Knowledge graph

标签： Abstract

1. [IJCAI 2016] Text-Enhanced Representation Learning for Knowledge Graph

The following issues:

• low performance to model 1-to-N, N-to-1, N-to-N relations.
• limited performance due to the structure sparseness of the knowledge graph

Methods: taking advantage of the rich context information: expand the semantic structure of knowledge graph and each relation

充分利用上下文信息来扩展知识图谱的语义结构；
使用不同的表示对不同的head和tail实体来更好的处理上面的问题
实验：link prediction; Triple Classification

2.[ICIKM 2012] Knowledge Graph Construction for Organizations utilizing Social and Concept Relationships

• Problems: In the traditional approach, to achieve enterprise knowledge management is mostly based on handcraft rules, limitations in understanding the complex knowledge dynamics of an organization such as the effect of social connections in knowledge transfer and the interrelatedness of concepts.

• Methods: to improve the knowledge mapping in an organization based on a graph representation that utilizes the concept and social relationships

在图表示的基础上，利用概念和社会关系探索了提高了知识映射的机理

3.[Microsoft Technical Report], Towards a Probabilistic Taxonomy of Many Concepts

We present a universal, probabilistic ontology that is more comprehensive than any of the existing ontologies

we present details of how the core ontology is constructed, and how it models knowledge's inherent uncertainty, ambiguity, and inconsistency.
Potential applications: understanding user intent
针对目前以概念为中心的方法来增强网页文本的理解，目前存在的三个问题

• 概念空间有限
• 在已有的本体和分类学中，知识非黑即白，太清晰（实际上人类的语言是多种多样的，概念模糊，并且有时候不一致）
• 没有一个评分的机理；特别是在网页搜索中，对实体和概念的排序是非常重要的

我们表示了一个全面的概率化的本体，它比现有的本体更加的全面和深刻;该本体可以对包含的信息产生概率化的解释