Tariq A., Ramsay A. Machine Learning for Emotion Analysis in Python: Build AI-powered tools for analyzing emotion using natural language processing

Packt Publishing, 2023. — 334 p.The AI winter has long thawed, but many organizations are still failing to harness the power of machine learning (ML). If you want to tap that potential and add value to your business with cutting-edge emotion analysis, you’ve found what you need in this trusty guide.In Machine Learning for Emotion Analysis, you’ll take your foundational data science skills and grow them in the exciting realm of emotion analysis. With its practical approach, you’ll be equipped with everything you need to give your company a clear insight into what your customers are thinking.This no-nonsense guide jumps right into the practicalities of emotion analysis, teaching you how to preprocess data, build a serviceable dataset, and ensure top-notch data quality. Once you’re set up for success, we get hands-on with complex ML techniques. This is where you go from the intermediate to the advanced, covering deep neural networks, support vector machines, conditional probabilities, and more, as you experience the full breadth of possibilities with emotion analysis. The book finally rounds out with a couple of in-depth use cases – a sort of sandbox for you to experiment with your newly acquired skill set.By the end of this book, you’ll be ready to present yourself as a valuable asset to any organization that takes data science seriously.Essentials.
Foundations.
Emotions.
Categorical.
Dimensional.
Sentiment.
Why emotion analysis is important.
Introduction to NLP.
Phrase structure grammar versus dependency grammar.
Rule-based parsers versus data-driven parsers.
Semantics (the study of meaning).
Introduction to machine learning.
Technical requirements.
A sample project.
Logistic regression.
Support vector machines (SVMs).
K-nearest neighbors (k-NN).
Decision trees.
Random forest.
Neural networks.
Making predictions.
A sample text classification problem.Building and Using a Dataset.
Building and Using a Dataset.
Ready-made data sources.
Creating your dataset.
Data from PDF files.
Data from web scraping.
Data from RSS feeds.
Data from APIs.
Other data sources.
Transforming data.
Non-English datasets.
Evaluation.Labeling Data.
Why labeling must be high quality.
The labeling process.
Best practices.
Labeling the data.
Gold tweets.
The competency task.
The annotation task.
Buy or build?
Results.
Inter-annotator reliability.
Calculating Krippendorff’s alpha.
Debrief.Preprocessing – Stemming, Tagging, and Parsing.Word parts and compound words.
Tokenizing, morphology, and stemming.
Spelling changes.
Multiple and contextual affixes.
Compound words.
Tagging and parsing.Approaches.
Sentiment Lexicons and Vector-Space Models.
Datasets and metrics.
Sentiment lexicons.
Extracting a sentiment lexicon from a corpus.
Similarity measures and vector-space models.
Vector spaces.
Calculating similarity.
Latent semantic analysis.Naïve Bayes.
Preparing the data for sklearn.
Naïve Bayes as a machine learning algorithm.
Naively applying Bayes’ theorem as a classifier.
Multi-label datasets.Support Vector Machines.
A geometric introduction to SVMs.
Using SVMs for sentiment mining.
Applying our SVMs.
Using a standard SVM with a threshold.
Making multiple SVMs.Neural Networks and Deep Neural Networks.
Single-layer neural networks.
Multi-layer neural networks.Exploring Transformers.
Introduction to transformers.
How data flows through the transformer model.
Input embeddings.
Positional encoding.
Encoders.
Decoders.
Linear layer.
Softmax layer.
Output probabilities.
Hugging Face.
Existing models.
Transformers for classification.
Implementing transformers.
Google Colab.
Single-emotion datasets.
Multi-emotion datasets.Multiclassifiers.
Multilabel datasets are hard to work with.
Confusion matrices.
Using “neutral” as a label.
Thresholds and local thresholds.
Multiple independent classifiers.Case Study.
Case Study – The Qatar Blockade.
The case study.
Short-term changes.
Long-term changes.
Proportionality revisited.Other Books You May Enjoy.