Exploring the Potential of BERT-BiLSTM-CRF and the Attention Mechanism in Building a Tourism Knowledge Graph

As an important infrastructure in the era of big data, knowledge graph can integrate and manage data resources. Therefore, construction tourism graphs with wide coverage high quality terms information from perspective tourists’ needs is effective solution to problem clutter field. This paper first analyzes current state domestic international research on constructing highlights problems associated graphs, which are that they time-consuming, laborious have a single function. In order make up for these shortcomings, this proposes set systematic methods build graph. method integrates BiLSTM BERT models combines attention mechanism. The steps as follows: First, preprocessing carried out by word segmentation removing stop words; second, after extracting features vectorization words, cosine similarity used classify text, text classification based naive Bayes being compared through experiments; third, popular words obtained popularity analysis model. two obtain words: One multi-dimensional product model principal component analysis; other emotion fourth, uses BiLSTM-CRF identify entities predict relationship between so extract high-quality triplets. improve effect entity recognition, recognition BiLSTM-LPT BiLSTM-Hanlp models. experimental results show effectively efficiency recognition; finally, was imported into Neo4j graphic database