| Titre : | A Deep Learning Approach for the analysis of feelings on social networks |
| Auteurs : | MOUATEZ BELLAH KARABAGHLI, Auteur ; Khaled Rezeg, Directeur de thèse |
| Type de document : | Monographie imprimée |
| Editeur : | Biskra [Algérie] : Faculté des Sciences Exactes et des Sciences de la Nature et de la Vie, Université Mohamed Khider, 2019 |
| Format : | 1 vol. (81 p.) / ill. / 29 cm |
| Langues: | Anglais |
| Mots-clés: | Analyze des sentiments,opinion mining,text mining,social networks,deep learning. |
| Résumé : |
Unstructured textual data produced on the Internet is growing rapidly and the analysis of feelings is becoming a challenge because of the limited contextual information they usually contain. Millions of people share daily real-time thoughts and opinions about everything, which generates an unstructured, informative and yet valuable information to data scientists. Traditional approaches are important to the world of consumer behavior because they require a large amount of time and resources, and lead to considerable losses for companies around the world. Text classification is an essential task for automatic natural language processing (NLP) with many applications, such as information retrieval, web search, ranking and spam filtering. The goal of the NLP is to process the text for analysis and extract information for decision support as a first step in our proposed work to propose an efficient and accurate approach for predicting sentiment from raw unstructured data in order to extract opinions from the Internet and predict online popular discussions using a deep learning approach, which can be valuable and decisive for economic and political researchers to serve the country and emerge from crises .In this work we present an approach for the classification of social media discussions about real-world events like popular mobility in Algeria, and we propose an approach to analyze the feelings of social media messages in Algeria. Applying the different stages of the NLP through the use of deep learning. |
| Sommaire : |
General introduction I – Part01 : Basic concepts……………..…………………... 11 1- Introduction……………………………………………….. 12 2-Artificial intelligence…………………..…………………...12 2-1AI problems…………………………………..……. 12 2-1-1Reasoning, problem solving……………….13 2-1-2Knowledge representation………................13 2-1-3Planning……………………………..…..…13 2-1-4Learning…………………………………....13 2-1-5Natural language processing(NLP)…………14 2-1-6Motion and manipulation…………………..14 2-1-7 Social intelligence……………………..…..14 3-Machine learning………….………………………………....14 3-1Machine learning tasks……………………………....15 3-1-1Supervised Learning………………….……..16 3-1-2Unsupervised Learning…………………..….18 3-1-3Reinforcement Learning…………...………..18 3-2Machine Learning algorithms……………………......20 3-2-1types of Machine Learning Algorithms……..21 3-2-2List of Machine Learning Algorithms……….21 3-2-2-1Linear Regression………………......22 3-2-2-2Logistic Regression………………….23 3-2-2-3Decision Tree………………………..23 3-2-2-4Naive Bayes………………………....24 3-2-2-5KNN (k- Nearest Neighbors)….....….25 3-2-2-6Neural networks…………….…….....26 4-Deep learning……………………………………….…............. 27 4-1The importance of deep learning……………….…..….27 4-2How Deep Learning Works……………………….…...28 4-3Difference between ML and DL…………………….....29 4-4How to Create and Train DL Models……………….....30 4-5Deep Neural Network…………………………….....…31 4-5-1Recurrent Neural Network (RNN)……………31 4-5-2Long Short Term Memory (LSTM)…………..33 4-5-3Convolutional Neural Network (CNN)……....33 4-5-4Types of NN to solve problems………...........34 5-Natural Language Processing………………………………....35 5-1Use of NLP………………………………………..…..35 5-2Difficulty of NLP…………………………………......35 5-3How does NLP Works…………………………..........36 5-4NLP techniques……………………………………....36 5-4-1Syntax…………………………………….....36 5-4-2Semantics………………………………...….37 6-Sentiment Analysis………………………………………...…37 6-1What is Sentiment Analysis………………………..38 6-2Sentiment Analysis Scope………………………….38 6-3Types of Sentiment Analysis………………………39 6-3-1Fine-grained Sentiment Analysis.…….…..39 6-3-2Emotion detection…………………….…..39 6-3-3Aspect-based Sentiment Analysis………..40 6-3-4Intent analysis…………………….……...40 6-3-5Multilingual sentiment analysis………….41 6-4Sentiment Analysis Algorithms…….…………….41 6-4-1Rule-based Approaches…………..….......41 6-4-2Automatic Approaches…………………..42 6-4-3The Training and Prediction Process........43 6-4-4Feature Extraction from Text……………44 6-4-5Classification Algorithms………………..44 7-Sentiment Analysis Challenges and importance..…………45 8-Conclusion…………………………………………….......45 II – Part02: Design and implementation …………..……..46 1-Problem…………………………………………..……..…47 2- Related works…………………………………………......48 3- Design………………………………………………..……49 3-1Collect……………………………………………..50 3-2 Retrieve posts……………………..……………....52 3-4 Preprocessing ………………………………….....52 3-5 Calculating vectors……………….………….…...54 3-6 Deep learning ‘LSTM’ training ………..……......54 3-7 Analyze and visualization……………….…..........55 3-8 Test………………………………………….…..…55 4- Conclusion………………………………………………...55 5-Proposed approach………………………………...............56 6-keras………………………………………….…….…..…..57 7- Proposed approach steps………………..………...............57 7-1 Collect and load data……………………………...57 7-2 NLP tasks (pre-processing)………………………..58 7-3 Vectorize dataset………………………………..…60 7-4 Deep learning step…………………………………62 7-4-1 Model building…………………………...65 7-4-1-1 Define Model……………….…...65 7-4-1-2 Compile Model…………….…….68 7-4-1-3 Fit model……………………..….69 7-5 Visualization the result……………………..……..70 3-5-1 Matplotlib…………………………………70 3-5-2 Discussion…………………………………73 3-5-3 Training validation………………….…….74 7-6 Test: Obtain the Sentiment………………………...75 7-7 Conclusion……………………………………...…77 |
| Type de document : | Mémoire master |
Disponibilité (1)
| Cote | Support | Localisation | Statut |
|---|---|---|---|
| MINF/472 | Mémoire master | bibliothèque sciences exactes | Consultable |
