But, see, accord: Generating "Blue Book" citations in HYPO
Citation: Kevin D. Ashley, Edwina L. Rissland (1987) But, see, accord: Generating "Blue Book" citations in HYPO. Proceedings of the 1st international conference on Artificial intelligence and law (RSS)
DOI (original publisher): 10.1145/41735.41744
Semantic Scholar (metadata): 10.1145/41735.41744
Sci-Hub (fulltext): 10.1145/41735.41744
Internet Archive Scholar (search for fulltext): But, see, accord: Generating "Blue Book" citations in HYPO
Download: http://portal.acm.org/citation.cfm?doid=41735.41744
Tagged: Computer Science
(RSS) argumentation (RSS), legal argumentation (RSS), citations (RSS), citation signals (RSS), HYPO (RSS), artificial intelligence and law (RSS), claims (RSS), legal citations (RSS), case-based reasoning (RSS)
Summary
HYPO is designed to support lawyers by modeling legal reasoning based on cases and hypothetical situations. It compares a "current fact situation" (cfs) to the most relevant prior cases to analyze and structure possible legal arguments.
Legal citations are stylized with introductory signals which indicate whether a prior case supports or weakens an argument, and whether it is an exact match to the current situation, or only partly analogous. Further, textual citations are expected to be given in a particular order as specified by the Blue Book:
- Legal proposition
- Introductory signal
- Authority
- Parenthetical explanation
- Related authority
Thus "the important linkages are in the introductory signals, the important facts in the parenthetical explanations."
HYPO's model
HYPO uses a Case Knowledge Base (CKB) and a library of dimensions.
CKB
The CKB describes aspects such as the plaintiff, defendant, and important facts in frames.
Dimensions
The dimensions represent the legal relevance, generalizing the facts which indicate the strengths and weaknesses. An example dimension would be "disclose-secrets": Plantiff did not voluntarily disclose secrets to outsiders. HYPO takes about 30 dimensions, drawn from legal scholarship. A dimension has six facets:
- factual prerequisites
- focal slots (the key prerequisites) and
- their range of values
- how to strengthen the focal slots
- the claims for which the dimension has relevance
- cases indexed from the CKB
HYPO's algorithm
HYPO produces a case-analysis-record with the facts, applicable dimensions, near-miss dimensions, potential claims, and relevant cases in the CKB. This case-analysis-record is used to construct the claim lattice, based on the applicable and near-miss dimensions; the most relevant cases (i.e. those sharing the greatest number of dimensions) are closest to the root of the claim lattice.
The claim lattice then indicates different ways to argue the case, and the strongest arguments can be read off the lattice. Including the near-miss dimensions is useful because they are very nearly "on-point"; additional facts could make them on-point, or these cases could be used to argue by analogy. Further, the missing facts point to possible ways the opponent could distinguish the cfs from the prior case.
Since the claim lattice shows the arguments from both the plantiff's and defendant's position, one can guess at the opponent's strongest arguments and counterarguments. (In this way the claim lattice is a bit like a map of possible moves in a boardgame.)
Citation Display
From the claims lattice, HYPO generates a "Cites Display" which uses introductory signals to represent the relevant cases from the CKB. This provices the key cases from the CKB which could be cited, which begins to make a skeleton argument.
Figure 9 shows the algorithm for generating an introductory signal from the claims lattice.
Theoretical and Practical Relevance
Further information about how the CKB was generated would be helpful.
The authors suggest that arguments could be evaulated with the Cites Display, since all Accord and no Contra indicate a strong argument. This point, while obvious, might be extended to interfaces beyond the legal domain, with suitable representations for the pro and con arguments.
