Hellixia User Guide
Hellixia is BayesiaLab’s subject-matter assistant powered by Generative AI. It helps users create, enrich, and review semantic, causal, and probabilistic model assets inside BayesiaLab. This guide focuses on how to use Hellixia functions in BayesiaLab.
Before You Begin
Confirm the following before running a Hellixia function:
- API provider configuration: set up the provider, model, API key, and endpoint in Settings.
- Knowledge Files: add source files when the function should use documents, reports, papers, transcripts, or other source material.
- General Context: provide domain guidance when the prompt alone is not enough.
- Selections: select the required nodes, arcs, comments, classes, or Knowledge Files before opening functions that operate on existing model elements.
- Model selection: choose a model/provider appropriate for the data sensitivity, task complexity, and token budget.
- Privacy and data handling: check provider terms and internal policy before sending confidential material to an external model.
Key Concepts
a file added as detailed LLM context or source material.
user-provided contextual guidance for a Hellixia operation.
a learned graph representing semantic proximity.
a graph that emphasizes sequential relationships among concepts.
a graph of entities or concepts connected by semantic relationships.
a graph of concepts connected by causal relationships, before full Bayesian-network quantification.
a probabilistic causal graph with conditional probability tables.
a causal model focused on triggering events, controls, consequences, and mitigation actions.
Function Index
The index below mirrors the Hellixia menu structure and links every visible function to its guide page.
Setup
Provider, key, endpoint, and connection settings required before Hellixia can call Generative AI services.
| Function | Best for | Inputs | Outputs | Status |
|---|---|---|---|---|
| Settings | Initial setup; Provider configuration; Endpoint troubleshooting | API provider account; API key; optional custom endpoint | Configured Hellixia connection | Complete |
Prompt-to-network generation
Functions that start with a question or prompt and generate semantic, causal, or probabilistic model structure.
| Function | Best for | Inputs | Outputs | Status |
|---|---|---|---|---|
| Automatic Causal Network Generator | Rapid causal model prototyping; Risk-centric causal modeling; Starting from a domain question | Prompt or question; optional General Context; optional modeling constraints | Nodes; causal arcs; arc comments; causal effects; conditional probability tables | Complete |
| Automatic Semantic Network Generator | Domain exploration; Theme discovery; Semantic proximity mapping | Question; keywords; optional General Context | Dimensions; embeddings; dataset; learned Semantic Network | Complete |
Document-to-model workflows
Functions that use Knowledge Files as source material for structured semantic and causal analysis.
| Function | Best for | Inputs | Outputs | Status |
|---|---|---|---|---|
| Document Analysis | Reports; articles; transcripts; multi-document source material | Knowledge File(s); optional selected nodes or text; optional General Context | Structured graph artifacts depending on the selected generator | Complete |
| Semantic Flowchart Generator | Narratives; procedures; process descriptions | Knowledge File content; optional selected node name, long name, or comment | Semantic Flowchart | Complete |
| Causal Semantic Diagram Generator | Causal narratives; mechanism extraction; pre-Bayesian causal mapping | Knowledge File content; optional selected node name, long name, or comment | Causal Semantic Diagram | Complete |
| Knowledge Graph Generator | Entity and relation extraction; source-material exploration | Knowledge File | Knowledge Graph | Complete |
| Causal Network Generator | Document-derived Causal Bayesian Networks; risk-centric models | Knowledge File content; optional General Context | Nodes; causal links; mechanism comments; causal effects; conditional probability tables; root priors | Complete |
| Semantic Network Generator | Semantic proximity among document-derived concepts; theme discovery in source material | Knowledge File; selected keywords; optional General Context | Extracted dimensions; embeddings; learned Semantic Network | Complete |
| Doc-to-Node Generator | Multi-document corpora; document-level semantic maps | Selected Knowledge Files | Document nodes; comments containing file content; optional embeddings; optional Semantic Network | Complete |
Semantic and relationship analysis
Functions that identify concepts, relationships, flow, and meaning before or beside probabilistic modeling.
| Function | Best for | Inputs | Outputs | Status |
|---|---|---|---|---|
| Semantic Flowchart Generator | Process descriptions; workflow narratives; ordered explanations | Selected node names, long names, comments, or supplied text; optional General Context | Semantic Flowchart | Complete |
| Causal Semantic Diagram Generator | Causal mechanism sketches; pre-model causal diagrams; hypothesis generation | Selected node names, long names, comments, or supplied text; optional General Context | Causal Semantic Diagram | Complete |
| Knowledge Graph Generator | Entity relationship mapping; domain inventories; conceptual exploration | Prompt, selected text, selected nodes, or optional General Context | Knowledge Graph | Complete |
| Entity Relationship Finder | Explaining links; building relationship candidates; checking entity pairs | Selected nodes or entities; optional General Context | Proposed relationships; relationship labels; reviewable explanations | Complete |
| Verbalize Relationships | Stakeholder communication; documentation; relationship review | Selected arcs or relationships; optional General Context | Relationship descriptions; comments or text suitable for review | Complete |
Causal knowledge mining and priors
Functions that elicit causal directions, mechanisms, effects, priors, and reviewable causal hypotheses.
