Academia.eduAcademia.edu

Outline

Title
Abstract
Figures
Introduction
Conclusions and Future Work
References

Mining Postal Addresses

Profile image of Jose Maria Gomez HidalgoJose Maria Gomez Hidalgo

2008

Abstract

This paper presents FuMaS (Fuzzy Matching System), a system capable of an efficient retrieval of postal addresses from noisy queries. The fuzzy postal addresses retrieval has many possible applications, ranging from datawarehouse dedumping, to the correction of input forms, or the integration within online street directories, etc. This paper presents the system architecture along with a series of experiments performed using FuMaS. The experimental results show that FuMaS is a very useful system when retrieving noisy postal addresses, being able to retrieve almost 85% of the total ones. This represents an improvement of the 15% when comparing with other systems tested in this set of experiments.

Figures (2)
Figure 1. Conceptual view of FuMaS architecture (subfigure A) and an abstraction level module decomposition (subfigure B). As can be seen, FuMaS presents a 3 abstraction levels based architecture, which are expert when working on a certain kind of entities (addresses, phrases and words).
Figure 1. Conceptual view of FuMaS architecture (subfigure A) and an abstraction level module decomposition (subfigure B). As can be seen, FuMaS presents a 3 abstraction levels based architecture, which are expert when working on a certain kind of entities (addresses, phrases and words).
After a small set of experiments, SOUNDEX has been ruled out because it has been designed for English names and when using Spanish words the similarity values has no real sense. Cosine similarity only uses the information of the number of equal words in the given 2 input strings, which is not very useful for this problem. Dice's similarity works similar to cosine similarity and was ruled out soon by the same reason. Then, the only two algorithms that really seems to fit to this problem are Levenshtein distance and Q-Grams’. The first set of experiments evaluates the adjustment of each algorithm to the given problem. Figure 2. Comparison of relevant addresses retrieved by FuMaS, Uniserv and 4 street directories that integrates some techniques to correct the user inputs. Dark bars show the results for recall 10 and light ones show the results for recall 1
After a small set of experiments, SOUNDEX has been ruled out because it has been designed for English names and when using Spanish words the similarity values has no real sense. Cosine similarity only uses the information of the number of equal words in the given 2 input strings, which is not very useful for this problem. Dice's similarity works similar to cosine similarity and was ruled out soon by the same reason. Then, the only two algorithms that really seems to fit to this problem are Levenshtein distance and Q-Grams’. The first set of experiments evaluates the adjustment of each algorithm to the given problem. Figure 2. Comparison of relevant addresses retrieved by FuMaS, Uniserv and 4 street directories that integrates some techniques to correct the user inputs. Dark bars show the results for recall 10 and light ones show the results for recall 1

Related papers

A Novel Approach to Address Validation using String Distance Metrics
IRJET Journal

IRJET, 2023

Address validation holds fundamental value in confirming the accuracy and geographical precision of addresses used by location-dependent and delivery-based organizations. Addresses often suffer from problems such as missing components and geographical inadequacies which can cause grave logistical issues if not validated adequately beforehand. The identification of missing or invalid address components to perform address validation can prove to be a helpful factor in saving time and cost for businesses and simultaneously reducing the chances of errors in service. Significant potential has been found in the usage of statistical measures such as correlation coefficients and measures of central tendency to perform the task of address validation. The system proposed in this paper uses a combination of different string-matching metrics to generate a normalized score based on statistical similarities. This score can then be used to filter out validated addresses according to the threshold of similarity required. Experiments have been conducted on a real-world healthcare dataset to demonstrate the effectiveness of the proposed approach in terms of accuracy and precision.

View PDFchevron_right
A data model for efficient address data representation-Lessons learnt from the Intiendo address matching tool
Serena Coetzee

2008

Geocoding refers to the process of assigning geographic identifiers and/or geographic coordinates to the description of a feature location, i.e. the words, codes or terms that describe a feature's location. Address matching is the specific case of geocoding where the description of a feature location comprises an address. Address matching is complicated by an incomplete or inaccurate incoming address or one that contains a misleading geographic identifier in its location type hierarchy. The Intiendo address matching tool is based on a data structure that is similar to the hierarchy of location types described in ISO 19112, but we have made some novel extensions to enable a spatial adjacency search. Intiendo does not rely purely on the alphanumeric match in the location type hierarchy but incorporates spatial proximity into the address matching. We also describe how the Intiendo address matching process can be configured and fine tuned, for example, by assigning weights to the location types in the hierarchy, and by specifying parameters for the spatial adjacency match. In this paper we present the hierarchical data structures of the Intiendo address matching tool and show how they are an extended implementation of the ISO 19112 general model. We show the similarities between the Intiendo and ISO 19112 models, and present the extensions that were implemented in Intiendo. By way of examples, we show that our extended model allows more efficient and accurate address matching incorporating spatial adjacency and hierarchical fine tuning.

View PDFchevron_right
Automatically Extracting Structure from Free Text Addresses
Radoslav Yankov

In this paper we present a novel way to automatically elementize postal addresses seen as a plain text string into atomic structured elements like "City" and "Street name". This is an essential step in all warehouse data cleaning activities. In spite of the practical importance of the problem and the technical challenges it offers, research effort on the topic has been limited. Existing commercial approaches are based on hand-tuned, rule-based approaches that are brittle and require extensive manual effort when moved to a different postal system. We present a Hidden Markov Model based approach that can work with just about any address domain when seeded with a small training data set. Experiments on real-life datasets yield accuracy of 89% on a heterogeneous nationwide database of Indian postal addresses and 99.6% on US addresses that tend to be more templatized.

View PDFchevron_right
A System for Recognition of Destination Address in Postal Documents of India
Umapada Pal

Malaysian Journal of Computer Science

Recognition of destination address is compulsory for automation of the postal system in India. Our observation found that such recognition becomes a very challenging task due to inter-mixing of three languages (Hindi, English and the official language of the particular state in which the postal document is supposed to reach).In this paper, our attempt towards development of a dynamic programming based system for city-name and pin code recognition of destination address in postal documents of India not only managed to address the difficulties related to identification of the scripts but also managed to get rid of those problems which is generated due to character touching in postal documents. For city-name recognition, lexicon information is used. However, no lexicon information is used for pin code recognition since an Indian pin code contains only 6 digits. We obtained 99.55% reliability from tri-lingual city-name recognition system where error rates are 0.20% and rejection rates a...

View PDFchevron_right
Applying a Part-of-Speech Tagger to Postal Address Detection on the Web
Nuno Marques

In this paper we show how a POS-tagger can be successfully adapted to a real world information retrieval system capable of extracting postal addresses from the Internet. We develop a particular tag-set for this system. Then we present and discuss the results acquired with the developed postal address tag-set. We conclude the paper by presenting a short description of the final IR system for Web postal address retrieval and drawing some conclusions.

View PDFchevron_right
Postal Address Identification and Sorting
Onkar Deshpande

International Journal for Research in Applied Science and Engineering Technology, 2021

In this fast-moving world, a normal man can take considerable time to find a postal card in a bunch of postcards with significant issues like unclear handwriting, having trouble recognizing some uncommon or ambiguous names. Also, in postal offices or industries, it negatively impacts the efficiency of the postal system. I am making a system for Indian postal automation based on recognizing pin-code on the postcard. In India, there are multiple languages were speak. Indian postcards are mainly written in three languages the state's official language, English, and Devanagari language. In India, more than 50% of people write Pincode digits in either English or Devanagari language, so I am making such a system that sorts both English and Devanagari language postcards. Moreover, the system is mature enough to recognize handwritten as well as printed digits. As a result, the system gets an accuracy of 92.59% on the English language postcards, 90% accuracy on the Devanagari language po...

View PDFchevron_right
Compounded Uniqueness Level: Geo-Location Indexing Using Address Parser
Tejbanta Chingtham

International Journal of Computer Theory and Engineering, 2009

Geo-location searching is an important feature for any search engine and research in this field is not new. The only issue that remains is how a search engine know whether a web page belongs to India or the USA? URLs ending with [.in] are the ultimate choice for India but not all web sites from India end with [.in]. This paper describes a technology known as the address parser. The address parser searches for patterns in a web page that communicates address information. The address parser does not parse every web page of a website for extracting the address but only works on those URLs where the probability of finding an address of the website owner is maximum, thereby eliminating false positives. A central knowledge base is built manually, which contains information like States of a country followed by their city names and other relevant information that may help the address parser do precise local indexing. It was observed that the address parser was not only able to recognize the address patterns in the web pages but also indexed them to city specific information. As a result, a person located in Gangtok, Sikkim, India searched for [universities]; the searching module showed the link of [Sikkim Manipal University] first, followed by other links from India. This work also focuses on the importance of the terms contained in the URLs for geographical based indexing and searching.

View PDFchevron_right
Symbolic Data Structure for Postal Address Representation and Address Validation Through Symbolic Knowledge Base
P. Nagabhushan

Pattern Recognition and Machine Intelligence, 2005

The postal address data and the domain information for address validation contain qualitative, numeric, interval and other types of data. The efficient processing of such data required for postal automation needs a robust data structure that facilitates their storage and access. A symbolic data structure is proposed to represent the postal address and the information relevant for validating the postal

View PDFchevron_right
Address Standardization using Supervised Machine Learning
Sohail Sarwar

Data mining has become an important task of today's rich information environments. Strong results are obtained through accurate historical reporting. Inaccurate and dirty records yield weak and wrong analysis. Unfortunately, organizations store addresses in unstructured formats resulting in multiple representations of same entities. These addresses need to be cleansed before they can be used in mining data. In this paper we present a supervised machine learning procedure, Hidden Markov Model (HMM). This automated probabilistic approach is used to segment a set of Asian addresses into their atomic units and standardize them. Results of this technique show that it can also be used to standardize even a large set of complex and un-formatted addresses.

View PDFchevron_right
Towards the automation of address identification
Luz marina Moreno Moreno

2013

A new semi-automatic method is presented to standardize or codify addresses, in order to produce bibliometric indicators from bibliographic databases. The hypothesis is that this new method is very trustworthy to normalize authors' addresses, easy and quick to obtain. As a way to test the method, a set of already hand-coded data is chosen to verify its reliability: 136,821 Spanish documents (2006-2008) downloaded previously from the Web of Science database. Unique addresses from this set were selected to produce a list of keywords representing various institutional sectors. Once the list of terms is obtained, addresses are standardized with this information and the result is compared to the previous hand-coded data. Some tests are done to analyze possible association between both systems (automatic and hand-coding), calculating measures of recall and precision, and some statistical directional and symmetric measures. The outcome shows a good relation between both methods. Although these results are quite general, this first overview of the address at the institutional sector level is a good way to develop a second approach for the selection of particular centers. This system has some new features because it provides a method based on the previous non-existence of master tables and it has a certain impact on the automation of tasks. The validity of the hypothesis has been proved taking into account not only the statistical measures, but also considering that the obtaining of general and detailed scientific output is less time-consuming and will be even less due to the feedback of the master tables reused for the same kind of data.

View PDFchevron_right

References (26)

  1. Batini, C. et al, 1986. A comparative analysis of methodologies for database schema integration. In ACM Computer Surveys, Vol. 18, No. 4,pp 323-364.
  2. Hernandez, M.A., Stolfo, S.J. 1998. Real-world data is dirty: Data cleansing and the merge/purge problem. In Data Mining and Knowledge Discovery, Vol. 2, No. 4, pp. 9-37.
  3. Fellegi, I.P., Sunter, A.B. 1969. A theory for record linkage. In Journal of the American Statistical Association, Vol. 64, No. 328, pp. 1138-1210.
  4. Gu, L., et al, Record linkage: Current practice and future directions.
  5. Dey, D., et al, A distance-based approach to entity reconciliation in heterogeneous databases. In IEEE Transactions on Knowledge and Data Engineering, Vol. 14, No. 3, pp. 567-582.
  6. Lim, E.P., et al, 1996. Entity identification in database integration. In Information Sciences, Vol. 89, No. 1, pp. 1-38.
  7. Wang, Y.R., Madnick, S.E. 1989, The inter-database instance identification problem in integrating autonomous systems. Proceedings of the Fifth International Conference on Data Engineering, pp. 46-55.
  8. Hernandez, M.A., Stolfo, S.J. 1995, The merge/purge problem for large databases. In Proceedings of the SIGMOD Conference, pp. 127-138.
  9. Christen, P., Churches, T. 2005, febrl-freely extensible biomedical record linkage. Sourceforge.net.
  10. Sauleau, E.A., et al. 2005, Medical record linkage in health information systems by approximate string matching and clustering. In BMC Medical Information Decision Making, Vol. 32, No. 5, pp. 5-32.
  11. Navarro, G., Raffinot, 2002, M. Flexible pattern matching in strings -Practical on-line search algorithms for texts and biological sequences. Cambridge University Press.
  12. Levenshtein, V. 1966, Binary codes capable of correcting deletions, insertions and reversals. In Soviet Physics Doklady, Vol. 10.
  13. Smith, T.F., Water, M.S., Identification of common molecular subsequences. Journal of Molecular Biology.
  14. Aho, A. 1990, Algorithms for finding patterns in string. In Handbook of Theoretical Computer Science: Algorithms and Complexity. MIT Press, pp. 255-300.
  15. Navarro, G., A guided tour to approximate string matching. ACM Computing Surveys, Vol. 33, No. 1, pp. 31-88.
  16. Chollet, G., 1994, Automatic speech and speaker recognition: overview, current issues and perspectives. Fundamentals of speech synthesis and speech recognition. pp. 129-147.
  17. Vintsyuk, T.K., 1968, Speech discrimination by dynamic programming. In Cybernetics and System Analysis. Baeza-Yates, R., Navarro, G. 1997, A practical index for text retrieval allowing errors, pp. 273-282.
  18. Navarro, G. et al, 2001. Matchsmile: A flexible approximate matching tool for personal names searching. In Proceedings of the SBBD'01, pp. 273-282.
  19. Wagner, R.A., Fischer, M.J. 1974, The string-to-string correction problem. In Journal of the ACM.
  20. Wagner, R.A., Lowrance, R., 1975, An extension of the string-to-string correction problem. In Journal of the ACM, Vol. 22, No. 2, pp. 177-183.
  21. Gusfield, D., 1997, Algorithms on strings, trees and sequences: computer science and computational biology. Cambridge University Press, New Yotk, USA.
  22. Needleman, S., Wunsch, C., 1970, A general method applicable to the search for similarities in the amino acid sequence of two proteins. In Journal of Molecular Biology, Vol. 48, pp. 444-453.
  23. Szucs, L.D., Hargreaves, S. 1996, The Source: A guidebook of American Genealogy. Ancestry.com.
  24. Kondrak, G.,et al. Cognates can improve statistical translation models. Proceedings of HLT-NAACL 2003, pp. 46-48.
  25. Salton, G., et al, 1975, A vector space model for automatic indexing. In Communications of the ACM.
  26. Gravano, L. et al, 2001, Using q-grams in a DBMS for approximate string processing. In IEEE Data Engineering Bulletin, Vol. 24, No. 4, pp. 28-34.

