Aspect based sentiment analysis using machine learning approach

Question

I am very new in machine learning. I have annotated data with category, aspect, opinion word and sentiment. for example, for the bellow text

"The apple was really tasty"

I have category->food, aspect-> apple, opinion word ->tasty and sentiment->positive. I have training data like this format.

How can I train a SVM classifier using this format of training set? How to extract features like n-gram, POS and sentiment word to train the classifier? Could you please suggest any beginning step for this aspect based sentiment analysis using machine learning algorithms?

Answer 1

Aleksandar Blekh has given some really nice links about the big picture of how to do sentiment analysis; I'll try to provide some links to software and talk about the nitty-gritty of how to make it work. I'll point you to the example using scikit-learn (http://scikit-learn.org/stable/ ), a machine learning library in Python.

You would first want to take your dataset and load it into scikit-learn in a sparse format. This link (http://scikit-learn.org/stable/tutorial/text_analytics/working_with_text_data.html ) gives examples of how to load text in a bag-of-words representation, and the same module (scikit-learn.feature_extraction.text) can also count n-grams. It then describes how to run Naive Bayes and SVM on that dataset. You can take that example and start playing with it.

Answer 2

I would recommend you to start from reading the draft of the introductory book "Sentiment analysis and opinion mining" by Bing Liu. The draft in a PDF document format is available for free here.

More details about the new upcoming book of this author, as well as comprehensive information on the topic of aspect-based sentiment analysis, with references and links to data sets, are available at this page: http://www.cs.uic.edu/~liub/FBS/sentiment-analysis.html.

Another interesting resource is a survey book "Opinion mining and sentiment analysis by Bo Pang and Lillian Lee. The book is available in print and as a downloadable PDF e-book in a published version or an author-formatted version, which are almost identical in terms of contents.