[GSoC 2024] DBpedia Hindi chapter

core/src/main/scala/org/dbpedia/extraction/config/mappings/DisambiguationExtractorConfig.scala

@@ -21,6 +21,7 @@ object DisambiguationExtractorConfig
          "ga" -> " (idirdhealáin)",
          "gl" -> " (homónimos)",
          "he" -> " (פירושונים)",
+         "hi" -> " (बहुविकल्पी)", // eg. https://hi.wikipedia.org/wiki/आयरलैण्ड_(बहुविकल्पी)
          "hu" -> " (egyértelműsítő lap)",
          "id" -> " (disambig)",
          "it" -> " (disambigua)",

core/src/main/scala/org/dbpedia/extraction/config/mappings/HomepageExtractorConfig.scala

@@ -21,6 +21,7 @@ object HomepageExtractorConfig
         "eu" -> Set("webgunea"),
         "fr" -> Set("website", "homepage", "web", "site", "siteweb", "site web"),/*cleanup*/
         "ga" -> Set("suíomh"),
+        "hi" -> Set("वेबसाइट"),
         "it" -> Set("homepage", "sito", "sito web"),
         "ja" -> Set("homepage", "website", "web", "siteweb", "HP", "ホームページ", "ウェブ", "サイト", "ウェブサイト", "公式サイト"),
         "mk" -> Set("Портал", "Мреж. место"),
@@ -50,6 +51,7 @@ object HomepageExtractorConfig
         "eu" -> "Kanpo loturak?",
         "fr" -> "(?:Lien externe|Liens externes|Liens et documents externes)",
         "ga" -> "(?:Naisc sheachtracha|Nasc sheachtrach)",
+        "hi" -> "बाहरी कड़ियाँ",
         "it" -> "Collegamenti esterni",
         "ja" -> "外部リンク",
         "mk" -> "Надворешни врски",
@@ -77,6 +79,7 @@ object HomepageExtractorConfig
         "eu" -> "ofiziala?",
         "fr" -> "officiel",
         "ga" -> "oifigiúil",
+        "hi" -> "आधिकारिक",
         "it" -> "ufficiale",
         "ja" -> "(?:公式|オフィシャル)",
         "mk" -> "официјален",
@@ -102,6 +105,7 @@ object HomepageExtractorConfig
         "es" -> Map("Página_web" -> "1"),
         "fr" -> Map("Site_officiel" -> "url"),
         "ga" -> Map("Páxina_web" -> "1"),
+        "hi" -> Map("आधिकारिक वेबसाइट" -> "1"),
         "ja" -> Map("Official website" -> "1"),
         "pt" -> Map("Oficial" -> "1"),
         "ru" -> Map("Официальный сайт" -> "1"),

core/src/main/scala/org/dbpedia/extraction/config/mappings/ImageExtractorConfig.scala

@@ -20,6 +20,7 @@ object ImageExtractorConfig
            "eu" -> """(?i)\{\{\s?(Cc-by-nc-sa-2.5|Wikimedia_logoa|Copyrightdun_logoa|Lizentzia_gabea|Album_azala|Aldizkari_azala|Fair_use|Bideo-zinta_azala|Dirua|DVD_azala|Egunkari_azala|Film_pantaila_irudia|Film_posterra|HQFL_logotipoa|Ikonoa|Ikurra|Irrati_logotipoa|Jatetxe_logotipoa|Joku_azala|Joku_pantaila_irudia|Kirol_logotipoa|Komiki_azala|Liburu_azala|Logotipoa|Mahai-joku_azala|Olinpiada_logotipoa|Politika_posterra|Propaganda|Software_azala|Software_pantaila_irudia|Zigilua|TB_pantaila_irudia|Web_pantaila_irudia)\s?\}\}""".r,
            "fr" -> """(?iu)\{\{\s?(Copyright by Wikimedia|Copyvio|Logo|Screenshot|Ordnance Survey Copyright|Fairuse|Noncommercial|PolandGov|nonderivative|NZCrownCopyright|PD-IndiaGov|ADRM2|Marque déposée)\s?\}\}""".r,
            "gl" -> """(?iu)\{\{\s?(non-free|Copyright by Wikimedia|Copyvio|Logo|Screenshot|PD-CAGov|Fairuse|Noncommercial|Nonderivative|NZCrownCopyright|PolandGov|PD-IndiaGov|ADRM2)\s?\}\}""".r,
+           "hi" -> """(?i)\{\{\s?non-free""".r,
            "id" -> """(?i)\{\{\s?(non-free|Fairuse|Logo|LogoOlahraga|LogoTV|FotoHistoris|GambarKarakter|Promophoto|Smithsonian|TampilanFilm|TampilanVideo|TampilanSitus|TampilanPermainan|GambarUang|GambarPerangko|SampulVideo|SampulPermainan|SampulAlbum|SampulBuku|Poster|GambarBerhakTayangBersyarat)\s?\}\}""".r,
            "it" -> """(?iu)\{\{\s?(Sconosciuto|Riservato|NonCommerciale|Unknown|Noncommercial|Nonderivative|Copyrighted|Screenshot|Ordinance Survey Copyright|Fairuse|Cc-nc|cc-by-nc|cc-by-nc-2.0|cc-nc-sa|cc-by-nc-sa|Cc-by-nc-sa-1.0|cc-by-nc-sa-2.0|cc-nd-nc|cc-by-nd-nc|cc-by-nd-nc-2.0|cc-nd|cc-by-nd|cc-by-nd-2.0|TW-cc-by-nc-nd-2.0|TW-cc-by-nc-sa-2.0|Copyright by Wikimedia|CopyrightbyWikimedia)\s?\}\}""".r,
            "ja" -> """(?iu)\{\{\s?(Copyright by Wikimedia|Copyvio|Logo|Screenshot|PD-CAGov|Fair use|Noncommercial|PolandGov|Nonderivative|NZCrownCopyright|PD-IndiaGov|ADRM2|RomanianGovernmentCopyright|FrenchMinistryOfForeignAffairs|IRFCA|Members of the Riksdag|Attribution-Ubisoft)\s?\}\}""".r,

dump/extraction.default.properties

@@ -65,7 +65,7 @@ extractors.ga=.MappingExtractor,.HomepageExtractor
 
 extractors.gl=.MappingExtractor
 
-extractors.hi=.MappingExtractor
+extractors.hi=.MappingExtractor,.HomepageExtractor,.DisambiguationExtractor,.TopicalConceptsExtractor,.ImageExtractorNew,.AnchorTextExtractor,.CommonsResourceExtractor
 
 extractors.hr=.MappingExtractor
 

dump/extraction.spark.properties

@@ -68,7 +68,7 @@ extractors.ga=.HomepageExtractor
 
 extractors.gl=
 
-extractors.hi=
+extractors.hi=.HomepageExtractor,.DisambiguationExtractor,.TopicalConceptsExtractor,.AnchorTextExtractor,.CommonsResourceExtractor
 
 extractors.hr=