Recognizing textual entailment: A review of resources, approaches, applications, and challenges

Abstract

The review aims to examine the current state of recognizing textual entailment (RTE) research and summarize the state-of-the-art methods in the development of natural language processing (NLP) applications, the various approaches, datasets, and future challenges. The main finding is that the availability of resources, i.e., datasets and sentence processing methods, is very important to note and that RTE can be widely applied to different text domains in NLP applications. The main challenges to be addressed in future RTE research are expanding RTE datasets, drawing inferences from more than one premise, and recognizing inferences from sentence fragments that use different languages.

Keywords

• Recognizing textual entailment
• Natural language inference
• Resources
• Applications
• Approaches
• Challenges RTE