Home
- Isabelle Eysseric
- David Poisson
- Nicolas Garde
PyCharm with Python 3.7 was used so that the Spacy library could work properly.
We wrote 2 methods to convert the sentences: one that writes the dataset with children/head (1h) and the other without (20 minutes).
In agreement with Professor Luc Lamontagne, to make task 3 go faster, we wrote the converted sentences of the method using the dependency tree of the negation_conversion.py file in text files.
These files were produced with the write_negated method found in the sentiment_analysis.py file. They are stored in the data folder.
Also, it is necessary to download the NLTK corpus, Sentiwordnet so that the sentiment_analyse.py file can work.
( See file negation_conversion.py )
- Step 1: Find the scope of the sentence
- Step 2: Capture the negative scope
- Step 3: Conversion and reconstruction of the sentence
Figure:
[0] SpaCy website, article "CORE MODELS : English : en_core_web_sm".
[1] SpaCy website, article "Annotation Specs : Dependencies : Part-of-speech tagging".
[2] SpaCy website, article "GUIDES: Linguistic Features : Dependency Parse: Visualizing dependencies".
[3] SpaCy website, article "GUIDES: Linguistic Features : Dependency Parse: Navigating the parse tree".
[4] SpaCy website, article "Annotation Specs : POS Tagging : Syntactic Dependency Parsing"
[7] Scikit-Learn website, article "Naive Bayes classifier for multinomial models"
[8] Scikit-Learn website, article "Logistic Regression (aka logit, MaxEnt) classifier"
[9] Scikit-Learn website, article "Multi-layer Perceptron classifier"
[10] Scikit-Learn website, article "6.2. Feature extraction"
[17] Scikit-Learn nwebsite, article "sklearn.feature_extraction.text.CountVectorizer.transform"
[11] SpaCy website, article "Annotation Specs : Named Entities : Lexical data for vocabulary"
[12] Universal Dependencies website, article "Universal POS tags"
[13] SpaCy website, article "GUIDES : Adding languages : Language data"
[14] NLTK website, article "SentiWordNet Interface"
[15] Stack Abuse website, article "Python for NLP: Creating Bag of Words Model from Scratch"
[16] SpaCy website, article "User guide"