Abstract: Sentiment analysis has been one of the most popular applications of NLP and text analytics fields due to its practical usefulness in real-world business environment. Specially, its popularity is due to the pervasiveness of various digital data such as social network, blogs, online reviews, and ratings in modern internet world. People express opinions, feelings, ideas, and their mood to the digital platforms. Sentiment analysis is the practice of applying natural language processing and text analysis techniques to identifying and classifying the sentiment information from a given text so that it is a great/beneficial tool to dig on the subjective text.
This talk is for those who want to learn the introduction of sentiment analysis task and its different model choices to implement it. By listening this talk, you will become familiar and comfortable with the sentiment analysis and its different methodological approaches. Plus, I will bring up more advanced topics in this area such as issues and challenges in sentiment analysis or Aspect-based Sentiment Analysis (ABSA).
Session 1 – Understanding of Sentiment Analysis
In this session, you will get the basic understanding of sentiment analysis and its specific task branches such as subjectivity, hate-speech, sarcasm, politeness, etc. You will also learn how its methodological approaches has been changed as the computational linguistics technologies has been evolved in NLP world. I will discuss about three main different approaches for SA – Lexicon-based approach, Machine-learning approach, and Hybrid approach.
Session 2 – Hands-on examples of SA
In this session, you will get the hands-on experience of sentiment analysis with existing related open datasets. You should be able to understand how to pre-process the given text data (such as Tokenization, Stemming, Part of Speech Tagging, Stopword, Regular Expression, etc.) before you apply the different SA approaches. Then you can learn how to design a SA model with your different model choices and how to implement it in computational ways.
Session 3 – Issues and Challenges in SA Task
In this session, we will discuss the issues and challenges in sentiment analysis area such as negation, bi-polar words, domain dependency, spam and fake detection. Plus, you will learn how the new approaches has been introduced to overcome these issues and challenges in this field.
Session 4 – Aspect-based Sentiment Analysis
In this session, aspect-based sentiment analysis (ABSA) will be introduced. ABSA is rapidly growing area of sentiment analysis that has gained prominence in recent years. You will learn the concepts and the major four elements of ABSA. You will also be able to get some brief understanding on the common approaches to apply ABSA on your dataset.
Python, basic understanding of data, data pre-processing, and data mining
Bio: Sunny is a seasoned professional data scientist, with over 15 years of relevant experience, and successful completion of significant company-onsite projects for many respected companies in South Korea and the US. Significant experience and dynamic practitioner in various domains, including NLP project lead, credit risk modeling, financial distress modeling, customer marketing prediction, and ML service provider consultation. She is passionate about creating and building AI solutions applying a variety of NLP technologies including sentiment analysis, conversational computing, topic modeling, etc. to support AI real-world usages for SME businesses. She is currently putting her efforts into her own AI start-up company – ReviewMind Inc. In 2020, her company was identified as an excellent start-up case by Korea Women in Science and Technology Support Center. Sunny and her team also won the best award in the 2021 Start-up Demo Day from the Korea Institute of Startup & Entrepreneurship Development. Sunny holds both a Masters in Data Science (Information Systems) and an MBA from the US and South Korea respectively.