app.py

# example 1 
from textwrap3 import wrap
import torch
import random
import numpy as np
import nltk
nltk.download('punkt')
nltk.download('brown')
nltk.download('wordnet')
from nltk.corpus import wordnet as wn
from nltk.tokenize import sent_tokenize
nltk.download('stopwords')
from nltk.corpus import stopwords
import string
import pke
import traceback
from flashtext import KeywordProcessor
from similarity.normalized_levenshtein import NormalizedLevenshtein
normalized_levenshtein = NormalizedLevenshtein()
from collections import OrderedDict
from sklearn.metrics.pairwise import cosine_similarity
import nltk
nltk.download('omw-1.4')
import gradio as gr
question_model = T5ForConditionalGeneration.from_pretrained('ramsrigouthamg/t5_squad_v1')
question_tokenizer = T5Tokenizer.from_pretrained('ramsrigouthamg/t5_squad_v1')
question_model = question_model.to(device)

# filter keywords 
!wget https://github.com/explosion/sense2vec/releases/download/v1.0.0/s2v_reddit_2015_md.tar.gz
!tar -xvf  s2v_reddit_2015_md.tar.gz
import numpy as np
from sense2vec import Sense2Vec
s2v = Sense2Vec().from_disk('s2v_old')
from sentence_transformers import SentenceTransformer


text = """Elon Musk has shown again he can influence the digital currency market with just his tweets. After saying that his electric vehicle-making company
Tesla will not accept payments in Bitcoin because of environmental concerns, he tweeted that he was working with developers of Dogecoin to improve
system transaction efficiency. Following the two distinct statements from him, the world's largest cryptocurrency hit a two-month low, while Dogecoin
rallied by about 20 percent. The SpaceX CEO has in recent months often tweeted in support of Dogecoin, but rarely for Bitcoin.  In a recent tweet,
Musk put out a statement from Tesla that it was “concerned” about the rapidly increasing use of fossil fuels for Bitcoin (price in India) mining and
transaction, and hence was suspending vehicle purchases using the cryptocurrency.  A day later he again tweeted saying, “To be clear, I strongly
believe in crypto, but it can't drive a massive increase in fossil fuel use, especially coal”.  It triggered a downward spiral for Bitcoin value but
the cryptocurrency has stabilised since.   A number of Twitter users welcomed Musk's statement. One of them said it's time people started realising
that Dogecoin “is here to stay” and another referred to Musk's previous assertion that crypto could become the world's future currency."""

for wrp in wrap(text, 150):
  print (wrp)
print ("\n")


# summerization with t5
from transformers import T5ForConditionalGeneration,T5Tokenizer
summary_model = T5ForConditionalGeneration.from_pretrained('t5-base')
summary_tokenizer = T5Tokenizer.from_pretrained('t5-base')

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
summary_model = summary_model.to(device)



def set_seed(seed: int):
    random.seed(seed)
    np.random.seed(seed)
    torch.manual_seed(seed)
    torch.cuda.manual_seed_all(seed)

set_seed(42)



def postprocesstext (content):
  final=""
  for sent in sent_tokenize(content):
    sent = sent.capitalize()
    final = final +" "+sent
  return final


def summarizer(text,model,tokenizer):
  text = text.strip().replace("\n"," ")
  text = "summarize: "+text
  # print (text)
  max_len = 512
  encoding = tokenizer.encode_plus(text,max_length=max_len, pad_to_max_length=False,truncation=True, return_tensors="pt").to(device)

  input_ids, attention_mask = encoding["input_ids"], encoding["attention_mask"]

  outs = model.generate(input_ids=input_ids,
                                  attention_mask=attention_mask,
                                  early_stopping=True,
                                  num_beams=3,
                                  num_return_sequences=1,
                                  no_repeat_ngram_size=2,
                                  min_length = 75,
                                  max_length=300)


  dec = [tokenizer.decode(ids,skip_special_tokens=True) for ids in outs]
  summary = dec[0]
  summary = postprocesstext(summary)
  summary= summary.strip()

  return summary


summarized_text = summarizer(text,summary_model,summary_tokenizer)


print ("\noriginal Text >>")
for wrp in wrap(text, 150):
  print (wrp)
print ("\n")
print ("Summarized Text >>")
for wrp in wrap(summarized_text, 150):
  print (wrp)
print ("\n")



# answer span extraction


def get_nouns_multipartite(content):
    out=[]
    try:
        extractor = pke.unsupervised.MultipartiteRank()
        extractor.load_document(input=content,language='en')
        #    not contain punctuation marks or stopwords as candidates.
        pos = {'PROPN','NOUN'}
        #pos = {'PROPN','NOUN'}
        stoplist = list(string.punctuation)
        stoplist += ['-lrb-', '-rrb-', '-lcb-', '-rcb-', '-lsb-', '-rsb-']
        stoplist += stopwords.words('english')
        # extractor.candidate_selection(pos=pos, stoplist=stoplist)
        extractor.candidate_selection(pos=pos)
        # 4. build the Multipartite graph and rank candidates using random walk,
        #    alpha controls the weight adjustment mechanism, see TopicRank for
        #    threshold/method parameters.
        extractor.candidate_weighting(alpha=1.1,
                                      threshold=0.75,
                                      method='average')
        keyphrases = extractor.get_n_best(n=15)
        

        for val in keyphrases:
            out.append(val[0])
    except:
        out = []
        traceback.print_exc()

    return out



def get_keywords(originaltext,summarytext):
  keywords = get_nouns_multipartite(originaltext)
  print ("keywords unsummarized: ",keywords)
  keyword_processor = KeywordProcessor()
  for keyword in keywords:
    keyword_processor.add_keyword(keyword)

  keywords_found = keyword_processor.extract_keywords(summarytext)
  keywords_found = list(set(keywords_found))
  print ("keywords_found in summarized: ",keywords_found)

  important_keywords =[]
  for keyword in keywords:
    if keyword in keywords_found:
      important_keywords.append(keyword)

  return important_keywords[:10]


imp_keywords = get_keywords(text,summarized_text)
print (imp_keywords)



def get_question(context,answer,model,tokenizer):
  text = "context: {} answer: {}".format(context,answer)
  encoding = tokenizer.encode_plus(text,max_length=384, pad_to_max_length=False,truncation=True, return_tensors="pt").to(device)
  input_ids, attention_mask = encoding["input_ids"], encoding["attention_mask"]

  outs = model.generate(input_ids=input_ids,
                                  attention_mask=attention_mask,
                                  early_stopping=True,
                                  num_beams=5,
                                  num_return_sequences=1,
                                  no_repeat_ngram_size=2,
                                  max_length=72)


  dec = [tokenizer.decode(ids,skip_special_tokens=True) for ids in outs]


  Question = dec[0].replace("question:","")
  Question= Question.strip()
  return Question



for wrp in wrap(summarized_text, 150):
  print (wrp)
print ("\n")

for answer in imp_keywords:
  ques = get_question(summarized_text,answer,question_model,question_tokenizer)
  print (ques)
  print (answer.capitalize())
  print ("\n")




# filter keywords 

# paraphrase-distilroberta-base-v1
sentence_transformer_model = SentenceTransformer('msmarco-distilbert-base-v3')





def filter_same_sense_words(original,wordlist):
  filtered_words=[]
  base_sense =original.split('|')[1] 
  print (base_sense)
  for eachword in wordlist:
    if eachword[0].split('|')[1] == base_sense:
      filtered_words.append(eachword[0].split('|')[0].replace("_", " ").title().strip())
  return filtered_words

def get_highest_similarity_score(wordlist,wrd):
  score=[]
  for each in wordlist:
    score.append(normalized_levenshtein.similarity(each.lower(),wrd.lower()))
  return max(score)

def sense2vec_get_words(word,s2v,topn,question):
    output = []
    print ("word ",word)
    try:
      sense = s2v.get_best_sense(word, senses= ["NOUN", "PERSON","PRODUCT","LOC","ORG","EVENT","NORP","WORK OF ART","FAC","GPE","NUM","FACILITY"])
      most_similar = s2v.most_similar(sense, n=topn)
      # print (most_similar)
      output = filter_same_sense_words(sense,most_similar)
      print ("Similar ",output)
    except:
      output =[]

    threshold = 0.6
    final=[word]
    checklist =question.split()
    for x in output:
      if get_highest_similarity_score(final,x)<threshold and x not in final and x not in checklist:
        final.append(x)
    
    return final[1:]

def mmr(doc_embedding, word_embeddings, words, top_n, lambda_param):

    # Extract similarity within words, and between words and the document
    word_doc_similarity = cosine_similarity(word_embeddings, doc_embedding)
    word_similarity = cosine_similarity(word_embeddings)

    # Initialize candidates and already choose best keyword/keyphrase
    keywords_idx = [np.argmax(word_doc_similarity)]
    candidates_idx = [i for i in range(len(words)) if i != keywords_idx[0]]

    for _ in range(top_n - 1):
        # Extract similarities within candidates and
        # between candidates and selected keywords/phrases
        candidate_similarities = word_doc_similarity[candidates_idx, :]
        target_similarities = np.max(word_similarity[candidates_idx][:, keywords_idx], axis=1)

