Add extractor configurations for Amharic

dbpedia · Aug 21, 2024 · fc4e376 · fc4e376
1 parent 2074df4
commit fc4e376
Show file tree

Hide file tree

Showing 6 changed files with 39 additions and 4 deletions.
diff --git a/...src/main/scala/org/dbpedia/extraction/config/mappings/DisambiguationExtractorConfig.scala b/...src/main/scala/org/dbpedia/extraction/config/mappings/DisambiguationExtractorConfig.scala
@@ -6,6 +6,7 @@ object DisambiguationExtractorConfig
     // For "ar" and "he" configurations, rendering right-to-left may seem like a bug, but it's not.
     // Don't change this unless you know what you're doing.
     val disambiguationTitlePartMap = Map(
+         "am" -> " (መንታ)",
          "ar" -> " (توضيح)",
          "bg" -> " (пояснение)",
          "ca" -> " (desambiguació)",

diff --git a/core/src/main/scala/org/dbpedia/extraction/config/mappings/GenderExtractorConfig.scala b/core/src/main/scala/org/dbpedia/extraction/config/mappings/GenderExtractorConfig.scala
@@ -6,6 +6,19 @@ object GenderExtractorConfig
     val pronounsMap = Map(
         "en" -> Map("she" -> "female", "her" -> "female", "he" -> "male", "his" -> "male", "him" -> "male", "herself" -> "female", "himself" -> "male",
                     "She" -> "female", "Her" -> "female", "He" -> "male", "His" -> "male", "Him" -> "male", "Herself" -> "female", "Himself" -> "male" //TODO why not just do case insensitive matches?
+        ),
+        "am" -> Map(
+          "እሷ" -> "ሴት",
+          "እሷን" -> "ሴት",
+          "የሷ" -> "ሴት",
+          "እራሷን" -> "ሴት",
+          "እራሷ" -> "ሴት",
+          "እሱ" -> "ወንድ",
+          "እሱን" -> "ወንድ",
+          "የእሱ" -> "ወንድ",
+          "የራሱ" -> "ወንድ",
+          "እራሱ" -> "ወንድ",
+          "እራሱን" -> "ወንድ"
         ),
          "pt" -> Map ("ela"-> "mulher", "dela" -> "mulher", "ele" -> "homem", "dele" -> "homem", "nela" -> "mulher", "nele" -> "homem",
                       "Ela"-> "mulher", "Dela" -> "mulher", "Ele" -> "homem", "Dele" -> "homem", "Nela" -> "mulher", "Nele" -> "homem"

diff --git a/core/src/main/scala/org/dbpedia/extraction/config/mappings/HomepageExtractorConfig.scala b/core/src/main/scala/org/dbpedia/extraction/config/mappings/HomepageExtractorConfig.scala
@@ -9,6 +9,23 @@ object HomepageExtractorConfig
     // Don't change this else if you know how it is done.
 
     private val propertyNamesMap = Map(
+        "am" -> Set(
+          "ድህረገፅ",
+          "ድህረ_ገፅ",
+          "ገጽ",
+          "ድህረ ገጽ",
+          "ድህረ_ገጽ",
+          "ድረ_ገፅ",
+          "ድረገፅ",
+          "ድረገጽ",
+          "ድረ ገጽ",
+          "ድረ_ገጽ",
+          "ዋና_ገጽ",
+          "ዌብሳይት",
+          "website",
+          "web",
+          "site"
+        ),
         "ar" -> Set("الموقع", "الصفحة الرسمية", "موقع", "الصفحة الرئيسية", "صفحة ويب", "موقع ويب"),
         "bg" -> Set("сайт", "уебсайт"),
         "ca" -> Set("pàgina", "web", "lloc"),
@@ -38,6 +55,7 @@ object HomepageExtractorConfig
     val supportedLanguages = propertyNamesMap.keySet
 
     private val externalLinkSectionsMap = Map(
+        "am" -> "(?:የውጭ ንባብ|የውጭ ማያያዣ)",
         "ar" -> "وصلات خارجية",
         "bg" -> "Външни препратки",
         "ca" -> "(?:Enllaços externs|Enllaço extern)",
@@ -65,6 +83,7 @@ object HomepageExtractorConfig
     }
 
     private val officialMap = Map(
+        "am" -> "ዋና",
         "ar" -> "رسمي",
         "bg" -> "официален",
         "ca" -> "oficial",

