Update ✨Entity Linking Application✨.py

✨Entity Linking Application✨.py

@@ -62,8 +62,8 @@ async def combination_method(name, session):
         x = itertools_combinations(new_name, 2)
         for i in x:
             new_word = (i[0] + " " + i[1])
-            url = f"{new_word} site:en.wikipedia.org inurl:/wiki/ -inurl:?"
-            s = search(url, num_results = 12, lang="en")
+            url = f"{new_word} site:en.wikipedia.org inurl:/wiki/ -inurl:? -inurl:Category: -inurl:Help: -inurl:Special: -inurl:File:"
+            s = search(url, num_results = 12, lang="en")l
             for i in s:
                 data.add(i.split("/")[-1])
     return data
@@ -73,7 +73,7 @@ async def single_method(name, session):
         data = set()
         new_name = name.replace("-", " ").replace("/", " ").split()
         for i in new_name:
-            url = f"{i} site:en.wikipedia.org inurl:/wiki/ -inurl:?"
+            url = f"{i} site:en.wikipedia.org inurl:/wiki/ -inurl:? -inurl:Category: -inurl:Help: -inurl:Special: -inurl:File:"
             s = search(url, num_results = 12, lang="en")
             for i in s:
                 data.add(i.split("/")[-1])
@@ -85,7 +85,7 @@ async def mains(name, single, combi):
     qids = set()
     
     async with aiohttp.ClientSession() as session:
-        url = f"{name} site:en.wikipedia.org inurl:/wiki/ -inurl:?"
+        url = f"{name} site:en.wikipedia.org inurl:/wiki/ -inurl:? -inurl:Category: -inurl:Help: -inurl:Special: -inurl:File:"
         s = search(url, num_results = 30, lang="en")
         for i in s:
             data.add(i.split("/")[-1])
@@ -267,17 +267,17 @@ async def main(name):
         with open(f"/home/user/app/info_extraction/{name}.json", "w", encoding="utf-8") as flast:
             json.dump(final_list, flast)
 
-def check_sentence(sentence):
-    two_consecutive_uppercase = r"[A-Z]{2}"
-    uppercase_followed_by_fullstop = r"[A-Z]\."
+#def check_sentence(sentence):
+#    two_consecutive_uppercase = r"[A-Z]{2}"
+#    uppercase_followed_by_fullstop = r"[A-Z]\."
 
-    if re.search(two_consecutive_uppercase, sentence):
-        return True
+#    if re.search(two_consecutive_uppercase, sentence):
+#        return True
 
-    if re.search(uppercase_followed_by_fullstop, sentence):
-        return True
+#    if re.search(uppercase_followed_by_fullstop, sentence):
+#        return True
 
-    return False 
+#    return False 
     
 chrome_driver_path = "chromedriver.exe"
 chrome_path = r'"C:\Program Files\Google\Chrome\Application\chrome.exe"'
@@ -428,7 +428,6 @@ def main_cli():
                     st.write(f"Applying Candidate Selection module... (4/5) [{number}/{len(list_with_full_names)}]")
                     with open(f"/home/user/app/info_extraction/{i}.json", "r") as f:
                         json_file = json.load(f)
-                        print(json_file)
                         lista = []
                         lista_1 = []
                         for element in json_file: