Multi Tasking Summarizer

Inspiration

The explosive growth of textual data on the web and the problem of obtaining desired information through this enormous volume of data has led to a dramatic increase in demand for developing automatic text summarization systems. For this multi-document text summarization approach was needed to save time and energy , which also increase productivity.

What it does

Used OpenAI GPT-3 API service to Generate Summary for any Text file / PDF file / Web Article/ YouTube Video which helped in saving time by 50% and increasing the productivity by 20%

How we built it

• To summarize text file we read the data of file and makes an API call for GPT -3 and it return the summary
• To summarize pdf files we extract the text data from it and summarize it
• To summarize web article we pass that link in prompt of API call and GPT -3 read it and return the summary
• To summarize youtube video we extract the subtitles from it, convert subtitle into text format and summarize it

Challenges we ran into

• It was challenging to work with prompt and decide a proper prompt for call
• It was challenging to filter the response data from GPT - 3 API call
• It was challenging to code different function for different type of files
• It was challenging to work with Youtube videos and generate their subtitles

Accomplishments that we're proud of

• Successfully generated summary not only for files and web article link but also youtube videos
• Successfully generated Key Terms & Key Points for input data

What we learned

• Learned about different library and virtual environment in python
• Learned about API and how they work
• Learned what GPT -3 is and how to use GPT -3
• Learned how read data from different sources

What's next for Multi Tasking Summarizer

• For now it can only summarize machine generated PDF files and can be upgrade to work for scanned pdf file
• For now it can only summarize youtube video in which subtitles are present and can be upgrade to work for any with or without subtitle youtube videos
• It generate output.json file as an output and it can be used anywhere to connect to any service

Built With

• api
• argparse
• cli
• openai
• python
• virtual-environment