AW: Lexico-Syntactic Patterns

Philipp, John, Ontolex-Members,

Thanks for your helpful comments.

I think the differentiation between general, i.e. Hearst, and domain-specific lexico-syntactic patterns is a vital one.
Regarding a concrete model, such as lemon, I think representing domain-specific patterns might be interesting for specific patterns applied to given ontologies.

I was thinking about patterns similar to the frames you exemplify, John, only in a more formal representation, which we would have to discuss.
The purpose of such a formalized representation might be to induce ontological properties/elements, on the other hand to document patterns applied to existing ontologies for evolution, matching, etc.

Regards,
Dagmar


________________________________
Von: Philipp Cimiano [mailto:cimiano@cit-ec.uni-bielefeld.de]
Gesendet: Donnerstag, 06. September 2012 08:08
An: public-ontolex@w3.org
Betreff: Re: Lexico-Syntactic Patterns

Dagmar, John, all,

of course these Hearst-style patterns are inherently ambiguous (this holds by the way for most of the entries in the lexicon).

Nevertheless, I agree with Dagmar that it might be useful to represent such patterns in a special ontology-lexicon, but this would not be a lexicon specific for a given ontology, but a domain-independent lexicon that captures the possible meaning of lexico-syntactic patterns with respect to OWL.

Just my two cents,

Philipp.


Von: johnmccrae@gmail.com [mailto:johnmccrae@gmail.com] Im Auftrag von John McCrae
Gesendet: Freitag, 31. August 2012 15:29
An: Gromann, Dagmar
Cc: public-ontolex@w3.org
Betreff: Re: Lexico-Syntactic Patterns

Hi Dagmar,

I was wondering what exactly you mean with lexico-syntactic patterns. In lemon we certainly had frames that could induce ontological properties, e.g.,

"X is the capital of Y" => X onto:capital Y

But these are fairly standard and similar to other semantic role/syntactic frames (and should clearly be a part of our work in this group).

On the other hand, we have Hearst patterns like

"X such as Y" => Y rdfs:subClassOf X

However, these kind of patterns would not traditionally belong in the lexicon as they do not represent a direct mapping, i.e.,

  1.  It is difficult to say when "such as" actually indicates a subclass relation (i.e., it is error prone)
  2.  It was not the intention of the speaker to use "such as" to express a subclass relation
Of course, it may be interesting to consider a representation of something like this in the group. Could you give some examples of the kind of mapping you are interested in, and how they might be represented (relative to an existing model like lemon).

Regards,
John

On Wed, Aug 22, 2012 at 3:29 PM, Gromann, Dagmar <Dagmar.Gromann@wu.ac.at<mailto:Dagmar.Gromann@wu.ac.at>> wrote:
Dear Ontolex members,

I would like to contribute some ideas on lexico-syntactic patterns to the specification of the requirements and the ongoing discussion.

I have two suggestions for the specification of patterns within the ontology-lexicon interface:
1) Apply lexico-syntactic patterns to axiomatizing semi-formal definitions (maybe particularly verbs)
2) Represent/trace applied patterns (in the event of ontology evolution/matching/design)

The current ontology-lexicon model refers to syntax in form of frames to represent the syntactic behavior of lemon entries. If I am not mistaken, frames may be defined as the elements used in patterns and within the world of ontology design are referred to as design patterns. In my mind, adding lexico-syntactic patterns to the ontology-lexicon interface may help to bridge the perceived gap between ontology and lexicon/linguistics. Such patterns help to establish various semantic relations and might contribute to axiomatizing semi-formal definitions. They can facilitate the formalization of lexical knowledge and/or matching of formalized knowledge on the basis of patterns. To some extent they contain knowledge about the usage of lexical entries, which might be matched to other domains/resources. Additionally, applying and representing lexico-syntactic patterns increases the reusability of the lexicon.

Such representation of patterns might help the modeling/interpretation of mappings across both resources, as has been suggested: http://www.w3.org/community/ontolex/wiki/Specification_of_Requirements/Lexicon-Ontology-Mapping (in reference to dynamic and static verbs).

By applying lexico-syntactic patterns to the process of ontology design/evolution, we might derive ontology elements. However, the choice of pattern on the basis of natural language text and/or definitions is usually not documented in existing ontologies. The representation of lexico-syntactic patterns constituting basic ontology design/evolution motivations and decisions can facilitate not only ontology evolution but also ontology-based information extraction, question answering, ontology alignment/matching, etc. Patterns recur across genres/domains, which might render their representation in the lexicon a useful resource for ontology matching. For matching ontological concepts by means of patterns lexical information about relations is essential, which is the reason why I see the representation of lexical patterns in an environment of rich lexical information as highly beneficial. For examply, [1] shows the matching of ontology elements based on lexical properties.

One major issue might be that lexico-syntactic patterns are language specific and might not be available in all languages.

I would be very grateful for any comments or specifications to these suggestions.

Kind regards,
Dagmar Gromann

[1] Nikitina, N., Rudolph, S., Blohm, S.: Refining Ontologies by Pattern-Based Completion. In: Blomquivst, E., Sandkuhl, K., Scharffe, F., Svateck, V. (eds): Proceedings of the Workshop on Ontology Patterns (WOP 2009)


******************************************************************
Mag. Dagmar Gromann
Universitätsassistentin prae doc
Research Assistant

Institut für Englische Wirtschaftskommunikation
Institute for English Business Communication

WU
Wirtschaftsuniversität Wien
Vienna University of Economics and Business
UZA 4 - Nordbergstraße 15, 5. Stock, A5.10
1090 Wien, AUSTRIA
Tel. + (43-1-)31336-4288<tel:%2B%20%2843-1-%2931336-4288>
Fax. + (43-1-)31336-747<tel:%2B%20%2843-1-%2931336-747>
E-Mail: dagmar.gromann@wu.ac.at<http://service.gmx.net/de/cgi/g.fcgi/mail/new?CUSTOMERNO=2968880&t=de286486608.1280834132.37638404&to=barbara.frodl%40wu.ac.at>
http://www.wu-wien.ac.at/ebc/<http://service.gmx.net/de/cgi/derefer?TYPE=3&DEST=http%3A%2F%2Fwww.wu-wien.ac.at%2Febc%2F>