5W1H+Which: Context-Valid Semantic Indexing with Progressive Ontology Binding

Yaxiao Liu; Pengbo Liu; Yiwen Liu; Yihua Guan; Jiaxing Song

cs.AISep 28, 2026

5W1H+Which: Context-Valid Semantic Indexing with Progressive Ontology Binding

Authors: Yaxiao Liu, Pengbo Liu, Yiwen Liu, Yihua Guan, Jiaxing Song

Organizations: PwC China AI Center · Tsinghua University

Abstract

Transforming raw data into queryable knowledge requires both early extraction of reusable information and explicit types, relations, and applicability conditions for particular tasks. If indexing selects content too early around a single business schema, later tasks may be unable to use information that was omitted. If the index retains only open-ended text, however, rule-based reasoning lacks checkable premises. We propose 5W1H+Which, a semantic indexing design that separates content extraction from ontology binding. The 5W1H questions organize source-grounded content units; Which points to versioned ontology elements and records mapping relations, scope, and validation status. Time, location, system environment, and participant roles are not merely retrieval labels: together, they constrain the contexts in which facts, bindings, and rules apply. Unbound content remains searchable, while bound content enters a formal reasoning path only after premise checks. The method further distinguishes business valid time, system knowledge time, and operational traces, and uses dependency records to support binding revalidation and the maintenance of derived conclusions. A worked example of migration from an on-premises server to a cloud environment illustrates the different treatment of world-state changes, ontology-version changes, and changes in rule applicability. We formulate three groups of falsifiable hypotheses concerning cross-task evidence coverage, control of contextual misuse, and incremental update cost. The planned evaluation includes a strong typed fact-graph baseline with the same evidence, temporal information, and budget, to test whether benefits arise from 5W1H organization, deferred binding, or additional information and engineering effort. The contribution is a testable indexing mechanism, not a claim to a new universal ontology or a demonstrated performance advantage.

Figures & tables

Prior work	Relevant mechanism	What our evaluation must isolate
SocraticKG; XPEventCore [R1, R2]	5W1H extraction and event modeling	Organization versus extra facts or better extraction prompts
OaK; OM4OV [R3, R5]	Dynamic ontology and mapping/version changes	Independent content retention and binding revalidation under equal information
DimMem [R12]	Dimensional memory and constrained retrieval	Additional value beyond a dimension-aware fact graph
RuleMem; InMind [R13, R14]	Rule reuse and implicit applicability	Candidate discovery versus validated admission; useful association versus scope leakage
Recuris [R15]	Experience, working state, and verified updates	Keep skill learning outside the core indexing claim
Fortunate Recall [R16]	Typed and generic lifecycle policies	Category labels versus metadata, routing, and enforced policies

Table 1

Direction	Content-layer representation	Meanings that must remain distinct
What	Objects, events, relations, or claims with exact payloads	A document topic is not a specific claim that can be assessed as true or false.
Who	References to participants with explicit roles	Author, operator, responsible party, approver, and current user are different roles.
When	Instants, intervals, precision, time zone, and the basis for resolving relative times	Event occurrence, planned validity, ingestion, and publication are different times.
Where	Typed physical, organizational, and computational scopes	A server room, department, cluster, and location within a source are not one kind of place.
Why	Source-stated purposes and reasons, or separately recorded causal hypotheses	A stated reason is not a demonstrated causal mechanism.
How	Methods, tools, procedures, parameters, and execution references	A planned method is not necessarily the method actually executed.

Table 2

Input	Source and system-record time	Business meaning	Index treatment
A	Verification record, 2026-09-01	S runs on on-premises device L on that date.	Record the evidenced state for that date; continued validity requires a separately declared persistence policy.
B	Planning record, 2026-09-10	S is scheduled to migrate to the cloud on 2027-01-01.	Mark as a plan; do not change actual deployment facts in advance.
C	Completion record ingested only on 2027-01-05	Migration is verified as effective on 2027-01-03; the former production instance is retired.	Preserve both actual effective time and late-arriving knowledge time.
D	Responsibility rule verified separately from C	After migration, team T manages the application layer; infrastructure responsibility follows the contract.	Do not infer all responsibilities automatically from the CloudDeployment type.

Table 3

Test	Hold fixed	Vary	Primary outcome
Extraction	Sources, model, token and storage budgets	5W1H prompt versus open atomic-fact prompt	Evidence coverage and qualifier retention
Binding	Same extracted content and mapper	Early schema filtering versus retained unbound content	New-task evidence coverage at matched precision
Categories	Same facts, lifecycle metadata, gates and routing	Named categories versus a flat typed representation	Incremental calibration and coverage; report equivalence margins if tested
Admission	Same facts, time/scope metadata and access controls	Stored metadata versus enforced semantic gates	Contextual misuse at matched answer coverage
Applicability	Same evidence and query templates	Positive indirect-use cases versus scope-mismatched negative pairs	Useful activation and false activation
Maintenance	Same event stream, semantics and measured cost accounting	Full recomputation versus incremental repair	Withdraw/add/retain agreement and total cost

Table 4

Explore similar work

May 7, 2026cs.DB

Anatomy of a Query: W5H Dimensions and FAR Patterns for Text-to-SQL Evaluation

Natural language interfaces to databases have gained popularity, yet the theoretical foundations for evaluating and designing these systems remain underdeveloped. We present QUEST (Query Understanding Evaluation through Semantic Translation), a framework resting on two independently motivated components: the FAR structural invariant, which holds that every well-formed query reduces to Filter, Aggregate, and Return operations; and the W5H dimensional framework, which holds that all filtering criteria map to six semantic dimensions (Who, What, Where, When, Why, and How). Validated across five text-to-SQL datasets (n = 120,464), FAR conformance is universal across all domains and schema types, while W5H dimensional profiles vary substantially. Healthcare queries are strongly concentrated in temporal (WHEN: 80.4%) and person-centric (WHO: 73.0%) dimensions far exceeding general-domain benchmarks, and causal (WHY) and mechanistic (HOW) reasoning are near-zero everywhere, with apparent HOW exceptions reflecting quantitative aggregation rather than genuine procedural reasoning. These results identify a frontier that must be crossed for genuine machine reasoning over structured data.

Vicki Stover Hertzberg, Eduardo Valverde, Joyce C. Ho

Emory University · Georgia Institute of Technology, Atlanta, USA

Text-To-Sql Queries Table Reasoning Datasets Reasoning Skills

Ask PDF

Aug 6, 2026cs.CL

Factorized Hypothesis Search for Evidence-to-Taxonomy Retrieval

Large-taxonomy retrieval often assumes that the input already expresses the target concept. In many settings, however, the input is indirect evidence, such as a table cell whose meaning depends on its row, column, datatype, and context. We call this mismatch the retrieval readiness gap. Our analysis shows that the current index retrieves the target reliably when its semantics are explicit, while raw evidence often leaves it deep in the ranking. We propose Factorized Hypothesis Search (FHS), which maintains multiple partial interpretations over named semantic dimensions. These hypotheses support structured query rendering, multi-hypothesis retrieval, and dimension-level candidate verification. On both financial taxonomy tagging and CodiEsp clinical coding tasks, FHS achieves the best Recall@1, MRR, and final accuracy among the non-oracle methods. Replacing the factorized hypothesis path with a free-text ensemble causes the largest drop in head-ranking performance, while sequential refinement provides no additional gain over FHS's strong parallel first round.

Linhai Ma, Ethan F. Wei, Xueqing Peng +3

The Fin AI, USA · Yale University, USA · Cardiff University, UK

Taxonomy Retrievers Hypothesis Exploratory Factor Analysis Readiness

Ask PDF

Jun 3, 2026cs.CL

Executable Schema Contracts: From Automatic Ingestion to Multi-Source Retrieval

Real-world data spans tables, documents, and semi-structured files with implicit semantics. Querying this data requires integrating evidence across inconsistent schemas and formats, yet existing approaches either demand costly manual engineering or bypass structure entirely. We present a system that automatically discovers an executable schema from raw multi-source data and uses it as a shared contract for knowledge graph construction and query-time retrieval. A closed-world field catalog constrains LLM-based schema discovery to attested fields; deterministic structural analysis infers identity keys, foreign keys, and source hierarchy; and the resulting schema drives extraction, deduplication, and cross-source linking into a provenance-aware knowledge graph. At query time the schema -- optionally extended via a monotonic protocol -- conditions a multi-tool agent routing retrieval across structured lookup, graph traversal, and vector search, returning grounded answers with traceable citations. In controlled zero-shot comparisons using the same LLM, data, and evaluation harness, the system improves over retrieval-only and decomposition-based baselines across four QA benchmarks, with ablations showing that schema-conditioned routing, structural intelligence, and schema-guided construction each contribute to the gains.

Padmaja Jonnalagedda, Yuguang Yao, Xiang Gao +2

Intuit AI Research

Schema Knowledge Graphs Discovery

Ask PDF