Press releases      MyWineStein       Dutch

Technology behind WineStein

WineStein's artificially intelligent software uses all ingredients entered to calculate a taste profile of a dish. Subsequently, this profile is used to calculate the best matching wine types. WineStein calculates a point score per wine type, and adds an indication of wether the wine type is in balance with the weight and complexity of the dish. WineStein uses some 700 wine types, thousands of ingredients and was trained with over 50,000 reference data. WineStein's software is also self learning and will provide a dynamic, well-balanced and varied wine & food pairing advice. The software is also able to advise best matching dishes with a wine of choice.

Self learning software

Winestein's concept was developed by advising wine community winewinewine.com. Based on this concept, its self learning algorithms and matching software were subsequently developed in cooperation with Smart Research a business enterprise based on Nijmegen University campus, The Netherlands, that specialises in neural networks and self learning techniques. All information and every advice provided by WineStein is being made available through an Application Interface (API). Linking your data with WineStein's database is done by means of an (XML) productfeed, that exchanges all relevant information regarding wines offered by a WineStein-client on a daily basis.

Dynamic advice

WineStein's pairing advice is based on wine type. A specific wine type is a gastronomic ideal-type of a certain wine, defined in terms of origin, grape varieties used, quality and taste profile. Wine types are commonly subdivided, taking differences into account that can be related to quality (e.g. Beaujolais and Beaujolais Crus), use or non-use of wood-ageing and stage of development (young, ready, aged). WineStein's advice is fully dynamic, made possible by the use of a special calculation model that will convert any combination of ingredients and preparations used into a single set of gastronomically relevant aspects.