The Semantic Knowledge Graph: A compact, auto-generated model for real-time traversal and ranking of any relationship within a domain

摘要

This paper describes a new kind of knowledge representation and mining systemwhich we are calling the Semantic Knowledge Graph. At its heart, the SemanticKnowledge Graph leverages an inverted index, along with a complementaryuninverted index, to represent nodes (terms) and edges (the documents withinintersecting postings lists for multiple terms/nodes). This provides a layer ofindirection between each pair of nodes and their corresponding edge, enablingedges to materialize dynamically from underlying corpus statistics. As aresult, any combination of nodes can have edges to any other nodes materializeand be scored to reveal latent relationships between the nodes. This providesnumerous benefits: the knowledge graph can be built automatically from areal-world corpus of data, new nodes - along with their combined edges - can beinstantly materialized from any arbitrary combination of preexisting nodes(using set operations), and a full model of the semantic relationships betweenall entities within a domain can be represented and dynamically traversed usinga highly compact representation of the graph. Such a system has widespreadapplications in areas as diverse as knowledge modeling and reasoning, naturallanguage processing, anomaly detection, data cleansing, semantic search,analytics, data classification, root cause analysis, and recommendationssystems. The main contribution of this paper is the introduction of a novelsystem - the Semantic Knowledge Graph - which is able to dynamically discoverand score interesting relationships between any arbitrary combination ofentities (words, phrases, or extracted concepts) through dynamicallymaterializing nodes and edges from a compact graphical representation builtautomatically from a corpus of data representative of a knowledge domain.