liveBook · Manning

Kg-based interpretability and discovery

As explained in Knowledge Graphs and LLMs in Action

A technique that uses knowledge-graph structure to make information interpretable and expose previously unrecognized connections or knowledge.

KG-based interpretability and discovery connects co-occurring entities through ontology paths. These paths explain why entities appear together, validate extracted entities, and reveal knowledge absent from the original sentence. section 7.7.3

A practical analysis pipeline:

  • Extract and disambiguate entities, then identify co-occurring pairs in the same sentence. section 7.7.3
  • Traverse ontology relationships to retrieve explanatory or newly informative paths. section 7.7.3
  • Query co-occurrence counts and sentence context to prioritize and ground findings. section 7.7.3

Medical entity analysis

AIDS and hepatitis co-occur in donation-risk sentences. SNOMED paths connect both to infectious disease and inflammatory liver disorder, explaining the wording; other paths expose hepatomegaly associated with AIDS and lupus hepatitis. Similar queries connect Zika virus with diseases, phenotypes, and related viruses.

How Knowledge Graphs and LLMs in Action covers it

How deep it goes
Provides a worked pipeline with ontology paths and Cypher queries, while deferring broader reliability issues to environment-dependent results.
Examples
Demonstrates AIDS–hepatitis paths, Zika co-occurrence queries, Dengue–Zika paths, and HPO disease–phenotype matching.
What it assumes
Assumes extracted and disambiguated entities connected to an ontology and knowledge graph.
What you can do after
Diagnose and explain entity relationships by querying co-occurrences, sentence context, and ontology paths.
How it connects
Connects named entity disambiguation, co-occurrence analysis, ontologies, SNOMED, HPO, and knowledge-graph search.
Cover of Knowledge Graphs and LLMs in Action
liveBookmanning.com
Read Knowledge Graphs and LLMs in Action on liveBook Get the book at manning.com

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Knowledge Graphs and LLMs in Action. The section links point to the parts of the book it draws on.