Interactive Query Answering on Knowledge Graphs with Soft Entity Constraints
Daniel Daza, Alberto Bernardi, Luca Costabello, Christophe Gueret, Masoud Mansoury, Michael Cochez, Martijn Schut
Key claim
Soft constraints enhance query answering without disrupting original results.
This paper presents a new approach to query answering in knowledge graphs that incorporates soft constraints, allowing users to express preferences. The key result is that the proposed methods maintain robust performance while adding minimal overhead, enabling more flexible interactions with graph databases.
The introduction of soft constraints in query answering is a meaningful extension to existing methods.
The methods are evaluated against extended benchmarks, though details on baselines are limited.
Deep reliability assessment
The methodology supports the incorporation of soft constraints into query answering on knowledge graphs, but the effectiveness of the approach may be overclaimed without extensive real-world testing.
Reproducibility
Yes, the paper provides open source code and datasets. The code is available at https://github.com/dfdazac/nqr.
Discussion questions
- How do soft constraints compare to hard constraints in terms of computational efficiency and accuracy?
- What are the practical implications of using soft constraints for real-time query answering systems?
- What specific scenarios or datasets would falsify the effectiveness of the proposed methods?
Key figure
Figure 1 illustrates a query over a knowledge graph with hard constraints and the incorporation of soft constraints, shown as shaded clusters of entities.