This paper introduces Explicit Semantic Ranking (ESR),
a new ranking technique to connect query and documents
using semantic information from a knowledge graph. We
First build an academic knowledge graph using S2’s corpus
and Freebase. The knowledge graph includes concept en-
tities, their descriptions, context correlations, relationships
with authors and venues, and embeddings trained from the
graph structure. We apply this knowledge graph and em-
beddings to our ranking task. Queries and documents are
represented by entities in the knowledge graph, providing
`smart phrasing’ for ranking. Semantic relatedness between
query and document entities is computed in the embedding
space, which provides a soft matching between related enti-
ties. ESR uses a two-stage pooling to generalize these entity-
based matches to query-document ranking features and uses
a learning to rank model to combine them.
本文引入显式语义排序（Explicit Semantic Ranking，ESR），这是一种利用知识图中的语义信息连接查询和文档的新排序技术。 我们首先使用S2的语料库和Freebase构建学术知识图。 知识图包括概念实体，它们的描述，上下文相关性，与作者和场所的关系以及从图结构训练的嵌入。 我们应用这个知识图并嵌入到我们的排名任务中。 查询和文档由知识图中的实体表示，为排序提供“智能表述”。 在嵌入空间中计算查询和文档实体之间的语义相关性，这提供了相关实体之间的软匹配。 ESR使用两阶段池来将这些基于实体的匹配概括为查询文档排名特征，并使用学习来对模型进行排名以将它们组合。

2.Related Work

之前的工作更多关注于学术图的分析而不是特别的排名。