Course details

8.3428

Practical NLP: Advanced Text Analysis of Surveys

SP
WS 2019 Prof. Dr. Michael Franke, Dr. rer. nat. Alexander Meier OFFLINE
6h/wk
12 ECTS
M.Sc modules:
CC-MP-SP - Study Project
CS-MP-SP - Study Project

This SP intends to draw the most insights out of large text corpora from e.g. employee surveys. Large companies regularly perform employee surveys not only to get a status update of their employees, but to care for their needs, foster their job satisfaction, commitment and engagement as well as improve the organization’s performance. Open questions with free, open comments are on the rise in surveys for their individuality and their high information content. However, the manual analysis is extremely cumbersome and time intensive. Fortunately, computational linguistics gives us new, countless opportunities and makes the use of open comments possible on a large scale, way beyond simple keyword analysis. Over the last years, a lot has happened in the field of automated text processing. We will use but are not limited to real-world (anonymized) survey data to apply state-of-the-art textanalysis algorithms. These (machine learning) algorithms reach from topic modeling over sentiment analysis to probabilistic models for anonymization. Beside extending and testing models, topic modeling will be enhanced e.g. through the integration of semantic WordNets.The project takes place in cooperation with the company deepsight GmbH. Deeper knowledge of the methods is not mandatory and can be acquired during the project. Additionally, the concept of the project is still in flux and may change depending on the skills and interests of participants. Important aspects are: Optionally: Using web scraping to acquire or expand text corpora; Extending and combining textanalysis algorithms for (better than human) performance/quality; From anonymization and data cleaning to topic modeling and wordnets; Developing/Expanding metrics and benchmarks, determining gold standards; Drafting visualizations with high usability What you learn: Web scraping; Practical natural language processing; Advanced text analysis with un-/semi-supervised learning; Sentiment analysis; Visualizations Prerequisites