miguel-kjh/Comment-Sentiment-Analysis

Sentiment analysis on an amazon food dataset.

Comment-Sentiment-Analysis

Sentiment analysis on an amazon food dataset.

Text mining

• Tokenizer: punkt
• WordNet: is a large lexical database of English. Nouns, verbs, adjectives and adverbs are grouped into sets of cognitive synonyms (synsets), each expressing a distinct concept. Synsets are interlinked by means of conceptual-semantic and lexical relations.

Machine Learning

• Bag of words(TF-idf and Boolean model)
• Decision Tree
• Random Forest
• AdaBoost

Deep Learning

• Word Embedding(Word2Vec)
• LSTM: Long short-term memory.

Results

Boolean Model

Model	Precision(%)	Macro Avg(%)	Weighted Avg(%)
DF + no processing data	33	20	38
RF + no processing data	63	17	51
AB + no processing data	64	18	51
DF + processing data	30	18	35
RF + processing data	53	20	49
AB + processing data	55	19	49
DF + processing data + balanced	20	20	20
RF + processing data + balanced	24	22	22
AB + processing data + balanced	22	17	17

TF-idf Matrix

Model	Precision(%)	Macro Avg(%)	Weighted Avg(%)
DF + no processing data	39	20	41
RF + no processing data	64	16	51
AB + no processing data	63	18	51
DF + processing data	34	19	38
RF + processing data	59	18	50
AB + processing data	55	19	49
DF + processing data + balanced	22	21	21
RF + processing data + balanced	22	17	17
AB + processing data + balanced	25	23	23

Deep Learning

Model	Precision(%)	Macro Avg(%)	Weighted Avg(%)
LTSM + no processing data	66	36	62
LTSM + processing data	65	41	64
LTSM + processing data + balanced	86	86	86

Technologies and Libraries

• Python 🐍
• Sklearn 🧮
• Keras ❤️
• NLTK 📝