Skip to content

How to View Results

This guide covers exploring analysis results and provenance in OntExtract.

Overview

After processing documents, OntExtract provides several ways to explore results:

  • Processing artifacts for individual documents
  • Experiment-level analysis and timelines
  • Provenance graphs showing processing history
  • Export options for further analysis

Viewing Processing Results

Processing results are accessed through experiment result pages, which aggregate findings across all documents in an experiment.

Access Results

  1. Go to Experiments > Select an experiment
  2. Expand the View Results section
  3. Click a result type (Definitions, Entities, Embeddings, Segments, or Temporal)

Artifact Types

Type Contents
Text Segments Paragraphs or sentences with character positions
Entities Named entities with types and confidence scores
Embeddings Vector representations (viewable as similarity scores)
Temporal Expressions Dates, periods, and durations found in text
Definitions Extracted concept definitions with pattern types

Definition Results

Definition artifacts include:

  • Term - The word or phrase being defined
  • Definition text - The extracted definition content
  • Pattern type - How it was detected (explicit_definition, explicit_reference, meaning, copula, acronym, also_known_as, ie_explanation, appositive)
  • Confidence - Score from 0.65-0.90 depending on pattern reliability
  • Source badge - "Pattern" (default) or "ZeroShot" (if enabled via ENABLE_ZERO_SHOT_DEFINITIONS=true)

Acronym definitions show strict validation: "IRA (Information Retrieval Agent)" passes because expansion letters match the acronym.

Note: Definition extraction works best on documents that explicitly define terminology, such as glossaries, textbook introductions, or standards documents. Research papers that use but do not define terms may return few or no results.

Artifact Details

Each artifact shows:

  • Operation type and timestamp
  • Source document reference
  • Processing parameters used
  • Tool version (for reproducibility)
  • Structured results

Experiment Results

Result Type Views

From the experiment page, use the View Results buttons to explore specific result types:

Button Shows
Definitions Extracted term definitions across all documents
Entities Named entities (people, organizations, places)
Embeddings Semantic similarity search interface
Segments Text segments with positions
Temporal Dates, periods, and temporal markers

Each view aggregates results from all documents in the experiment, enabling comparison of findings across the corpus.

Definition Results

Entity Results

Temporal Results

Timeline View

For temporal evolution experiments:

  1. Go to Experiments > Select experiment
  2. View the timeline showing documents by period
  3. See term usage patterns across time

LLM Synthesis Results

When using LLM Orchestration:

  1. Go to Document Pipeline
  2. Click View Results on the orchestration status banner (shown when orchestration completes)
  3. See cross-document patterns and term cards

Note: The synthesis organizes findings but does not interpret them. Analytical conclusions remain with the researcher.

Provenance Tracking

OntExtract records complete PROV-O provenance for all operations.

Viewing Provenance

Each artifact links to its provenance chain showing:

  • wasDerivedFrom - Source document(s)
  • wasGeneratedBy - Processing activity
  • wasAssociatedWith - Tool and version used
  • used - Input entities consumed

Why Provenance Matters

  • Reproducibility - Recreate exact processing conditions
  • Transparency - Understand how results were generated
  • Debugging - Trace unexpected results to their source
  • Scholarly citation - Document analytical methodology

Export Options

Export Formats

Format Use Case
CSV Tabular data for spreadsheets
JSON Structured data for programming

What Can Be Exported

  • Processing artifacts with metadata
  • Entity extraction results
  • Segment text with positions
  • Provenance records

Planned Features

The following results features are under consideration for future releases:

  • Interactive timeline visualizations
  • Semantic drift graphs
  • Comparative period analysis
  • Formatted report export (PDF)

Troubleshooting

No results showing

  • Verify processing operations completed
  • Check the experiment's View Results section
  • Review experiment status

Missing artifacts

  • Ensure the operation was selected during processing
  • Check for errors in the processing log
  • Verify document had extractable content