diff --git a/core/src/main/scala/org/dbpedia/extraction/config/mappings/ImageExtractorConfig.scala b/core/src/main/scala/org/dbpedia/extraction/config/mappings/ImageExtractorConfig.scala
@@ -12,6 +12,7 @@ object ImageExtractorConfig
     // Don't change this else if you know how it is done.
     val NonFreeRegex = Map(
            "ar" -> """(?i)\{\{\s?غير حر""".r,
+           "am" ->"""(?i)\{\{\s?(non-free|Logo|Screenshot|Noncommercial|ነፃ_ያልሆነ)""".r,
            "bg" ->"""(?i)\{\{\s?non-free""".r,
            "de" -> """(?iu)\{\{\s?(Dateiüberprüfung/benachrichtigt_\(Kategorie\)|Geschützt|Geschützt-Ungeklärt|Bild-LogoSH|Bild-PD-alt-100|Bild-PD-alt-1923|Bild-WikimediaCopyright)\s?\}\}""".r ,
            "el" -> """(?iu)\{\{\s?(εύλογη χρήση|σήμα|σήμα αθλητικού σωματείου|αφίσα ταινίας|σκηνή από ταινία|γραφικά υπολογιστή|εξώφυλλο άλμπουμ|εξώφυλλο βιβλίου|μη ελεύθερο έργο τέχνης|σελίδα κόμικς|σελίδα εφημερίδας|εικόνα-βιντεοπαιχνίδι|ιδιοκτησία Wikimedia)\s?\}\}""".r ,
@@ -29,9 +30,9 @@ object ImageExtractorConfig
            "ru" -> """(?iu)\{\{\s?(CopyrightByWikimedia|Fairuse|несвободный файл|несвободная лицензия|запрещенная лицензия)\s?\}\}""".r
     )
 
-    val flagRegex = """(?iu)s?^([^a-zA-Z0-9]*|[\w\s]*[^a-zA-Z0-9]+)(flag|banner|pavillon|drapeau|bandera|pabellón|bandiera|флаг)([^\w]*|[_\s]+)""".r
-    val mapRegex = """(?iu)s?^([^a-zA-Z0-9]*|[\w\s]*[^a-zA-Z0-9]+)(map|karte|location|position|carte|carta|lage)([^\w]*|[_\s]+)""".r
-    val signatureRegex = """(?iu)s?^([^a-zA-Z0-9]*|[\w\s]*[^a-zA-Z0-9]+)(signature|unterschrift)""".r
+    val flagRegex = """(?iu)s?^([^a-zA-Z0-9]*|[\w\s]*[^a-zA-Z0-9]+)(flag|banner|pavillon|drapeau|bandera|pabellón|bandiera|флаг|ባንዲራ|ሰንደቅ_ዓላማ)([^\w]*|[_\s]+)""".r
+    val mapRegex = """(?iu)s?^([^a-zA-Z0-9]*|[\w\s]*[^a-zA-Z0-9]+)(map|karte|location|position|carte|carta|lage|ካርታ)([^\w]*|[_\s]+)""".r
+    val signatureRegex = """(?iu)s?^([^a-zA-Z0-9]*|[\w\s]*[^a-zA-Z0-9]+)(signature|unterschrift|ፊርማ)""".r
     val cOARegex = """(?iu)s?^([^a-zA-Z0-9]*|[\w\s]*[^a-zA-Z0-9]+)(coat_of_arms|emblem|crest|wappen|grandes_armes|blason|armoiries)([^\w]*|[_\s]+)""".r
 
 

diff --git a/core/src/main/scala/org/dbpedia/extraction/config/mappings/InfoboxExtractorConfig.scala b/core/src/main/scala/org/dbpedia/extraction/config/mappings/InfoboxExtractorConfig.scala
@@ -12,6 +12,7 @@ object InfoboxExtractorConfig
 
     val ignoreProperties = Map (
         "en"-> Set("image", "image_photo", "map"),
+        "am"-> Set("ምስል", "ፎቶ", "ስዕል", "ካርታ", "አርማ"),
         "ar"-> Set("صورة"),
         "id"-> Set("foto", "gambar"),
         "el"-> Set("εικόνα", "εικονα", "Εικόνα", "Εικονα", "χάρτης", "Χάρτης"),

diff --git a/...rc/main/scala/org/dbpedia/extraction/config/mappings/TopicalConceptsExtractorConfig.scala b/...rc/main/scala/org/dbpedia/extraction/config/mappings/TopicalConceptsExtractorConfig.scala
@@ -10,7 +10,7 @@ object TopicalConceptsExtractorConfig
     val catMainTemplates = Set(
       "مزيد"   ,// ar
                                "Infocat", "Infocatm", // ca
-                               "Catmore", // el,ja
+                               "Catmore", // el,ja,am
                                "Cat main", // en
                                "AP", // es
                                "Nagusia", // eu