ASL

.gitignore

@@ -1,4 +1,2 @@
 .DS_Store
-.env
-__pycache__/*
-gensim-data/*
+.env

.python-version

@@ -1 +0,0 @@
-3.11.13

vectorizer.py

@@ -1,6 +1,5 @@
 import gensim
 import gensim.downloader
-from gensim.models import KeyedVectors
 import numpy as np
 import pandas as pd
 import os
@@ -19,16 +18,8 @@ class Vectorizer:
         """
         Returns a KeyedVector object loaded from gensim
         """
-        model_path = os.path.join(os.getcwd(), 'gensim-data', 'GoogleNews-vectors-negative300.bin.gz')
         try:
-            print(f"Loading model from {model_path}")
-            kv = KeyedVectors.load_word2vec_format(model_path, binary=True)
-            print("Word2Vec model loaded successfully as KeyedVectors object.")
-            return kv
-        except FileNotFoundError:
-            print(f"Error: Model file not found at {model_path}. Trying to download...")
             kv = gensim.downloader.load(model_name)  # returns a keyedvector
-            print("Word2Vec model loaded successfully as KeyedVectors object.")
             return kv
         except Exception as e:
             print(f"Unable to load embedding model from gensim: {e}")
@@ -52,24 +43,10 @@ class Vectorizer:
 
     def encode(self, word):
         print(f"encoding {word}")
-        if self.kv is None:
-            print("KeyedVectors not loaded")
-            return None
-        if word in self.kv.key_to_index:
+        if self.kv is not None and word in self.kv.key_to_index:
             return self.kv[word]
         else:
             print(f"Error: {word} is not in the KeyedVector's vocabulary")
-            # Try to find closest match
-            try:
-                closest_matches = self.kv.most_similar(word, topn=3)
-                if closest_matches:
-                    closest_word = closest_matches[0][0]
-                    print(f"Using closest match '{closest_word}' for '{word}'")
-                    return self.kv[closest_word]
-                else:
-                    print(f"No similar words found for '{word}'")
-            except Exception as e:
-                print(f"Error finding similar words: {e}")
             return None
     
     def encode_and_format(self, word):
@@ -84,11 +61,10 @@ class Vectorizer:
         try:
             await self.ensure_supabase_initialized()
             query_embedding = self.encode(query)
-            
             if query_embedding is None:
                 return {
                     "match": False,
-                    "error": f"'{query}' not in vocabulary and no similar words found"
+                    "error": f"'{query}' not in vocabulary"
                 }
 
             query_embedding = query_embedding.tolist()
@@ -154,11 +130,8 @@ def load_filtered_kv(model_name='word2vec-google-news-300', vocab=None):
 async def main():
     vectorizer = Vectorizer()
 
-    # Test exact word match
     vector = vectorizer.encode("test")
     print(vector)
-    
-    # Test words not in vocabulary with closest match fallback
     result = await vectorizer.vector_query_from_supabase("dog")
     print(result)
     result = await vectorizer.vector_query_from_supabase("cat")