Title Checker Investigation — "No Index2 Data Relevant"

Session: 910ba5f0-d344-4f88-8d5f-5bd360c749d4
URL: https://app.bhume.in/title-checker/sessions/910ba5f0-d344-4f88-8d5f-5bd360c749d4
Target parcel: Gat 42/2, village Wagdara / वागदरा, Hingana taluka, Nagpur district
Investigated: 2026-07-15 (read-only, via prod cloud-sql-proxy)

TL;DR

There is no error. Index2 scraping ran successfully and found 94 documents. The relevance classifier then discarded all 92 classified documents because none of them are about the target parcel Gat 42/2 — they are all about plot #42 / row-house #42 sitting inside other Gat numbers in the same village. This is a legitimate "no relevant Index2 transactions found" outcome, not a pipeline failure.

Evidence

Pipeline status — clean, no failures

Field	Value
status	completed
stop_reason	None
retry_count	0
index2_failed_years	[]
index2_not_found_years	[]
index2_failure_details	[]
index2_total_scraped	126 (pages)
Index2 documents extracted	94
seven_twelve_raw	present (target 7/12 fetched OK)

Relevance classification — everything discarded

index2_relevance:
  directly_relevant: 0
  parent_chain:      0
  discarded:        92

Why every document was discarded (root cause)

In Wagdara, "42" is an extremely common layout plot number and row-house number that appears inside many different Gat numbers. Every one of the 92 documents references a "42" that is not the target agricultural parcel Gat 42/2. Representative examples:

Reg. No	Parcel in document	Discard reason
986/2025	Gat 102/103, Plot 42	Layout plot within Gat 102/103, not Gat 42/2
5613/2025	Gat 62/1, Plots 1–140	Bulk layout; Plot 42 is within Gat 62/1
2956/2026	Gat 64/2, Plot 42	Plot 42 within Gat 64/2
843/2026	Gat 52, Plot 42	Plot 42 within Gat 52
468/2023	Gat 9, Row House 41 & 42	Row house in Gat 9
5268/2011	Sukali village 42, 41/A, 41/B	Different village entirely
…	(Gats 17, 21, 24, 50, 85, 105, 121–127, etc.)	Plot/house #42 within other Gats

A scan of all extracted gat_numbers / survey_number fields found no document referencing root Gat 42 or sub-parcel 42/2 in Wagdara. The classifier's per-document reasoning is consistent and correct: "Plot 42 within Gat X ≠ target Gat 42/2."

Conclusion

The pipeline behaved correctly:

Index2 scraped cleanly across 2003–2026 (0 failures).
It surfaced the large volume of noise created by plot #42 being popular across many layouts.
It correctly excluded all of it and kept nothing, because the actual target parcel Gat 42/2 has no registered transactions in Index2.

Recommended follow-ups (product judgment, not code fixes)

Confirm the target really is Gat "42/2". If the customer's parcel is actually plot 42 within a layout Gat (common for developed / NA land near Nagpur), then the "correct" record is one of the 92 discarded docs — you'd need to identify the parent Gat. The fetched 7/12 will indicate whether 42/2 is an agricultural survey-gat or an NA layout plot.
A re-scrape would not change the result. With 0 failures and full year coverage, forcing an Index2 re-scrape (index2_failed_years + resume parallel_fetch) would return the same documents.

Generated by Claude Code. Source data read read-only from production via cloud-sql-proxy.