Tourism Local Knowledge Acquisition via Pre-Trained Language Model

Local knowledge which is rich in travelogue provides differentiation, competitiveness and sustainability for tourism entrepreneurs and tourist, that can be well expressed by knowl-edge graph. Traditional tourism knowledge graph construction method constructed automatically by structured data facing the difficulties in extracting entity and complex relationship in travelogue from UGC data. To address these limitations, an tourism local knowledge information extraction framework for constructing a knowledge graph via pre-trained language model is proposed in this paper. A fine-tuned smaller language model (6B) is employed to extract tourism local knowledge entity, property and relationship from the travelogue generated by user on the online tourism agency platform, enabling automated and universal analysis of textualized data.