H. Westermann et al. Toward an Intelligent Tutoring System for Argument Mining in Legal Texts

2025-05-06 0 0 760.38KB 10 页 10玖币

侵权投诉

H. Westermann et al. /

Toward an Intelligent Tutoring System for

Argument Mining in Legal Texts

Hannes WESTERMANN a,1, Jarom´

ır ˇ

SAVELKA b, Vern R. WALKER c,

Kevin D. ASHLEY dand Karim BENYEKHLEF a

aCyberjustice Laboratory, Facult´

e de droit, Universit´

e de Montr´

eal

bSchool of Computer Science, Carnegie Mellon University

cLLT Lab, Maurice A. Deane School of Law, Hofstra University

dSchool of Computing and Information, University of Pittsburgh

Abstract. We propose an adaptive environment (CABINET) to support caselaw

analysis (identifying key argument elements) based on a novel cognitive computing

framework that carefully matches various machine learning (ML) capabilities to the

proﬁciency of a user. CABINET supports law students in their learning as well as

professionals in their work. The results of our experiments focused on the feasibility

of the proposed framework are promising. We show that the system is capable of

identifying a potential error in the analysis with very low false positives rate (2.0-

3.5%), as well as of predicting the key argument element type (e.g., an issue or a

holding) with a reasonably high F1-score (0.74).

Keywords. Intelligent tutoring system, caselaw analysis, case brief, legal education,

legal annotation, legal text classiﬁcation, argument mining, human-computer

interaction.

1. Introduction

In this paper we examine the application of cognitive computing [16] to support both

a law student learning how to extract key arguments from a court opinion and a legal

expert performing the same. We propose an adaptive environment that evolves from a

tutoring system to a production annotation tool, as a user transitions from a learner to

an expert. The concept is based on a novel cognitive computing framework where (1)

the involvement of machine learning (ML) based components is carefully matched to

the proﬁciency level of a human user; and (2) the involvement respects the limitations

of the state-of-the-art of automated argument mining in legal cases. We experimentally

conﬁrm feasibility of the key ML components by testing the following two hypotheses:

Given a sentence in a case brief, it is possible (H1) to detect if the sentence is placed in

an incorrect section, and (H2) to predict the correct section for the sentence.

1Corresponding Author: Hannes Westermann, E-mail: hannes.westermann@umontreal.ca

arXiv:2210.13635v1 [cs.CL] 24 Oct 2022

H. Westermann et al. /

2. Background

Lawyers routinely analyze case decisions (i.e., court opinions) to gain insight into what

is a persuasive or binding precedent (typically common law countries) and/or what is the

established decision-making practice in a given matter (typically civil law countries). As

the list of relevant cases may be long and the opinions might be sizeable, a principled

approach to the analysis is necessary to make the task feasible and as efﬁcient/effective

as possible. Such an approach requires knowing how to read an opinion, which parts to

focus on, and which information to identify as crucial for understanding the case.

In U.S. law schools, case briefs are widely employed to teach law students how to

analyze a case and how to use prior decisions to create new arguments or analyses [6].

Writing a case brief involves reading and understanding a case, and identifying text pas-

sages that contain the key aspects of the decision. These are then extracted and arranged

in a structured format that often includes the following sections:

• Facts - Events and actions relevant to the dispute.

• Issue - Main questions (points of contention) the court must address.

• Holding - Legal rulings when the law is applied to a particular set of facts.

• Procedural History - The treatment the dispute has received from the courts.

• Reasoning - The analysis of the court leading to the outcome.

• Rule - The ofﬁcial rules the court must adhere to (e.g., statutory provisions).

Interestingly, many professors never ask students to turn in their briefs and, hence, do

not provide a learner with much needed feedback. [18] However, practice and feedback

are essential for learning. When it comes to practice, the research clearly shows that

it should be focused and deliberate [8], at the appropriate level of challenge [8], and

in sufﬁcient quantity [14]. Such practice should be coordinated with targeted feedback

on speciﬁc aspects of students’ performance in order to promote the greatest learning

gains. [2,5] Feedback should also be timely, i.e., immediate and frequent [13]. These

elements do not seem present when it comes to learning to brief cases. As a result, while

law students tend to start out by dutifully brieﬁng cases, they usually switch to a less

detailed approach after a few weeks, focused on color-coding sentences or taking notes

in the margins of the case texts. Due to the lack of feedback and practice, it is thus unclear

whether the crucial skill of brieﬁng cases has been acquired.

To address the issue we propose CABINET, an intelligent tutoring system that grad-

ually evolves from a platform aimed at learners to a powerful annotation environment

to support an expert. In a nutshell, CABINET allows a user to select a sentence and as-

sign it to one of the case brief’s sections. More importantly, the system provides varying

levels of scaffolding (i.e., varying levels of challenge) and timely feedback appropriate

for the learner’s level of proﬁciency to maximize the learning outcomes. The tool thus

adapts with the user, teaching them how to brief cases at ﬁrst and later supporting them

in brieﬁng and understanding cases more efﬁciently.

3. Related Work

Numerous researchers have proposed frameworks where a human and a computer com-

plement each other in performing tasks in the legal domain. For example, human-aided

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

H.Westermannetal./TowardanIntelligentTutoringSystemforArgumentMininginLegalTextsHannesWESTERMANNa;1,Jarom´rSAVELKAb,VernR.WALKERc,KevinD.ASHLEYdandKarimBENYEKHLEFaaCyberjusticeLaboratory,Facult´ededroit,Universit´edeMontr´ealbSchoolofComputerScience,CarnegieMellonUniversitycLLTLab,MauriceA.DeaneSc...

展开>> 收起<<

H. Westermann et al. Toward an Intelligent Tutoring System for Argument Mining in Legal Texts.pdf

共10页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

H. Westermann et al. Toward an Intelligent Tutoring System for Argument Mining in Legal Texts

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: