How rejonehridoy/emotion_detection_from_text_using_machine_learning works
Project type and purpose
• This is a standalone machine‐learning/text‐analysis pipeline implemented in a Jupyter notebook (Google Colab) for detecting emotions in tweets.
• It reads a labeled dataset (SemEval-2018 AIT-2018), applies text preprocessing, trains and evaluates six different classifiers, and outputs comparative results.
Main components
a. Data Sources
• emojis.txt (emoji‐to‐text mapping)
• text_emotions.csv (AIT-2018 tweets + labels)
b. Preprocessing Module
• Punctuation removal
• Emoji conversion (using emojis.txt)
• Tokenization
• Stopword removal
• Lemmatization
c. Feature Extraction
• TF–IDF vectorizer (or count vectorizer)
d. Model Training & Selection
• Support Vector Machine
• Logistic Regression
• Random Forest
• XGBoost
• Multinomial Naive Bayes
• Decision Tree
e. Evaluation & Comparison
• Accuracy calculation for each model
• Tabular/graphical comparison
f. Artifacts
• Jupyter Notebook (.ipynb)
• Presentation (.pptx)
• Report (.pdf)
Relationships and data flows
• Preprocessing → Feature Extraction
• Each Model → Evaluation module
• Evaluation → Final Results (notebook outputs, report, slides)
Architectural patterns and technologies
• Linear pipeline pattern: data ingestion → transformation → model training → evaluation
• Batch processing (all data processed at once in Colab)
• Key libraries: pandas, scikit-learn, XGBoost, NLTK or spaCy (for tokenization/lemmatization)
• Runtime: Google Colab (Python 3 environment)
Detailed elements
• Dataset directory and files under Data Sources
• Output artifacts: Notebook (.ipynb), Report (.pdf), Presentation (.pptx)