The web aimed to enable universal contribution and access, using the Internet’s decentralized architecture as its foundation. However, the current web trends away from this ideal as data is being centralized in silos controlled by major companies like Facebook, Google, and Amazon. Decentralization efforts seek to return control to users and reverse this trend. These efforts commonly revolve around (personal) decentralized data vaults. However, current querying approaches cannot handle the scale of decentralization that will occur when using personal data vaults.
An effective solution to this issue could be Link Traversal-based Query Processing (LTQP). LTQP is an integrated querying approach that enables the query engine to execute queries with zero knowledge of the data and discover sources on the fly. These properties make LTQP a viable candidate to query the envisioned scale of decentralization on the web. However, as the engine does not know what data will be queried in advance, creating an optimized query plan before executing the query is challenging.
LTQP is currently employed for client-side querying, where one engine instance services a single client. Despite engines serving a single client, current engines do not utilize client-specific engine usage patterns to implement personalized query optimization algorithms. By personalizing query optimization, historical query usage patterns can be used to optimize future queries and act as the missing prior knowledge during query optimization.