知识图谱旨在描述客观世界的概念、实体、事件及其之间的关系，是目前人工智能领域的研究热点之一。知识图谱补全基于知识图谱的现有三元组数据，建模预测真实存在的新三元组，是实现知识图谱演化的关键技术。现有的知识图谱补全方法缺乏人易理解的针对预测结果的解释机制，导致后续三元组真实性验证的人工成本很高。为此，项目拟采用人易理解的文本和逻辑信息解释知识图谱补全方法的预测结果，利用基于逻辑的溯因推理实现对应的解释机制。该机制的实现需要解决以下挑战性问题：（1）如何快速计算具有对应文本信息的溯因解释？（2）如何产生和应用不确定逻辑规则？（3）如何改进现有知识图谱补全方法，提高可解释三元组在前n个预测的未知三元组中的比例？项目将对上述问题进行深入研究，为知识图谱补全提供一种高效可靠的机制。项目涉猎知识表示学习和溯因推理等研究领域，能大力促进这些领域的理论发展，并为可解释人工智能时代的到来提供一些有价值的参考。

Knowledge graph is one of the research hotspots in the field of artificial intelligence (AI), which is designed to describe the concepts, entities, events and their relationships in the objective world. Knowledge graph completion is the key technology to evolvement of knowledge graphs, which learns a model and predicts the new realistic triples based on the existing triples in a knowledge graph. Existing methods for knowledge graph completion lack an intelligible mechanism for explaining the prediction results, making the subsequent manual verification of realistic triples very costly. Therefore, this project plans to exploit the intelligible information on text and logics to explain prediction results of a method for knowledge graph completion, and to apply logic-based abductive reasoning to realize such an explanation mechanism. The mechanism comes with the following challenging problems: (1) How to promptly compute abductive explanations that have corresponding text information? (2) How to generate and apply uncertain logic rules? (3) How to enhance existing methods for knowledge graph completion to increase the ratio of interpretable triples in the top-n predicted unknown triples? The project will study these problems carefully to provide an efficient and reliable mechanism for knowledge graph completion. The project involves knowledge representation learning, abductive reasoning and other research fields. It can greatly promote the theoretical development of these fields and can also provide some valuable references for the coming of explainable AI era.