國立虎尾科技大學 |

Natural Language Processing for Justifiable Legal Practitioner Assistance.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Natural Language Processing for Justifiable Legal Practitioner Assistance./
Author:	Bhambhoria, Rohan.
Description:	1 online resource (122 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-11, Section: B.
Contained By:	Dissertations Abstracts International85-11B.
Subject:	Computer engineering. -
Online resource:	click for full text (PQDT)
ISBN:	9798382230399

Natural Language Processing for Justifiable Legal Practitioner Assistance.
Bhambhoria, Rohan.

Natural Language Processing for Justifiable Legal Practitioner Assistance. - 1 online resource (122 pages)

Source: Dissertations Abstracts International, Volume: 85-11, Section: B.

Thesis (Ph.D.)--Queen's University (Canada), 2024.

Includes bibliographical references

The significant advancement of artificial intelligence (AI) has had a profound impact on many areas of applications. With Natural Language Processing (NLP) being a key player in this technological progress, exploring NLP within vital areas represents a pathway for immediate practical benefits while also challenging the existing AI solutions and helping to develop better models. This dissertation is centered around a comprehensive examination of NLP in the legal domain, a realm where the interpretation of language is of critical importance.In this dissertation, we investigate three main aspects of challenges: (i) The interpretability of NLP models, which is essential for legal practitioners to derive meaningful insights. To solve this challenge, we utilize post-hoc and ante-hoc methods of interpretability and show that the latter can be utilized to develop versatile interpretable solutions for several tasks in the legal domain. Specifically, we introduce a system that contains model-agnostic interpretable intermediate layers, a technique that proves to be effective for legal documents. (ii) The evaluation of existing datasets while identifying the disparity between the controlled experimental settings in academic research and the real-world legal problems. For this challenge, we propose a systematic annotation protocol that involves the expertise of legal professionals in dataset curation. We utilize weakly supervised learning by means of a curriculum learning strategy, effectively demonstrating the improved performance of a deep learning model. Then, we design a task with parallels to the thought-process of lawyers, i.e., with reference to both precedents and legislative provisions. Additionally, we introduce a prototype architecture to add interpretability, achieving strong performance while adhering to decision parameters used by lawyers. (iii) Lastly, the task of adapting general-purpose NLP models for specific legal tasks needs to be examined. We refactor conventional tasks on hierarchical datasets into a more indicative long-tail prediction task. We observe LLMs are prone to failure in these cases. To address these limitations, we propose the use of entailment-contradiction prediction in conjunction with LLMs, which allows for strong performance in a strict zero-shot setting. Our method does not require any parameter updates, a resource-intensive process and achieves strong performance across multiple datasets.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798382230399Subjects--Topical Terms:

569006
Computer engineering.
Subjects--Index Terms:

Natural Language ProcessingIndex Terms--Genre/Form:

554714
Electronic books.

Natural Language Processing for Justifiable Legal Practitioner Assistance.
LDR:03787ntm a22003737 4500 001 1145009
005 20240617111739.5
006 m o d
007 cr mn ---uuuuu
008 250605s2024 xx obm 000 0 eng d
020 $a 9798382230399
035 $a (MiAaPQ)AAI31252217
035 $a (MiAaPQ)QueensUCan_197432803
035 $a AAI31252217
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Bhambhoria, Rohan. $3 1470207
245 1 0 $a Natural Language Processing for Justifiable Legal Practitioner Assistance.
264 0 $c 2024
300 $a 1 online resource (122 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-11, Section: B.
500 $a Advisor: Zhu, Xiaodan.
502 $a Thesis (Ph.D.)--Queen's University (Canada), 2024.
504 $a Includes bibliographical references
520 $a The significant advancement of artificial intelligence (AI) has had a profound impact on many areas of applications. With Natural Language Processing (NLP) being a key player in this technological progress, exploring NLP within vital areas represents a pathway for immediate practical benefits while also challenging the existing AI solutions and helping to develop better models. This dissertation is centered around a comprehensive examination of NLP in the legal domain, a realm where the interpretation of language is of critical importance.In this dissertation, we investigate three main aspects of challenges: (i) The interpretability of NLP models, which is essential for legal practitioners to derive meaningful insights. To solve this challenge, we utilize post-hoc and ante-hoc methods of interpretability and show that the latter can be utilized to develop versatile interpretable solutions for several tasks in the legal domain. Specifically, we introduce a system that contains model-agnostic interpretable intermediate layers, a technique that proves to be effective for legal documents. (ii) The evaluation of existing datasets while identifying the disparity between the controlled experimental settings in academic research and the real-world legal problems. For this challenge, we propose a systematic annotation protocol that involves the expertise of legal professionals in dataset curation. We utilize weakly supervised learning by means of a curriculum learning strategy, effectively demonstrating the improved performance of a deep learning model. Then, we design a task with parallels to the thought-process of lawyers, i.e., with reference to both precedents and legislative provisions. Additionally, we introduce a prototype architecture to add interpretability, achieving strong performance while adhering to decision parameters used by lawyers. (iii) Lastly, the task of adapting general-purpose NLP models for specific legal tasks needs to be examined. We refactor conventional tasks on hierarchical datasets into a more indicative long-tail prediction task. We observe LLMs are prone to failure in these cases. To address these limitations, we propose the use of entailment-contradiction prediction in conjunction with LLMs, which allows for strong performance in a strict zero-shot setting. Our method does not require any parameter updates, a resource-intensive process and achieves strong performance across multiple datasets.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Computer engineering. $3 569006
650 4 $a Electrical engineering. $3 596380
653 $a Natural Language Processing
653 $a AI solutions
655 7 $a Electronic books. $2 local $3 554714
690 $a 0544
690 $a 0464
690 $a 0800
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a Queen's University (Canada). $3 1148613
773 0 $t Dissertations Abstracts International $g 85-11B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31252217 $z click for full text (PQDT)