知识图谱以其强大的语义处理能力和开放组织能力，为大数据时代的知识化组织和智能应用奠定了基础。对于互联网旅游行业，如何构建和应用一个旅游领域知识图谱成为非常有价值的问题。本项目组将充分利用在自然语言处理方面积累的研究经验，并借鉴已有先进的研究方法，拟搭建一个面向内蒙古旅游业服务的蒙古文知识图谱。本项目重点研究如下内容：(1)构建符合蒙文文本特点的蒙古文词向量；(2)针对旅游领域实体的类别和结构特征，研究结合语言模型与Attention机制的深度学习实体识别模型；(3)基于端到端网络的领域属性关系抽取方法；(4)基于上述技术构建蒙古文旅游图谱原型，并借助翻译技术和结构化程度较好的外源数据，对蒙古文旅游知识图谱进行补全和融合；(5)研究旅游领域知识图谱中实体和标签的翻译与消歧技术。本项目研究成果可以为语义查询、个性化推荐及旅游领域问答系统提供数据支持，研究内容新颖并具有广泛的应用前景和学术价值。

Knowledge graph, with its powerful semantic processing ability and open organization ability, laid the foundation for the knowledge organization and intelligent application for big data. For the Internet travel industry, how to construct and apply a tourism knowledge graph has become a very valuable issue..In this project, we will make full use of our research experience in natural language processing and learn from available advanced research technologies to build a serviceable Mongolian knowledge graph oriented to the Inner Mongolian tourism domain. It focuses on following several aspects: (1) Distributed representation which conforms to the characteristics of Mongolian word structures; (2) Entity recognition method based on language model and attention mechanism will be researched; we apply the powerful semantic mining ability of the deep learning model to identify named entities with high-precision. (3) End-to-End model are researched to extract entity-attribute relation in tourism domain. (4) Constructing the prototype of Mongolian tourism knowledge graph based on the above-mentioned techniques, and further-processing the external structured data in order to integrate into the tourism knowledge graph as a supplement. (5) Translation and disambiguation techniques for short texts, like entities and labels in tourism knowledge graph. The research results of this project can provide data support for semantic query, personalized recommendation and QA system in tourism field. The research on this project is highly significance in both theory and practice.