OCR correction based on document level knowledge
Identifieur interne :
000596 ( PascalFrancis/Corpus );
précédent :
000595;
suivant :
000597
OCR correction based on document level knowledge
Auteurs : T. Nartker ;
K. Taghva ;
R. Young ;
J. Borsack ;
A. ConditSource :
-
SPIE proceedings series [ 1017-2653 ] ; 2003.
RBID : Pascal:03-0421337
Descripteurs français
English descriptors
Abstract
For over 10 years, the Information Science Research Institute (ISRI) at UNLV has worked on problems associated with the electronic conversion of archival document collections. Such collections typically have a large fraction of poor quality images and present a special challenge to OCR (Optical Character Recognition) systems. Frequently, because of the size of the collection, manual correction of the output is not affordable. Because the output text is used only to build the index for an information retrieval (IR) system, the accuracy of non-stopwords is the most important measure of output quality. For these reasons, ISRI has focused on using document level knowledge as the best means of providing automatic correction of non-stopwords in OCR output. In 1998, we developed the MANICURE post-processing system that combined several document level corrections. Because of the high cost of obtaining accurate ground-truth text at the document level, we have never been able to quantify the accuracy improvement achievable using document level knowledge. In this report, we describe an experiment to measure the actual number (and percentage) of non-stopwords corrected by the MANICURE system. We believe this to be the first quantitative measure of OCR conversion improvement that is possible using document level knowledge.
Notice en format standard (ISO 2709)
Pour connaître la documentation sur le format Inist Standard.
| pA |
| A01 | 01 | 1 | | @0 1017-2653 |
|---|
| A05 | | | | @2 5010 |
|---|
| A08 | 01 | 1 | ENG | @1 OCR correction based on document level knowledge |
|---|
| A09 | 01 | 1 | ENG | @1 Document recognition and retrieval X : Santa Clara CA, 22-24 January 2003 |
|---|
| A11 | 01 | 1 | | @1 NARTKER (T.) |
|---|
| A11 | 02 | 1 | | @1 TAGHVA (K.) |
|---|
| A11 | 03 | 1 | | @1 YOUNG (R.) |
|---|
| A11 | 04 | 1 | | @1 BORSACK (J.) |
|---|
| A11 | 05 | 1 | | @1 CONDIT (A.) |
|---|
| A12 | 01 | 1 | | @1 KANUNGO (Tapas) @9 ed. |
|---|
| A12 | 02 | 1 | | @1 SMITH (Elisa H. Barney) @9 ed. |
|---|
| A12 | 03 | 1 | | @1 JIANYING HU @9 ed. |
|---|
| A12 | 04 | 1 | | @1 KANTOR (Paul B.) @9 ed. |
|---|
| A14 | 01 | | | @1 UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy @2 Las Vegas, NV @3 USA @Z 1 aut. @Z 2 aut. @Z 3 aut. @Z 4 aut. @Z 5 aut. |
|---|
| A18 | 01 | 1 | | @1 International Society for Optical Engineering @2 Bellingham WA @3 USA @9 patr. |
|---|
| A20 | | | | @1 103-110 |
|---|
| A21 | | | | @1 2003 |
|---|
| A23 | 01 | | | @0 ENG |
|---|
| A26 | 01 | | | @0 0-8194-4810-9 |
|---|
| A43 | 01 | | | @1 INIST @2 21760 @5 354000108522010120 |
|---|
| A44 | | | | @0 0000 @1 © 2003 INIST-CNRS. All rights reserved. |
|---|
| A45 | | | | @0 5 ref. |
|---|
| A47 | 01 | 1 | | @0 03-0421337 |
|---|
| A60 | | | | @1 P @2 C |
|---|
| A61 | | | | @0 A |
|---|
| A64 | 01 | 1 | | @0 SPIE proceedings series |
|---|
| A66 | 01 | | | @0 USA |
|---|
