Update app.py

app.py

@@ -9,11 +9,6 @@ from nltk.corpus import wordnet
 # Initialize the English text classification pipeline for AI detection
 pipeline_en = pipeline(task="text-classification", model="Hello-SimpleAI/chatgpt-detector-roberta")
 
-# Function to predict the label and score for English text (AI Detection)
-def predict_en(text):
-    res = pipeline_en(text)[0]
-    return res['label'], res['score']
-
 # Ensure necessary NLTK data is downloaded for Humanifier
 nltk.download('wordnet')
 nltk.download('omw-1.4')
@@ -25,64 +20,35 @@ except OSError:
     subprocess.run(["python", "-m", "spacy", "download", "en_core_web_sm"])
     nlp = spacy.load("en_core_web_sm")
 
-# Function to get synonyms using NLTK WordNet (Humanifier)
-def get_synonyms_nltk(word, pos):
-    synsets = wordnet.synsets(word, pos=pos)
-    if synsets:
-        lemmas = synsets[0].lemmas()
-        return [lemma.name().replace('_', ' ') for lemma in lemmas]
-    return []
-
-# Function to capitalize the first letter of sentences and proper nouns (Humanifier)
-def capitalize_sentences_and_nouns(text):
-    doc = nlp(text)
-    corrected_text = []
-
-    for sent in doc.sents:
-        sentence = []
-        for token in sent:
-            if token.i == sent.start:  # First word of the sentence
-                sentence.append(token.text.capitalize())
-            elif token.pos_ == "PROPN":  # Proper noun
-                sentence.append(token.text.capitalize())
-            else:
-                sentence.append(token.text)
-        corrected_text.append(' '.join(sentence))
-
-    return ' '.join(corrected_text)
-
-# Function to correct tense errors in a sentence (Tense Correction)
-def correct_tense_errors(text):
-    doc = nlp(text)
+# Function to check subject-verb agreement
+def check_subject_verb_agreement(doc):
     corrected_text = []
-
     for token in doc:
-        if token.tag_ in {"VBD", "VBN"} and token.lemma_:
-            # Convert past tense verbs to their base form
-            corrected_text.append(token.lemma_)
-        else:
-            corrected_text.append(token.text)
-
+        if token.dep_ == "nsubj":  # Check if the token is a subject
+            subject = token
+            verb = token.head  # Find the associated verb
+            if verb.tag_ in {"VBZ", "VBP"}:  # Singular/plural verb forms
+                if subject.tag_ == "NNS" and verb.tag_ == "VBZ":  # Plural subject with singular verb
+                    corrected_text.append(verb.lemma_)  # Convert verb to plural form
+                elif subject.tag_ == "NN" and verb.tag_ == "VBP":  # Singular subject with plural verb
+                    corrected_text.append(verb.lemma_ + 's')  # Convert verb to singular form
+                else:
+                    corrected_text.append(verb.text)  # No correction needed
+            else:
+                corrected_text.append(verb.text)
+        corrected_text.append(token.text)
     return ' '.join(corrected_text)
 
-# Function to correct singular/plural errors (Singular/Plural Correction)
-def correct_singular_plural_errors(text):
-    doc = nlp(text)
+# Function to correct singular/plural errors using dependency parsing
+def correct_singular_plural_errors(doc):
     corrected_text = []
     
     for token in doc:
         if token.pos_ == "NOUN":
-            if token.tag_ == "NN":  # Singular noun
-                if any(child.text.lower() in {'many', 'several', 'few', 'a', 'one'} for child in token.head.children):
-                    corrected_text.append(token.text if token.text.endswith('s') else token.text + 's')
-                else:
-                    corrected_text.append(token.text)
-            elif token.tag_ == "NNS":  # Plural noun
-                if any(child.text.lower() in {'a', 'one'} for child in token.head.children):
-                    singular = token.lemma_
-                    corrected_text.append(singular)
-                else:
-                    corrected_text.append(token.text)
+            if token.tag_ == "NN" and token.head.pos_ == "VERB" and token.head.tag_ == "VBP":
+                corrected_text.append(token.lemma_ + 's')  # Singular noun, plural verb
+            elif token.tag_ == "NNS" and token.head.pos_ == "VERB" and token.head.tag_ == "VBZ":
+                corrected_text.append(token.lemma_)  # Plural noun, singular verb
             else:
                 corrected_text.append(token.text)
         else:
@@ -90,70 +56,54 @@ def correct_singular_plural_errors(text):
     
     return ' '.join(corrected_text)
 
-# Function to check and correct article errors
-def correct_article_errors(text):
-    doc = nlp(text)
-    corrected_text = []
-    tokens = list(doc)
-    
-    for i, token in enumerate(tokens):
-        if token.text.lower() in {'a', 'an'}:
-            if i + 1 < len(tokens):
-                next_token = tokens[i + 1]
-                if next_token.text[0].lower() in 'aeiou':
-                    corrected_text.append('an')
-                else:
-                    corrected_text.append('a')
-            else:
-                corrected_text.append(token.text)
-        else:
-            corrected_text.append(token.text)
-    return ' '.join(corrected_text)
-
 # Paraphrasing function using SpaCy and NLTK (Humanifier)
 def paraphrase_with_spacy_nltk(text):
     doc = nlp(text)
     paraphrased_words = []
     
     for token in doc:
+        # Map SpaCy POS tags to WordNet POS tags
         pos = None
-        if token.pos_ == "NOUN":
+        if token.pos_ in {"NOUN"}:
             pos = wordnet.NOUN
-        elif token.pos_ == "VERB":
+        elif token.pos_ in {"VERB"}:
             pos = wordnet.VERB
-        elif token.pos_ == "ADJ":
+        elif token.pos_ in {"ADJ"}:
             pos = wordnet.ADJ
-        elif token.pos_ == "ADV":
+        elif token.pos_ in {"ADV"}:
             pos = wordnet.ADV
         
         synonyms = get_synonyms_nltk(token.text.lower(), pos) if pos else []
         
-        # Replace with a synonym only if it's more common and fits the context
-        if synonyms and token.pos_ in {"NOUN", "VERB", "ADJ", "ADV"}:
-            # Avoid replacing with the same word or rare synonyms
-            synonym = synonyms[0]
-            if synonym != token.text.lower() and len(synonym.split()) == 1:
-                paraphrased_words.append(synonym)
-            else:
-                paraphrased_words.append(token.text)
+        # Replace with a synonym only if it makes sense
+        if synonyms and token.pos_ in {"NOUN", "VERB", "ADJ", "ADV"} and synonyms[0] != token.text.lower():
+            paraphrased_words.append(synonyms[0])
         else:
             paraphrased_words.append(token.text)
     
-    paraphrased_sentence = ' '.join(paraphrased_words)
-    return paraphrased_sentence
+    return ' '.join(paraphrased_words)
 
 # Combined function: Paraphrase -> Grammar Correction -> Capitalization (Humanifier)
 def paraphrase_and_correct(text):
     # Step 1: Paraphrase the text
     paraphrased_text = paraphrase_with_spacy_nltk(text)
     
-    # Step 2: Apply grammatical corrections on the paraphrased text
-    corrected_text = correct_article_errors(paraphrased_text)
+    # Step 2: Parse the text with spaCy
+    doc = nlp(paraphrased_text)
+    
+    # Step 3: Apply grammatical corrections on the paraphrased text
+    corrected_text = correct_article_errors(doc)
+    
     corrected_text = capitalize_sentences_and_nouns(corrected_text)
-    corrected_text = correct_singular_plural_errors(corrected_text)
-    corrected_text = correct_tense_errors(corrected_text)
     
-    return corrected_text
+    corrected_text = check_subject_verb_agreement(nlp(corrected_text))
+    
+    corrected_text = correct_singular_plural_errors(nlp(corrected_text))
+    
+    # Step 4: Capitalize sentences and proper nouns (final correction step)
+    final_text = correct_tense_errors(nlp(corrected_text))
+    
+    return final_text
 
 # Gradio app setup with two tabs
 with gr.Blocks() as demo:
@@ -161,16 +111,16 @@ with gr.Blocks() as demo:
         t1 = gr.Textbox(lines=5, label='Text')
         button1 = gr.Button("🤖 Predict!")
         label1 = gr.Textbox(lines=1, label='Predicted Label 🎃')
-        score1 = gr.Textbox(lines=1, label='Probability')
-        
+        score1 = gr.Textbox(lines=1, label='Prob')
+
         # Connect the prediction function to the button
         button1.click(predict_en, inputs=[t1], outputs=[label1, score1], api_name='predict_en')
     
     with gr.Tab("Humanifier"):
-        text_input = gr.Textbox(lines=10, label="Input Text")
+        text_input = gr.Textbox(lines=5, label="Input Text")
         paraphrase_button = gr.Button("Paraphrase & Correct")
         output_text = gr.Textbox(label="Paraphrased Text")
-        
+
         # Connect the paraphrasing function to the button
         paraphrase_button.click(paraphrase_and_correct, inputs=text_input, outputs=output_text)