Skip to main content

Sentiment Analysis

In this course, you learn the essentials of Sentiment Analysis. It will go live by September 2020.

About This Course

In this course, participants learn the essentials of Sentiment Analysis. Much of today’s data is available in the form of text. Business documents, email, blog posts, consumer reviews and other forms of social media communication are just a few examples of the many forms of text data that we produce and interact with in our personal and professional life. The field of Text Analytics is all about gaining insight from textual data using computational methods. Sentiment analysis is a specific form of Text Analytics. It comprises approaches to extract the opinion or feelings of the author with regard to the subject s/he is writing about.

Sentiment Analysis is one of the most popular ways to use text data. Scanning social media data and online product reviews, manufacturing companies have an opportunity to learn how about customers’ opinions on their offers and which product attributes are appreciated or criticized. Similar techniques help companies to enhance customer service by analyzing email communication with consumers. Sentiment Analysis is also a mega-topic in the scope of political marketing and has been used to anticipate future developments in financial markets.

The course is targeted at professionals who want to familiarize themselves with the possibilities and potential of sentiment analysis and available approaches to extract sentiment information from text. We will begin with the foundations of sentiment analysis, identify relevant forms and sub-streams, and elaborate on use cases in business and society. We will also learn about service providers and implementation options in the form of commercial and open-source solutions.

The second chapter will focus on one specific form of sentiment analysis, which is called the lexicon-based approach. In this approach, we use dictionaries to determine the polarity of a piece of text. Such dictionaries exist for many languages. We will also examine sentiment dictionaries that were developed for specific applications. Consider the word ‘bullish’. We do not use it in spoken language much. In a financial context, however, it has a clear-defined meaning and signals an expectation of stock prices to increase. This kind of domain knowledge is available in special-purpose dictionaries such as the famous Loughran-McDonald dictionary for applications in finance and accounting. The chapter will conclude with a discussion of techniques to develop sentiment dictionaries. Afterward, participants will be able to generate a sentiment dictionary that optimally addresses the characteristics and requirements of your applications.

Chapter 3 is concerned with machine learning. We will begin with a brief recap of core machine learning principles and then dive into ways for using machine learning to extract sentiment. Especially more advanced use cases in which you seek detailed insight which sentiment a piece of text expresses about different aspects need machine learning. Coming back to the previous example of a product review. It is nice to know that consumers value your product. It is much more valuable though to understand which specific product features they like or like less. This is called aspect-based sentiment analysis and a good part of the chapter will be devoted to this highly relevant field.

In the last chapter of the course, we will experience the very latest developments in deep learning-based sentiment analysis. Deep learning has revolutionized the field of natural language processing and contributed many exciting models for sentiment analysis. Participants will have an opportunity to familiarize themselves with relevant developments and ready for working with cutting- edge sentiment classifiers.

The course features more than 4 hours of video lectures, more than 50 multiple choice questions , and various references to background literature. A certificate signed by the instructors is provided upon successful completion.

Example real-life cases of Sentiment Analysis in Python are also provided and extensively discussed.

We can also come and teach this course on-site in classroom format. If interested, please mail us at:


The enrollment fee for this course is EUR 250 (VAT excl.) per participant. Payments are securely handled by PayPal. If you are a company in the European Union, then we can apply VAT reverse charge. For this, please mail your VAT number to Part of our course revenue is used towards funding organizations involvement in protecting and cleaning our oceans. See our about page to learn more about our mission statement.

After enrollment, participants will get 1 year unlimited access to all course material (videos, code scripts, quizzes and certificate).


Before subscribing to this course, you should have a basic understanding of descriptive statistics (e.g., mean, median, standard deviation, histograms, scatter plots, etc.) and inference (e.g., regression analysis). While the basics of machine learning will be revised in the course, prior experience in that field, for example through completing the Machine Learning Essentials BlueCourses course, will come in handy. Likewise, sentiment analysis is a specific form of text data analysis. Thus, an understanding of the broader scope of natural language processing will be useful but is not essential. Our recent Text Analytics course provides an excellent foundation.

Course Outline

  • Chapter 1: Introduction to Sentiment Analysis
    • Foundations of sentiment analysis
    • Problem setting, motivation, and a little bit of history
    • A simple sentiment classifier in Python
    • Approaches toward sentiment analysis
    • Business applications
    • A formal perspective on sentiment and different levels of analysis
    • A short primer in natural language processing
    • Challenges in sentiment analysis
    • Tooling: Software and service providers
    • Sentiment analysis research at a glance
    • Quiz
  • Chapter 2: Lexicon-Based Sentiment Analysis
    • What is lexicon-based sentiment analysis
    • Overview of sentiment lexica
    • Sentiment analysis process with examples in Python
      • Mark sentiment expressions
      • Appl sentiment shifters
      • Handle but-clauses
      • Aggregate sentiment scores
    • Limitations of the lexicon-based approach
    • Special purpose lexica
      • Classifying tweets
      • Sentiment analysis for finance applications
    • Sentiment lexicon generation
      • Dictionary-based approach
      • Corpus-based approach
      • Recent developments
    • Quiz
  • Chapter 3: Machine Learning-Based Sentiment Analysis
    • A primer in machine learning
      • Supervised vs. unsupervised learning
      • Classification vs. regression
      • A versatile learner: the artificial neural network
    • Evaluating a sentiment model
      • Data organization
      • Indicators of predictive performance
        • Error measures for regression
        • Confusion matrix and derived metrics
        • Precision-Recall curves
    • Machine learning-based sentiment analysis
      • From structured data to text data
      • Feature representation for sentiment analysis
        • The bag of words approach
        • Word vectors
      • A neural network-based sentiment classifier in Python
      • Sentiment rating prediction and emotion classification
    • From document to sentence level
      • Peculiarities and challenges of sentence-level sentiment analysis
      • Sentence subjectivity classification and sentiment classification
    • The holy grail: aspect-level sentiment analysis
      • Why focusing on aspects
      • Use cases of aspect-based sentiment analysis
      • Opinion target extraction
        • Aspect extraction
          • Frequency-based techniques
          • Leveraging syntactic relations
          • Supervised learning-based approach
        • Entity extraction
        • Sentiment classification
        • Opinion holder and time extraction
    • Quiz
  • Chapter 4: Advanced Text Classification Using Deep Transfer Learning
    • A bird’s eye view on deep learning
    • Deep transfer learning for natural language processing
      • Language models
      • Word vectors revisited
      • Building blocks of modern language models
        • LSTMs, GRUs, and bidirectionality
        • Attention and transformers
      • State-of-the-art models
        • ULMfit and ELMO
        • The BERT family
      • Quo Vadis sentiment analysis
    • Quiz

Course Staff

Prof. dr. Stefan Lessmann

Prof. dr. Stefan Lessmann

Stefan was born in Hamburg (northern Germany) on the 18th of March 1975. He still lives in Hamburg, which is actually one of the best places in Germany (given the context of the course, you may want to note that this statement is subjective and carries a strong positive sentiment), together with his wife Tanja and his beautiful kids Linus and Lucie. Both of them being younger than ten at the time of writing, there is little to say about hobbies or ways in which he would spend leisure time (what time?). However, Stefan is eager to improve his very underdeveloped cooking skills and to do a bit of exercising. Readers who seek a great way to exercise may want to run a Google search for suspension training, which will return some highly recommended results. As many fellowmen and women born in Hamburg, Stefan is a very maritime person. Hamburg is unfortunately not at the seaside but has the river Elbe and its harbor of which all locals are extremely fond. Stefan used to do a lot of time when he was younger (i.e., ages ago) and still loves spending time at the sea, for example in holidays. This also explains his commitment to ; together with some great collaboration with Bart Baesens since 2006.

In his professional life, Stefan holds a professorship at the Humboldt-University of Berlin. He is heading the Chair of Information Systems at the School of Business and Economics, and very much appreciates the honor to work with a great international team of highly talented data scientists and AI researchers. Focus areas of the group include deep neural networks, natural language processing, and causal machine learning. With regard to business applications, Stefan is particularly interested in marketing and credit risk analytics and has many years of experience in these domains. He actively participates in collaboration, knowledge transfer, and consulting projects with industry partners; from start-up companies to global players and not-for-profit organizations. Stefan is also very active in the field of professional education and has delivered numerous workshops in his research areas to industry partners.

Prior to joining the Humboldt-University in 2014, Stefan was affiliated with the University of Hamburg, where he completed his Ph.D. and habilitation. During that time, Stefan was also a guest lecturer at the University of Southampton, UK, and taught various under- and postgraduate courses on quantitative methods, electronic business, and web application development. Stefan serves as an associate editor for the International Journal of Business Analytics, Digital Finance, and the International Journal of Forecasting, as well as department editor of Business and Information System Engineering (BISE). He is also very active in peer-review processes for various journals and firmly believes that this is an important task for every academic. Stefan secured substantial amounts of research funding and published close to hundred papers in leading international journals and conferences, including the European Journal of Operational Research, Decision Support Systems, and the IEEE Transactions of Software Engineering. Authoritative papers on the use of machine learning in credit scoring and software defect predictions received prestigious awards and many hundred citations. An almost complete overview of Stefan’s research work is available at ResearchGate .