IDChase: Mitigating Identifier Migration Trap in Biological Databases

Improving Integration Effectiveness of ID Mapping Based Biological Record Linkage

“…The maintenance cost of such materialized mappings is significant. In [11], we have presented the complexities associated with ID mapping changes and migration. It was observed that without a very close collaboration with the source databases, it is very difficult to maintain such mappings.…”

“…In Section 4.4.1, we will discuss how provenance computation depends on the change logs and the query being processed. Change logs, on the other hand, are polled from a set of databases identified as authorities, and organized in a priority order using a scheme similar to IDChase [11]. This warehouse maintenance process ensures that no invalid IDs are ever mapped to ensure accuracy of mappings returned as a query response.…”

“…Research in change management has mainly focused on shallow web [14], [15], [16], [17] and general view maintenance [2], [18], and is not directly applicable to ID maintenance. In [11], we pursued a more active approach that blended continual querying and view maintenance to discover such migrations with an aim to minimally and incrementally maintain materialized views of ID maps as one of our ongoing efforts.…”

“…We identify two principal factors that impact the quality of integrationmaterialized view maintenance and access methods to mapping for information aggregation. We have addressed the issue of materialized view maintenance in an earlier paper [11] as part of an ongoing research, and hence, though tangentially relevant, it will not be discussed here. The main focus is the access method to ID maps and developing a declarative approach to record linkage for data integration.…”

“…These warehouses often materialize relationships among objects represented using such IDs, which are either gathered from other databases or computed using ID mapping tools such as DAVID [3], OntoTranslate [4], Synergizer [5], GeneID Converter [6], IdBean [7], IDMapper [8], g:Profiler [9] and MatchMiner [10]. The materialized views must then be maintained by monitoring possible ID changes, migration, and relationship changes among the objects, occurring in databases they do not control, or risk producing inconsistent and obsolete results, perhaps even facing loss of functionality [11]. These seemingly lower level ID management issues contribute to a much larger and even more important information management problem in life sciences, called ID mapping and object linkage.…”