move everything to a single funding-acknowledgement model

doc/Benchmarking-biorxiv.md

@@ -263,7 +263,6 @@ Evaluation on 2000 random PDF files out of 2000 PDF (ratio 1.0).
 |--- |--- |--- |--- |--- |
 | availability_stmt | 0 | 0 | 0 | 0 |
 | figure_title | 4.24 | 2.01 | 2.72 | 22978 |
-| funding_stmt | 0 | 0 | 0 | 0 |
 | reference_citation | 71.04 | 71.33 | 71.18 | 147470 |
 | reference_figure | 70.59 | 67.74 | 69.13 | 47984 |
 | reference_table | 48.12 | 83.06 | 60.94 | 5957 |
@@ -282,7 +281,6 @@ Evaluation on 2000 random PDF files out of 2000 PDF (ratio 1.0).
 |--- |--- |--- |--- |--- |
 | availability_stmt | 0 | 0 | 0 | 0 |
 | figure_title | 69.47 | 32.89 | 44.65 | 22978 |
-| funding_stmt | 0 | 0 | 0 | 0 |
 | reference_citation | 83.03 | 83.37 | 83.2 | 147470 |
 | reference_figure | 71.21 | 68.34 | 69.75 | 47984 |
 | reference_table | 48.57 | 83.83 | 61.51 | 5957 |

grobid-core/src/main/java/org/grobid/core/GrobidModels.java

@@ -48,8 +48,8 @@ public enum GrobidModels implements GrobidModel {
  ASTRO("astro"),
  SOFTWARE("software"),
  DATASEER("dataseer"),
- ACKNOWLEDGEMENT("acknowledgement"),
- FUNDING("funding"),
+ //ACKNOWLEDGEMENT("acknowledgement"),
+ FUNDING_ACKNOWLEDGEMENT("funding-acknowledgement"),
  INFRASTRUCTURE("infrastructure"),
  DUMMY("none");
 

grobid-core/src/main/java/org/grobid/core/data/Funder.java

@@ -149,4 +149,13 @@ public void setLayoutTokens(List<LayoutToken> layoutTokens) {
  public void addLayoutTokens(List<LayoutToken> layoutTokens) {
  this.layoutTokens.addAll(layoutTokens);
  }
+
+ public String toString() {
+ StringBuilder builder = new StringBuilder();
+ if (fullName != null)
+ builder.append(fullName);
+ if (abbreviatedName != null)
+ builder.append(abbreviatedName);
+ return builder.toString();
+ }
 }

grobid-core/src/main/java/org/grobid/core/data/Funding.java

@@ -182,4 +182,25 @@ public boolean isValid() {
  else
  return false;
  }
+
+ public String toString() {
+ StringBuilder builder = new StringBuilder();
+ if (funder != null)
+ builder.append("funder: " + funder.toString() + "\n");
+ if (grantName != null)
+ builder.append("grant name: " + grantName.toString() + "\n");
+ if (grantNumber != null)
+ builder.append("grant number: " + grantNumber.toString() + "\n");
+ if (projectFullName != null)
+ builder.append("project name: " + projectFullName.toString() + "\n");
+ if (projectAbbreviatedName != null)
+ builder.append("project abbreviated name: " + projectAbbreviatedName.toString() + "\n");
+ if (programFullName != null)
+ builder.append("program name: " + programFullName.toString() + "\n");
+ if (programAbbreviatedName != null)
+ builder.append("program abbreviated name: " + programAbbreviatedName.toString() + "\n");
+ if (url != null)
+ builder.append("url: " + url.toString() + "\n");
+ return builder.toString();
+ }
 }

grobid-core/src/main/java/org/grobid/core/engines/EngineParsers.java

@@ -24,6 +24,7 @@ public class EngineParsers implements Closeable {
  private FigureParser figureParser = null;
  private TableParser tableParser = null;
  private MonographParser monographParser = null;
+ private FundingAcknowledgementParser fundingAcknowledgementParser = null;
 
  public AffiliationAddressParser getAffiliationAddressParser() {
  if (affiliationAddressParser == null) {
@@ -170,6 +171,17 @@ public MonographParser getMonographParser() {
  return monographParser;
  }
 
+ public FundingAcknowledgementParser getFundingAcknowledgementParser() {
+ if (fundingAcknowledgementParser == null) {
+ synchronized (this) {
+ if (fundingAcknowledgementParser == null) {
+ fundingAcknowledgementParser = new FundingAcknowledgementParser();
+ }
+ }
+ }
+ return fundingAcknowledgementParser;
+ } 
+
  /**
  * Init all model, this will also load the model into memory
  */
@@ -186,6 +198,7 @@ public void initAll() {
  figureParser = getFigureParser();
  tableParser = getTableParser();
  //MonographParser monographParser = getMonographParser();
+ fundingAcknowledgementParser = getFundingAcknowledgementParser();
  }
 
  @Override
@@ -268,7 +281,12 @@ public void close() throws IOException {
  LOGGER.debug("CLOSING monographParser");
  }
 
- LOGGER.debug("==> All resources closed");
+ if (fundingAcknowledgementParser != null) {
+ fundingAcknowledgementParser.close();
+ fundingAcknowledgementParser = null;
+ LOGGER.debug("CLOSING fundingAcknowledgementParser");
+ }
 
+ LOGGER.debug("==> All resources closed");
  }
 }

grobid-core/src/main/java/org/grobid/core/engines/FullTextParser.java

@@ -15,6 +15,8 @@
 import org.grobid.core.data.Figure;
 import org.grobid.core.data.Table;
 import org.grobid.core.data.Equation;
+import org.grobid.core.data.Funding;
+import org.grobid.core.data.Funder;
 import org.grobid.core.document.Document;
 import org.grobid.core.document.DocumentPiece;
 import org.grobid.core.document.DocumentPointer;
@@ -1909,7 +1911,7 @@ protected List<Figure> processFigures(String rese, List<LayoutToken> layoutToken
  for (TaggingTokenCluster cluster : Iterables.filter(clusteror.cluster(),
  new TaggingTokenClusteror.LabelTypePredicate(TaggingLabels.FIGURE))) {
  List<LayoutToken> tokenizationFigure = cluster.concatTokens();
- Figure result = parsers.getFigureParser().processing(
+ Figure result = this.parsers.getFigureParser().processing(
  tokenizationFigure,
  cluster.getFeatureBlock()
  );
@@ -2470,6 +2472,42 @@ private void toTEI(Document doc,
 
  tei.append("\t\t<back>\n");
 
+ // funding in header
+ StringBuilder fundingStmt = new StringBuilder();
+ if (StringUtils.isNotBlank(resHeader.getFunding())) {
+ List<LayoutToken> headerFundingTokens = resHeader.getLayoutTokens(TaggingLabels.HEADER_FUNDING);
+
+ List<Funding> fundings = this.parsers.getFundingAcknowledgementParser().processing(headerFundingTokens);
+ for (Funding funding : fundings) {
+ System.out.println(funding.toString());
+ }
+
+ Pair<String, List<LayoutToken>> headerFundingProcessed = processShort(headerFundingTokens, doc);
+ if (headerFundingProcessed != null) {
+ fundingStmt = teiFormatter.processTEIDivSection("funding",
+ "\t\t\t",
+ headerFundingProcessed.getLeft(),
+ headerFundingProcessed.getRight(),
+ resCitations,
+ config);
+ }
+ if (fundingStmt.length() > 0) {
+ tei.append(fundingStmt.toString());
+ }
+ }
+
+ // funding statements in non-header part
+ fundingStmt = getSectionAsTEI("funding",
+ "\t\t\t",
+ doc,
+ SegmentationLabels.FUNDING,
+ teiFormatter,
+ resCitations,
+ config);
+ if (fundingStmt.length() > 0) {
+ tei.append(fundingStmt);
+ }
+
  // acknowledgement is in the back
  StringBuilder acknowledgmentStmt = getSectionAsTEI("acknowledgement", "\t\t\t", doc, SegmentationLabels.ACKNOWLEDGEMENT,
  teiFormatter, resCitations, config);
@@ -2508,36 +2546,6 @@ private void toTEI(Document doc,
  tei.append(availabilityStmt.toString());
  }
 
- // funding in header
- StringBuilder fundingStmt = new StringBuilder();
- if (StringUtils.isNotBlank(resHeader.getFunding())) {
- List<LayoutToken> headerFundingTokens = resHeader.getLayoutTokens(TaggingLabels.HEADER_FUNDING);
- Pair<String, List<LayoutToken>> headerFundingProcessed = processShort(headerFundingTokens, doc);
- if (headerFundingProcessed != null) {
- fundingStmt = teiFormatter.processTEIDivSection("funding",
- "\t\t\t",
- headerFundingProcessed.getLeft(),
- headerFundingProcessed.getRight(),
- resCitations,
- config);
- }
- if (fundingStmt.length() > 0) {
- tei.append(fundingStmt.toString());
- }
- }
-
- // funding statements in non-header part
- fundingStmt = getSectionAsTEI("funding",
- "\t\t\t",
- doc,
- SegmentationLabels.FUNDING,
- teiFormatter,
- resCitations,
- config);
- if (fundingStmt.length() > 0) {
- tei.append(fundingStmt);
- }
-
  tei = teiFormatter.toTEIAnnex(tei, reseAnnex, resHeader, resCitations,
  tokenizationsAnnex, markerTypes, doc, config);
 

grobid-core/src/main/java/org/grobid/core/engines/FundingParser.java → grobid-core/src/main/java/org/grobid/core/engines/FundingAcknowledgementParser.java

@@ -37,12 +37,12 @@
 
 import static org.grobid.core.engines.label.TaggingLabels.*;
 
-public class FundingParser extends AbstractParser {
+public class FundingAcknowledgementParser extends AbstractParser {
 
- private static final Logger LOGGER = LoggerFactory.getLogger(FundingParser.class);
+ private static final Logger LOGGER = LoggerFactory.getLogger(FundingAcknowledgementParser.class);
 
- protected FundingParser() {
- super(GrobidModels.FUNDING);
+ protected FundingAcknowledgementParser() {
+ super(GrobidModels.FUNDING_ACKNOWLEDGEMENT);
  }
 
  /**
@@ -76,7 +76,7 @@ private List<Funding> getExtractionResult(List<LayoutToken> tokenizations, Strin
  // first funding
  Funding funding = new Funding();
 
- TaggingTokenClusteror clusteror = new TaggingTokenClusteror(GrobidModels.FUNDING, result, tokenizations);
+ TaggingTokenClusteror clusteror = new TaggingTokenClusteror(GrobidModels.FUNDING_ACKNOWLEDGEMENT, result, tokenizations);
  List<TaggingTokenCluster> clusters = clusteror.cluster();
  TaggingLabel previousLabel = null;
 

grobid-core/src/main/java/org/grobid/core/engines/label/TaggingLabels.java

@@ -121,6 +121,7 @@ public class TaggingLabels {
  public final static String PROJECT_NAME_LABEL = "<projectName>";
  public final static String PROJECT_ABBRV_NAME_LABEL = "<projectAbbrv>";
  public final static String URL_LABEL = "<url>";
+ public final static String PERSON_LABEL = "<person>";
 
  /* title page (secondary title page)
  * publisher page (publication information, including usually the copyrights info) 
@@ -260,16 +261,19 @@ public class TaggingLabels {
  public static final TaggingLabel MONOGRAPH_BACK = new TaggingLabelImpl(GrobidModels.MONOGRAPH, BACK_LABEL);
  public static final TaggingLabel MONOGRAPH_OTHER = new TaggingLabelImpl(GrobidModels.MONOGRAPH, OTHER_LABEL);
 
