Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Structured Query Construction via Knowledge Graph Embedding

Ruijie Wang, Meng Wang, Jun Liu, Michael Cochez, Stefan Decker

TL;DR

This work addresses constructing graph-structured queries from natural language questions over large knowledge graphs by introducing a GL-KG based embedding framework. The offline embedding learns representations that capture generalized local contexts and preserve a translation-like structure, enabling online phrase mapping, structure computing, and query generation to assemble accurate queries efficiently. Empirical results on DBpedia and QALD-6 demonstrate competitive effectiveness and superior end-to-end efficiency compared with baselines, while failure analyses highlight the dominant impact of phrase mapping quality. The approach offers a scalable pathway for semantic question answering over large KGs by tightly integrating embedding-based structure inference with graph-structured query generation.

Abstract

In order to facilitate the accesses of general users to knowledge graphs, an increasing effort is being exerted to construct graph-structured queries of given natural language questions. At the core of the construction is to deduce the structure of the target query and determine the vertices/edges which constitute the query. Existing query construction methods rely on question understanding and conventional graph-based algorithms which lead to inefficient and degraded performances facing complex natural language questions over knowledge graphs with large scales. In this paper, we focus on this problem and propose a novel framework standing on recent knowledge graph embedding techniques. Our framework first encodes the underlying knowledge graph into a low-dimensional embedding space by leveraging generalized local knowledge graphs. Given a natural language question, the learned embedding representations of the knowledge graph are utilized to compute the query structure and assemble vertices/edges into the target query. Extensive experiments were conducted on the benchmark dataset, and the results demonstrate that our framework outperforms state-of-the-art baseline models regarding effectiveness and efficiency.

Structured Query Construction via Knowledge Graph Embedding

TL;DR

This work addresses constructing graph-structured queries from natural language questions over large knowledge graphs by introducing a GL-KG based embedding framework. The offline embedding learns representations that capture generalized local contexts and preserve a translation-like structure, enabling online phrase mapping, structure computing, and query generation to assemble accurate queries efficiently. Empirical results on DBpedia and QALD-6 demonstrate competitive effectiveness and superior end-to-end efficiency compared with baselines, while failure analyses highlight the dominant impact of phrase mapping quality. The approach offers a scalable pathway for semantic question answering over large KGs by tightly integrating embedding-based structure inference with graph-structured query generation.

Abstract

In order to facilitate the accesses of general users to knowledge graphs, an increasing effort is being exerted to construct graph-structured queries of given natural language questions. At the core of the construction is to deduce the structure of the target query and determine the vertices/edges which constitute the query. Existing query construction methods rely on question understanding and conventional graph-based algorithms which lead to inefficient and degraded performances facing complex natural language questions over knowledge graphs with large scales. In this paper, we focus on this problem and propose a novel framework standing on recent knowledge graph embedding techniques. Our framework first encodes the underlying knowledge graph into a low-dimensional embedding space by leveraging generalized local knowledge graphs. Given a natural language question, the learned embedding representations of the knowledge graph are utilized to compute the query structure and assemble vertices/edges into the target query. Extensive experiments were conducted on the benchmark dataset, and the results demonstrate that our framework outperforms state-of-the-art baseline models regarding effectiveness and efficiency.

Paper Structure

This paper contains 20 sections, 18 equations, 15 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Knowledge Graph
  4. KG Embedding Techniques
  5. Proposed Framework
  6. Overview of Proposed Framework
  7. KG Embedding Learning
  8. Phrase Mapping
  9. Structure Computing
  10. Query Generation
  11. Time Complexity Analysis
  12. Discussion
  13. Experiments
  14. Dataset
  15. Effectiveness Evaluation
  16. ...and 5 more sections

Figures (15)

  • Figure 1: The general query construction process of the example NLQ.
  • Figure 2: An example of the semantic gap between NLQs and KGs.
  • Figure 3: Illustrations of the translation mechanisms of TransE and TransR.
  • Figure 4: An overview of our framework.
  • Figure 5: The L-KG and GL-KG of the entity vertex Tim Burton.
  • ...and 10 more figures

Theorems & Definitions (3)

  • Definition 3.1: Local Knowledge Graph
  • Definition 3.2: Generalized Local Knowledge Graph
  • Definition 3.3: Structure Matrix