| Function | Best for | Inputs | Outputs | Status |
|---|---|---|---|---|
| Causal Network Generator | Causal model bootstrapping; risk-centric model generation | Prompt, selected nodes, or General Context | Nodes; causal arcs; causal effects; conditional probability tables; root priors | Complete |
| Causal Relationship Finder | Causal link screening; mechanism elicitation; existing-node causal mapping | Selected nodes; optional General Context | Proposed causal links; causal mechanisms; arc comments | Complete |
| Multi-Engine Causal Relationship Finder | Cross-checking causal hypotheses; reducing single-model dependence; expert review preparation | Selected nodes; multiple configured LLM engines; optional General Context | Engine-specific causal proposals; comparison material for analyst review | Complete |
| Causal Structural Priors | Constrained structure learning; causal orientation review; expert prior elicitation | Selected arcs or nodes; optional General Context | Structural priors; causal direction explanations; review material | Complete |
| Pairwise Causal Link | Two-variable causal checks; arc-level review; focused causal elicitation | Two selected variables; optional General Context | Proposed causal direction; mechanism explanation; optional arc comment | Complete |
| Root Priors Elicitor | Initial quantification; root-node priors; expert review before inference | Selected root nodes; state definitions; optional General Context | Prior probability distributions; elicitation comments | Complete |
| ICI Local Effects Elicitor | Noisy-OR style quantification; causal effect elicitation; parameter review | Selected ICI relationships; node states; optional General Context | Independent local effect estimates; reviewable parameter notes | Complete |
Authoring, enrichment, translation, and visualization
Functions that enrich nodes, arcs, comments, classes, embeddings, images, and translated labels.
| Function | Best for | Inputs | Outputs | Status |
|---|---|---|---|---|
| Dimension Elicitor | Keyword-based exploration; survey-item ideation; semantic scaffolding | Keywords; optional selected nodes; optional General Context | Dimension nodes; node comments; optional classes | Complete |
| Embedding Generator | Semantic proximity learning; clustering; semantic network construction | Selected nodes; node names, long names, or comments | Embedding variables or data suitable for learning and clustering | Complete |
| Comments | Model documentation; stakeholder communication; comment cleanup | Selected nodes or arcs; optional General Context | Node comments; arc comments; long names | Complete |
| Keyword-Based Node Comment Generator | Fast node documentation; keyword-grounded descriptions; semantic enrichment | Selected nodes; keywords; optional General Context | Node comments | Complete |
| Definition-Based Comment Generator | Concept definitions; node documentation; domain vocabulary cleanup | Selected nodes; optional definitions; optional General Context | Node comments | Complete |
| Node Comment Condenser | Dense comments; tooltip-ready summaries; readability cleanup | Selected nodes with comments | Condensed node comments | Complete |
| Node Comment Elaborator | Stakeholder-facing explanations; documentation enrichment; concept clarification | Selected nodes with comments; optional General Context | Elaborated node comments | Complete |
| Long Name Generator from Node Comment | Readable labels; model cleanup; semantic documentation | Selected nodes with comments | Node long names | Complete |
| Arc Comment Condenser | Arc-comment cleanup; mechanism summaries; readability review | Selected arcs with comments | Condensed arc comments | Complete |
| Arc Comment Elaborator | Mechanism explanations; causal review; stakeholder communication | Selected arcs with comments; optional General Context | Elaborated arc comments | Complete |
| Class Description Generator | Cluster interpretation; factor naming; class documentation | Selected classes or grouped nodes; optional General Context | Class descriptions; suggested class names | Complete |
| Semantic Variable Clustering | Large network organization; survey item grouping; semantic class creation | Selected nodes; names, long names, comments, or embeddings | Semantic clusters; classes; reviewable groupings | Complete |
| Image Generator | Visual model communication; node illustration; presentation support | Selected nodes; node names, long names, or comments; optional prompt guidance | Generated images associated with nodes | Complete |
| Translator | Multilingual models; localized documentation; cross-language review | Selected nodes; source and target languages; names, states, or comments | Translated labels, state names, and comments | Complete |
Workflow Guides
I Have a Question or Prompt
Start with Automatic Causal Network Generator or Automatic Semantic Network Generator, then review the proposed structure.
I Have Documents
Use Document Analysis to turn Knowledge Files into graph artifacts. Common next steps include Document Analysis > Causal Network Generator and Document Analysis > Doc-to-Node Generator.
I Already Have Nodes
Enrich selected nodes with Dimension Elicitor, Embedding Generator, Semantic Variable Clustering, or Comments.
I Need Causal Orientation or Priors
Use Causal Relationship Finder, Causal Structural Priors, Root Priors Elicitor, or ICI Local Effects Elicitor to create reviewable hypotheses.
I Need to Communicate or Enrich a Model
Use Verbalize Relationships, Class Description Generator, Image Generator, or Translator.
Troubleshooting
| Symptom | What to check |
|---|---|
| API key not configured | Open Settings and verify the provider, key, endpoint, and model access. |
| No nodes or arcs selected | Select the required graph elements before opening functions that operate on existing nodes or arcs. |
| No Knowledge File loaded | Add one or more Knowledge Files before using Document Analysis functions. |
| Context is too broad | Narrow the prompt, add General Context, or split the task into smaller runs. |
| Output is too dense | Reduce requested scope, constrain the number of concepts, or condense generated comments after review. |
| Causal directions look weak | Treat them as hypotheses, compare against source material, and use expert review before creating structural priors. |
| Provider or rate-limit errors | Check provider status, API limits, billing, model availability, and endpoint configuration. |
Examples and Tutorials
- Hellixia ExamplesBrowse examples by domain, input type, output type, and workflow.
- Webinar: Discovering Complex Causal StructuresSee Hellixia causal analysis functions in a webinar workflow.
- Webinar: Harnessing Hellixia InnovationsReview Bayesian-network construction workflows with Hellixia.
- Webinar: From Narratives to NetworksFollow a narrative-text-to-causal-network workflow.