"I realise that some of my criticisms may be mistaken; but to refuse to criticize judgements for fear of being mistaken is to abandon criticism altogether... If any of my criticisms are found to be correct, the cause is served; and if any are found to be incorrect the very process of finding out my mistakes must lead to the discovery of the right reasons, or better reasons than I have been able to give, and the cause is served just as well."

-Mr. HM Seervai, Preface to the 1st ed., Constitutional Law of India.

Wednesday, July 15, 2020

Alt+Law: Technological Innovation and the Law

[The legal profession is regarded as one of the least innovative professions (see, here, for instance). Lawyers do not accept technological improvements easily. Given this, Ms. Aarvi Singh's Alt+law bot comes as a pleasant surprise. She is a student of RGNUL, Punjab and an avid researcher with several papers published in various law journals/ reviews.

Ms. Singh's bot in Telegram has the potential to help stressed out law students. The bot purports to summarise pdf files and provide summaries of these files. Apparently, the bot runs even on pdfs containing more than 500 pages. In this post, Ms Singh explains to the world her bot, "Alt+Law". Do test the bot and give your feedback to Ms. Aarvi Singh at aarvi03singh@gmail.com ]

Introducing my project Alt+Law 
by Aarvi Singh

During my first internship, I was assigned the task to give a summary of a number of judgments and one such judgment was LIC v Escort, (1986) 1 SCC 264. It was way too long a judgment to summarize in half an hour. At that very moment of fear, the anxiety of being judged as poor reading capacity loomed over me. I completed the task, but always wished for some tools to assist students in researching especially when time is all that matters.
Finally, I took this task to create some tools that would assist students in creating a summary of bulky papers. I created a bot on Telegram which will summarize any judgment, paper, or any pdf with content.

Technical Aspect of the bot

@altpluslaw_bot, currently is a Telegram bot, which receives and sends pdf file to an apache tika grobid document parser (open-source software). The software extracts citations and references (if any) with fair accuracy (currently 60%). Once the citations and the overall text is received then it is sent to a summarizer. Currently, I have deployed different types of summarizers which gives the summary of the text by reducing the number of word counts to approximately 10% of the original words.
Image as available on Telegram

Scope of Improvements

I am working to improve the model and soon the model will be able to extract and summarize text from websites. The next version will come with options of summarization ratios. A new bot will be publishing three distinct newsletters on various topics but as of now topics are alternative investment, technology law and energy law updates. 

No comments: