Judicial Authorship attribution and Influence Detection (2020) [PDF]

Task

We try to solve two research questions:

• Can we predict which judge wrote which opinion text (Yes)
• Can we detect influence of some judges on the decision of other judges (Maybe)

Method

Architecture: After some data cleaning, we use a bag of words with n-Grams approach and then apply an SVM to classify
Performance: On the Authorship Attribution task we achieve an accuracy up to 95% (Average for the newest time-split: 0.85)

This Repository contains:

• Report 'Judicial Authorship attribution and Influence Detection'
• Folders 85 - 97 for each circuit court (See list below) which each contains:
  • Jupyter notebook with replication code
  • Results of the Notebooks

Number to american circuit court mapping

85: 1st Circuit Court
86: 2nd Circuit Court
87: 3rd Circuit Court
88: 4th Circuit Court
89: 5th Circuit Court
90: 6th Circuit Court
91: 7th Circuit Court
92: 8th Circuit Court
93: 9th Circuit Court
94: 10th Circuit Court
95: 11th Circuit Court
96: 12th Circuit Court
97: Supreme Court