2 1.3. RESEARCH OBJECTIVES AND CONTRIBUTIONS.............................................................................. 2 1.4. THESIS OUTLINE................................................................. 5 PART 1: LITERATURE REVIEW AND RELATED WORK...................................................................................... 6 CHAPTER 2: THEORETICAL FOUNDATIONS OF EMOTION DETECTION............................................................... 6 2.1. INTRODUCTION ............................................. 6 2.2. EMOTION DEFINITION....................................................... 7 2.3. TYPES OF EMOTION ............................................................................................................... 8 2.4. EMOTION MODELS ................................................................................................................ 8 2.4.1. CATEGORICAL EMOTION MODELS............................. 9 2.4.2. DIMENSIONAL EMOTION MODELS........................................ 10 2.5. EMOTION DETECTION ........................................................................................................... 12 2.6. EMOTION DETECTION VERSUS SENTIMENT ANALYSIS................................................................. 13 2.7. EMOTION DETECTION MODALITIES.......................................................................................... 14 2.7.1. EMOTION DETECTION FROM FACIAL EXPRESSION....................................................................... 14 2.7.2. EMOTION DETECTION FROM BODY POSTURE AND GESTURES ....................................................... 15 2.7.3. EMOTION DETECTION FROM PHYSIOLOGICAL SIGNALS................................................................ 15 2.7.4. EMOTION DETECTION FROM SPEECH ....................................................................................... 16 2.7.5. EMOTION DETECTION FROM TEXT....................................... 16 2.7.6. MULTIMODALITY ............................................................................ 17 2.8. CONCLUSION .......................................................................................... 18 CHAPTER 3: TEXT BASED EMOTION DETECTION...................................................... 20 3.1. INTRODUCTION ..................................................................................... 20 3.2. EMOTION IN TEXT ............................................................................ 21 3.3. APPLICATION DOMAIN OF TBED ............................................................................................ 21 3.4. TBED EVOLUTION...................................................................................... 23 3.5. ARTIFICIAL INTELLIGENCE FOR TBED ...................................................................................... 25 3.6. NATURAL LANGUAGE PROCESSING FOR TBED ......................................................................... 25 3.7. TEXT BASED EMOTION DETECTION SYSTEMS............................................................................ 26 3.8. TEXT BASED EMOTION DETECTION PROCESS............................................................................ 27 3.9. RESOURCES AND APPROACHES FOR TBED ............................................................................... 29 3.9.1. RESOURCES FOR DETECTING EMOTIONS IN TEXT....................................................................... 29 3.9.2. TEXT BASED EMOTION DETECTION TECHNIQUES........................................................................ 33II 3.10. CHALLENGES AND DIFFICULTIES CONCERNING TBED ................................................................. 36 3.11. CONCLUSION ............................................................. 38 PART 2: CONTRIBUTIONS .................................................. 39 CHAPTER 4: AN ENSEMBLE CLUSTERING BASED FRAMEWORK FOR TEXT BASED EMOTION DETECTION FOR TWITTER DATA (CONTRIBUTION 1) ........................................................... 39 4.1 INTRODUCTION........................................................ 39 4.2 BASIC CONCEPTS................................................... 41 4.2.1. CLUSTERING DEFINITION ....................................................................................................... 41 4.2.2. THE MAIN STEPS OF THE CLUSTERING PROCESS.......................................................................... 41 4.2.3. CLUSTERING METHODOLOGIES ............................................. 43 4.2.4. TEXT CLUSTERING............................................................ 47 4.2.5. LIMITATIONS OF SINGLE CLUSTERING ALGORITHMS ................................................................... 47 4.2.6. FUNDAMENTALS OF CLUSTERING ENSEMBLES........................................................................... 48 4.2.7. EVALUATION METRICS FOR THE CLUSTERING ALGORITHMS .......................................................... 50 4.3. RELATED WORKS.................................................... 52 4.4. METHODOLOGY.............................. 4.4.1. DATA COLLECTION........................................... 55 4.4.2. DATA PREPROCESSING ................................ 56 4.4.3. EMBEDDING TOKENIZED TWEETS WITH BERT MODEL ............................................................... 57 4.4.4. ENSEMBLE CLUSTERING ................................. 58 4.5. RESULTS AND DISCUSSIONS....................................................................................................... 62 4.5.1. RESULTS OF USING DIFFERENT OBJECTS REPRESENTATIONS.......................................................... 63 4.5.2. RESULTS OF USING DIFFERENT CLUSTERING ALGORITHMS.................................. 65 4.5.3. IN DIFFERENT PARAMETER INITIALIZATION ............................... 67 4.6. CONCLUSION......... 69 CHAPTER 5: EMOTION DETECTION FROM YOUTUBE COMMENTS (CONTRIBUTION 2) ..................................... 70 5.1. INTRODUCTION........ 70 5.2. RELATED WORKS.............................................................................. 70 5.3. METHODOLOGY........ 72 5.3.1. DATA COLLECTION.................................................. 73 5.4. RESULTS AND DISCUSSION .............................................. 76 5.5. CONCLUSION ................................................ 81 CHAPTER 6: BIG DATA FRAMEWORK FOR EMOTION DETECTION (CONTRIBUTION3) ........................................ 83 6.1. INTRODUCTION..................................... 83 6.2. BASIC CONCEPTS........................................... 84 6.2.1. BIG DATA.............................................. 84 6.2.2. APACHE SPARK ..................................... 86 6.3. RELATED WORKS ............. 89 6.4. METHODOLOGY....... 93III 6.4.1. DATASET CREATION........................ 93 6.4.2. DATASET PREPARATION.................................... 94 6.4.3. CLUSTERING TECHNIQUE ....................................................................................................... 95 6.4.4. MACHINE LEARNING TECHNIQUES............................................ 95 6.5. RESULTS AND DISCUSSIONS....................................... 95 6.6. CONCLUSION..... 97 CHAPTER 7: GENERAL CONCLUSION AND PERSPECTIVES ....................... 98 REFERENCES ....... 100 |
| Type de document : | Thése doctorat |
Disponibilité (1)
| Cote | Support | Localisation | Statut |
|---|---|---|---|
| TINF/218 | Théses de doctorat | bibliothèque sciences exactes | Consultable |
