Legal Document Summary Extraction

Industry: Law Firm

Summary Extration

Overview

Text Summarization using a custom built deep learning algorithm. Generating effective summary for any large document using Deep Learning is a hot research topic. In the market we have seen many keywords based extraction methods which gets your job partially done but the summary is not like we humans write. We were tasked to build a model which would be used to generate summary as a human does using an abstractive summarization method.

Technology

  • Deep Learning
  • Machine Learning

Technology

  • Python

Key Technical Challenges:

  • A custom deep learning model which can understand the meaning of document & can modify certain statements based on vocabulary given.
  • Gathering enough data & re-structuring it in a way that model can understand it.
  • Generating summary the way the client needed.

Business + Technical Points:

  • Accurate enough summary which has the meaning of document kept & dynamic in nature based on user’s desired characters.
  • System should be scalable automatically to get new kind of data into training & consideration.

Result:

  • We are able to summarize a long document into a short summary.