Natural Language Processing for Narcissism Detection

Description A recently published study examines essays written by participants regarding the severity of narcissism. These are to be analysed using modern NLP methods to predict the NPI. However, applying state-of-the-art deep learning methods to small data sets of only a few hundred participants is a particular challenge. One way to overcome this challenge is to use simplified embeddings[1] which represent only parts of the sentences and thus make text structure easier to learn even with less data.
The aim of this study is to develop a machine learning approach to analyze these essays and compare the method of Linguistic Inquiry and Word Count (LIWC) commonly used in psychology.  
Task In this thesis, the student(s) develops narcissism detection based on newly developed word embeddings.
Utilises Tensorflow, Word Embeddings, Deep Neural Networks.
Requirements Advanced knowledge in Machine Learning and Natural Language Processing, Good programming skills (e.g. Python, C++).
Languages English or German.
Supervisor Lukas Stappen, M. Sc.
Manuel Milling, M. Sc.