From 1eb865b088e376e2cafa23904dc8a6e542fd7df5 Mon Sep 17 00:00:00 2001
From: Cyprian Laskowski <cyp@trojina.si>
Date: Fri, 21 Feb 2020 20:47:54 +0100
Subject: [PATCH] IssueID #959: filled in valency schema and put in manual
 example

---
 examples/valency_lexicon/valency1.xml         |  39 ----
 .../valency_manual_example.xml                | 180 ++++++++++++++++++
 resources/schema/inventory.xsd                | 111 ++++++++++-
 resources/schema/valency_lexicon.xsd          |  14 +-
 4 files changed, 301 insertions(+), 43 deletions(-)
 delete mode 100644 examples/valency_lexicon/valency1.xml
 create mode 100644 examples/valency_lexicon/valency_manual_example.xml
diff --git a/examples/valency_lexicon/valency1.xml b/examples/valency_lexicon/valency1.xml
deleted file mode 100644
index 2c0d229..0000000
--- a/examples/valency_lexicon/valency1.xml
+++ /dev/null
@@ -1,39 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<lexicon>
-  <entry>
-    <head>
-      <status>testing</status>
-      <headword>
-        <lemma>blahblahblahati</lemma>
-      </headword>
-      <grammar>
-        <category>glagol</category>
-      </grammar>
-      <measureList>
-        <measure type="frequency">88888</measure>
-      </measureList>
-    </head>
-    <body>
-      <senseList/>
-    </body>
-  </entry>
-  <entry>
-    <head>
-      <status>testing</status>
-      <headword>
-        <lemma>bluhbluhbluhati</lemma>
-      </headword>
-      <grammar>
-        <category>glagol</category>
-      </grammar>
-      <measureList>
-        <measure type="frequency">999999</measure>
-      </measureList>
-    </head>
-    <body>
-      <senseList/>
-    </body>
-  </entry>
-</lexicon>
-      
-
diff --git a/examples/valency_lexicon/valency_manual_example.xml b/examples/valency_lexicon/valency_manual_example.xml
new file mode 100644
index 0000000..78ef4b0
--- /dev/null
+++ b/examples/valency_lexicon/valency_manual_example.xml
@@ -0,0 +1,180 @@
+﻿<dictionary>
+  <entry>
+    <head>
+      <status>confirmed</status>
+      <headword>
+        <lemma>aktivirati</lemma>
+      </headword>
+      <lexicalUnit id="317" type="single">
+        <lexeme lexical_unit_lexeme_id="317">aktivirati</lexeme>
+      </lexicalUnit>
+      <grammar>
+        <category>glagol</category>
+        <grammarFeature name="Aspect">biaspectual</grammarFeature>
+        <!-- iz Sloleksa -->
+      </grammar>
+      <measureList>
+        <measure type="frequency" source="Gigafida 2.0">11410</measure>
+        <!-- iz Gigafide -->
+        <measure type="frequency" source="ssj500k 2.2">5</measure>
+        <!-- iz ssj500k -->
+      </measureList>
+      <labelList>
+        <label type="Levin_class" feature_id="55">"begin" verbs</label>
+        <!-- iz Levin classes list http://www-personal.umich.edu/~jlawler/levin.verbs -->
+        <label type="Žele_class" feature_id="56">glagol ravnanja</label>
+        <!-- Žele: vezljivostni slovar -->
+      </labelList>
+    </head>
+    <body>
+      <statisticsContainerList>
+        <statisticsContainer>
+          <semanticRole>PAT</semanticRole>
+          <measureList>
+            <measure type="valency_pattern_ratio" ratio_base="148">50</measure>
+            <!-- odstotek vzorcev, v katerih se pojavlja PAT -->
+            <measure type="valency_sentence_ratio">100</measure>
+            <!-- odstotek stavkov (v kateremkoli vzorcu), v katerih se pojavlja PAT -->
+          </measureList>
+        </statisticsContainer>
+        <statisticsContainer>
+          <semanticRole>AIM</semanticRole>
+          <measureList>
+            <measure type="valency_pattern_ratio" ratio_base="1">1</measure>
+            <!-- odstotek vzorcev, v katerih se pojavlja AIM -->
+            <measure type="valency_sentence_ratio">100</measure>
+            <!-- odstotek stavkov (v kateremkoli vzorcu), v katerih se pojavlja AIM -->
+          </measureList>
+        </statisticsContainer>
+      </statisticsContainerList>
+      <senseList>
+        <sense id="317">
+          <definitionList/>
+          <valencyPatternList>
+            <valencyPattern id="2">
+              <measureList>
+                <measure type="frequency_all" source="Gigaida 2.0">7</measure>
+                <!-- število stavkov s tem vzorcem -->
+                <measure type="frequency_all" source="ssj500k 2.2">1</measure>
+              </measureList>
+              <semanticRoleContainerList>
+                <semanticRoleContainer>
+                  <semanticRole>PAT</semanticRole>
+                  <syntacticStructureList>
+                    <syntacticStructure id="8"/>
+                    <!-- iz datoteke: Kolokacije_strukture_JOS-32-representation.xml -->
+                  </syntacticStructureList>
+                </semanticRoleContainer>
+                <semanticRoleContainer>
+                  <semanticRole>AIM</semanticRole>
+                  <syntacticStructureList>
+                    <syntacticStructure id="36">
+                      <comp structure_component_index="3">za</comp>
+                    </syntacticStructure>
+                  </syntacticStructureList>
+                </semanticRoleContainer>
+              </semanticRoleContainerList>
+              <patternRepresentation>aktivirati KOGA/KAJ S KAKŠNIM NAMENOM</patternRepresentation>
+              <exampleContainerList>
+                <exampleContainer>
+                  <corpusExample corpus_id="1" modified="false">
+                    Glede vabe ušata res ni izbirčna
+                    <tree role="AIM">
+                      <comp syntactic_structure_id="36" structure_component_index="3">za</comp>
+                      uspešen
+                      <comp syntactic_structure_id="36" structure_component_index="4">ribolov</comp>
+                    </tree>
+                    pa
+                    <tree role="PAT">
+                      <comp syntactic_structure_id="8" structure_component_index="4">jo</comp>
+                    </tree>
+                    moramo
+                    <comp role="headword">aktivirati</comp>,
+                    jih obnoreti, preslepiti.
+                  </corpusExample>
+                </exampleContainer>
+                <exampleContainer>
+                  <corpusExample modified="false">
+                    Glede vabe ušata res ni izbirčna
+                    <tree role="AIM">
+                      <comp syntactic_structure_id="36" structure_component_index="3">za</comp>
+                      uspešen
+                      <comp syntactic_structure_id="36" structure_component_index="4">ribolov</comp>
+                    </tree>
+                    pa
+                    <tree role="PAT">
+                      <comp syntactic_structure_id="8" structure_component_index="4">jo</comp>
+                    </tree>
+                    moramo
+                    <comp role="headword">aktivirati</comp>,
+                    jih obnoreti, preslepiti.
+                  </corpusExample>
+                </exampleContainer>
+              </exampleContainerList>
+            </valencyPattern>
+            <valencyPattern id="3">
+              <measureList>
+                <measure type="frequency_all" source="GF2.0">20</measure>
+                <!-- število stavkov s tem vzorcem -->
+                <measure type="frequency_all" source="ssj500k">0</measure>
+              </measureList>
+              <semanticRoleContainerList>
+                <semanticRoleContainer>
+                  <semanticRole>ACT</semanticRole>
+                  <syntacticStructureList>
+                    <syntacticStructure id="9"/>
+                    <!-- iz datoteke: Kolokacije_strukture_JOS-32-representation.xml -->
+                  </syntacticStructureList>
+                </semanticRoleContainer>
+                <semanticRoleContainer>
+                  <semanticRole>PAT</semanticRole>
+                  <syntacticStructureList>
+                    <syntacticStructure id="8"/>
+                    <syntacticStructure id="33">
+                      <comp structure_component_index="2">z/s</comp>
+                    </syntacticStructure>
+                  </syntacticStructureList>
+                </semanticRoleContainer>
+                <semanticRoleContainer>
+                  <semanticRole>MANN</semanticRole>
+                  <syntacticStructureList system="JOS">
+                    <syntacticStructure id="19"/>
+                    <syntacticStructure id="36">
+                      <comp structure_component_index="3">na</comp>
+                      <comp structure_component_index="3">pred</comp>
+                    </syntacticStructure>
+                    <syntacticStructure id="33">
+                      <comp structure_component_index="2">z/s</comp>
+                    </syntacticStructure>
+                    <syntacticStructure id="34">
+                      <comp structure_component_index="3">preko</comp>
+                    </syntacticStructure>
+                  </syntacticStructureList>
+                </semanticRoleContainer>
+              </semanticRoleContainerList>
+              <patternRepresentation>KDO/KAJ aktivira KOGA/KAJ KAKO</patternRepresentation>
+              <exampleContainerList>
+                <exampleContainer>
+                  <corpusExample modified="false">
+                    <tree role="ACT">
+                      <comp syntactic_structure_id="9" structure_component_index="1">Poljubljanje</comp>
+                    </tree>
+                    <tree role="MANN">
+                      <comp syntactic_structure_id="19" structure_component_index="1">hitro</comp>
+                    </tree>
+                    <comp role="headword">aktivira</comp>
+                    <tree role="PAT">
+                      <comp syntactic_structure_id="8" structure_component_index="4">hormon</comp>
+                      sreče
+                    </tree>
+                    in krepi imunski sistem.
+                  </corpusExample>
+                </exampleContainer>
+              </exampleContainerList>
+            </valencyPattern>
+          </valencyPatternList>
+        </sense>
+      </senseList>
+    </body>
+  </entry>
+</dictionary>
diff --git a/resources/schema/inventory.xsd b/resources/schema/inventory.xsd
index e13f984..9ad95a6 100644
--- a/resources/schema/inventory.xsd
+++ b/resources/schema/inventory.xsd
@@ -103,6 +103,7 @@
       <xsd:extension base="xsd:decimal">
         <xsd:attribute name="type" type="xsd:string" use="required"/>
         <xsd:attribute name="source" type="xsd:string"/>
+        <xsd:attribute name="ratio_base" type="xsd:int"/>
       </xsd:extension>
     </xsd:simpleContent>
   </xsd:complexType>
@@ -185,8 +186,10 @@
     <xsd:simpleContent>
       <xsd:extension base="xsd:string">
         <xsd:attribute name="lexical_unit_lexeme_id" type="xsd:int"/>
+        <xsd:attribute name="syntactic_structure_id" type="xsd:int"/>
+        <xsd:attribute name="structure_component_index" type="xsd:int"/>
         <xsd:attribute name="lexeme_id" type="xsd:int"/>
-        <xsd:attribute name="role" use="required">
+        <xsd:attribute name="role">
           <xsd:simpleType>
             <xsd:restriction base="xsd:string">
               <xsd:enumeration value="headword"/>
@@ -201,10 +204,21 @@
     </xsd:simpleContent>
   </xsd:complexType>
 
+  <xsd:complexType name="exampleTreeType" mixed="true">
+    <xsd:sequence>
+      <xsd:element name="comp" type="componentLexemeType" maxOccurs="unbounded"/>
+    </xsd:sequence>
+    <xsd:attribute name="role" type="xsd:string"/>
+  </xsd:complexType>
+
   <xsd:complexType name="corpusExampleType" mixed="true">
     <xsd:sequence>
-      <xsd:element name="comp" type="componentLexemeType" minOccurs="0" maxOccurs="unbounded"/>
+      <xsd:choice minOccurs="0" maxOccurs="unbounded">
+        <xsd:element name="comp" type="componentLexemeType"/>
+        <xsd:element name="tree" type="exampleTreeType"/>
+      </xsd:choice>
     </xsd:sequence>
+    <xsd:attribute name="corpus_id" type="xsd:int"/>
     <xsd:attribute name="example_id" type="xsd:int"/>
     <xsd:attribute name="modified" type="xsd:boolean"/>
     <xsd:attribute name="lexical_unit_id" type="xsd:int"/>
@@ -264,4 +278,97 @@
     </xsd:sequence>
   </xsd:complexType>
 
+  <xsd:simpleType name="semanticRoleType">
+    <xsd:restriction base="xsd:string">
+      <xsd:enumeration value="ACT"/>
+      <xsd:enumeration value="PAT"/>
+      <xsd:enumeration value="REC"/>
+      <xsd:enumeration value="ORIG"/>
+      <xsd:enumeration value="RESLT"/>
+      <xsd:enumeration value="TIME"/>
+      <xsd:enumeration value="DUR"/>
+      <xsd:enumeration value="FREQ"/>
+      <xsd:enumeration value="LOC"/>
+      <xsd:enumeration value="SOURCE"/>
+      <xsd:enumeration value="GOAL"/>
+      <xsd:enumeration value="EVENT"/>
+      <xsd:enumeration value="AIM"/>
+      <xsd:enumeration value="CAUSE"/>
+      <xsd:enumeration value="CONTR"/>
+      <xsd:enumeration value="COND"/>
+      <xsd:enumeration value="REG"/>
+      <xsd:enumeration value="ACMP"/>
+      <xsd:enumeration value="RESTR"/>
+      <xsd:enumeration value="MANN"/>
+      <xsd:enumeration value="MEANS"/>
+      <xsd:enumeration value="QUANT"/>
+      <xsd:enumeration value="MWPRED"/>
+      <xsd:enumeration value="MODAL"/>
+      <xsd:enumeration value="PHRAS"/>
+    </xsd:restriction>
+  </xsd:simpleType>
+
+  <xsd:complexType name="statisticsContainerType">
+    <xsd:sequence>
+      <xsd:choice>
+        <xsd:element name="semanticRole" type="semanticRoleType"/>
+      </xsd:choice>
+      <xsd:element name="measureList" type="measureListType"/>
+    </xsd:sequence>
+  </xsd:complexType>
+
+  <xsd:complexType name="statisticsContainerListType">
+    <xsd:sequence>
+      <xsd:element name="statisticsContainer" type="statisticsContainerType" minOccurs="0" maxOccurs="unbounded"/>
+    </xsd:sequence>
+  </xsd:complexType>
+
+  <xsd:complexType name="syntacticStructureType">
+    <xsd:sequence>
+      <xsd:element name="comp" type="componentLexemeType" minOccurs="0" maxOccurs="unbounded"/>
+    </xsd:sequence>
+    <xsd:attribute name="id" type="xsd:int"/>
+  </xsd:complexType>
+
+  <xsd:complexType name="syntacticStructureListType">
+    <xsd:sequence>
+      <xsd:element name="syntacticStructure" type="syntacticStructureType" minOccurs="0" maxOccurs="unbounded"/>
+    </xsd:sequence>
+    <xsd:attribute name="system" type="xsd:string"/>
+  </xsd:complexType>
+
+  <xsd:complexType name="semanticRoleContainerType">
+    <xsd:sequence>
+      <xsd:element name="semanticRole" type="semanticRoleType"/>
+      <xsd:element name="syntacticStructureList" type="syntacticStructureListType"/>
+    </xsd:sequence>
+    <xsd:attribute name="cluster" type="xsd:int" use="optional"/>
+  </xsd:complexType>
+
+  <xsd:complexType name="semanticRoleContainerListType">
+    <xsd:sequence>
+      <xsd:element name="semanticRoleContainer" type="semanticRoleContainerType" minOccurs="0" maxOccurs="unbounded"/>
+    </xsd:sequence>
+  </xsd:complexType>
+
+  <xsd:simpleType name="patternRepresentationType">
+    <xsd:restriction base="xsd:string"/>
+  </xsd:simpleType>
+
+  <xsd:complexType name="valencyPatternType">
+    <xsd:sequence>
+      <xsd:element name="measureList" type="measureListType"/>
+      <xsd:element name="semanticRoleContainerList" type="semanticRoleContainerListType"/>
+      <xsd:element name="patternRepresentation" type="patternRepresentationType"/>
+      <xsd:element name="exampleContainerList" type="exampleContainerListType"/>
+    </xsd:sequence>
+    <xsd:attribute name="id" type="xsd:int"/>
+  </xsd:complexType>
+      
+  <xsd:complexType name="valencyPatternListType">
+    <xsd:sequence>
+      <xsd:element name="valencyPattern" type="valencyPatternType" minOccurs="0" maxOccurs="unbounded"/>
+    </xsd:sequence>
+  </xsd:complexType>
+
 </xsd:schema>
diff --git a/resources/schema/valency_lexicon.xsd b/resources/schema/valency_lexicon.xsd
index 6ecf0de..be5b93d 100644
--- a/resources/schema/valency_lexicon.xsd
+++ b/resources/schema/valency_lexicon.xsd
@@ -5,7 +5,7 @@
               schemaLocation="http://www.w3.org/2001/xml.xsd" />
   <xsd:include schemaLocation="inventory.xsd"/>
 
-  <xsd:element name="lexicon">
+  <xsd:element name="dictionary">
     <xsd:complexType>
       <xsd:sequence>
         <xsd:element name="entry" minOccurs="0" maxOccurs="unbounded">
@@ -19,16 +19,26 @@
 	            <xsd:element name="lexicalUnit" type="lexicalUnitType" minOccurs="0"/>
                     <xsd:element name="grammar" type="grammarType" minOccurs="0"/>
                     <xsd:element name="measureList" type="measureListType" minOccurs="0"/>
+                    <xsd:element name="labelList" type="labelListType" minOccurs="0"/>
                   </xsd:sequence>
                 </xsd:complexType>
               </xsd:element>
               <xsd:element name="body">
                 <xsd:complexType>
                   <xsd:sequence>
+                    <xsd:element name="statisticsContainerList" type="statisticsContainerListType"/>
                     <xsd:element name="senseList">
                       <xsd:complexType>
                         <xsd:sequence>
-                          <xsd:element name="sense" minOccurs="0" maxOccurs="unbounded"/>
+                          <xsd:element name="sense" minOccurs="0" maxOccurs="unbounded">
+                            <xsd:complexType>
+                              <xsd:sequence>
+                                <xsd:element name="definitionList" type="definitionListType" minOccurs="0"/>
+                                <xsd:element name="valencyPatternList" type="valencyPatternListType" minOccurs="0"/>
+                              </xsd:sequence>
+                              <xsd:attribute name="id" type="xsd:int"/>
+                            </xsd:complexType>
+                          </xsd:element>
                         </xsd:sequence>
                       </xsd:complexType>
                     </xsd:element>
