Category Intersection is the point at which two or more categories meet or join, especially in regards to profiles.

Current Operations

Below is a fictional example to explain the concept, highlighting key words that will be resused:

Let's say we have the profile of Paulette_Bernège (1896-1973) who was a French journalist, publicist and author^[1]. Through our current system of categorization, it is generally accepted that if a profile could fit under the topics of "category A" and "category B", that "category A and B" could (or should) be created and used.

This can be further complicated when multiple topics apply. If a profile could fit under the topics of “category A”, “category B”, and “category C”, then this could lead to four categories "category A and B", "category B and C", "category A and C" as well as "category A, B and C".

Since we do not have any reliable methods to search or automatically intersect categories, this level of categorization works and is what would be considered a manual implementation of Category Intersection. However, this method of categorization eventually leads to a massive amount of individual (and narrow) categories, and can also segregate profiles into silos (placed in one category, but not another). As of this writing, there are currently 254,458 categories on WikiTree!

Current Pros and Cons

1. Profiles are listed in narrow categories that provide detailed/specific information (such as a specific City/Town) and are not grouped together at larger non-specific levels such as Country. This is useful for those who are experienced with a specific Country's Administrative Structure.

1. Profiles are only listed in narrow categories that provide detailed/specific information (such as a specific City/Town) and do not include high level information such as Country. This can be a struggle for users who are not experienced with a specific Country's Administrative Structure.
2. The number of categories needed in this method of categorization are exponential in order to accomodate the endless amount of intersections that can be (or will be) needed in the future.
   • France Authors
   • England Authors
   • Italy Authors
   • ... etc ...
3. This method is only useful when all profiles are consistently being placed in their applicable categories.

New Concept

By installing a MediaWiki Extension^[2] we could open up endless possibilities for WikiTree and how we group, search and categorize profiles:

“The Multi-Category Search extension allows users to find pages that are included in several specified categories at once”

Using the same example of Paulette Bernège, we could now categorize her by the high-level "traits and features" from her profile:

|Category: Authors||(occupation) |- |Category: France||(location) |- |Category: 20th Century||(time period based on when her work occurred) |}

Using this new MediaWiki Extension, we could now look up category intersections automatically through a new Special:MultiCategorySearch page. Using the image below to help illustrate, we could perform the following advanced searches:

Given:

Possible Results:

According to our general rules, “If a category could contain millions of people, we should create a narrower subcategory”. This means that for this new extension to be useful, we would need to change some of the fundamental guidelines of categories moving forward, without affecting the current benefits. By retiring templates (and the associated functions/suggestions) like identifying "Top Level" categories where we do not want profiles to reside.

To expand further, one of the major benefits of the current structure is that profiles are listed in narrow categories that provide detailed information. As an example, right now we would not categorize a profile to Category:France, but to a narrower category such as Category: Betschdorf, Bas-Rhin. By retiring some guidelines and templates and their associated functions/erros/suggestions (i.e., identifying "Top Level" categories where we do not want profiles to reside because they "could contain millions of profiles") we could now list a profile in all applicable categories, such as Category: Betschdorf, Bas-Rhin, Category: Bas-Rhin, France, and Category:France in order to use the features of automatic category intersection.

Now given:

Possible Results:

Using transclusion (the inclusion of part or all of an electronic document into one or more other documents), the multi-category search results could also be dynamically shown on a page by providing a simple statement: {{Special:MultiCategorySearch/include=Authors/include=20th Century/include=France}}

Click the link for a working Implementation of Extension:Multi-CategorySearch. On the page, you will see a search form where you can either select categories using a dropdown, or manually type in category names. Fill out the manual search form as shown below (using Peace + 1963) and click "Search" to return all pages categorized to "Category:Peace" and "Category:1963":

Other example search options:

• Christmas + Australia
• Flowers + 1983
• Dogs + United States of America

Concept Pros and Cons

1. Manually intersected categories are no longer needed.
2. Provides unlimited ways to search for profiles, based on their defining characteristics (as identified by the included categories).

1. Retiring of a fundamental guideline for many category structures (but not all - see Note). This guideline effectively manages the placement of profiles in large categories such as Category:France.
   • If a category could contain millions of people create a narrower subcategory. Ideally, bottom-level categories should be much smaller, with no more than a few thousand people in them. This isn't always possible, e.g. with towns and cities, but we should aim to create narrow categories when we can.
   • Note: This guideline could not be completely retired since this would allow levels of categorization that do not make sense (i.e., to Category:Categories.
2. Introduces a "new way of thinking" for categorization, and how profiles can be categorized.
3. Would exponentially expand the number of categories that can be (should be) applied to profiles.
4. This method is only useful when all profiles are consistently being placed in their applicable categories.