- public static final TaggingLabel FUNDING_FUNDER_NAME = new TaggingLabelImpl(GrobidModels.FUNDING, FUNDER_NAME_LABEL);
- public static final TaggingLabel FUNDING_FUNDER_ABBRV_NAME = new TaggingLabelImpl(GrobidModels.FUNDING, FUNDER_ABBRV_NAME_LABEL);
- public static final TaggingLabel FUNDING_PROGRAM_NAME = new TaggingLabelImpl(GrobidModels.FUNDING, PROGRAM_NAME_LABEL);
- public static final TaggingLabel FUNDING_PROGRAM_ABBRV_NAME = new TaggingLabelImpl(GrobidModels.FUNDING, PROGRAM_ABBRV_NAME_LABEL);
- public static final TaggingLabel FUNDING_GRANT_NUMBER = new TaggingLabelImpl(GrobidModels.FUNDING, GRANT_NUMBER_LABEL);
- public static final TaggingLabel FUNDING_GRANT_NAME = new TaggingLabelImpl(GrobidModels.FUNDING, GRANT_NAME_LABEL);
- public static final TaggingLabel FUNDING_PROJECT_NAME = new TaggingLabelImpl(GrobidModels.FUNDING, PROJECT_NAME_LABEL);
- public static final TaggingLabel FUNDING_PROJECT_ABBRV_NAME = new TaggingLabelImpl(GrobidModels.FUNDING, PROJECT_ABBRV_NAME_LABEL);
- public static final TaggingLabel FUNDING_URL = new TaggingLabelImpl(GrobidModels.FUNDING, URL_LABEL);
- public static final TaggingLabel FUNDING_OTHER = new TaggingLabelImpl(GrobidModels.FUNDING, OTHER_LABEL);
+ public static final TaggingLabel FUNDING_FUNDER_NAME = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, FUNDER_NAME_LABEL);
+ public static final TaggingLabel FUNDING_FUNDER_ABBRV_NAME = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, FUNDER_ABBRV_NAME_LABEL);
+ public static final TaggingLabel FUNDING_PROGRAM_NAME = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, PROGRAM_NAME_LABEL);
+ public static final TaggingLabel FUNDING_PROGRAM_ABBRV_NAME = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, PROGRAM_ABBRV_NAME_LABEL);
+ public static final TaggingLabel FUNDING_GRANT_NUMBER = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, GRANT_NUMBER_LABEL);
+ public static final TaggingLabel FUNDING_GRANT_NAME = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, GRANT_NAME_LABEL);
+ public static final TaggingLabel FUNDING_PROJECT_NAME = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, PROJECT_NAME_LABEL);
+ public static final TaggingLabel FUNDING_PROJECT_ABBRV_NAME = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, PROJECT_ABBRV_NAME_LABEL);
+ public static final TaggingLabel FUNDING_URL = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, URL_LABEL);
+ public static final TaggingLabel FUNDING_PERSON = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, PERSON_LABEL);
+ public static final TaggingLabel FUNDING_INSTITUTION = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, INSTITUTION_LABEL);
+ public static final TaggingLabel FUNDING_AFFILIATION = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, AFFILIATION_LABEL);
+ public static final TaggingLabel FUNDING_OTHER = new TaggingLabelImpl(GrobidModels.FUNDING_ACKNOWLEDGEMENT, OTHER_LABEL);
 
  protected static void register(TaggingLabel label) {
  cache.putIfAbsent(new Pair<>(label.getGrobidModel(), label.getLabel()), label);
@@ -404,7 +408,7 @@ protected static void register(TaggingLabel label) {
  register(MONOGRAPH_BACK);
  register(MONOGRAPH_OTHER);
 
- // funding
+ // funding-acknowledgement
  register(FUNDING_FUNDER_NAME);
  register(FUNDING_FUNDER_ABBRV_NAME);
  register(FUNDING_PROGRAM_NAME);
@@ -414,6 +418,9 @@ protected static void register(TaggingLabel label) {
  register(FUNDING_PROJECT_NAME);
  register(FUNDING_PROJECT_ABBRV_NAME);
  register(FUNDING_URL);
+ register(FUNDING_PERSON);
+ register(FUNDING_AFFILIATION);
+ register(FUNDING_INSTITUTION);
  register(FUNDING_OTHER);
  }
 

grobid-home/config/grobid.yaml

@@ -237,7 +237,7 @@ grobid:
  epsilon: 0.0001
  window: 20
 
- - name: "funding"
+ - name: "funding-acknowledgement"
  engine: "wapiti"
  #engine: "delft"
  wapiti: