Design and Implementation of Queries for Model-Driven Spreadsheets

Abstract

This paper presents a domain-specific querying language for model-driven spreadsheets. We briefly show the design of the language and present in detail its implementation, from the denormalization of data and translation of our user-friendly query language to a more efficient query, to the execution of the query using Google. To validate our work, we executed an empirical study, comparing QuerySheet with an alternative spreadsheet querying tool, which produced positive results.

Embedding model-driven spreadsheet queries in spreadsheet systems

“…Although constructing queries on a model-driven spreadsheet environment seems to improve efficiency and effectiveness, our empirical observations also suggest that building SQL-like queries may still be too complex for end users. In fact, our studies showed that advanced spreadsheet users with basic SQL experience still felt it difficult to write spreadsheet queries [16]. We observed syntactic and semantic difficulties: first, spreadsheet users were not comfortable to express queries on a textual syntactic SQL-like notation that is very different from the spreadsheet programming paradigm.…”

“…The SQL-like model-driven query is more descriptive than some other approaches, taking advantage of the attribute names used in the model [15], [16]. By doing so, the query itself becomes more human-friendly and understandable to both read and write.…”

“…In the past, we have proposed the first model-driven approach to spreadsheet querying [15], [16]. This initial proposal includes a language and a tool to construct queries over spreadsheets with models.…”

“…In previous work we proposed QuerySheet: a textual SQLlike query language and system for model-driven spreadsheets [15]. An empirical study showed that QuerySheet greatly improves the productivity of end users when compared to the use of another querying system [16]. Regardless, we also observed that although our technique facilitates querying spreadsheets for users with experience in SQL, those with less experience still felt it difficult to write queries (this was observed in both our system and the competing one), showing that traditional query systems are too complex for end users.…”

Model-driven Query Generation for Elasticsearch

“…The framework is enriched with a model-driven query language [20] which supports most of the SQL standards. The language is also structured as a DSL and the related MDE framework is integrated with Google Query function in [26]. Similarly, FDL [27] is a description language for spreadsheets, which is empowered with visualization and analysis tool for constructing the separation between the input of formulas and the output of calculation results.…”