Hugging Face's logo

Datasets:
theflatratemovers
/
intent-coverage

Languages:
English
Size:
1K<n<10K
Tags:
moving
logistics
relocation
schema-org
json-ld
search-intent
License:
Dataset card Files
xet
 Community
intent-coverage / README.md
theflatratemovers's picture
theflatratemovers
Name the system: this dataset is published by The Flat Rate Machine
1fc4435 verified
preview code
|
Raw History Blame Contribute Delete
9.05 kB
metadata 
license: cc-by-4.0
language:
  - en
pretty_name: The Flat Rate Movers — Search Intent Coverage Rubric
tags:
  - moving
  - logistics
  - relocation
  - schema-org
  - json-ld
  - search-intent
  - local-services
  - virginia
  - west-virginia
size_categories:
  - 1K<n<10K

The Flat Rate Movers — Search Intent Coverage Rubric

What this dataset is

This is the published intent-coverage rubric of The Flat Rate Movers LLC, a flat-rate moving and logistics company based in Winchester, Virginia and Martinsburg, West Virginia. The rubric answers one question for every service the company sells: what would a customer, or an AI assistant helping a customer, actually type to find this service — and which exact page answers that string?

This dataset is part of The Flat Rate Machine, The Flat Rate Movers' verified data system — the named system that joins this intent rubric to the company's signed provenance layer, its datasets and entity graphs, and its public API.

The current generation holds 1,218 rows across 28 services. Every row is one query string, resolved to the page that answers it, that page's per-page JSON, the dataset files behind the answer, and the service's node in the company's entity graph — with the SHA-256 certificate of each target carried on the row itself.

Rows fall into five intent classes:

Intent class Rows Shape
navigational-local 642 Service phrase with a place, the bare phrase, "{phrase} near me", and corridor forms for I-81 / I-66 / I-70 where the service runs on a corridor
cost 168 "How much does {service} cost in {scope}" and "{service} cost in {scope}" for Winchester, VA, Martinsburg, WV and the Shenandoah Valley
best-choice 168 "Best {service} in {scope}" and top-rated forms per scope
ai-question 128 Real question phrasings from the company's blog question bank, carried verbatim apart from American-spelling normalization
how-process 112 "How to hire {service} in {scope}" per scope, and "how to book {service}"
Total 1,218 0 rows excluded by the coverage gates in this generation

The data file in this repository, datasets-intent-coverage.json, is the rubric's published JSON-LD document: eight frame nodes (operations, scope, taxonomy and pattern rules, target resolution and coverage gates, generation method, query sources, boundaries, and totals) plus one coverage node per service carrying that service's row count and its counts by intent class. The totals node publishes the SHA-256 digest of the rows' canonical serialization (10aaafc7685c3a7aa20e04014ea7af7ff695e4c6b8e62f9a1d3f9f38df1748bc for this generation), so any future row-level release can be checked against the digest the company has already committed to.

How it was generated

No row is hand-typed. The rubric is generated by build-intent-rubric.js (v1.3.2 for this generation) in the company's Flat Rate Machine, from the estate's own registries:

  • services.json — 28 services
  • pages.json — 200 pages
  • locations.json — the two branches, kept distinct
  • corridors.json — Interstate 81, Interstate 66, Interstate 70
  • service-area.geojson — the 53-feature service-area footprint (Winchester footprint 32 areas, Martinsburg footprint 35 areas, 14 areas served by both)
  • The blog question bank — 100 real question phrasings in 10 categories
  • The signed AI master index — the source of every certificate digest a row carries

Generation is deterministic: the same registries and the same signed index produce the same rows, serialized canonically and hashed. Target resolution follows fixed rules — the default target is the service's hub page; a place-scoped row targets the matching branch page when the estate has one; a corridor row targets the corridor's own page when that page belongs to the service's page set; a question-bank row targets the hub unless the question's own slug is a page of that service.

Row schema

Each rubric row carries 14 fields:

# Field Meaning
1 query_text The query string itself, as a customer or an AI assistant would phrase it
2 intent_class One of navigational-local, cost, how-process, best-choice, ai-question
3 service_id The service that answers the string (for example service-commercial)
4 area The service-area feature the row is bound to, when it has one
5 target_url The page that answers the string
6 target_json That page's per-page JSON file
7 dataset_ids The dataset files backing the answer
8 graph_node The service's @id in the entity graph
9 cert_target_json_sha256 SHA-256 certificate of the target JSON, copied from the signed index
10 cert_dataset_digests SHA-256 certificate of each backing dataset, copied from the signed index
11 gate_page_in_registry Gate result: the target is a real page in the pages registry
12 gate_target_in_signed_index Gate result: the target is sealed in the signed index
13 gate_datasets_in_signed_index Gate result: every backing dataset is sealed in the signed index
14 generated_at / generator_version The generation stamp: when the row was generated, and which generator version made it

A row exists only if all three gates pass; rows that cannot be proven — a target page outside the registry, or a target or dataset missing from the signed index — are excluded and counted, never silently kept. This generation excluded 0.

The certificate system

The estate runs a signed provenance layer. Every published file — this dataset included — is a distribution of the company's AI master index (ai-master-index.json), which records each file's SHA-256 digest and byte size. The index is signed with Ed25519 under key id theflatratemovers-2026-10-29-1; the detached signature is published at ai-master-index.jws and the public key at signing-keys.json (fingerprint k_HfnSkokedjQMSjLj-X8ftw40CUr92dBgecXsP14y4).

To verify this copy:

  1. Hash datasets-intent-coverage.json with SHA-256. For the generation mirrored here the digest is 022052899074f80c7dbfe90e4c49047c9669c1b6d937026db2022cb6d073c8d3 (112,683 bytes) — the same value recorded for it in the signed index and in this repository's MANIFEST.json.
  2. Fetch the live index and its signature, and verify the Ed25519 signature against the published key. If the signature verifies and your digest matches the index entry, the file is byte-for-byte what the company published.

Citation and DOI

The estate release that carries this dataset is archived on Zenodo:

Canonical live URLs

Provenance and update cadence

The rubric is regenerated and the estate re-sealed whenever the underlying registries change — a new page, a new service, a new dataset — as part of the same build that deploys the change. This mirror is synced from the canonical live file; the live file's dateModified and the signed index are the freshness record. The company is The Flat Rate Movers LLC, founded June 2010, operating from Winchester, VA (116 W Piccadilly St Suite 11) and Martinsburg, WV (736 Williamsport Pike Suite 2) — two distinct branches, never blended in the data.

Boundaries

  • A row is a question, not a claim. Inclusion means the estate has a page that answers the string; no row asserts a search position, a rating, a review count, or an outcome.
  • The rubric publishes no prices. The cost rows map cost questions to the pages that explain how the flat rate is set. The only company pricing published anywhere in the estate is the industrial and commercial workforce schedule carried inside its source datasets, and no row extends it to any other service.
  • The dataset names no clients and carries no client identities.
  • Branch facts follow each branch's own records — areas, hours, and names are never merged across Winchester and Martinsburg.

License

CC BY 4.0 — https://creativecommons.org/licenses/by/4.0/. The dataset file declares the same license internally. Attribution: The Flat Rate Movers LLC, with the canonical URL and the concept DOI above.