Vol.:(0123456789)
Artifcial Intelligence and Law
https://doi.org/10.1007/s10506-020-09273-1
1 3
ORIGINAL RESEARCH
Scalable and explainable legal prediction
L. Karl Branting
1
· Craig Pfeifer
2
· Bradford Brown
1
· Lisa Ferro
3
·
John Aberdeen
3
· Brandy Weiss
1
· Mark Pfaf
3
· Bill Liao
1
© Springer Nature B.V. 2020
Abstract
Legal decision-support systems have the potential to improve access to justice,
administrative efciency, and judicial consistency, but broad adoption of such sys-
tems is contingent on development of technologies with low knowledge-engineer-
ing, validation, and maintenance costs. This paper describes two approaches to an
important form of legal decision support—explainable outcome prediction—that
obviate both annotation of an entire decision corpus and manual processing of
new cases. The frst approach, which uses an attention network for prediction and
attention weights to highlight salient case text, was shown to be capable of predict-
ing decisions, but attention-weight-based text highlighting did not demonstrably
improve human decision speed or accuracy in an evaluation with 61 human subjects.
The second approach, termed semi-supervised case annotation for legal explana-
tions, exploits structural and semantic regularities in case corpora to identify textual
patterns that have both predictable relationships to case decisions and explanatory
value.
Keywords Artifcial intelligence and law · Machine learning · Human language
technology · Explainable prediction
1 Introduction
Recent advances in artifcial intelligence (AI) and human language technology
(HLT) have created new opportunities to automate routine aspects of case manage-
ment and adjudication, freeing human experts to focus on aspects of these tasks that
most require human judgment and knowledge. An important application of this tech-
nology is decision support for routine administrative decision-making and adjudica-
tion. In nations across the world, benefts adjudications, resolution of commercial
conficts, criminal defense, and other forms of access to justice are often impeded by
* L. Karl Branting
lbranting@mitre.org
Extended author information available on the last page of the article