通俗的理解,所谓推理,就是从现有的知识出发,运用逻辑思维能力,得出一些隐性的结论。具体到知识图谱中,所谓的知识推理,就是利用图谱中现有的知识(三元组),得到一些新的实体间的关系或者实体的属性(三元组)。设想一下,假如机器推理做的完备,一方面,它能够帮我们填充知识图谱中大量的空缺,使得知识更为完备;另一方面,对于知识问答、推荐系统等任务也有非常大的加成。
小编整理了以下几篇论文供大家参考:
1.RNNLogic: Learning Logic Rules for Reasoning on Knowledge Graphs
https://www.aminer.cn/pub/5f80379891e01119a5df7419/?f=cs
2.Query2box: Reasoning over Knowledge Graphs in Vector Space Using Box Embeddings
https://www.aminer.cn/pub/5e5e189a93d709897ce1e391/?f=cs
3.Reasoning on Knowledge Graphs with Debate Dynamics
https://www.aminer.cn/pub/5e5e190e93d709897ce4b5a1/?f=cs
4.Learning Collaborative Agents with Rule Guidance for Knowledge Graph Reasoning
https://www.aminer.cn/pub/5eb78919da5629cf24430384/?f=cs
5.Learning Reasoning Strategies in End-to-End Differentiable Proving
https://www.aminer.cn/pub/5ede0553e06a4c1b26a84060/?f=cs