I am lawyer from SK. I also like programming and Maths. Which technology should i use for this task? Automatic legal document reviewing. I have a big data set of specific contracts in slovak. Now I want to build my model. Model will tell good/bad contract + bold probable errors based on data set. I don´t know much about NLP, so I´am just asking for advice. Which tech should I study for this task? Word2Vec, keywords extraction algos etc. Thank you guys for your time.
I am not a lawyer but I assume a contract is good if it is comprehensive and covers certain aspects extensively. An "easy" solution would be the following:
- You could probably train a model which could classify a contract to which area it belongs (aka topic classification). If the probability of belonging with the rest of its the peer contracts is high then it could be a good contract.
Then again I am not sure how effective this would be and what pre-processing steps should be taken and would certainly need a big amount of labelled contracts. I would encourage you to dive into NLP and check the various NLP tasks that may leverage the process with your domain expertise.
Lastly, you could ask for help from a company like Lawgeex to make the NLP (or LLP I think they call it for Law Language abbreviation) stuff for you. But then again, not sure if companies like this could fit the Slovakian market. It would be cool for them to have Language Model that could fine tune in other languages!