Building Simulation Software tools support designers to analyse and identify certain users´ behavioural patterns; besides, they can predict future trends about the energy demand and consumption in buildings, as well as CO₂ emissions, design analysis, energy efficiency, or lighting. These tools allow to collect and report information about such processes. However, understanding the results from simulations usually implies interpreting an extremely large amount of data or graphs, which can be a complex task. Therefore, there is a need of alternatives that ease this interpretation of results, hence complementing classic simulation tools. Under the widespread EN 15251 model criteria, this paper presents a novel technology to improve reporting tools of building simulation software by using linguistic description of data and timespan computational perceptions. A data-driven software architecture for automatically generating linguistic reports is here proposed, which provides designers with a better understanding of the data from building simulation tools. In order to show and explore the possibilities of this technology, a software application has been designed, implemented and evaluated by experts. The survey showed that usefulness and clarification were better evaluated than simplicity and time-saving for the three kinds of report, though always above 7 points out of 10, being most of p-values of contingency below 0.05.