Sense Axis introduced by the multilingual notation ex- tension, because LMF assumes that lexicons modeled with the framework are essentially monolingual. Wordnet-LMF also fully makes use of the Sense Axis to interrelate lex- icons in different languages. It serves as a pivot to connect Senses/Synsets in different languages (hence, in dif- ferent lexicons) that share an equivalent meaning. In fact, a synset in WN-ja is associated with the corresponding synset in PWN via a Sense Axis node. Figure 25 overviews the core structure of LMF that is relevant for describing Wordnet-LMF. The top-most ele- ment is Lexical Resource accompanied by Global Information which essentially stores profile-like in- formation. A lexical resource consists of one or more Lexicons and zero or more Sense Axes. A lexicon consists of one or more Lexical Entries and zero or more Synsets. Observing Saussure’s tradition, a lexical entry consists of form (Lemma) and Sense. A Sense node functions as a pointer to the corresponding Synset node. An instance of Synset Relation represents a conceptual relationship between synsets. The core infor- mation structure plays a crucial role in designing the Web service API.
Since the primary functionality of the lexicon access, given a query, is to present a partial lexicon extracted from the target lexicon, REST’s resource-centric approach is highly applicable. That is:
Access queries are structured in the sense that they assume particular information structure in the target lexicon, thus can be mapped to URIs that reflect the information structure;
The resulted partial lexicon should be represented in a pre-shared standard data format, providing an oppor- tunity to apply LR-related international standards.
It is LMF that not only provides the standardized data for- mat for representing access results but also presents a solid structural specification for the target lexicon type.
4.2. Access URI Pattern and Examples
We designed the URI pattern for invoking the Web ser- vice primarily based on the lexicon structure specified by Wordnet-LMF, and secondarily by considering possible us- age patterns. Figure 3 shows the basic structure of the pro- posed URI pattern, which consists of the core part and the optional part.
path in the URI
pwn, wn-ja, edr, ...
Figure 3: Structure of the proposed URI pattern.
The organization of the core part is summarized as shown in Table 1 and Table 2. The [qterm] part designates a query term string.
Figure 2: LMF core structure relevant to Wordnet-LMF.
Table 1: Organization of the core part.
[target] part word
[search by] part
form pos form-pos
writtenForm partOfSpeech writtenForm, partOfSpeech id <Definition> <Statement>
id definition statement
LMF-based Web Service for WordNet-type Semantic Lexicons
REST (Representational State Transfer) (Richardson and Ruby, 2007) is a style of software architecture to realize Web services. With this architecture, inputs (requests) and outputs (responses) of a Web service are built around the transfer of representations of resources. Here, a resource is coherent and meaningful object that should be designated by URI (Uniform Resource Identifier); the resources should be represented in the format that is shared by the possible clients in advance.
5A diamond shape in the diagram indicates ”aggregation” as in the formal UML class diagram. Cardinalities are however omit- ted.
Table 2: Organization of the search by part.
Notice here that according to the partial
Figure 4 shows the result
with the showing