Research On The Construction And Application Of Knowledge Graph In The Field Of Chinese Tourism

As a major change brought about by the age of big data Artificial Intelligence(AI)is quickly integrated into every aspect of our lives.Meanwhile,as the most important infrastructure of AI technology,the Knowledge Graph will inevitably continue to rise.The Knowledge Graph is a highly structured sematic knowledge base that uses logical symbols to describe concepts and interrelationships in the form of logical symbols in the real world,and serves as the foundation for supporting applications such as next-generation search engines and question answering systems.It is like a key that can open the treasure trove of human knowledge and will gradually lead more disciplines(including tourism,finance,education,telecommunications,medical care,etc.)to start a new era of development.This thesis is based on the field of Chinese tourism,focusing on the practical issues(such as the human and time costs in the process of Knowledge Graph construction,the completion of Knowledge Graph,the entity alignment,and the presentation methods of knowledge services)that operators and tourists pay close attention to.A series of improvements have been made to the construction and application of the Chinese Tourism Knowledge Graph,hoping to enable tourists to quickly and accurately capture the information they need in massive tourism data.The specific work is concluded as follows:(1)Aiming at the problems of time-consuming,labor-intensive and Knowledge Graph completion in the process of Knowledge Graph construction,we propose a new construction method of the Knowledge Graph on Chinese tourism domain.Here,the tourism knowledge was selected from the existing high-quality Chinese encyclopedia knowledge base,and then the Knowledge Graph on Chinese tourism domain was constructed in the form of tourism knowledge base.In this way,various labor and time costs are reduced.In addition,by introducing the external knowledge base XLore,we have also successfully infoboxed the Chinese Knowledge Graph on tourism-domain.Then,based on the constructed Knowledge Graph on domain-tourism by us,aiming at the shortcomings of existing representation learning models such as TransE,which only learn triple information in isolation,and ignore a lot of other information related to knowledge,this paper proposes a new TransE model that fuses Knowledge Graph entity description text information.We here use a BERT model to obtain a vector of entity description text,and fuses this vector with the TransE model through a neural network with a single hidden layer.In this paper,the above method is applied to the constructed Knowledge Graph on Chinese tourism domain,and the entities and relationships of the Knowledge Graph are projected into the same low-dimensional continuous vector space in an embedded manner,thereby learning a low-dimensional vector for each entity and relationship.At last,we validated the effective of the method in link prediction and triple classification experiments.The effectiveness of the method is verified in experiments of link prediction and triplet classification.(2)Aiming at the problem of entity alignment in the knowledge integration stage during the construction of Knowledge Graph,this paper proposes a new method of entity alignment based on neural network word vector representation technology by studying the technique of word vector representation in the field of natural language processing.The core point of this method is the fine-tunning BERT model and modeling the entity alignment as a binary classification problem.Finally,by comparison with the CBoW model,Skip-gram model,DSG model,and ELMo model,this method does have a clear advantage in the performance of entity alignment task.(3)To better serve tourists with high-quality knowledge services and realize the linking and sharing of knowledge,we have further established a knowledge service prototype system based on the completion of the construction of the Knowledge Graph on Chinese tourism domain.This solution implements many functions including semantic search of tourism entities,visualization of tourism Knowledge Graph,knowledge cards of tourism entities,display of InfoBox,physical links in the knowledge base,recommendation of tourist attractions,API release of tourism knowledge,and update of tourism knowledge base.