Visual Interfaces for Querying Semantic Knowledge Graphs

You are reading version 1 of this direction, frozen 2026-10-04. The latest version is version 2.

Version 1 of 2120 papers151 conceptsfrozen 2026-10-04

In many specialized fields—from oncology and ecology to human trafficking investigation—the most valuable data is locked behind complex, high-dimensional structures like Resource Description Framework (RDF) triples or billion-node knowledge graphs. While these formats are powerful for machines, they create a "semantic gap" for the domain experts who need them. A doctor might know exactly what patient trajectory they are looking for, but they likely do not know how to write a 20-line SPARQL or Cypher query to find it.

Research into visual query interfaces aims to bridge this gap. By treating the user interface as a translation layer, researchers are moving away from code-heavy interactions toward "graphical inquiry," where sketches, facets, and direct manipulation are automatically compiled into formal database logic.

The Translation Layer

The core intuition behind this research is that a graphical interface can serve as a functional compiler. Instead of requiring a user to master the syntax of a language like MOQL or SQL, systems like VisualMOQL by Vincent Oria and colleagues allow users to draw spatial configurations that the system then translates into object-oriented queries.

This mechanism is particularly effective for temporal and sequential data. In Portinari, Sagar Sen and their team allow epidemiologists to construct patient "trajectories" on a drag-and-drop canvas. The system takes these visual event sequences and generates the underlying graph queries necessary to search through millions of screening records. By offloading the "how" of the query to the software, the expert can focus entirely on the "what."

Abstracting Schema Complexity

A major barrier to data access is not just the query language, but the schema itself. Real-world databases often involve "shredded" data—information spread across dozens of tables or complex ontologies.

The Optique platform, described by Evgeny Kharlamov, Ahmet Soylu, and others, addresses this through Ontology-Based Data Access (OBDA). It presents users with familiar business or scientific terms rather than table names. When a user navigates these concepts in a graph-based interface, the system uses "mappings" to translate those high-level clicks into optimized SQL.

Similarly, VizBrick by SangKeun Lee uses visual metaphors to help building scientists author semantic metadata. By providing class suggestions and rule-based relationship mapping, it allows experts to build valid RDF models without ever seeing a line of Turtle or JSON-LD code.

Query by Sketch and Example

Sometimes, a user’s query is "fuzzy"—they are looking for a trend or a shape rather than a specific value. This has led to the development of "Query by Sketch" interfaces.

  • Pattern Matching: Tarique Siddiqui introduced Zenvisage, which allows users to literally sketch a desired trend (like a "hockey-stick" growth curve). The system then performs a similarity search across massive datasets to find visualizations that match that visual intuition.
  • Structural Motifs: In neuroscience, Jakob Troidl and colleagues developed Vimo, where researchers sketch neuronal connectivity "motifs." These sketches are translated into graph queries that scan petascale brain datasets for recurring structural building blocks.

This approach works because it leverages human pattern recognition—our ability to "know it when we see it"—and maps it to the precision of graph theory.

Managing High Dimensionality with Facets

When dealing with billion-node graphs, even a visual search can become overwhelming. To manage this, researchers like Jason Slepicka and Amandeep Singh have applied faceted browsing to knowledge graphs.

In the DIG system, used to combat human trafficking, complex graph traversals are transformed into a series of simple attribute filters. This makes exploring a 1.4-billion-node graph feel as intuitive as filtering for shoes on an e-commerce site. By shifting the cognitive load from "constructing a path" to "filtering a result set," these interfaces democratize access to high-stakes investigative data.

The Future: Closing the Loop

While much of the work has focused on retrieving data, newer systems like CleanGraph by Tyler Bikaun are exploring how direct manipulation can be used to refine and complete knowledge graphs. The goal is a bidirectional flow: a visual interface that not only lets you ask questions of the data but also allows you to correct and expand it through intuitive mouse and keyboard actions, effectively turning domain experts into "human-in-the-loop" data engineers.

Papers in this version

Showing 60 of 120 papers.

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 75 papers
  2. Version 1 · 2026-10-04 · 120 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_visual_query_formulation_b2b621bd,
  title = {visual query formulation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/visual-query-formulation/rdr_b2b621bd9de547c5bfe4671533bb9bbd},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_visual_interfaces_for_querying_semantic__b2b621bd_v1,
  title = {Visual Interfaces for Querying Semantic Knowledge Graphs},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/visual-interfaces-for-querying-semantic-knowledge-graphs/rdr_b2b621bd9de547c5bfe4671533bb9bbd/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}