| 4d5acc33f00c52f9… | For provenance, naive RAG is usually document-level, whereas the Pyrana Context Engine provides claim-level provenance with supporting evidence so that answers can be traced to specific claims. | active | contextual | yes | |
| 8cd4607fd2e8a3af… | In the Pyrana versus naive RAG comparison, naive RAG uses text chunks as its retrieval unit, whereas the Pyrana Context Engine uses atomic CXUs so that retrieval operates on finer-grained knowledge objects. | active | contextual | yes | |
| e7119b1d68f8a84e… | The fused result is a ranked answer with provenance and policy priority, and this matters because it improves trust and usability. | active | contextual | yes | |
| 4135ae0cf23af02e… | In the semantic lane, the example result is a supporting paragraph on attributable record changes, and this lane matters because it adds explanatory evidence and nearby context. | active | contextual | yes | |
| 5f86469f0e920d3c… | In the scoped graph lane, the example result is 'Packaging Line 3 operator review procedure,' and this lane matters because it keeps the answer within the correct operational scope. | active | contextual | yes | |
| 4642862d52d20357… | In the template graph lane, the example result is 'Audit trail required for EBR changes,' and this lane matters because it provides deterministic policy-critical retrieval. | active | contextual | yes | |
| d7267be6b2d5ef41… | The resulting answer set is a governed bundle of requirements, procedures, and supporting context tied to the relevant entities rather than merely a collection of similar text. | active | contextual | yes | |
| 24739b8d7e838280… | The Context Engine fuses results, promotes mandatory high-priority claims, and suppresses claims that should not surface in the active mode so that the final answer reflects policy priority and mode constraints. | active | contextual | yes | |
| dbe06e73c2f6b2f1… | The Context Engine runs semantic retrieval in parallel during the walkthrough so that nearby supporting material is captured alongside graph-derived governance content. | active | contextual | yes | |
| f95d48dbe0837b8a… | The Context Engine traverses from the line and system entities to governing CXUs and related procedural claims so that it can collect governance knowledge connected to the identified entities. | active | contextual | yes | |
| 5ad46953b908ca8b… | The Context Engine identifies that the example question is governance-oriented and triggers a graph-led retrieval strategy so that the search method matches the governance intent of the request. | active | contextual | yes | |
| 651bbd0541f926b9… | The Context Engine first resolves the relevant entities as the Electronic Batch Record System and Packaging Line 3 when answering the example governance question so that retrieval is anchored to the correct objects. | active | contextual | yes | |
| 883ea1285fccffb2… | A naive RAG system would retrieve paragraphs mentioning terms like batch record, change, or Packaging Line 3 and leave the model to infer which statements are mandatory requirements versus local operating notes. | active | contextual | yes | |
| 1bf0d95680b1efe3… | In the retrieval walkthrough, the example user question asks what governs changes to electronic batch records on Packaging Line 3, establishing a governance-oriented retrieval scenario tied to a specific system and line. | active | contextual | yes | |
| 45f99003c56f6250… | Semantic vector retrieval runs in parallel with the selected graph strategy and its results are fused using weighted ranking so that structural relevance and semantic relevance reinforce each other rather than replacing one another. | active | contextual | yes | |
| be23a964f7babafe… | The system can generate free-form graph search behavior dynamically for exploratory or less structured requests so that retrieval extends beyond the template library while still operating on the governed graph rather than raw text alone. | active | contextual | yes | |
| 2b95ee5c63bf9c80… | A naive RAG system might retrieve paragraphs mentioning 'batch record,' 'change,' or 'Packaging Line 3' and leave the result at keyword-level similarity rather than governed graph-based reasoning. | active | contextual | yes | |
| 70bc825165cbe5b2… | The retrieval walkthrough uses the question 'What governs changes to electronic batch records on Packaging Line 3?' as an example for illustrating the retrieval process in a pharma operations context. | active | contextual | yes | |
| b2e170f0ae086e62… | After graph and semantic retrieval execute, the system fuses results using weighted ranking so that structural relevance and semantic relevance reinforce rather than replace each other. | active | contextual | yes | |
| 3efe1e716e32f6ca… | For exploratory or less structured requests, the system uses Free-form graph search by generating graph search behavior dynamically so that retrieval can extend beyond the template library while still operating on the governed graph. | active | contextual | yes | |
| 44a4028a7d68fec6… | Scoped graph traversal is useful when the answer must be limited to a specific policy regime, department, region, or tenant context so that retrieval honors explicit scope restrictions. | active | contextual | yes | |
| a686b38479a369aa… | When a request includes set, domain, or relationship constraints, the system uses Scoped graph traversal by constructing a scoped graph query that respects those boundaries so that answers remain within the required context. | active | contextual | yes | |
| f7bf7abe653c5912… | Template graph retrieval handles known query intents including governance, dependency, lineage, and requirements lookups so that recurring enterprise questions can be answered through bounded deterministic traversals. | active | contextual | yes | |
| 7839d3d9b8f93fb2… | For common enterprise question patterns, the system uses pre-defined graph templates as Template graph retrieval so that known query intents can be handled with deterministic structure and bounded traversal. | active | contextual | yes | |
| 5703980518020c3f… | Retrieval is the system's most differentiated subsystem because Pyrana does not treat every question as a similarity search over text and instead selects among three complementary retrieval strategies before fusing their outputs with semantic retrieval. | active | contextual | yes | |
| 4a677f149c690fdc… | The graph is the structural model that makes scoped, explainable knowledge retrieval possible rather than merely serving as decoration around vector search. | active | contextual | yes | |
| cdaa7ed938efeef3… | In the example subgraph, the system can distinguish a regulated requirement from a local operational note, connect both to the same entity, and determine which should dominate under governance so that governed prioritization is possible. | active | contextual | yes | |
| bd57ab09a65c35a9… | The resulting corpus can be traversed by policy, by entity, by lineage, and by semantic similarity rather than by similarity alone so that retrieval uses structural and semantic paths together. | active | contextual | yes | |
| 46884dbad302ef84… | Scope relationships express membership in sets, domains, or operating boundaries so that the graph can enforce contextual limits on retrieval and interpretation. | active | contextual | yes | |
| 9262c8387341f6f0… | Knowledge relationships express dependency, contradiction, specification, or governance so that the graph can represent substantive interactions among knowledge units. | active | contextual | yes | |
| bb5741c1a6668ac6… | Context relationships link a CXU to the entity or entities it concerns so that knowledge units are explicitly attached to their relevant real-world subjects. | active | contextual | yes | |
| fa5ab8a74663afed… | The graph relationship taxonomy distinguishes three classes of relationships: context relationships, knowledge relationships, and scope relationships, so that graph structure captures different kinds of linkage explicitly. | active | contextual | yes | |
| 1f94a6807d32450c… | Sets create scoped subgraphs that allow the system to answer the same question differently when different domains, tenants, or policy frames are active so that retrieval is context-sensitive. | active | contextual | yes | |
| 69bb9e37b54f5608… | A CXU may belong to multiple sets, including universal sets and sets representing regulatory regimes, business units, customer environments, regions, or internal operating boundaries, so that the same knowledge unit can participate in multiple scoped contexts. | active | contextual | yes | |
| 2522f4aba4da595c… | A naive RAG system might retrieve paragraphs mentioning 'batch record,' 'change,' or 'Packaging Line 3' and leave the user with text matches rather than governed structural resolution. | active | contextual | yes | |
| e6fbbf2e2b6873d0… | The retrieval walkthrough uses the question 'What governs changes to electronic batch records on Packaging Line 3?' as an example so that the architecture can be explained through a concrete governance query. | active | contextual | yes | |
| dc283ff370721784… | The system fuses retrieval results using weighted ranking so that structural relevance and semantic relevance reinforce rather than replace each other. | active | contextual | yes | |
| 5616b1ea6f8db7f7… | Semantic vector retrieval runs in parallel with whichever graph strategy is selected so that semantic matching complements graph-based retrieval during the same query execution. | active | contextual | yes | |
| 396cdfb085ead13e… | Free-form graph search generates graph search behavior dynamically for exploratory or less structured requests so that the architecture can reach beyond the template library while still using the governed graph. | active | contextual | yes | |
| 2443c55410fb93e9… | Scoped graph traversal is useful when an answer must be limited to a specific policy regime, department, region, or tenant context so that retrieval remains within the required operating scope. | active | contextual | yes | |
| b5bebd39b60bef03… | Scoped graph traversal constructs a scoped graph query when a request includes set, domain, or relationship constraints so that the query respects those boundaries during retrieval. | active | contextual | yes | |
| 4c9e8cf6daf9b2b3… | Template graph retrieval uses pre-defined graph templates for common enterprise question patterns so that known query intents can be handled with deterministic structure and bounded traversal. | active | contextual | yes | |
| 792698681367a6cc… | Pyrana selects among three complementary retrieval strategies and then fuses their outputs with semantic retrieval so that structural and semantic evidence are combined in the final answer set. | active | contextual | yes | |
| 062f622d61120540… | Retrieval is the system's most differentiated subsystem, and Pyrana does not treat every question as a similarity search over text so that retrieval can use multiple specialized strategies instead of text similarity alone. | active | contextual | yes | |
| 9cbb16f37a03aaea… | The graph is not merely a decoration around vector search; it is the structural model that makes scoped, explainable knowledge retrieval possible so that retrieval can be governed and interpretable. | active | contextual | yes | |
| 5b544c0f107572cd… | In the example subgraph, the system can distinguish a regulated requirement from a local operational note, connect both to the same entity, and decide which one should dominate under governance so that governed knowledge takes precedence appropriately. | active | contextual | yes | |
| 5dc823be51c8b17b… | The example subgraph is a stylized illustration of how a small corpus might look for a pharma operations question so that readers can visualize the graph model in a domain-specific scenario. | active | contextual | yes | |
| 634c7cdd2a93391d… | The resulting corpus can be traversed by policy, by entity, by lineage, and by semantic similarity rather than by similarity alone so that retrieval uses graph structure in addition to text resemblance. | active | contextual | yes | |
| 6cd57cd409112f17… | Scope relationships express membership in sets, domains, or operating boundaries so that the graph can enforce contextual limits on retrieval and governance. | active | contextual | yes | |
| 14102908b532fca9… | Knowledge relationships include dependency, contradiction, specification, or governance so that the graph can represent how knowledge units interact or constrain one another. | active | contextual | yes | |