Abstract

This thesis aims to identify current trends in Privacy-Preserving Natural Language Processing (NLP) and potential research gaps that need to be explored. For this purpose, we will apply a systematic mapping study. 

Motivation

Why is Privacy-Preserving NLP important? It protects our primary means of communication: Text and Speech. Nowadays, we exchange our data in different forms especially text and speech, for allegedly „free” service on third-party platforms. However, an eye-opening event for Privacy was the Cambridge Analytica/Facebook scandal showing what is done with our data we don't know about. In 2017, 2.6 billion records were breached (76% accidentally, 23% due to malicious outsiders). Data Breaches can lead to Identity Thefts, Credit Card Frauds, etc. 

Research Questions

RQ1: What privacy-related challenges exist in the area of Natural Language Processing (NLP)?

RQ2: What approaches are used to preserve privacy in NLP tasks, and how can they be classified?

RQ3: What are the current research gaps and possible future research directions in the area of privacy-preserving NLP?