Constructing Legal Arguments with Rules in the Legal Knowledge Interchange Format (LKIF)

tags

LKIF

Notes

open, vendor-neutral standard for exchanging formal models of the law, suitable for use in legal knowledge-based systems.

NOTER_PAGE: (1 . 0.5590062111801243)

legal rules are interpreted as policies for reasoning in resource-limited, decision-making processes.

NOTER_PAGE: (2 . 0.10559006211180125)

use LKIF rule bases to construct and visu- alize arguments about cases.

NOTER_PAGE: (2 . 0.5058661145617668)

Rules are reified in this language, with an identifier and a set of properties

NOTER_PAGE: (2 . 0.8322981366459627)

unlike Horn clause logic, rules may have negative conclusions.

NOTER_PAGE: (3 . 0.3947550034506556)

A party who wants to use some rule need not show that no exception applies. The burden of proof for exceptions is on those interesting in showing the rule does not apply.

NOTER_PAGE: (3 . 0.5169082125603864)

In knowledge bases for particular legal domains, rules can be prioritized both extensionally, by asserting facts about which rules have priority over which other rules, and intensionally, using meta-rules about priorities.

NOTER_PAGE: (3 . 0.8661145617667357)

syntax of LKIF rules by noting that it can be viewed as a generalization of the syntax of Horn clause logic

NOTER_PAGE: (6 . 0.4748102139406487)

Three premises, implicit in each rule, are made explicit here.

NOTER_PAGE: (8 . 0.7066942719116632)

The semantics of negation is dialectical, not classical negation or negation-as-failure.

NOTER_PAGE: (9 . 0.5617667356797792)

The closed-world assumption is not made.

NOTER_PAGE: (9 . 0.5762594893029676)

In Carneades, a negated sentence, (not p), is acceptable just when the complement of the proof standard assigned to p is satisfied

NOTER_PAGE: (9 . 0.6004140786749482)

two ways to represent- ing arguments, rules, ontologies and cases in XML. One uses OWL

NOTER_PAGE: (11 . 0.8461007591442374)

rules and arguments cannot be conveniently written or maintained using generic OWL editors

NOTER_PAGE: (12 . 0.17805383022774326)

implementing a translator for LKIF documents encoded in OWL requires the document to first be preprocessed into some canonical concrete syntax

NOTER_PAGE: (12 . 0.3588681849551415)

LKIF offers an alternative, more compact, XML syntax

NOTER_PAGE: (12 . 0.40993788819875776)

‘Carneades’ is the name of both a computational model of argumentation [19] and an implementation of this model

NOTER_PAGE: (14 . 0.6721877156659766)

The ESTRELLA module for reasoning with ontologies is still being designed.

NOTER_PAGE: (15 . 0.7895100069013112)

ontologies, rules and cases are all at the same layer in our system, as they are all interpreted as knowledge representa- tion formalisms from which arguments can be constructed

NOTER_PAGE: (16 . 0.3077984817115252)

we are interested in finding argument graphs which provide sufficient grounds, or reasons, for ‘accepting’ the statement of the query, presumptively

NOTER_PAGE: (17 . 0.5935127674258109)

legal reasoning cannot be viewed, in general, as the application of some deductive logic, such as first-order predicate logic, to some theory of the facts and relevant legal domain.

NOTER_PAGE: (19 . 0.8274672187715666)

Mechanical jurisprudence, as this model has been called, is somewhat of a strawman. It was soundly rejected by rule skeptics like the realists. As Gardner puts it, law is more “rule-guided” than “rule- governed.”

NOTER_PAGE: (20 . 0.17529330572808835)

The process of determining whether the facts of a case can be “subsumed” under some legal concept is one of argumentation.

NOTER_PAGE: (20 . 0.48309178743961356)

Argumentation cannot be reduced to logic.

NOTER_PAGE: (21 . 0.28502415458937197)

Legal rules are neither material implications nor procedures for updating variables in working memory, but rather schemes for constructing legal arguments.

NOTER_PAGE: (21 . 0.8198757763975155)