Back to Blog
![]() Vader Sentiment Analysis is lexicon and rule-based so it uses a dictionary to score individual lexical features and sums these up to give an overall polarity. I wanted to try both rule-based sentiment analysis and a model-based technique. This the model used was trained on modern web data while the text is several hundred years old, it would be an interesting exercise to try and train the Spacy model on more specific data and see if that improves the entity extraction. For example, Duke of York (the future King James II of England) appears around 2000 times according to the term frequency analysis, but does appear in the top 25 person entities here. The Spacy out-the-box model worked far from perfectly here, many entities were categorised as the wrong type and many weren’t extracted at all. Pepys’ maid Jane Edwards and his brother Tom. ![]() Pen, who is Admiral William Penn, a politician and father of the founder of the modern American state of Pennsylvania. ![]() Batten who was Sir William Batten, a high ranking navy colleague of Pepys’. ![]() We see here various people Pepys mentioned, such as W. ![]()
0 Comments
Read More
Leave a Reply. |