Skip to content
SIH Buddyby Ganeev Singh
Dev
All problem statements
SIH26099Worth consideringacceptance 4/5

AI-Driven Standardization and Harmonization of Material Codes Across CPSEs

Ministry of Petroleum & Natural Gas · Smart Automation · Software

A genuinely hard matching problem in a domain nobody else will touch, with an inspectable demo — the drag is that the promised dataset probably will not arrive, so build your own realistic corpus early and make the near-miss handling your pitch.

Data: CPSE Material Master Data / Sample Material Master Dataset – to be provided by participating CPSEs.

What it actually is

Two public sector companies buying the identical bolt will have logged it under different codes, different descriptions and different units, so nobody can tell the two entries are the same thing. That means duplicate stock, no shared purchasing and no visibility across the sector. The ask is a system that recognises when different records describe the same material and proposes one common code.

What to build

An entity resolution engine over material master records, which is the real problem hiding behind the platform language: parsing short, abbreviation-heavy technical descriptions into structured attributes such as type, dimensions, material grade and standard, then matching records across organisations that describe the same item despite entirely different code schemes, word orders, abbreviations and units — distinguishing genuinely identical items from functionally equivalent ones and from similar-looking items that differ in a specification that matters — with a confidence score per proposed match, a reviewer workflow where a materials engineer accepts or rejects each cluster, generation of a standardised description and common code per accepted cluster with mapping retained back to each organisation's original code, and an audit trail over every change.

Smallest thing that wins the room

Load two organisations' material masters with no shared coding scheme, watch the engine cluster the duplicates with confidence scores, and open a near-miss pair where a single differing specification correctly prevented a match.

How crowded this one gets

A guess, projected from the 2025 statements — the last year where both the submission counts and the winners were published.

Moderate120–270 teams expectedroughly 1 in 99–230 wins it

Quieter than 55% of the 226 · #103 of 226 by expected field

A normal-sized field. Your idea has to be good, not miraculous.

Why: central ministry statements sat below the average.

This is a guess, not a fact

Nobody has published 2026’s numbers yet. This is an analysed estimate from last year’s pattern, so please do not take it as the truth — check the live counter on the SIH portal before you decide anything. The range covers the middle half of likely outcomes, so one statement in two lands outside it. Entry closes at 500 ideas per statement, so no range goes past that — a statement that reaches the cap fills and shuts rather than drawing an unlimited crowd. The model reads only three things a team can see before choosing — software or hardware, the theme, and what kind of body posted it — and those explain about a quarter of the variation in last year’s field sizes (R² 0.25 on held-out statements). Trust the band more than the number, and the ordering more than either. It cannot see how good your idea is, which is the part that actually decides it.

The scores

The number is the shorthand. The line under it is the reason.

What you will be writing

  • entity resolution over short technical descriptions
  • abbreviation normalisation and attribute parsing
  • sentence embedding plus blocking for candidate generation
  • UNSPSC or ECCMA classification anchoring
  • confidence-scored human review workflow
  • code mapping with retained lineage
  • Master data management
  • Public procurement
  • Entity resolution

Prior art to read before you start

duplicate and near-duplicate record detection · technical description normalisation · cross-organisation code harmonisation

Analysed by Claude Opus. Every score above is a judgment call with its reasoning attached — kindly cross-check this against the official statement on the SIH portal before your team commits to it.