Related papers

US address classification based on text processing and machine learning
Vuk Batanović

Proceedings of the 29th Telecommunications Forum (TELFOR 2021), 2021

Addresses represent a crucial type of textual data for real estate companies. In order to identify, fix, or remove incorrect entries, we categorize addresses into one of six predefined classes. In this context, we explore the effects of different text processing and classification methods. The best results are obtained by using non-linear classifiers with a combination of unigram and bigram features.

View PDFchevron_right
Automatic Identification of Addresses: A Systematic Literature Review
Paulo Rita

ISPRS International Journal of Geo-Information, 2021

Address matching continues to play a central role at various levels, through geocoding and data integration from different sources, with a view to promote activities such as urban planning, location-based services, and the construction of databases like those used in census operations. However, the task of address matching continues to face several challenges, such as non-standard or incomplete address records or addresses written in more complex languages. In order to better understand how current limitations can be overcome, this paper conducted a systematic literature review focused on automated approaches to address matching and their evolution across time. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed, resulting in a final set of 41 papers published between 2002 and 2021, the great majority of which are after 2017, with Chinese authors leading the way. The main findings revealed a consistent move from more traditional a...

View PDFchevron_right
A Combined Semantic Search and Machine Learning Approach for Address Entity Resolution
Sujit Pal

2019

We have developed a comprehensive prototype solution for a specific use case involving entity resolution for mailing addresses of financial institutions. Our objective was to find matches between user entry of misspelled or inaccurate addresses of business entities and their corresponding entries in a “gold copy” of complete and accurate mailing addresses (dictionary). Three distinct matching methods (PySolr, SoDA and Record Linkage) were used for a preliminary, yet diverse scheme of lookups in finding matches. These lookup processes may optionally be followed by search via a hybrid machine learning (ML) model via regularized logistic regression and hierarchical clustering using Dedupe. Our experimental results of elapsed times for searches using the three lookup methods on a variety of match types suggest that majority of the simpler matches are detected extremely fast (elapsed times: ~ 6 – 48 milliseconds) at the lookup stage, making it suitable for detecting simple and possibly m...

View PDFchevron_right
A data model for efficient address data representation–Lesson learnt from the Intiendo address matching tool
Serena Coetzee

Geocoding refers to the process of assigning geographic identifiers and/or geographic coordinates to the description of a feature location, i.e. the words, codes or terms that describe a feature's location. Address matching is the specific case of geocoding where the description of a feature location comprises an address. Address matching is complicated by an incomplete or inaccurate incoming address or one that contains a misleading geographic identifier in its location type hierarchy. The Intiendo address matching tool is based on a data structure that is similar to the hierarchy of location types described in ISO 19112, but we have made some novel extensions to enable a spatial adjacency search. Intiendo does not rely purely on the alphanumeric match in the location type hierarchy but incorporates spatial proximity into the address matching. We also describe how the Intiendo address matching process can be configured and fine tuned, for example, by assigning weights to the location types in the hierarchy, and by specifying parameters for the spatial adjacency match. In this paper we present the hierarchical data structures of the Intiendo address matching tool and show how they are an extended implementation of the ISO 19112 general model. We show the similarities between the Intiendo and ISO 19112 models, and present the extensions that were implemented in Intiendo. By way of examples, we show that our extended model allows more efficient and accurate address matching incorporating spatial adjacency and hierarchical fine tuning.

View PDFchevron_right
Postal Address Detection from Web Documents
zhang qian

2005

An approach to postal address detection from webpages is proposed. The webpages are first segmented into text blocks based on their visual similarity. The text content in each block undergoes the recognition process, which employs a syntactic approach. The grammars of almost all possible patterns of postal addresses are built for this purpose. The results of our preliminary experiments on 44 webpages with 56 true addresses show that our approach can detect the postal addresses with a high precision (89.3%) and a low false alarms rate (3.8%).

View PDFchevron_right
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved