README.md

metadata

title: Question Generator
emoji: 🔑
colorFrom: yellow
colorTo: yellow
sdk: streamlit
sdk_version: 1.10.0
app_file: app.py
pinned: false

Internship-IVIS-labs

• The Intelligent Question Generator app is an easy-to-use interface built in Streamlit which uses KeyBERT, Sense2vec, T5
• It uses a minimal keyword extraction technique that leverages multiple NLP embeddings and relies on Transformers 🤗 to create keywords/keyphrases that are most similar to a document.
• sense2vec (Trask et. al, 2015) is a nice twist on word2vec that lets you learn more interesting and detailed word vectors.

Repository Breakdown

src Directory

---

• src/Pipeline/QAhaystack.py: This file contains the code of question answering using haystack.
• src/Pipeline/QuestGen.py: This file contains the code of question generation.
• src/Pipeline/Reader.py: This file contains the code of reading the document.
• src/Pipeline/TextSummariztion.py: This file contains the code of text summarization.
• src/PreviousVersionCode/context.py: This file contains the finding the context of the paragraph.
• src/PreviousVersionCode/QuestionGenerator.py: This file contains the code of first attempt of question generation.

Installation

$ git clone https://github.com/HemanthSai7/Internship-IVIS-labs.git

$ cd Internship-IVIS-labs

pip install -r requirements.txt

• For the running the app for the first time locally, you need to uncomment the the lines in src/Pipeline/QuestGen.py to download the models to the models directory.

streamlit run app.py

• Once the app is running, you can access it at http://localhost:8501

  You can now view your Streamlit app in your browser.

  Local URL: http://localhost:8501
  Network URL: http://192.168.0.103:8501

Tech Stack Used

Timeline

Week 1-2:

Tasks

• Understanding and brushing up the concepts of NLP.
• Extracting images and text from a pdf file and storing it in a texty file.
• Exploring various open source tools for generating questions from a given text.
• Read papers related to the project (Bert,T5,RoBERTa etc).
• Summarizing the extracted text using T5 base pre-trained model from the pdf file.

Week 3-4:

Tasks

• Understanding the concept of QA systems.
• Created a basic script for generating questions from the text.
• Created a basic script for finding the context of the paragraph.

Week 5-6:

Tasks

• Understanding how Transformers models work for NLP tasks Question answering and generation
• Understanding how to use the Haystack library for QA systems.
• Understanding how to use the Haystack library for Question generation.
• PreProcessed the document for Haystack QA for better results .

Week 7-8:

Tasks

• Understanding how to generate questions intelligently.
• Explored wordnet to find synonyms
• Used BertWSD for disambiguating the sentence provided.
• Used KeyBERT for finding the keywords in the document.
• Used sense2vec for finding better words with high relatedness for the keywords generated.

Week 9-10:

Tasks

• Create a streamlit app to demonstrate the project.