Browse Prior Art Database

A method and system is disclosed for automatically analyze tone of court opinions/legal cases.

IP.com Disclosure Number: IPCOM000247900D
Publication Date: 2016-Oct-11
Document File: 4 page(s) / 54K

Publishing Venue

The IP.com Prior Art Database

Abstract

System and Method for Analyzing the Tone of Legal Cases

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 39% of the total text.

Page 01 of 4

A method and system is disclosed for automatically analyze tone of court opinions/legal cases.

Disclosed is a method and system for automatically analyze tone of court opinions/legal cases.

In accordance with the method and system disclosed herein, there are five main tones: Anger, Sadness, Joy, Disgust, Fear. These main tones are expanded so as to have gradients within them. The main tones are expanded as shown in the following:

• Anger - Frustration, Outrage, Protest, Piqued • Sadness - Disappointment, Disapproval • Joy - Approval, Compassion, Support, Esteem • Disgust - Dislike, Objection, Oppose, Reject, Dissatisfied, Dispute
• Fear - Concerned, Uneasiness, Doubt, Dismay

The system is configured to automatically retrieve published cases and apply natural language processing to detect entities and perform tone analysis to distinguish the tones used in reference to each case. The system maintains tones for each case in a database for subsequent usage.

The entire case opinion can be analyzed. Consider the following implementation:

(1) automatically retrieve published cases, (2) apply natural language processing to cases
(2.1) Apply an entity extraction program on each case X on the database

         (2.2) Detect the "entities" (i.e. Legal cases) mentioned in the document X, with their surrounding context. (This is not limited to "N sentences before and after"; existing Natural Language Processing tools can provide an accurate context in reference to an object).

         (2.3) Determine from these entities when the object is a case, call them Y_1, Y_2, …; so a "case treatment" on each case Y_i is obtained.

(3) on each case treatment, apply a tone analyzer to distinguish the tones used in reference to each case Y_i, (4) Combine the obtained tones and apply the trained models to derive a "tone treatment" on case Y_i.

    (5) persist these tones in storage
Later on, when a given case Y needs to be analyzed by a user of the system (e.g. a legal practitioner), "tone treatment" flags are retrieved from the system, obtaining a list such as (example)


- Case X_1 (AA vs BB, this court, this date) treats case Y as "frustrated (anger)" with score 0.8
- Case X_2 treats case Y as "supported (joy)" with score 0.9 and "concerned (fear)" with score "0.2"

1


Page 02 of 4

These flags, accompanied by their scores (gradients) and the original text of each case, can solve the problem of finding out what type of case treatment has been received in each case. For example, if Anger and Disgust are high and Joy is low, then the case if flagged "Criticized" with high probability.

Further, the tones of the entire opinion are compared with previous opinions from the same judge. The system keeps track of the judge's emotions on a case by case basis. The system also compares the tones of judges' opinions to get a better idea of how the judges feel with each issue.

Consider the following examples:


A) "In Milkovich v. Lorrain, the United States Supreme Court rejected the notio...