Interrogation de données structurellement hétérogènes dans les bases de données orientées documents
In EGC 2018, vol. RNTI-E-34, pp.155-166
Documents oriented DB can store any document, whatever its data schema. This facility generates potential structural heterogeneity of documents which makes it difficult to query data because the same entity can be described according to different schemas. This paper presents an approach of transparent querying of documents oriented DB with structural heterogeneity. For that purpose, we suggest building a dictionary which associates every field with all partial schemas (paths) of the DB allowing to reach it. This dictionary is used to rewrite the user query in disjunctions of partial schemas in order to find all the documents, whatever their real schema. Our experiments show acceptable costs of execution of the rewritten query when compared to the cost of equivalent query on homogeneous schemas.