RE: Idiomatic LegalRuleML samples for the GitHub repository

Dear all,

As agreed in the last Skype meeting, I have added the Right in ps44, in the selected examples:

<lrml:Right iri="devoOntology">

<ruleml:Atom>

<ruleml:Rel>applyToMakeNotEnforceable</ruleml:Rel>

<ruleml:Var>penalty</ruleml:Var>

</ruleml:Atom>

</lrml:Right>

And then I added at the beginning:

<lrml:Prefix pre="devoOntology" refID="https://docs.oasis-open.org/legalruleml/devoOntology#" />

The new file is attached.

Nevertheless, it does not validate it. Specifically, it does not like the "iri" attribute:

col:37 - riga:457 - Errore:cvc-complex-type.3.2.2: Attribute 'iri' is not allowed to appear in element 'lrml:Right'.

I'm validating with respect to "lrml-compact.xsd":

xsi:schemaLocation="http://docs.oasis-open.org/legalruleml/ns/v1.0/ http://docs.oasis-open.org/legalruleml/legalruleml-core-spec/v1.0/csprd02/xsd-schema/compact/lrml-compact.xsd"

Yes, I saw in the documentation that the "iri" attribute is allowed on deontic operators, e.g.:

<lrml:Obligation

key="oblig1"

iri="ex:achievementObligation">

...

</lrml:Obligation>

So I don't understand what's going on...

From: Robaldo Livio.
Sent: 22 November 2020 15:16
To: legalruleml@lists.oasis-open.org
Subject: Idiomatic LegalRuleML samples for the GitHub repository

Dear LegalRuleMLs,

As agreed on the Skype chat, we have worked on a corpus for regulating smoking in public in Scotland (the "smoking corpus").

Livio tagged one file from this corpus in idiomatic LegalRuleML, and we discussed it. The result is the file "IDIOMATIC - asp_20160003_en_EO_AF_AW_AA (1).xml", attached.

Since this file is not linked to any logic or ontology, Livio basically "invented" the predicates. The list of all predicates used is in the file "predicates.txt", likewise attached. Some of them are of course rather weird, as they would need to be decomposed into smaller predicates; predicates would also need to be related of one another; this is a problem that goes all the way back to the first studies of the British Nationality Act. However, these are the same problems as would arise with the RuleML part of the XML format, so are not LegalRuleML problems per se. Here, we focus on the LegalRuleML tags that are concerned with legal/deontic reasoning.

The file "IDIOMATIC - asp_20160003_en_EO_AF_AW_AA (1).xml" includes 52 norms. Since these are too many, we made a selection. We selected only the following sections of the file, which includes all main kinds of norm we can find in legal text. The selection has been copied into the file "SELECTED EXAMPLES FROM IDIOMATIC - asp_20160003_en_EO_AF_AW_AA (1).xml".

For the time being, we propose to first address these main LegalRuleML tags: Obligation, Permission, Prohibition, ConstitutiveStatement, Overriding, Right, PenaltyStatement, ReparationStatement (other than Right, this is the same set as were analysed in the initial smoking corpus). Note that there are lot of additional LegalRuleML tags that are not in this document, but we should better work on these separately, as it is not easy to find sample occurrences in existing legislation (in fact, we didn't find any in the smoking corpus). In the future, we can try to find other real-world occurrences, from other legislative documents, to exemplify the missing tags.

==========================================================================================

stmts1

1 Permission

1 Prohibition

1 Overriding

stmts2

1 PenaltyStatement

1 ReparationStatement

stmts12

1 Permission

2 Obligations

1 ConstitutiveStatement

stmts14

1 ConstitutiveStatement

stmts16

1 Right => WE DON'T KNOW HOW TO REPRESENT IT!!!

1 Permission

==========================================================================================

So, these should be the real-world examples that we should upload on the GitHub and try to formalize in other logical frameworks (DDL, I/O logic, etc.).

There are three problems that we would like to discuss in the next Skype meeting or in this mailing list. The first two are about these tags and the selected examples. The third is about the previous representation:

As you can see above, in stmts16 there is a Right. We don't know how to formalize it. Any help from you is more than welcome.
In stmts2, there is a PenaltyStatement and a ReparationStatement. In the former, we used the LegalRuleML tag "<lrml:SuborderList>". This is not fully clear to us and we would like to talk about this.

We read in the LegalRuleML core specification: "One of the characteristics of norms is that having violated them, a penalty can be introduced to compensate for the violation, where a penalty is understood to also be a deontic formula. To model this feature of norms and legal reasoning, Governatori and Rotolo [25] introduced what is called here a suborder list"
We think the suborder list needs to have something like: suborder [ProhibitionOnSmoking; ObligationToPayPenalty]. But, exactly how this is stated to maintain all the conditions on the prohibition is unclear to us at the moment. Also how to place the PenaltyStatement is unclear to us. Is it: PenaltyStatement [suborder [ProhibitionOnSmoking; ObligationToPayPenalty]]? Or suborder [ProhibitionOnSmoking; PenaltyStatement[ObligationToPayPenalty]]? Or: [ProhibitionOnSmoking; ObligationToPayPenalty[PenaltyStatement]]? The latter makes most sense to us, but is it valid and related to other's intuitions.

The current representation loses all the advantages for which the smoking corpus was designed. That is, it can no longer facilitate visualisation and querying with respect to full string search within a sentence and with respect to a norm classification. So, searching for 'powers of Scottish Ministers' is no longer feasible, though this was feasible with the previous analysis. This satisfied colleagues in legislative counsel's office of the Scottish Government. This could be resolved simply by incorporating a Paraphrase element within the scope of the norm and containing the full string of the source sentence along the lines of the previous version. We'd need to test if this still works with the previous query and visualisation tool, but something along these lines has proven to be very useful and a good communication tool.

Looking forward to discussion.

Thanks,

Livio & Adam

legalruleml message