Construction of Kazakh Knowledge Graph in Tourism

Abstract

The construction of knowledge graph is to organize and store knowledge from the objective world to form aknowledge system. However, building a high-quality, large-scale knowledge graph is an important challenge for the low-resource Kazakh language. For the needs of the tourism field, this paper proposes a method for constructing Kazakh knowledge graphs based on named entity recognition and relation extraction. First, we use the deep learning method to obtain the semantic information of the Kazakh entity features, use Bi-LSTM to learn from different directions to obtain effective context information. Through multi-feature fusion, the word and word location features and entity tags are merged to obtain the association between the word and the target entity. Second, we use R-BERT to capture the semantic information of sentences and target entities to better adapt to relation classification tasks. Finally, we use the graph database neo4j to store the triple knowledge of the model layer, the ontology and the data layer to construct the Kazakh tourism knowledge graph。