Design and implementation of popular quality analysis system for dry red wines originating from Northwest China

Abstract

SummaryChinese wine, especially those from the Northwest region, has received worldwide attention and consumer recognition. However, mislabelling the year and origin of wine is a major fraud challenging the wine industry, such as the deception on year or origin. Inspired by the Appellation d'Origine Contrôlée (A.O.C) system, a feasible wine popular quality analysis system (WPQAS) has been established to supervise the selling. First, an investigation for all potential users was carried out to analyse their requirements for WPQAS. Then, the system framework was designed with the Browser/Server (B/S) 3‐tier architecture, including the data layer for saving data, the business logic layer for running the models, and the presentation layer for interacting with users. Meanwhile, a unified modelling language (UML) sequence diagram was used to monitor the dynamic behaviour of the WPQAS. The core for driving system operation was generated based on a series of machine learning algorithms, including K‐nearest neighbour (KNN) for detecting outlines, random forest (RF) for weighting the chemical indexes, and artificial neural network (ANN) for training the models. Finally, the WPQAS, implemented as a software, was applied to estimate the popular quality of 600 dry red wines from Northwest China. The mode of decoupling models from the system could ensure WPQAS simplicity and scalability, and make the models reusable. Thus, the present work proposes a lightweight system that can be easily popularised and applied in diverse scenarios, and would play a significant positive impact on the correct meeting the challenges associated with counterfeit wine.