Skip to content
SemantaSemantaDocs

Explore PSPP knowledge graphs

Filter by paper and node type, then open related evidence.

Last updated

Knowledge graphs and PSPP

Open a knowledge base in Knowledge Assets and select Knowledge graph to view its PSPP graph.

PSPP organizes relationships among Processing, Structure, Properties and Performance. The graph may also contain device, layer, material, procedure and hyperedge nodes.

Select papers before exploring relationships

  1. Confirm graph building has completed for the document.
  2. Use the paper selector to focus on a paper or include all papers.
  3. Filter node types to focus on materials, layers, devices, performance or procedures.
  4. Inspect the loaded node and relationship counts at the bottom.

Start with one paper to understand its layers and processes, then widen the scope for comparison.

Use zoom and fit controls to navigate. For a dense graph, narrow the scope before zooming into the area of interest.

Open the node-type legend to identify categories. Read both endpoints and the relationship label to distinguish a material belonging to a layer, a process acting on a layer, and a device reporting a performance metric.

Open the paper behind a node or relationship

Select a node or relationship to inspect its details. Use the related-paper action to open the paper's extracted parameters and evidence chunks.

For example, start from a layer or process relationship, identify its device, and check the corresponding fabrication description. For performance relationships, also confirm the sample and test conditions.

The graph helps locate evidence. Verify values and conditions in the source text. A missing relationship can also reflect extraction coverage; inspect the document before concluding that the paper contains no relevant information.

Refresh and continue research

After adding documents or retrying processing, wait for graph building and refresh. If the graph is empty, check paper and node-type filters, then document status.

Continue with Knowledge-base retrieval in a conversation to summarize papers in natural language. For broader database relationships, explore the device hypergraph.