Machine Learning for Coding Retail Product Names to Consumer-Price Categories: A Rule-plus-Bag-of-Words Pipeline with Reliability-Weighted Human-in-the-Loop Labeling
arXiv:2606.02004v1 Announce Type: new Abstract: Consumer-price measurement increasingly draws on alternative data sources -- scanner, web-scraped, and transaction/receipt data. A recurring obstacle is that product descriptions in such sources are short, noisy, and abbreviated, with no standard product code, so each item must first be mapped to a consumption classification (e.g., the UN COICOP scheme) before prices can be compared. This paper studies that mapping as a general, reproducible method. The pipeline is: (i) text normalization and tokenization of noisy item names; (ii) a prefix-tree (
The increasing availability of alternative data sources for price measurement is driving the need for robust, automated methods to process unstructured product data for economic analysis.
Accurate and automated classification of product names is critical for more refined inflation measurement and understanding consumer spending patterns, impacting monetary policy and retail strategy.
This machine learning pipeline offers a more granular and efficient method for converting raw product data into standardized economic classifications, reducing manual effort and improving data quality.
- · Economic statistical agencies
- · Retail analytics firms
- · AI researchers (NLP)
- · Governments
- · Manual data coders
- · Legacy statistical methods
Improved and faster measurement of consumer prices and inflation metrics.
More agile economic policy responses due to higher-fidelity data inputs.
Potential for new financial products or market indicators derived from real-time, granular consumer price data.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL