database - Modeling complex hierarchies -

To get some experience, I am trying to create an expert system that answers questions about the animal kingdom can give. However, I have gone into a problem model domain. I originally considered a hierarchy of animals such as

  - the only -bird-karnavar-hawk-hierbivore-bozai-mammals-cranavara-herbivores  

< P> I thought that would allow me to easily create questions like "give me all the birds", but it would be very expensive to say "give me all non-vegetarian", so I can see the hierarchy again Written:

  -command -Carnivore -Bird-Hawk-Mammals -Assign -Hiribers-Bird-Blaz -Salalfpers  

But now I will be very slow to "give to all the birds".

This is a simple example, but I got it when I was thinking that I really do not know how to model complex relationships, in terms of answering a specialist system, so strictly in nature As per the above given questions, a guided, cyclical graph seems like it can solve a mathematical problem, but it is stored in a relational database and it is Maintaining (updates) seems like a nightmare to me. I would like to know how people generally model such things, for further reading, explanations or points for resources will be acceptable and appreciated. You can deal with taxonomies on one of the problems (only one, or even worst one, in fact). Multiple Inheritance avoids many problems with taxonomies as a conceptual tool - another way to put it, a classification defines a tree, the Mi-based classification plan is more generalized. Esaikik defines the graph, and therefore an additional degree of freedom is provided in your modeling.

The perspective of a relational database will be different (especially not thinking about hierarchy or inheritance) but very conceptual results come in the form of "multi-part": "class" (meaning lenius / Square / order / family / genus / species) is an area of ​​record, diet (carnivorous, herbivor, ubiquitous) a different one - they do not interrupt each other, neither the concept nor the searches / retrieval In.

If you are forced to model with tools, which restrict you to taxonomies (aka tree, single succession, & amp; nbsp;), then due to their pain Some tricks (up to a modest degree), but they depend on the specific restrictions of each device, so it is difficult to 'generalize'.