| C01 | 01 | | ENG | @0 For over 10 years, the Information Science Research Institute (ISRI) at UNLV has worked on problems associated with the electronic conversion of archival document collections. Such collections typically have a large fraction of poor quality images and present a special challenge to OCR (Optical Character Recognition) systems. Frequently, because of the size of the collection, manual correction of the output is not affordable. Because the output text is used only to build the index for an information retrieval (IR) system, the accuracy of non-stopwords is the most important measure of output quality. For these reasons, ISRI has focused on using document level knowledge as the best means of providing automatic correction of non-stopwords in OCR output. In 1998, we developed the MANICURE post-processing system that combined several document level corrections. Because of the high cost of obtaining accurate ground-truth text at the document level, we have never been able to quantify the accuracy improvement achievable using document level knowledge. In this report, we describe an experiment to measure the actual number (and percentage) of non-stopwords corrected by the MANICURE system. We believe this to be the first quantitative measure of OCR conversion improvement that is possible using document level knowledge. |
|---|
| C02 | 01 | X | | @0 001A01F |
|---|
| C02 | 02 | X | | @0 205 |
|---|
| C03 | 01 | X | FRE | @0 Reconnaissance optique caractère @5 01 |
|---|
| C03 | 01 | X | ENG | @0 Optical character recognition @5 01 |
|---|
| C03 | 01 | X | SPA | @0 Reconocimento óptico de caracteres @5 01 |
|---|
| C03 | 02 | X | FRE | @0 Précision @5 02 |
|---|
| C03 | 02 | X | ENG | @0 Accuracy @5 02 |
|---|
| C03 | 02 | X | SPA | @0 Precisión @5 02 |
|---|
| C03 | 03 | X | FRE | @0 Correction erreur @5 03 |
|---|
| C03 | 03 | X | ENG | @0 Error correction @5 03 |
|---|
| C03 | 03 | X | SPA | @0 Corrección error @5 03 |
|---|
| C03 | 04 | X | FRE | @0 Traitement document @5 04 |
|---|
| C03 | 04 | X | ENG | @0 Document processing @5 04 |
|---|
| C03 | 04 | X | SPA | @0 Tratamiento documento @5 04 |
|---|
| C03 | 05 | X | FRE | @0 Conversion donnée @5 05 |
|---|
| C03 | 05 | X | ENG | @0 Data conversion @5 05 |
|---|
| C03 | 05 | X | SPA | @0 Conversión datos @5 05 |
|---|
| C03 | 06 | X | FRE | @0 Système @5 06 |
|---|
| C03 | 06 | X | ENG | @0 System @5 06 |
|---|
| C03 | 06 | X | SPA | @0 Sistema @5 06 |
|---|
| C03 | 07 | X | FRE | @0 Evaluation performance @5 07 |
|---|
| C03 | 07 | X | ENG | @0 Performance evaluation @5 07 |
|---|
| C03 | 07 | X | SPA | @0 Evaluación prestación @5 07 |
|---|
| C03 | 08 | X | FRE | @0 MANICURE (système traitement document) @2 NI @4 INC @5 27 |
|---|
| C03 | 09 | X | FRE | @0 Mot vide @2 NI @4 CD @5 96 |
|---|
| C03 | 09 | X | ENG | @0 Stop word @2 NI @4 CD @5 96 |
|---|
| N21 | | | | @1 293 |
|---|
| N82 | | | | @1 PSI |
|---|
|
|---|
| pR |
| A30 | 01 | 1 | ENG | @1 Document recognition and retrieval. Conference @2 10 @3 Santa Clara CA USA @4 2003-01-22 |
|---|
|
|---|
Format Inist (serveur)
| NO : | PASCAL 03-0421337 INIST |
|---|
| ET : | OCR correction based on document level knowledge |
|---|
| AU : | NARTKER (T.); TAGHVA (K.); YOUNG (R.); BORSACK (J.); CONDIT (A.); KANUNGO (Tapas); SMITH (Elisa H. Barney); JIANYING HU; KANTOR (Paul B.) |
|---|
| AF : | UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy/Las Vegas, NV/Etats-Unis (1 aut., 2 aut., 3 aut., 4 aut., 5 aut.) |
|---|
| DT : | Publication en série; Congrès; Niveau analytique |
|---|
| SO : | SPIE proceedings series; ISSN 1017-2653; Etats-Unis; Da. 2003; Vol. 5010; Pp. 103-110; Bibl. 5 ref. |
|---|
| LA : | Anglais |
|---|
| EA : | For over 10 years, the Information Science Research Institute (ISRI) at UNLV has worked on problems associated with the electronic conversion of archival document collections. Such collections typically have a large fraction of poor quality images and present a special challenge to OCR (Optical Character Recognition) systems. Frequently, because of the size of the collection, manual correction of the output is not affordable. Because the output text is used only to build the index for an information retrieval (IR) system, the accuracy of non-stopwords is the most important measure of output quality. For these reasons, ISRI has focused on using document level knowledge as the best means of providing automatic correction of non-stopwords in OCR output. In 1998, we developed the MANICURE post-processing system that combined several document level corrections. Because of the high cost of obtaining accurate ground-truth text at the document level, we have never been able to quantify the accuracy improvement achievable using document level knowledge. In this report, we describe an experiment to measure the actual number (and percentage) of non-stopwords corrected by the MANICURE system. We believe this to be the first quantitative measure of OCR conversion improvement that is possible using document level knowledge. |
|---|
| CC : | 001A01F; 205 |
|---|
| FD : | Reconnaissance optique caractère; Précision; Correction erreur; Traitement document; Conversion donnée; Système; Evaluation performance; MANICURE (système traitement document); Mot vide |
|---|
| ED : | Optical character recognition; Accuracy; Error correction; Document processing; Data conversion; System; Performance evaluation; Stop word |
|---|
| SD : | Reconocimento óptico de caracteres; Precisión; Corrección error; Tratamiento documento; Conversión datos; Sistema; Evaluación prestación |
|---|
| LO : | INIST-21760.354000108522010120 |
|---|
| ID : | 03-0421337 |
|---|
Links to Exploration step
Pascal:03-0421337
Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" level="a">OCR correction based on document level knowledge</title>
<author><name sortKey="Nartker, T" sort="Nartker, T" uniqKey="Nartker T" first="T." last="Nartker">T. Nartker</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><author><name sortKey="Taghva, K" sort="Taghva, K" uniqKey="Taghva K" first="K." last="Taghva">K. Taghva</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><author><name sortKey="Young, R" sort="Young, R" uniqKey="Young R" first="R." last="Young">R. Young</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><author><name sortKey="Borsack, J" sort="Borsack, J" uniqKey="Borsack J" first="J." last="Borsack">J. Borsack</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><author><name sortKey="Condit, A" sort="Condit, A" uniqKey="Condit A" first="A." last="Condit">A. Condit</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><publicationStmt><idno type="wicri:source">INIST</idno>
<idno type="inist">03-0421337</idno>
<date when="2003">2003</date>
<idno type="stanalyst">PASCAL 03-0421337 INIST</idno>
<idno type="RBID">Pascal:03-0421337</idno>
<idno type="wicri:Area/PascalFrancis/Corpus">000596</idno>
</publicationStmt><sourceDesc><biblStruct><analytic><title xml:lang="en" level="a">OCR correction based on document level knowledge</title>
<author><name sortKey="Nartker, T" sort="Nartker, T" uniqKey="Nartker T" first="T." last="Nartker">T. Nartker</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><author><name sortKey="Taghva, K" sort="Taghva, K" uniqKey="Taghva K" first="K." last="Taghva">K. Taghva</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><author><name sortKey="Young, R" sort="Young, R" uniqKey="Young R" first="R." last="Young">R. Young</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><author><name sortKey="Borsack, J" sort="Borsack, J" uniqKey="Borsack J" first="J." last="Borsack">J. Borsack</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><author><name sortKey="Condit, A" sort="Condit, A" uniqKey="Condit A" first="A." last="Condit">A. Condit</name>
<affiliation><inist:fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</inist:fA14><series><title level="j" type="main">SPIE proceedings series</title>
<idno type="ISSN">1017-2653</idno>
<imprint><date when="2003">2003</date>
</imprint><seriesStmt><title level="j" type="main">SPIE proceedings series</title>
<idno type="ISSN">1017-2653</idno>
</seriesStmt><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Accuracy</term>
<term>Data conversion</term>
<term>Document processing</term>
<term>Error correction</term>
<term>Optical character recognition</term>
<term>Performance evaluation</term>
<term>Stop word</term>
<term>System</term>
</keywords><keywords scheme="Pascal" xml:lang="fr"><term>Reconnaissance optique caractère</term>
<term>Précision</term>
<term>Correction erreur</term>
<term>Traitement document</term>
<term>Conversion donnée</term>
<term>Système</term>
<term>Evaluation performance</term>
<term>MANICURE (système traitement document)</term>
<term>Mot vide</term>
</keywords><front><div type="abstract" xml:lang="en">For over 10 years, the Information Science Research Institute (ISRI) at UNLV has worked on problems associated with the electronic conversion of archival document collections. Such collections typically have a large fraction of poor quality images and present a special challenge to OCR (Optical Character Recognition) systems. Frequently, because of the size of the collection, manual correction of the output is not affordable. Because the output text is used only to build the index for an information retrieval (IR) system, the accuracy of non-stopwords is the most important measure of output quality. For these reasons, ISRI has focused on using document level knowledge as the best means of providing automatic correction of non-stopwords in OCR output. In 1998, we developed the MANICURE post-processing system that combined several document level corrections. Because of the high cost of obtaining accurate ground-truth text at the document level, we have never been able to quantify the accuracy improvement achievable using document level knowledge. In this report, we describe an experiment to measure the actual number (and percentage) of non-stopwords corrected by the MANICURE system. We believe this to be the first quantitative measure of OCR conversion improvement that is possible using document level knowledge.</div>
</front><inist><standard h6="B"><pA><fA01 i1="01" i2="1"><s0>1017-2653</s0>
</fA01><fA05><s2>5010</s2>
</fA05><fA08 i1="01" i2="1" l="ENG"><s1>OCR correction based on document level knowledge</s1>
</fA08><fA09 i1="01" i2="1" l="ENG"><s1>Document recognition and retrieval X : Santa Clara CA, 22-24 January 2003</s1>
</fA09><fA11 i1="01" i2="1"><s1>NARTKER (T.)</s1>
</fA11><fA11 i1="02" i2="1"><s1>TAGHVA (K.)</s1>
</fA11><fA11 i1="03" i2="1"><s1>YOUNG (R.)</s1>
</fA11><fA11 i1="04" i2="1"><s1>BORSACK (J.)</s1>
</fA11><fA11 i1="05" i2="1"><s1>CONDIT (A.)</s1>
</fA11><fA12 i1="01" i2="1"><s1>KANUNGO (Tapas)</s1>
<s9>ed.</s9>
</fA12><fA12 i1="02" i2="1"><s1>SMITH (Elisa H. Barney)</s1>
<s9>ed.</s9>
</fA12><fA12 i1="03" i2="1"><s1>JIANYING HU</s1>
<s9>ed.</s9>
</fA12><fA12 i1="04" i2="1"><s1>KANTOR (Paul B.)</s1>
<s9>ed.</s9>
</fA12><fA14 i1="01"><s1>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy</s1>
<s2>Las Vegas, NV</s2>
<s3>USA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
<sZ>5 aut.</sZ>
</fA14><fA18 i1="01" i2="1"><s1>International Society for Optical Engineering</s1>
<s2>Bellingham WA</s2>
<s3>USA</s3>
<s9>patr.</s9>
</fA18><fA20><s1>103-110</s1>
</fA20><fA21><s1>2003</s1>
</fA21><fA23 i1="01"><s0>ENG</s0>
</fA23><fA26 i1="01"><s0>0-8194-4810-9</s0>
</fA26><fA43 i1="01"><s1>INIST</s1>
<s2>21760</s2>
<s5>354000108522010120</s5>
</fA43><fA44><s0>0000</s0>
<s1>© 2003 INIST-CNRS. All rights reserved.</s1>
</fA44><fA45><s0>5 ref.</s0>
</fA45><fA47 i1="01" i2="1"><s0>03-0421337</s0>
</fA47><fA60><s1>P</s1>
<s2>C</s2>
</fA60><fA64 i1="01" i2="1"><s0>SPIE proceedings series</s0>
</fA64><fA66 i1="01"><s0>USA</s0>
</fA66><fC01 i1="01" l="ENG"><s0>For over 10 years, the Information Science Research Institute (ISRI) at UNLV has worked on problems associated with the electronic conversion of archival document collections. Such collections typically have a large fraction of poor quality images and present a special challenge to OCR (Optical Character Recognition) systems. Frequently, because of the size of the collection, manual correction of the output is not affordable. Because the output text is used only to build the index for an information retrieval (IR) system, the accuracy of non-stopwords is the most important measure of output quality. For these reasons, ISRI has focused on using document level knowledge as the best means of providing automatic correction of non-stopwords in OCR output. In 1998, we developed the MANICURE post-processing system that combined several document level corrections. Because of the high cost of obtaining accurate ground-truth text at the document level, we have never been able to quantify the accuracy improvement achievable using document level knowledge. In this report, we describe an experiment to measure the actual number (and percentage) of non-stopwords corrected by the MANICURE system. We believe this to be the first quantitative measure of OCR conversion improvement that is possible using document level knowledge.</s0>
</fC01><fC02 i1="01" i2="X"><s0>001A01F</s0>
</fC02><fC02 i1="02" i2="X"><s0>205</s0>
</fC02><fC03 i1="01" i2="X" l="FRE"><s0>Reconnaissance optique caractère</s0>
<s5>01</s5>
</fC03><fC03 i1="01" i2="X" l="ENG"><s0>Optical character recognition</s0>
<s5>01</s5>
</fC03><fC03 i1="01" i2="X" l="SPA"><s0>Reconocimento óptico de caracteres</s0>
<s5>01</s5>
</fC03><fC03 i1="02" i2="X" l="FRE"><s0>Précision</s0>
<s5>02</s5>
</fC03><fC03 i1="02" i2="X" l="ENG"><s0>Accuracy</s0>
<s5>02</s5>
</fC03><fC03 i1="02" i2="X" l="SPA"><s0>Precisión</s0>
<s5>02</s5>
</fC03><fC03 i1="03" i2="X" l="FRE"><s0>Correction erreur</s0>
<s5>03</s5>
</fC03><fC03 i1="03" i2="X" l="ENG"><s0>Error correction</s0>
<s5>03</s5>
</fC03><fC03 i1="03" i2="X" l="SPA"><s0>Corrección error</s0>
<s5>03</s5>
</fC03><fC03 i1="04" i2="X" l="FRE"><s0>Traitement document</s0>
<s5>04</s5>
</fC03><fC03 i1="04" i2="X" l="ENG"><s0>Document processing</s0>
<s5>04</s5>
</fC03><fC03 i1="04" i2="X" l="SPA"><s0>Tratamiento documento</s0>
<s5>04</s5>
</fC03><fC03 i1="05" i2="X" l="FRE"><s0>Conversion donnée</s0>
<s5>05</s5>
</fC03><fC03 i1="05" i2="X" l="ENG"><s0>Data conversion</s0>
<s5>05</s5>
</fC03><fC03 i1="05" i2="X" l="SPA"><s0>Conversión datos</s0>
<s5>05</s5>
</fC03><fC03 i1="06" i2="X" l="FRE"><s0>Système</s0>
<s5>06</s5>
</fC03><fC03 i1="06" i2="X" l="ENG"><s0>System</s0>
<s5>06</s5>
</fC03><fC03 i1="06" i2="X" l="SPA"><s0>Sistema</s0>
<s5>06</s5>
</fC03><fC03 i1="07" i2="X" l="FRE"><s0>Evaluation performance</s0>
<s5>07</s5>
</fC03><fC03 i1="07" i2="X" l="ENG"><s0>Performance evaluation</s0>
<s5>07</s5>
</fC03><fC03 i1="07" i2="X" l="SPA"><s0>Evaluación prestación</s0>
<s5>07</s5>
