NALEA

The deadline for submitting papers to the NaLELA workshop has been extended to:

Friday April 10

For further information, see the link under Pages.

We look forward to your participation at the workshop.

The following is a post by Nick Bassiliades on 12/12/2008 with a reply by Tom van Engers

Concerning the NaLea project proposal I have suggestions regarding the role of onologies.

I believe that the role of ontologies should be leveraged, in order to stress out that the project builds upon existing results of EU projects on Semantic Web. It should be made more clear (although it is already mnetioned in the 5-page document) that there will be two types of ontologies:

• one domain-independent that concerns the argumentation process itself
• many domain-dependent that concern the argument of discourse of each application

The former guides argument identification (debate structure), while the latter guide the recognition of argument/debate content. Classical NLP techniques (guided by grammars/vocabularies) could be boosted by organizing templates of grammars/vocabularies around ontologies. For example, when a general-purpose grammar fails to correctly identify some arguments, whereas dedicated/specific grammars could perform better. So, ML classification could quickly identify the specific situation/sub-domain we are dealing with, and them specific grammar could be employed.

Another issue could be the distributed/heterogeneous nature of argumentation according to specific sub-groups of the community. For example, concerning policy setting, there might be different groups of people which may use specific vocabularies or may have specific argumentation requirements. The framework of Nalea could support specific argumentation sub-groups by making customized ontology versions, that are compatible with the general ontologies. Then groups of people could use their own sub-domain ontologies for arguing among themselves and when consensus is reached they could propagate their sub-group resolution to the larger community. For example, imagine a dispute among a national labor union and the national society of employers. Each union needs to reach a consensus among themselves before arguing against the other. So both of them could use the Nalea framework, possibly using a subset of the ontology or the vocabulary. Maybe people from different industries also need to reach a consensus among themselves before entering the dialog with people from different industries. So they may need a slightly more specific vocabulary/ontology than the general ontology.

This may call for ontology modules that are used in addition to top-level/common ontologies for each different application or sub-group of people using the system.

Reply by Tom van Engers:

I completely agree with the perspective that ontologies should play an important role. Not just for the sake of it that they provide us with the terminological basis for the argumentation part of the automated system’s components. We need also other ontologies and knowledge layers (some of them can be expressed in OWL DL btw) such as mereology, norms and the generic cases therein, etc.. not to forget some core and top ontologies expressing common sense knowledge. Especially challenging for me would using a priori knowledge in combination with statistical and pattern matching oriented techniques typical for current NLP approaches. The Leibniz Center for Law has a long lasting background in creating such ontologies and in the Harness approach (see the Jurix proceeding 2008) we also have a tool for integrating ontologies and normative reasoning. We hope to reuse this and expand it in the Nalea project.

Held at the International Conference on AI and Law
June 8-12 2009, Barcelona, Spain

Workshop date: June 12, 2009

The aim of this workshop is to draw together researchers around the issues of the empirical analysis, formalisation, and implementation of legal argumentation in natural language. Such a system would be a decision-support tool which translates natural language arguments into and out of an argumentation framework or logic which supports reasoning and inference. As the interface is in natural language, the tool would be accessible to a wide range of end-users. The workshop builds on recent advances in natural language engineering and argumentation including: controlled languages, predictive editors, text mining and corpus analysis, natural language parsing, ontology construction, translation of natural language sentences into first order logic, logical inference, linguistic analysis of argumentation, and computational theories of argumentation. It draws on an interdisciplinary community in Computer Science, Linguistics, and the Law.

While argumentation can be addressed in a broad range of areas, the workshop focusses particularly on the language, logic, and computation of legal argumentation such as that found between lawyers arguing a case before a court or found in legal briefs and decisions where justifications are given for and against a decision.

For further information and submission, see NaLELA 09

This blog has been set up for discussion of topics related to natural language engineering of argumenation — corpora, text-mining, ontology development, linguistics, logic, and computation.