Owner Identification from Postal Correspondence Using Machine Learning

The postal service is a crucial part of information transmission in today's world. Heavily relying on the delivery to the correct owner, it faces the problem of distinguishing the current and past owners. Our research seeks to identify the current and past recipients of postal letters by examining address frequencies and patterns. It incorporates various machine learning models such as classification models and clustering analysis. Using statistical and machine learning methods, ownership is determined by parsing postal envelope contents and extracting addresses and names by image processing. This approach provides a comprehensive understanding of change in ownership over time. The methodology used in the research integrates Named Entity Recognition (NER), Optical Character Recognition (OCR), and Address Frequency Analysis for the purpose. The machine learning models accurately identify the owner from postal correspondence with 75% accuracy.