</fC03><fC03 i1="08" i2="X" l="FRE"><s0>MANICURE (système traitement document)</s0>
<s2>NI</s2>
<s4>INC</s4>
<s5>27</s5>
</fC03><fC03 i1="09" i2="X" l="FRE"><s0>Mot vide</s0>
<s2>NI</s2>
<s4>CD</s4>
<s5>96</s5>
</fC03><fC03 i1="09" i2="X" l="ENG"><s0>Stop word</s0>
<s2>NI</s2>
<s4>CD</s4>
<s5>96</s5>
</fC03><fN21><s1>293</s1>
</fN21><fN82><s1>PSI</s1>
</fN82><pR><fA30 i1="01" i2="1" l="ENG"><s1>Document recognition and retrieval. Conference</s1>
<s2>10</s2>
<s3>Santa Clara CA USA</s3>
<s4>2003-01-22</s4>
</fA30><server><NO>PASCAL 03-0421337 INIST</NO>
<ET>OCR correction based on document level knowledge</ET>
<AU>NARTKER (T.); TAGHVA (K.); YOUNG (R.); BORSACK (J.); CONDIT (A.); KANUNGO (Tapas); SMITH (Elisa H. Barney); JIANYING HU; KANTOR (Paul B.)</AU>
<AF>UNLV/Information Science Research Institute, Box 4021, 4505 Maryland Pkwy/Las Vegas, NV/Etats-Unis (1 aut., 2 aut., 3 aut., 4 aut., 5 aut.)</AF>
<DT>Publication en série; Congrès; Niveau analytique</DT>
<SO>SPIE proceedings series; ISSN 1017-2653; Etats-Unis; Da. 2003; Vol. 5010; Pp. 103-110; Bibl. 5 ref.</SO>
<LA>Anglais</LA>
<EA>For over 10 years, the Information Science Research Institute (ISRI) at UNLV has worked on problems associated with the electronic conversion of archival document collections. Such collections typically have a large fraction of poor quality images and present a special challenge to OCR (Optical Character Recognition) systems. Frequently, because of the size of the collection, manual correction of the output is not affordable. Because the output text is used only to build the index for an information retrieval (IR) system, the accuracy of non-stopwords is the most important measure of output quality. For these reasons, ISRI has focused on using document level knowledge as the best means of providing automatic correction of non-stopwords in OCR output. In 1998, we developed the MANICURE post-processing system that combined several document level corrections. Because of the high cost of obtaining accurate ground-truth text at the document level, we have never been able to quantify the accuracy improvement achievable using document level knowledge. In this report, we describe an experiment to measure the actual number (and percentage) of non-stopwords corrected by the MANICURE system. We believe this to be the first quantitative measure of OCR conversion improvement that is possible using document level knowledge.</EA>
<CC>001A01F; 205</CC>
<FD>Reconnaissance optique caractère; Précision; Correction erreur; Traitement document; Conversion donnée; Système; Evaluation performance; MANICURE (système traitement document); Mot vide</FD>
<ED>Optical character recognition; Accuracy; Error correction; Document processing; Data conversion; System; Performance evaluation; Stop word</ED>
<SD>Reconocimento óptico de caracteres; Precisión; Corrección error; Tratamiento documento; Conversión datos; Sistema; Evaluación prestación</SD>
<LO>INIST-21760.354000108522010120</LO>
<ID>03-0421337</ID>
</server>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Ticri/CIDE/explor/OcrV1/Data/PascalFrancis/Corpus
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 000596 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/PascalFrancis/Corpus/biblio.hfd -nk 000596 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Ticri/CIDE
|area= OcrV1
|flux= PascalFrancis
|étape= Corpus
|type= RBID
|clé= Pascal:03-0421337
|texte= OCR correction based on document level knowledge
}}
| This area was generated with Dilib version V0.6.32.
Data generation: Sat Nov 11 16:53:45 2017. Site generation: Mon Mar 11 23:15:16 2024 |  |