        # Calculate MMR
        mmr = (lambda_param) * candidate_similarities - (1-lambda_param) * target_similarities.reshape(-1, 1)
        mmr_idx = candidates_idx[np.argmax(mmr)]

        # Update keywords & candidates
        keywords_idx.append(mmr_idx)
        candidates_idx.remove(mmr_idx)

    return [words[idx] for idx in keywords_idx]

def get_distractors_wordnet(word):
    distractors=[]
    try:
      syn = wn.synsets(word,'n')[0]
      
      word= word.lower()
      orig_word = word
      if len(word.split())>0:
          word = word.replace(" ","_")
      hypernym = syn.hypernyms()
      if len(hypernym) == 0: 
          return distractors
      for item in hypernym[0].hyponyms():
          name = item.lemmas()[0].name()
          #print ("name ",name, " word",orig_word)
          if name == orig_word:
              continue
          name = name.replace("_"," ")
          name = " ".join(w.capitalize() for w in name.split())
          if name is not None and name not in distractors:
              distractors.append(name)
    except:
      print ("Wordnet distractors not found")
    return distractors

def get_distractors (word,origsentence,sense2vecmodel,sentencemodel,top_n,lambdaval):
  distractors = sense2vec_get_words(word,sense2vecmodel,top_n,origsentence)
  print ("distractors ",distractors)
  if len(distractors) ==0:
    return distractors
  distractors_new = [word.capitalize()]
  distractors_new.extend(distractors)
  # print ("distractors_new .. ",distractors_new)

  embedding_sentence = origsentence+ " "+word.capitalize()
  # embedding_sentence = word
  keyword_embedding = sentencemodel.encode([embedding_sentence])
  distractor_embeddings = sentencemodel.encode(distractors_new)

  # filtered_keywords = mmr(keyword_embedding, distractor_embeddings,distractors,4,0.7)
  max_keywords = min(len(distractors_new),5)
  filtered_keywords = mmr(keyword_embedding, distractor_embeddings,distractors_new,max_keywords,lambdaval)
  # filtered_keywords = filtered_keywords[1:]
  final = [word.capitalize()]
  for wrd in filtered_keywords:
    if wrd.lower() !=word.lower():
      final.append(wrd.capitalize())
  final = final[1:]
  return final

sent = "What cryptocurrency did Musk rarely tweet about?"
keyword = "Bitcoin"

# sent = "What did Musk say he was working with to improve system transaction efficiency?"
# keyword= "Dogecoin"


# sent = "What company did Musk say would not accept bitcoin payments?"
# keyword= "Tesla"


# sent = "What has Musk often tweeted in support of?"
# keyword = "Cryptocurrency"

print (get_distractors(keyword,sent,s2v,sentence_transformer_model,40,0.2))




context = gr.inputs.Textbox(lines=10, placeholder="Enter paragraph/content here...")
output = gr.outputs.HTML(  label="Question and Answers")
radiobutton = gr.inputs.Radio(["Wordnet", "Sense2Vec"])

def generate_question(context,radiobutton):
  summary_text = summarizer(context,summary_model,summary_tokenizer)
  for wrp in wrap(summary_text, 100):
    print (wrp)
  # np = getnounphrases(summary_text,sentence_transformer_model,3)
  np =  get_keywords(context,summary_text)
  print ("\n\nNoun phrases",np)
  output=""
  for answer in np:
    ques = get_question(summary_text,answer,question_model,question_tokenizer)
    if radiobutton=="Wordnet":
      distractors = get_distractors_wordnet(answer)
    else:
      distractors = get_distractors(answer.capitalize(),ques,s2v,sentence_transformer_model,40,0.2)
    # output= output + ques + "\n" + "Ans: "+answer.capitalize() + "\n\n"
    output = output + "<b style='color:blue;'>" + ques + "</b>"
    output = output + "<br>"
    output = output + "<b style='color:green;'>" + "Ans: " +answer.capitalize()+  "</b>"+"<br>"
    if len(distractors)>0:
      for distractor in distractors[:4]:
        output = output + "<b style='color:brown;'>" + distractor+  "</b>"+"<br>"
    output = output + "<br>"

  summary ="Summary: "+ summary_text
  for answer in np:
    summary = summary.replace(answer,"<b>"+answer+"</b>" + "<br>")
    summary = summary.replace(answer.capitalize(),"<b>"+answer.capitalize()+"</b>")
  output = output + "<p>"+summary+"</p>"
  output = output + "<br>"
  return output


iface = gr.Interface(
  fn=generate_question, 
  inputs=[context,radiobutton], 
  outputs=output)
iface.launch(debug=True)