3.2 Philosophical Issues

In order to make good metadata to create the “SFU Online Data Dictionary,” it is very important to understand semantics and ontologies.  Solid understanding and proper application of these fields of thought make the composition of good metadata possible. Also, understanding Semantic Query Language and Wordnet is necessary, since these search engines introduce reducing the semantic differences and ontological problems.

Semantics

The term semantics can be defined as words which has “different meaning for different purposes for different people.” In creating definitions for elements in metadata, the semantic difference can often create problems since the definitions are often very short and simple words that could be interpreted variably depending on prospective users’ own experience and knowledge. This leads us to the problem of query metadata with regards to the semantics of the search parameter.  In querying many data sources, how do you know that your search parameter has been interpreted the same way?  Handled with the same meaning: same context?  This problem associated with semantic differences can be solved when ontology is properly used.  In our project, we have addressed this issue using XML tags (refer to technical issue).  Semantic interoperability has been identified as a key issue concerning geographic data sharing between different geo-spatial information communities

Ontology

Ontology is a process of “achieving a clear and concise description of terms and concepts.” Also, ontologies can be characterized as shared vocabularies or conceptualizations of a specific subject matter. Therefore, production of ontology for terms with semantic difference can help reduce and reduce the problems with interpretation.  By properly using ontology, it will promote integration of data from different sources into a single system and improvement of “access to and sharing of existing geographical information resources.” It provides a logical definition of concepts and their properties and argues for the benefit of this approach in typical application scenarios.  A means to improve access to and sharing of existing geographical information resources is by standardizing a set of ontologies which we see fit to describe the semantics of a dataset.  In our project, we need to address ontologies to improve the metadata search engine by using XML tags.  Because our group is responsible for determining the search keywords, we must identify a set of ontologies that best suits the dataset in order to achieve an efficient metadata search engine.

Semantic Query Language and Wordnet

SemQL is a semantic version of Standard Query Language (SQL).  It is similar to SQL, except that it has no FROM clause, and it is being developed to search multi-database systems, such that it considers the context of the query parameters. For such a query to be successful, it must know two important things: where to find relevant information on the component databases, and which entities, elements and attributes within the component databases meet the semantic requirements of the search. With the use of Wordnet, semantic networks can be established for use in the search, also known as semantic heterogeneity classification. Wordnet and semantic heterogeneity are important factors to examine when developing search engines of this nature.  “Because meaningful sentences are composed of meaningful words, any system that hopes to process natural languages as people do must have information about words and their meanings” Wordnet searches for many semantic relationships in determining the meaning of a group of words.  Such factors include synonymy (comparision for synonyms), antonymy (opposites), troponomy (manner), and several others.

Wordnet is a very powerful tool in finding semantic relationships within groups of words, but the validity of the Wordnet results may be of concern.  Questions arise out of how Wordnet deals with differences in cultural expression.  Consider the language ontologies that exist within the English language alone.  For example, in England, large trucks are not known as trucks; rather, they are known as “lorries.”  Many other culture-language discrepancies exist that could result in Wordnet producing output that may not be useful by the search engine, thus lowering its reliability. More research needs to be undertaken on Wordnet, especially on how the inner working of its code actually performs what it does.

The process resulting from the conflation of SemQL and semantic heterogeneity is known as “semantic query processing,” and has proven to be quite successful in finding specific entries in groups of small and moderately sized databases.  However, It is unclear whether or not this search process would work in a very large environment, especially where the key fallacy is the construction of semantic heterogeneity.  Semantic relationships would be much harder to form, thus would probably not be formed correctly at all in some cases.  SemQL is a very intriguing language, and it deserves more attention from the geo-spatial community as a possible replacement for SQL as the query language of choice for search engines dealing with geo-spatial data.