From d2b64ba054c7e2f5b8ad64021169c4b7a8808388 Mon Sep 17 00:00:00 2001
From: Simon Gray <simongray@gmail.com>
Date: Tue, 30 May 2023 16:19:01 +0200
Subject: [PATCH] #3 - add the remaining old English synonym links as ILI
 references

---
 .gitignore                              |  1 +
 src/main/dk/cst/dannet/bootstrap.clj    | 31 ++++++++++++++++++++-----
 src/main/dk/cst/dannet/web/section.cljc | 11 +++++++--
 3 files changed, 35 insertions(+), 8 deletions(-)

diff --git a/.gitignore b/.gitignore
index 91d777be..e468bea4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -28,6 +28,7 @@ bootstrap/other/sentiment/sense_polarities.tsv
 # The open English WordNet
 bootstrap/other/english/english-wordnet-2022.ttl
 bootstrap/other/english/ili.ttl
+bootstrap/other/english/ili-map-pwn20.tab
 bootstrap/other/english/yaml
 bootstrap/other/english/senseidx.edn
 
diff --git a/src/main/dk/cst/dannet/bootstrap.clj b/src/main/dk/cst/dannet/bootstrap.clj
index 77030fd6..084d2b32 100644
--- a/src/main/dk/cst/dannet/bootstrap.clj
+++ b/src/main/dk/cst/dannet/bootstrap.clj
@@ -30,6 +30,9 @@
             [dk.cst.dannet.query :as q]
             [dk.cst.dannet.query.operation :as op]))
 
+(declare read-triples)
+(declare cor-k-pos)
+
 ;; TODO: missing labels
 ;;       http://localhost:3456/dannet/data/synset-48454
 ;;       http://localhost:3456/dannet/data/synset-49086
@@ -557,6 +560,17 @@
 (def senseidx->english-synset
   (delay (edn/read-string (slurp "bootstrap/other/english/senseidx.edn"))))
 
+(def wn20-id->ili
+  (delay (->> (read-triples [identity
+                             "bootstrap/other/english/ili-map-pwn20.tab"
+                             :encoding "UTF-8"
+                             :separator \tab])
+              (filter (fn [[_ _ confidence]]
+                        (= confidence "1")))
+              (map (fn [[ili-id wn-id _]]
+                     [(str "ENG20-" wn-id) (keyword "ili" ili-id)]))
+              (into {}))))
+
 (h/defn ->english-link-triples
   "Convert a `row` from 'relations.csv' to triples.
 
@@ -564,10 +578,12 @@
   [[subj-id _ rel obj-id _ _ :as row]]
   ;; Ignores eq_has_hyponym and eq_has_hyperonym, no equivalent in GWA schema.
   ;; This loses us 123 of the original 5000l links to the Princton WordNet.
+  ;; TODO: implement dns relations for this as we apparently use those the new data too...
   (when (= "eq_has_synonym" rel)
-    ;; TODO: need backup for IDs that match e.g. "ENG20-07945291-n"
-    (when-let [obj (get @senseidx->english-synset obj-id)]
-      #{[(synset-uri subj-id) :wn/eq_synonym obj]})))
+    (if-let [obj (get @senseidx->english-synset obj-id)]
+      #{[(synset-uri subj-id) :wn/eq_synonym obj]}
+      (when-let [ili-obj (get @wn20-id->ili obj-id)]
+        #{[(synset-uri subj-id) :wn/ili ili-obj]}))))
 
 ;; TODO: can we create new forms/words/synsets rather than overload writtenRep?
 (defn explode-written-reps
@@ -600,9 +616,6 @@
   [s & after]
   (apply str "\"" s "\"" after))
 
-(declare read-triples)
-(declare cor-k-pos)
-
 (def sense-properties
   (let [row->kv (fn [[dannetsemid lemma hom pos dn_lemma id gloss]]
                   (let [[_ pos'] (re-matches #"([^\.]+)\.?" pos)
@@ -1103,6 +1116,7 @@
                  :separator \tab
                  :preprocess (comp mark-duplicate-senses rest)]
 
+    ;; TODO: publish as a separate dataset?
     ;; Links to the Open English WordNet
     :oewn-links [->english-link-triples "bootstrap/dannet/DanNet-2.5.1_csv/relations.csv"]}
 
@@ -1258,6 +1272,11 @@
   (->> (read-triples (get-in imports [prefix/dn-uri :relations]))
        (take 10))
 
+  ;; Example English WordNet or ILI links
+  (->> (read-triples (get-in imports [prefix/dn-uri :oewn-links]))
+       (remove nil?)
+       (take 10))
+
   ;; Example links to the Open English WordNet
   (->> (read-triples (get-in imports [prefix/dn-uri :en-links]))
        (take 10))
diff --git a/src/main/dk/cst/dannet/web/section.cljc b/src/main/dk/cst/dannet/web/section.cljc
index 5c7b905e..603296c7 100644
--- a/src/main/dk/cst/dannet/web/section.cljc
+++ b/src/main/dk/cst/dannet/web/section.cljc
@@ -11,7 +11,9 @@
   [[nil [:rdf/type
          :owl/sameAs
          :skos/definition
+         :wn/definition
          :rdfs/comment
+         :wn/partOfSpeech
          :lexinfo/partOfSpeech
          :lexinfo/senseExample
          :dns/sentiment
@@ -29,10 +31,15 @@
      :ontolex/sense
      :ontolex/isSenseOf
      :ontolex/lexicalizedSense
-     :ontolex/isLexicalizedSenseOf]]
+     :ontolex/isLexicalizedSenseOf
+     :wn/ili
+     :wn/eq_synonym]]
    [#{(->LangStr "Semantic relations" "en")
       (->LangStr "Betydningsrelationer" "da")}
-    (some-fn (prefix/with-prefix 'wn :except #{:wn/partOfSpeech})
+    (some-fn (prefix/with-prefix 'wn :except #{:wn/partOfSpeech
+                                               :wn/definition
+                                               :wn/ili
+                                               :wn/eq_synonym})
              (comp #{:dns/usedFor
                      :dns/usedForObject
                      :dns/nearAntonym