Looking at Vector Space and Language Models for IR using Density Matrices
This work addresses the challenge of unifying different IR models for researchers and practitioners, though it appears incremental as it builds on existing frameworks without demonstrating broad SOTA results.
The paper tackled the problem of integrating Vector Space Models (VSM) and Language Models (LM) for Information Retrieval (IR) by using density matrices from Quantum Theory as a unified representation, showing that this approach can combine the strengths of both models to enable new retrieval methods.
In this work, we conduct a joint analysis of both Vector Space and Language Models for IR using the mathematical framework of Quantum Theory. We shed light on how both models allocate the space of density matrices. A density matrix is shown to be a general representational tool capable of leveraging capabilities of both VSM and LM representations thus paving the way for a new generation of retrieval models. We analyze the possible implications suggested by our findings.