Operadores binários para consulta de similaridade em banco de dados multimídia / Binary operators in multimidia data base
AUTOR(ES)
Enzo Seraphim
DATA DE PUBLICAÇÃO
2006
RESUMO
The present databases managers are not adequated for complex data manipulation and among them we point out the multimedia data that to speed up the query use the equality operation on the index structure. Equality operations are not much used in operations that involve complex data sence the presence of two elements extremely equal is rare. An operator class that better manipulates these data are the similarity operators. Selection operators by similarity are the range query and the nearest neighbors query. For example, the selection operator to the nearest neighbors answers the queries like: ?select five proteins more similar by the alignment of the Sparc protein (responsible for the skin cancer)?. There are many works developed to provide similarity by selection operators envvolving structures based in trees. However, few studies have been done involving the use of different operators selection, for example, the join. A join operator compares pairs of objects of the elements belonging to the domain of the data, whereas a selection operator receives a constant to make the comparison of the elements. We can have three similarity join operators: the range join operators, the nearest neighbor and the closest neighbors pair. For instance, a query using the range join answers these kind of queries: ?Select the proteins restrained in the Hepatitis B virus that differ up to two unities from the alignment of the protein found in the Heapatitis C virus?. This work presents a new metric access method with an extreme amount of data as well as implementations forms of the join in metric structures
ASSUNTO(S)
similarity infex metric structures indexação por similaridade estruturas métricas similarity join junção por similaridade
