Classification systems are everywhere in everyday life. At a young age, most people were introduced to their first one in grade school. To spark a memory, think of the classification systems such as the Linnaean Taxonomy or the Dewey Decimal System. Linnaean Taxonomy created organisms into species and kingdoms while the Dewey Decimal System categorizes books within libraries by assigning each with a series of numbers. By referencing these two classification structures can help highlight the similarities between the current standard parts classification featured in CAD systems today.
The basis of parts classification is to start with a relatively small number of main categories. From there, parts are then divided into basic groups and then continually subdivided into smaller more specific groups with repetition. Even though there is not a streamlined way to classify parts, here are some best practices for it.
Use a classification structure to group similar parts -- To cleanse and normalize data for similar parts, they need to share the same naming conventions, attribute profiles, and attribute value lists. A classification structure can be used to address this issue. Normalizing is the process of making similar items look similar. It is best to not cleanse similar parts independently of each other. In turn, this means they should be classified together under a single data model or classification structure. The classification structure will help enforce that all similar parts will be described in the same way. Not only will parts share the same attributes, but the values for each attribute will also have the same format. For example, material attributes will have the same list of materials for each part. The descriptions will have the same format.
Avoid getting flustered on creating the classification structure -- To limit the debates on what your classification structure should resemble, limit the number of people involved in making those decisions. For commonly purchased parts, try to use industry-standard structures use by consortiums like ECCMA, eClass, PIDX or reference popular online catalogs and mimic their structures. At this point, it is not worth the time recreating the wheel. The most important thing is not where a category belongs in the structure but the attributes that define that structure, as well. Most searchers tend to be keyword searches and not based on classification structure.
Identify the most critical part/raw materials -- It is very easy to get carried away and want to have an attribute for every characteristic of a part. The more attributes that are used will then increase the data cleansing effort. At this point, it is important to determine which attributes are the most critical. Once realizing which is the most critical, it is important to shrink the list or make the less important attributes optional. To determine the most critical attributes, simulate a search scenario – what is the minimal information you need to find a critical part or raw material. When cross-referencing a drawing, users can always look at that drawing to find out more detailed information on that part. Determine the most critical parts/materials first - begin with the end in mind -- To obtain the most value for your organization, identify the parts that are the most important. .
Those are typically parts or materials shared in every product across multiple business groups. They can also be the parts that represent the biggest spend. When starting with those parts, will allow the data cleansing initiative to have the biggest impact.
Leverage resources such as smart part numbers, description fields, commodity codes, suppliers, etc. -- There are many techniques, tools, and resources that one can leverage to help expedite data cleansing. Smart part numbers and commodity codes are some examples of description fields that can be used to help group similar parts and extract data. Reverse engineering a smart part number to obtain critical attribute data can also be a substantial help. Parse descriptions provide a jump start to data cleansing. As a result, leverage the services of data cleansing and content providers to obtain cleansed data quickly. It may be more economical and timelier to pay a third party versus using internal resources, especially if the data already exists e.g. standard purchased parts containing manufacturing part numbers