Skip to content

Latest commit

 

History

History
86 lines (52 loc) · 2.32 KB

README.md

File metadata and controls

86 lines (52 loc) · 2.32 KB
Raw

legalAI

Some useful modules for processing legal documents in Python

How to use

Refer to README.md in each module

Package Requirements

  • stanza 1.2.3 or above

    The common module will start a Stanford CoreNLP server. This server will listen at port 9010.

    • The server requires Java Running Environment (JRE).

  • nltk 3.6.5 or above

    Also install corresponding nltk data according to error messages.

Sentence Simplification

Split a compound or complex sentence into simple ones.

API Functions:

split_sentence

Split the sentence based on two gates.

Argument Type Usage Remark
sentence string The sentence to be spliced
[complex] boolean If recursively split a complex sentence By default, True
[compound] boolean If recursively split a compound sentence By default, True

Return:

A list of simple sentences

Remarks:

  • The resultant sentences will not have ending punctuations.
  • Will consider proper names. For example, the sentence "I ate Fish and Chip." will not be splitted, because "Fish and Chip" can be intelligently identified as proper names.

Legal Concept Tree

API Functions:

get_concepts

Get the indexed tree data structure of legal concepts.

Return:

A map of tree nodes, where keys are legal terms and values are corresponding tree node.

Definition of node:

class Node:
    def __init__(self, parent, value):
        self.parent = parent
        self.value = value
        if parent:
            self.level = parent.level + 1
        else:
            self.level = 0

Sentence to Label

API Functions:

sent2label

Generate several legally important tags from a sentence.

Argument Type Usage Remark
sentence string The target sentence
[mode] HYBRID/CON/DEP/WORD Which tag generation method to use By default, HYBRID

Return:

A list of phrases