A Study on an Integrated NER and Coreference Resolution Framework for Travel and Tourism Social Media Texts: Performance Comparison Using Advanced NLP Models

Abstract

The study presents an integrated framework combining Named Entity Recognition (NER) and coreference resolution tailored for travel and tourism social media texts. Social media content is often noisy, informal, and multilingual, posing significant challenges for traditional NER methods. Leveraging advanced transformer based NLP models, including BERT and its variants, the framework addresses these challenges by improving entity extraction and contextual linking of references. This paper shows data collection, preprocessing techniques and model architectures adapted for the travel and tourism context. Evaluation on a manually curated social media dataset demonstrates notable improvements in entity recognition accuracy and coreference resolution consistency. The results highlight the potential of this approach for enhancing tourism analytics, trend monitoring, and recommendation systems. This work contributes to the growing body of research on applying deep learning techniques to domain specific social media text analysis in the tourism sector.