Sanja Štajner, Uni Mannheim

Automated Text Simplification: the Strengths and Weaknesses of the Existing Systems

Syntactically and lexically complex texts and sentences pose difficulties both for humans (especially people with various reading or cognitive impairments, or non-native speakers) and for natural language processing systems (e.g. information extraction, machine translation, summarization, semantic role labeling). In the last 30 years, many systems have been proposed that attempt at automatically simplifying vocabulary and sentence structure of complex sentences. A few of them even operate on the text level trying to simplify discourse structure.
As there are no reliable strategies for comparative evaluation of different text simplification (TS) systems, and many systems are not publicly available, it is very difficult to compare the systems among themselves and know where we stand at the moment. This talk will present some of the most influential TS systems, ranging from rule-based syntactic simplifiers and hybrid lexico-syntactic TS systems, over the state-of-the-art lexical simplifiers based on the use of word embeddings and machine translation-based TS systems, to the newest neural TS systems. The emphasis will be on comparative evaluation of those systems and discussion about possible avenues to improve them.