Query Synthesis

Interactive Query Synthesis

Many novice data analysts have trouble understanding and crafting queries in database query languages like SQL. Our goal is to simplify querying for such analysts, by guiding them to the query that they have in mind without much effort.

In our ICDT’10 paper, given that an analyst has selected a subset of tuples of interest, we developed algorithms that automatically generate the most succinct and accurate query that captures that subset. We examined this problem as a tug-of-war between three forces:

  1. Query family (e.g., Conjunctive queries)

  2. Approximation (e.g., I'm fine with capturing 80% of the right tuples)

  3. Succintness (e.g., Give me a query that is the smallest) and tried to evaluate the complexity of solving the problem.

In our VLDB’11 paper, we designed algorithms for interacting with the analyst (by asking as few questions as possible) to identify the predicates of interest.

Relevant tech reports / publications:

  1. Human-assisted Graph Search: It's Okay to Ask Questions, pdf talk
    Aditya Parameswaran, Anish Das Sarma, Hector Garcia-Molina, Neoklis Polyzotis and Jennifer Widom
    37th International Conf. on Very Large Data Bases (VLDB), Seattle, USA, Sep 2011

  2. Synthesizing View Definitions from Data, pdf pptx
    Anish Das Sarma, Aditya Parameswaran, Hector Garcia-Molina and Jennifer Widom
    13th International Conf. on Database Theory (ICDT), Lausanne, Switzerland, Mar 2010