Upload 89 files

app/engine/lexical.py

@@ -452,6 +452,19 @@ def _unglossed_verb_allowed(
     return False
 
 
+def _polish_headword_upgrade(synset, source_lemma: str, cand_lemma: str) -> bool:
+    """Polish-only: allow source→headword upgrades in small verb synsets.
+
+    Example: maintain→keep on keep/maintain/hold. Blocks identify→place where
+    the source is already the headword.
+    """
+    lemmas = _synset_single_word_lemmas(synset)
+    if not lemmas or lemmas[0] != cand_lemma:
+        return False
+    source_rank = _source_lemma_rank(synset, source_lemma)
+    return 1 <= source_rank <= 2 and 2 <= len(lemmas) <= 4
+
+
 def _pick_verb_candidate(
     token,
     doc,
@@ -461,11 +474,20 @@ def _pick_verb_candidate(
     *,
     effective_min: float,
     gain_floor: float,
+    polish: bool = False,
+    aggressive: bool = False,
 ) -> tuple[float, str, float, Any] | None:
     """Choose a verb synonym from a few top senses with collocational safety."""
     obj = _object_lemma(token)
     src_lemma = token.lemma_.lower()
     weak_gloss = max(float(effective_min), 0.25)
+    # Prenominal/predicative participles often act like adjectives
+    # ("satisfying customer experience") — never trust unglossed verb senses.
+    adjectival_participle = token.tag_ == "VBG" and token.dep_ in {
+        "amod",
+        "acomp",
+        "oprd",
+    }
     options: list[tuple[float, int, float, float, str, Any]] = []
     ordered = sorted(
         synsets[:6],
@@ -483,14 +505,25 @@ def _pick_verb_candidate(
         if score < effective_min and simplicity_gain < gain_floor:
             continue
         cand_lemma = _candidate_lemma_for_replacement(synset, token, replacement)
-        if score < weak_gloss and not _unglossed_verb_allowed(
-            synset,
-            source_lemma=src_lemma,
-            cand_lemma=cand_lemma,
-            stop_words=stop_words,
-        ):
-            continue
+        high_gain_ok = False
         if score < weak_gloss:
+            if adjectival_participle:
+                continue
+            unglossed_ok = _unglossed_verb_allowed(
+                synset,
+                source_lemma=src_lemma,
+                cand_lemma=cand_lemma,
+                stop_words=stop_words,
+            )
+            # Polish may upgrade to the synset headword (maintain→keep) when
+            # the source is a secondary lemma. Aggressive/ensure stay stricter.
+            high_gain_ok = (
+                polish
+                and simplicity_gain >= 0.70
+                and _polish_headword_upgrade(synset, src_lemma, cand_lemma)
+            )
+            if not unglossed_ok and not high_gain_ok:
+                continue
             multiword_parent = False
             try:
                 for word in synset.words():
@@ -508,6 +541,20 @@ def _pick_verb_candidate(
         # (blocks create→make results while allowing purchase→buy).
         if obj and src_colloc >= 4.7 and score < weak_gloss and score < effective_min:
             continue
+        if (
+            score < weak_gloss
+            and polish
+            and high_gain_ok
+            and obj
+            and (
+                cand_colloc + 0.05 < src_colloc
+                # Large collocation jumps often mark a different sense
+                # (identify areas → name areas), while modest ones are OK
+                # (maintain attitude → keep attitude).
+                or (src_colloc >= 4.2 and cand_colloc - src_colloc >= 0.45)
+            )
+        ):
+            continue
         options.append(
             (
                 score,
@@ -679,6 +726,8 @@ def refine_sentence(
                     stop_words,
                     effective_min=effective_min,
                     gain_floor=verb_gain_floor,
+                    polish=polish,
+                    aggressive=aggressive,
                 )
                 if not picked_verb:
                     continue
@@ -922,6 +971,7 @@ def _force_one_safe_swap(
                 stop_words,
                 effective_min=0.08,
                 gain_floor=0.30,
+                aggressive=True,
             )
             if not picked_verb:
                 continue