Machine Learning / NLP
SentiPhrase - Phrase Level Sentiment Analysis
Machine Learning / NLP Developer
Details
RoleMachine Learning / NLP Developer
TypeMachine Learning / NLP
StatusCompleted
Tech Stack
PythonStreamlit
Project Overview
SentiPhrase is a phrase-level sentiment analysis application designed to analyze Indonesian text at a more granular level than traditional sentence-level sentiment analysis. The application extracts meaningful phrases using Stanza and rule-based sentence structures, then applies multiple sentiment analysis approaches including Naive Bayes, SVM, LSTM, and IndoBERT. TF-IDF, cosine similarity, and lexicon-based methods are also incorporated to support feature extraction and phrase identification. The application was built with Streamlit and deployed as an interactive web application.
Key Highlights
- Built an interactive Indonesian phrase-level sentiment analysis application using Streamlit and Python
- Compared four sentiment analysis approaches: Naive Bayes, SVM, LSTM, and IndoBERT
- Implemented phrase extraction using Stanza and rule-based sentence structures
- Applied TF-IDF, cosine similarity, and lexicon-based methods for text processing and phrase identification
- Integrated multiple NLP and machine learning models into a single interactive web application