Exploring Query Results

by   Murali Mani, et al.

Users typically interact with a database by asking queries and examining results provided by the database system. We refer to the user examining the results of a query and asking follow-up questions as query result exploration. Our work builds on two decades of database community research on provenance. One of the contributions of this paper is that we identify constraints in the context of query result exploration that have previously been unexplored in provenance research. These constraints are useful in optimizing performance and thus improving the user experience during query result exploration. Further, three approaches for computing provenance have been described in the literature: (a) in the lazy approach, no additional data is materialized and provenance is computed from the base tables as needed; (b) in the eager approach, additional data is materialized and provenance is computed from this additional materialized data; (c) in the hybrid approach, some additional data is materialized, and provenance is computed using this additional materialized data as well as the base tables. In this paper, we investigate lazy and eager approaches that utilize constraints in the context of query result exploration, as well as novel hybrid approaches, where the keys for some of the base tables are materialized as selected by a cost model. For the TPC-H benchmark, we find that the constraints that we identify are applicable to 19 out of the 22 queries, and result in a better performance of the lazy approach for 7 out of these 19 queries. With materialization, our hybrid approach resulted in a better performance for all but one single table TPC-H query with no joins (where existing approach performed as good as our approach). Further, the performance benefit from our approaches are significant, sometimes several orders of magnitude, compared to previous research.


page 1

page 2

page 3

page 4


Query Significance in Databases via Randomizations

Many sorts of structured data are commonly stored in a multi-relational ...

NeuroCard: One Cardinality Estimator for All Tables

Query optimizers rely on accurate cardinality estimates to produce good ...

IDEBench: A Benchmark for Interactive Data Exploration

Existing benchmarks for analytical database systems such as TPC-DS and T...

Cognitive Visual-learning Environment for PostgreSQL

PostgreSQL is an object-relational database (ORDBMS) that was introduced...

Towards Deterministic Decomposable Circuits for Safe Queries

There exist two approaches for exact probabilistic inference of UCQs on ...

Using clarification questions to improve software developers' Web search

Context: Recent research indicates that Web queries written by software ...

Prioritizing Technical Debt in Database Normalization Using Portfolio Theory and Data Quality Metrics

Database normalization is the one of main principles for designing relat...

Please sign up or login with your details

Forgot password? Click here to reset