Fitment-Based Part Clustering for Interchangeable Item Retrieval

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing item retrieval systems on listing platforms struggle to efficiently identify and return interchangeable parts due to the absence of part numbers (MPNs) on item listings, leading to increased computing resource consumption and user frustration from repetitive queries, resulting in high network latency, storage device I/O, and low conversion rates.

Innovation Solution

An item retrieval system uses a machine learning model to generate item embeddings based on fitment data, clustering interchangeable parts based on similarity scores and constraints, reducing repetitive user queries by providing relevant item listings directly.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional item retrieval systems are used without part number data, then the system can operate with simpler data requirements, but the system cannot accurately identify interchangeable parts leading to repetitive user queries and high computing resource consumption

Engineering Contradiction:
Improveaccuracy of interchangeable parts identificationVSAvoidcomplexity of retrieval system
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary component (part number extraction and matching system) that bridges the gap between simple item listings and accurate interchangeable parts identification. This intermediary layer processes item data to extract or infer part numbers, enabling precise matching without requiring part numbers to be explicitly present in all listings.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system performs preliminary actions by pre-processing item data to extract, infer, or store part number information before retrieval operations. This preliminary processing creates a structured foundation that enables accurate interchangeable parts identification during actual user queries, reducing computational overhead during runtime.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If the system processes all item listings individually without clustering, then it maintains simplicity in data structure, but it consumes excessive computing resources and storage I/O due to repetitive queries

Engineering Contradiction:
Improveretrieval efficiencyVSAvoidcomputing resource consumption
Core Design Contradiction:
ProductivityVSLoss of energy

Solution Approach 1:

The patent merges multiple item listings that represent interchangeable parts into clustered groups based on part number matching. Instead of processing each item listing individually, the system clusters compatible items together, allowing a single retrieval operation to return multiple interchangeable alternatives and eliminating repetitive processing of equivalent items.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The system creates simplified representative copies or pointers to clustered item groups, allowing efficient retrieval through reference to cluster identifiers rather than processing full item data structures repeatedly. This copying approach reduces storage I/O and computing resources by working with compact cluster representations.

Inventive Principle:
Principle #26Copying

3Ease of operation

If the system returns all matching item listings without intelligent filtering, then it ensures complete coverage of available parts, but it increases network latency and user frustration from overwhelming results

Engineering Contradiction:
Improveuser experience qualityVSAvoidnetwork latency
Core Design Contradiction:
Ease of operationVSLoss of time

Solution Approach 1:

The patent extracts and prioritizes the most relevant interchangeable parts from clustered groups based on criteria such as popularity, compatibility confidence, or user preferences. By taking out only the most useful results rather than returning all matching items, the system reduces network latency and presents manageable, high-quality results to users.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system applies different quality filters and sorting criteria to different clusters or regions of results based on local characteristics such as part category, compatibility certainty, or historical user behavior. This local quality approach tailors the presentation of results to specific contexts, improving ease of operation without sacrificing comprehensive coverage.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP4668194A1Item retrieval using fitment data-based part clustering
Publication Date: 2025.12.24 EBAY INC
  • EP4668194A1 patent drawingFigure 1
  • EP4668194A1 patent drawingFigure 2
  • EP4668194A1 patent drawingFigure 3

AI summary

Some aspects relate to technologies for performing item retrieval on a listing platform using clusters of interchangeable parts formed using fitment data. In accordance with some aspects, item data is accessed for each of a plurality of part item listings on a listing platform, where the item data for each part item listing includes fitment data. An item embedding is generated for each part item listing using the item data. The item embeddings are clustered to generate a plurality of clusters, wherein each cluster includes one or more item embeddings. Cluster data is stored for the plurality of clusters. The cluster data for each cluster associates a cluster identifier and an item listing identifier for each item embedding in the cluster. The cluster data can be leveraged to perform item retrieval for the listing platform.