Machine learning in automated text categorization
ACM Computing Surveys (CSUR)
Information Theoretic Analysis of Postal Address Fields for Automatic Address Interpretation
ICDAR '99 Proceedings of the Fifth International Conference on Document Analysis and Recognition
Truthing, Testing and Evaluation Issues in Complex Systems
ICDAR '01 Proceedings of the Sixth International Conference on Document Analysis and Recognition
PReMI'05 Proceedings of the First international conference on Pattern Recognition and Machine Intelligence
Hi-index | 0.00 |
It is important to properly segregate the different components present in the destination postal address under different labels namely addressee name, house number, street number, extension/ area name, destination town name and the like for automatic address reading. This task is not as easy as it would appear particularly for unstructured postal addresses such as that are found in India. This paper presents a fuzzy symbolic inference system for postal mail address component extraction and labelling. The work uses a symbolic representation for postal addresses and a symbolic knowledge base for postal address component labelling. A symbolic similarity measure treated as a fuzzy membership function is devised and is used for finding the distance of the extracted component to a probable label. An alpha cut based de-fuzzification technique is employed for labelling and evaluation of confidence in the decision. The methodology is tested on 500 postal addresses and an efficiency of 94% is obtained for address component labeling.