knowledge/semantic/chunks
Source: knowledge/semantic/chunks.parquet. Source vintage: Package vintage 2026-10-01; source path partition not dated; inspect record and rule clocks separately. Grain: Packaged source record; inspect schema and linked world method before aggregation. Snapshot: 7a53af6a9beb58928d4a64dc92db09c9db65433f9909f1a10d16e302adaf86ba.
Read the method for this source
Explore this source with typed OData queries · Filtering, grouping and evidence method
Schema and filter instructions · Count this selection · First records
| node_id Int64 | document_node_id Int64 | doc_id Utf8 | ordinal Int64 | source_sha256 Utf8 | chunk_sha256 Utf8 | byte_start Int64 | byte_end Int64 | char_start Int64 | char_end Int64 | text Utf8 | extractor_version Utf8 | chunker_version Utf8 | __row_id Int64 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 60 Filter to this value Explore graph node | 60 Filter to this value | mcd-article-contractor Filter to this value | 0 Filter to this value | 5d1812d181448308b85dd84e34c141e7c2ed5b8ef8df2e3bd1459ecf20f4cc5e Filter to this value | 6f2c83bae87eb6fea47e9f5c82922d493ce6a3c31b9a74cce32a7e6e1345e74c Filter to this value | 0 Filter to this value | 628 Filter to this value | 0 Filter to this value | 628 Filter to this value | source: mcd-article-contractor state: shipped table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet named by: pipelines/coverage-intelligence/build_coverage_policy.py names the object carried in: data/policy gavel row count: 31866 note: Article to MAC. LOAD-BEARING, not decoration: the three IDTF matrices naming 76705 belong to Novitas, First Coast and Palmetto separately, so a row read without its contractor reads as a national rule and is not one 9 columns, including: article_id, article_version, article_type, contractor_id, contractor_type_id, contractor_version, last_updated, filename, vintage | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 60 |
| 61 Filter to this value Explore graph node | 61 Filter to this value | mcd-article-hcpc Filter to this value | 0 Filter to this value | 2276ee2dfb4d815b1588a7e974e6d6c8cb295345b8fc846a2e8481bff6dbbe27 Filter to this value | 266077d368b814439d6a9b462abf60ab5f0e49372689e59913c0ca95e6a9fcef Filter to this value | 0 Filter to this value | 536 Filter to this value | 0 Filter to this value | 536 Filter to this value | source: mcd-article-hcpc state: shipped table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet named by: pipelines/coverage-intelligence/build_coverage_policy.py names the object carried in: data/policy gavel row count: 23439 note: Article to HCPCS join, CPT included. 5.5 MB against the LCD table's 401 KB. This is the join that works 11 columns, including: article_id, article_version, hcpc_code_id, hcpc_code_version, hcpc_code_group, long_description, short_description, range, last_updated, filename, vintage | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 61 |
| 62 Filter to this value Explore graph node | 62 Filter to this value | mcd-article-text Filter to this value | 0 Filter to this value | 8f29483093c187057f2d1c0b6b40f42bb46cac4c9fd5462818de19fc5930ce8f Filter to this value | 75e81160db7f10ea7b190d120a5f4e4b38a01c51e536c88b17e563636e041d26 Filter to this value | 0 Filter to this value | 851 Filter to this value | 0 Filter to this value | 851 Filter to this value | source: mcd-article-text state: shipped table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet named by: pipelines/coverage-intelligence/build_coverage_policy.py names the object carried in: data/policy gavel row count: 2269 note: Article full text. Carries the IDTF Credentialing Matrix, whose per-code rows state supervision level, supervising physician qualification and technician credential 29 columns, including: article_id, article_version, article_type, title, article_pub_date, article_eff_date, article_end_date, description, other_comments, sad_url, thirty_percent, status, last_updated, history_exp, key_article, icd9_covered_para, icd9_noncovered_para, revenue_para, article_rev_end_date, source_article_id, date_retired, keywords, icd10_doc, add_icd10_info, cms_cov_policy, display_id, reference_article and 2 more | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 62 |
| 63 Filter to this value Explore graph node | 63 Filter to this value | mcd-contractor-names Filter to this value | 0 Filter to this value | 223fb35a74cb14789cc1f7206d47a7568cbd3a04cff87edf8dc881361f1fb2cd Filter to this value | 195a1355640181ac416e651a750d15b930a084637d390c921946145a8a5115ee Filter to this value | 0 Filter to this value | 725 Filter to this value | 0 Filter to this value | 725 Filter to this value | source: mcd-contractor-names state: shipped table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet named by: pipelines/coverage-intelligence/build_coverage_policy.py names the object carried in: data/policy gavel row count: 144 note: Contractor id to name. Without it every downstream answer names a MAC as "314", which is not something a reader can act on and not something they can check 25 columns, including: contractor_id, contractor_type_id, contractor_version, contractor_bus_name, contractor_number, dmerc_rgn, address1, address2, address3, city, state_id, zipcode, phone, fax, status, last_updated, url, email, ignore, status_flag, cmd_name, cmd_title, contractor_subtype_id, filename, vintage | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 63 |
| 64 Filter to this value Explore graph node | 64 Filter to this value | mcd-lcd Filter to this value | 0 Filter to this value | 0b831cf7a928ca2e88cccaddb4ba406affa138b8992b1cb802057086d28ddd3c Filter to this value | f7a1e3e4a140510857019934c6faec720a8c3e89d54fc396f7814f29bde70d7f Filter to this value | 0 Filter to this value | 930 Filter to this value | 0 Filter to this value | 930 Filter to this value | source: mcd-lcd state: available table_state: unpacked gavel vintage: 2026-09-24 upstream files: all_lcd.zip, contractor.csv, contractor_jurisdiction.csv, lcd.csv, lcd_x_contractor.csv, lcd_x_hcpc_code.csv, lcd_x_urls.csv why not shipped: this world does not carry it, and gavel holds no `data.parquet` for it either. gavel opened the archive and holds its members under `_unpacked/`, and did not consolidate them into a `data.parquet`. The tables exist and are one join away, which is a smaller task than a conversion from scratch. At vintage 2026-09-24 the source holds: all_lcd.zip, contractor.csv, contractor_jurisdiction.csv, lcd.csv, lcd_x_contractor.csv, lcd_x_hcpc_code.csv, lcd_x_urls.csv. That is a conversion task in gavel and it is a different fact from a decision not to ship. `available` means fetched and staged by gavel. It does not mean validated, does not mean converted, and does not mean read by anything here. | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 64 |
| 65 Filter to this value Explore graph node | 65 Filter to this value | mcd-lcd-contractor Filter to this value | 0 Filter to this value | c26ec243e414dda9836cbbdcc0da245bd8656214632250e410f9805042fcf265 Filter to this value | 058d57faa7960b9edda2e8c661776287d8bae8a0630ba315f6ac7eb7ce612e3e Filter to this value | 0 Filter to this value | 649 Filter to this value | 0 Filter to this value | 649 Filter to this value | source: mcd-lcd-contractor state: available table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet gavel row count: 15199 note: LCD to MAC. An LCD binds one contractor, so a hit without this join reads as a national rule and is not one why not shipped: this world does not carry it. The source is converted and readable in gavel and no tool here reads it. `available` means fetched and staged by gavel. It does not mean validated, does not mean converted, and does not mean read by anything here. 8 columns, including: lcd_id, lcd_version, contractor_id, contractor_type_id, contractor_version, last_updated, filename, vintage | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 65 |
| 66 Filter to this value Explore graph node | 66 Filter to this value | mcd-lcd-contractor-jurisdiction Filter to this value | 0 Filter to this value | f3ca917994181c6e7a69eb02a0d01c69d3b8d7d661f6d9cac012eb790e32bd0c Filter to this value | 27ca205dece16a689613e22970766ed9714e09e9d84e789c075a4263e9ff6cb3 Filter to this value | 0 Filter to this value | 671 Filter to this value | 0 Filter to this value | 671 Filter to this value | source: mcd-lcd-contractor-jurisdiction state: available table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet gavel row count: 337 note: Which states each MAC covers. Turns a contractor id into the geography a claim is actually adjudicated in why not shipped: this world does not carry it. The source is converted and readable in gavel and no tool here reads it. `available` means fetched and staged by gavel. It does not mean validated, does not mean converted, and does not mean read by anything here. 9 columns, including: contractor_id, contractor_type_id, contractor_version, state_id, last_updated, active_date, term_date, filename, vintage | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 66 |
| 67 Filter to this value Explore graph node | 67 Filter to this value | mcd-lcd-hcpc Filter to this value | 0 Filter to this value | cd9a44ed950250fcff457dc54f4c394f8ac6082f3b99518eba8ad1999596b903 Filter to this value | d1ba40b3dfbcd6be1c1a0676560cb3ea978236847eea0c53eaaf9aa82d5dbd42 Filter to this value | 0 Filter to this value | 864 Filter to this value | 0 Filter to this value | 864 Filter to this value | source: mcd-lcd-hcpc state: available table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet gavel row count: 2452 note: LCD to HCPCS join. Measured 2026-08-22: this table does NOT carry every code an LCD names, and returned zero rows for all twelve [redacted] codes AND for 76700 and 93000, which are ubiquitous. Whatever it is, it is not the complete join, so a code absent here is not a code no LCD mentions why not shipped: this world does not carry it. The source is converted and readable in gavel and no tool here reads it. `available` means fetched and staged by gavel. It does not mean validated, does not mean converted, and does not mean read by anything here. 11 columns, including: lcd_id, lcd_version, hcpc_code_id, hcpc_code_version, hcpc_code_group, long_description, short_description, last_updated, range, filename, vintage | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 67 |
| 68 Filter to this value Explore graph node | 68 Filter to this value | mcd-lcd-policy Filter to this value | 0 Filter to this value | 26140b36058db1108ca764d2ccacdf8758153257e918c8930c52268ee030ed45 Filter to this value | 0805500df41fd531cce477837dd0c9dd6373760d73074a7ad4d76829ee95b69c Filter to this value | 0 Filter to this value | 954 Filter to this value | 0 Filter to this value | 954 Filter to this value | source: mcd-lcd-policy state: available table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet gavel row count: 1070 note: LCD full text: indications, limitations, and who may furnish or interpret. The 52 MB table and the reason this corpus exists why not shipped: this world does not carry it. The source is converted and readable in gavel and no tool here reads it. `available` means fetched and staged by gavel. It does not mean validated, does not mean converted, and does not mean read by anything here. 50 columns, including: lcd_id, lcd_version, display_id, title, determination_number, cms_cov_policy, orig_det_eff_date, ent_det_end_date, rev_eff_date, rev_end_date, indication, diagnoses_support, icd9_dont_support_para, icd9_dont_support_ast, diagnoses_dont_support, coding_guidelines, doc_reqs, appendices, util_guide, source_info, adv_meeting, comment_start_dt, comment_end_dt, notice_start_dt, rev_hist_num and 25 more | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 68 |
| 69 Filter to this value Explore graph node | 69 Filter to this value | mcd-lcd-urls Filter to this value | 0 Filter to this value | 07e30b9acf3efd45d992d090e3ef37fbe1ce246702463ab696b305cd8ec198f1 Filter to this value | 37cf19c880634b8b87da74aab5820fd9c5c418e02c7ad6af7142b6ab7d8626cf Filter to this value | 0 Filter to this value | 672 Filter to this value | 0 Filter to this value | 672 Filter to this value | source: mcd-lcd-urls state: available table_state: converted gavel vintage: 2026-09-24 upstream files: data.parquet gavel row count: 129 note: Per-document source URLs. This is the linkback CMS's clearance is conditional on, which is why it is a table and not a constant why not shipped: this world does not carry it. The source is converted and readable in gavel and no tool here reads it. `available` means fetched and staged by gavel. It does not mean validated, does not mean converted, and does not mean read by anything here. 11 columns, including: lcd_id, lcd_version, url_type_id, url_id, url, url_name, sort_order, last_updated, url_description, filename, vintage | coverage-source-register/v1 Filter to this value | hammer-register-chunker-v3 Filter to this value | 69 |