Quantum-Inspired Keyword Search on Multi-Model Databases

TL;DR

This paper introduces a quantum-inspired keyword search method for multi-model databases, leveraging quantum formalism to improve search accuracy and efficiency over existing methods.

Contribution

It applies quantum physics concepts to vector space modeling for keyword search in multi-model databases, incorporating pattern mining and dimensionality reduction.

Findings

Outperforms state-of-the-art keyword search methods

Demonstrates improved accuracy in retrieving relevant results

Shows efficiency gains through dimensionality reduction

Abstract

With the rising applications implemented in different domains, it is inevitable to require databases to adopt corresponding appropriate data models to store and exchange data derived from various sources. To handle these data models in a single platform, the community of databases introduces a multi-model database. And many vendors are improving their products from supporting a single data model to being multi-model databases. Although this brings benefits, spending lots of enthusiasm to master one of the multi-model query languages for exploring a database is unfriendly to most users. Therefore, we study using keyword searches as an alternative way to explore and query multi-model databases. In this paper, we attempt to utilize quantum physics's probabilistic formalism to bring the problem into vector spaces and represent events (e.g., words) as subspaces. Then we employ a density matrix to encapsulate all the information over these subspaces and use density matrices to measure the divergence between query and candidate answers for finding top-\textit{k} the most relevant results. In this process, we propose using pattern mining to identify compounds for improving accuracy and using dimensionality reduction for reducing complexity. Finally, empirical experiments demonstrate the performance superiority of our approaches over the state-of-the-art approaches.