Application of baking knowledge in software systems

Abstract

In the baking industry the use of computer systems has become integral to the efficient operation of the plant and smooth running of the commercial aspects of the bakery. In the plant, computers have been used for the in-line control and reporting (for accountability) mechanisms built into equipment. The use of programmable logic cards (PLC) to ensure that the equipment performs as required is widespread. For example, PLCs are used in mixing machines to ensure that the ingredients are mixed at the correct speed for the requisite time period or, in the case of ovens, that each zone has the desired temperature and humidity profile. There is much software, for example spreadsheets, word processing and database packages, to assist the baker, large or small to perform the business function and to ensure the smooth running of materials requisition, stock control, accounting, production scheduling and communicating with the customer. An example of this type of software is SAP which integrates databases, applications, operating systems and hardware. There is also another area where computer technology can be used by the baker. This is for assistance with the technology of the product. Achieving quality products consistently, whether new or existing ones, must be the goal of any baker wanting to remain in or improve the business. Computer programs which advise on recipe formulation, on optimal equipment operation and assist in perfecting a faulty or substandard product can be useful tools to the bakery technologist in all aspects of product development, maintenance and production. They can help the baker to understand and apply the underlying technology of the products and their quality for existing or new products. They can assist bakers to meet their goals more effectively. However, such programs cannot be developed unless the knowledge exists about the technology of the product and is made available. Embedding knowledge into a computer program requires a different approach from that used for the development of spreadsheets and databases. The systems used are commonly called 'knowledge-based' or 'expert' systems. It is useful to give some definitions: An expert system is a computer program that seeks to model the expertise of a human expert within a specific domain; A knowledge-based system embodies heuristic knowledge (rules-of-thumb, best guess, etc.) captured from intelligent sources. For this chapter, I shall use the term 'knowledge-based systems' or 'KBS' to include both of these definitions as in reality the terms are used interchangeably. It is easier to consider how KBSs differ from data and information processing, using an example as follows. If a baker has a recipe with ingredients and quantities, then using a data processing program the cost of the recipe can be determined simply by multiplying the unit quantity cost by the quantity of an ingredient and then summing all the costs of all the individual ingredients. Each time a new recipe is input the calculations can be done and the route or set of logic/numeric equations executed by the program is the same. If the baker then follows a recipe and processing method, say for pan bread, and finds that the product is faulty or below specification, then in order to obtain some advice for corrective actions the computer program that the baker uses must contain some rules and facts about the ingredients, the processing and the combination and interactions between them. When another recipe, perhaps for rolls instead of pan bread, is input to the program, the path and sets of rules are different. The route through such a program can take many paths. Each path is determined from information given and decisions made by the user and the knowledge contained and defined within the system. An explanation of the reasons behind the conclusions reached by the program and corrective action for rectifying the problem can be given to the baker. In order to develop programs which can deal with the technology of the product, that knowledge must be available and be accessible. The integrity and validity of the knowledge must be sound. There are different sources from which the knowledge can be gathered. First, there is knowledge in the form of the written word together with the numerical data to support it. Second, there are the human sources-the experts. For example, the methods which are typically used to assess bread characters have been described in Chapter 1 but assessment of bread quality, whether subjective or objective, has no value unless the user has the ability to change bread character in a particular way. Traditionally this is a role for an 'expert'. Experts are hard to find and when they are found the very fact that they are recognized as experts means that they are busy people. Currently experts who thoroughly understand the technology of breadmaking and can apply their knowledge of breadmaking are in limited supply. Without them KBSs for the fermented products industry cannot be developed. The knowledge, once captured and structured into a KBS program, can be used by technologists with different levels of skills. Whether used by an expert or novice, the KBS never forgets or becomes tired, and the advice or information returned to the user is consistent. From that information, users can learn and expand their own knowledge of the baking technology concerned. The areas where KBSs have been most successful in the baking industry have been in advisory and fault diagnostic systems. The first important criterion is that the 'domains' or subject areas must be well defined before such systems are developed. Attempting to produce in one system a tool which covers the whole of a subject area is a mammoth task and will probably be doomed to failure. It is better to take a domain and split that domain, say bread technology, into manageable chunks. These chunks can be linked together to form the larger picture. The second important criterion when developing such systems is that the knowledge should exist and be sound. Where experts are used to provide the knowledge input and where there may be conflict between experts in the knowledge given, then care must be taken in sorting the 'fact' from the 'opinion'. In many instances there may be a case for performing further work to determine the facts to support the existing knowledge. With advances in software programming languages and in the platforms in which they operate, encoding the knowledge has become much easier. However, the structuring of the knowledge for such purposes is still as important as ever. The systemization of knowledge has to be done before encoding can start and can be used independently of the software if the funds for programming are not available. The user interface or dialogue is designed to make the querying of that knowledge easy and meaningful to the user. © 2007 Springer Science+Business Media, LLC.