Improving IR Performance from OCRed Text Using Cooccurrence

Kripabandhu Ghosh; Anirban Chakraborty; Swapan Kumar Parui

doi:10.1145/2701336.2701648

Improving IR Performance from OCRed Text Using Cooccurrence

Kripabandhu Ghosh, Anirban Chakraborty, Swapan Kumar Parui

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Information Retrieval performance is hurt to a great extent by OCR errors. Much research has been reported on modelling and correction of OCR errors. However, all the existing systems make use of language dependent resources or training texts to study the nature of errors. No research has been reported on improving retrieval performance from erroneous text when no training data is available. We propose a novel algorithm for automatic detection of OCR errors and improvement of retrieval performance from the erroneous corpus. Our algorithm does not use any training data or any language specific resources like thesaurus. It also does not use any knowledge about the language except that the word delimiter is blank space. We have tested our algorithm on erroneous OCRed Bangla FIRE collection offered in the RISOT 2012 track and obtained about 9% improvement over the OCRed baseline. However, the improvement is not statistically significant.

Original language	English
Title of host publication	Post-Proceedings of the 4th and 5th Workshops of the Forum for Information Retrieval Evaluation
Editors	Prasenjit Majumder, Mandar Mitra, Madhulika Agrawal, Parth Mehta
Place of Publication	New York, NY, USA
Publisher	Association for Computing Machinery (ACM)
Number of pages	7
ISBN (Print)	9781450328302
DOIs	https://doi.org/10.1145/2701336.2701648
Publication status	Published - 4 Dec 2013
Event	5th Workshop of the Forum for Information Retrieval Evaluation (FIRE) 2013 - New Delhi, India Duration: 4 Dec 2013 → 6 Dec 2013 Conference number: 5

Publication series

Name	FIRE '12 & '13
Publisher	Association for Computing Machinery

Workshop

Workshop	5th Workshop of the Forum for Information Retrieval Evaluation (FIRE) 2013
Abbreviated title	FIRE 2013
Country/Territory	India
City	New Delhi
Period	4/12/13 → 6/12/13

Keywords / Materials (for Non-textual outputs)

Query Expansion
Cooccurrence

Access to Document

10.1145/2701336.2701648Licence: All Rights Reserved

https://dl.acm.org/doi/10.1145/2701336.2701648Licence: All Rights Reserved

Cite this

Ghosh, K., Chakraborty, A., & Parui, S. K. (2013). Improving IR Performance from OCRed Text Using Cooccurrence. In P. Majumder, M. Mitra, M. Agrawal, & P. Mehta (Eds.), Post-Proceedings of the 4th and 5th Workshops of the Forum for Information Retrieval Evaluation Article 7 (FIRE '12 & '13). Association for Computing Machinery (ACM). https://doi.org/10.1145/2701336.2701648

Ghosh, Kripabandhu ; Chakraborty, Anirban ; Parui, Swapan Kumar. / Improving IR Performance from OCRed Text Using Cooccurrence. Post-Proceedings of the 4th and 5th Workshops of the Forum for Information Retrieval Evaluation. editor / Prasenjit Majumder ; Mandar Mitra ; Madhulika Agrawal ; Parth Mehta. New York, NY, USA : Association for Computing Machinery (ACM), 2013. (FIRE '12 & '13).

@inproceedings{f900e2e0a2e04550a4e7b2809933d385,

title = "Improving IR Performance from OCRed Text Using Cooccurrence",

abstract = "Information Retrieval performance is hurt to a great extent by OCR errors. Much research has been reported on modelling and correction of OCR errors. However, all the existing systems make use of language dependent resources or training texts to study the nature of errors. No research has been reported on improving retrieval performance from erroneous text when no training data is available. We propose a novel algorithm for automatic detection of OCR errors and improvement of retrieval performance from the erroneous corpus. Our algorithm does not use any training data or any language specific resources like thesaurus. It also does not use any knowledge about the language except that the word delimiter is blank space. We have tested our algorithm on erroneous OCRed Bangla FIRE collection offered in the RISOT 2012 track and obtained about 9% improvement over the OCRed baseline. However, the improvement is not statistically significant.",

keywords = "Query Expansion, Cooccurrence",

author = "Kripabandhu Ghosh and Anirban Chakraborty and Parui, {Swapan Kumar}",

year = "2013",

month = dec,

day = "4",

doi = "10.1145/2701336.2701648",

language = "English",

isbn = "9781450328302",

series = "FIRE '12 & '13",

publisher = "Association for Computing Machinery (ACM)",

editor = "Prasenjit Majumder and Mandar Mitra and Madhulika Agrawal and Parth Mehta",

booktitle = "Post-Proceedings of the 4th and 5th Workshops of the Forum for Information Retrieval Evaluation",

address = "United States",

note = "5th Workshop of the Forum for Information Retrieval Evaluation (FIRE) 2013, FIRE 2013 ; Conference date: 04-12-2013 Through 06-12-2013",

}

Ghosh, K, Chakraborty, A & Parui, SK 2013, Improving IR Performance from OCRed Text Using Cooccurrence. in P Majumder, M Mitra, M Agrawal & P Mehta (eds), Post-Proceedings of the 4th and 5th Workshops of the Forum for Information Retrieval Evaluation., 7, FIRE '12 & '13, Association for Computing Machinery (ACM), New York, NY, USA, 5th Workshop of the Forum for Information Retrieval Evaluation (FIRE) 2013, New Delhi, India, 4/12/13. https://doi.org/10.1145/2701336.2701648

Improving IR Performance from OCRed Text Using Cooccurrence. / Ghosh, Kripabandhu; Chakraborty, Anirban; Parui, Swapan Kumar.
Post-Proceedings of the 4th and 5th Workshops of the Forum for Information Retrieval Evaluation. ed. / Prasenjit Majumder; Mandar Mitra; Madhulika Agrawal; Parth Mehta. New York, NY, USA: Association for Computing Machinery (ACM), 2013. 7 (FIRE '12 & '13).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Improving IR Performance from OCRed Text Using Cooccurrence

AU - Ghosh, Kripabandhu

AU - Chakraborty, Anirban

AU - Parui, Swapan Kumar

N1 - Conference code: 5

PY - 2013/12/4

Y1 - 2013/12/4

N2 - Information Retrieval performance is hurt to a great extent by OCR errors. Much research has been reported on modelling and correction of OCR errors. However, all the existing systems make use of language dependent resources or training texts to study the nature of errors. No research has been reported on improving retrieval performance from erroneous text when no training data is available. We propose a novel algorithm for automatic detection of OCR errors and improvement of retrieval performance from the erroneous corpus. Our algorithm does not use any training data or any language specific resources like thesaurus. It also does not use any knowledge about the language except that the word delimiter is blank space. We have tested our algorithm on erroneous OCRed Bangla FIRE collection offered in the RISOT 2012 track and obtained about 9% improvement over the OCRed baseline. However, the improvement is not statistically significant.

AB - Information Retrieval performance is hurt to a great extent by OCR errors. Much research has been reported on modelling and correction of OCR errors. However, all the existing systems make use of language dependent resources or training texts to study the nature of errors. No research has been reported on improving retrieval performance from erroneous text when no training data is available. We propose a novel algorithm for automatic detection of OCR errors and improvement of retrieval performance from the erroneous corpus. Our algorithm does not use any training data or any language specific resources like thesaurus. It also does not use any knowledge about the language except that the word delimiter is blank space. We have tested our algorithm on erroneous OCRed Bangla FIRE collection offered in the RISOT 2012 track and obtained about 9% improvement over the OCRed baseline. However, the improvement is not statistically significant.

KW - Query Expansion

KW - Cooccurrence

U2 - 10.1145/2701336.2701648

DO - 10.1145/2701336.2701648

M3 - Conference contribution

SN - 9781450328302

T3 - FIRE '12 & '13

BT - Post-Proceedings of the 4th and 5th Workshops of the Forum for Information Retrieval Evaluation

A2 - Majumder, Prasenjit

A2 - Mitra, Mandar

A2 - Agrawal, Madhulika

A2 - Mehta, Parth

PB - Association for Computing Machinery (ACM)

CY - New York, NY, USA

T2 - 5th Workshop of the Forum for Information Retrieval Evaluation (FIRE) 2013

Y2 - 4 December 2013 through 6 December 2013

ER -

Improving IR Performance from OCRed Text Using Cooccurrence

Abstract

Publication series

Workshop

Keywords / Materials (for Non-textual outputs)

Access to Document

Fingerprint

Cite this