Although RecipeCategory and IngredientCategory have very similar names and attributes, they are, in fact, two different types of entity, because each of them (a) has a specific business domain meaning, (b) holds distinct kinds of interrelationships and (c) entails a particular set of constraints.
If the intention is to implement a relational dabatase (RDB), it is quite helpful to perform an analyzis of the business domain of interest in terms of entity types (i.e., types or prototypes of entity occurrences), their attributes and concerning interrelationships before thinking in terms of tables (which should be a derivation of the aforementioned analysis). Proceeding in this way, it is easier to capture the meaning of said business domain with accuracy and then reflect it in an actual RDB structure.
#Business rules
Recipe and Recipe Category
Let us start working with two entity types: Recipe and RecipeCategory. In the scenario in question, it appears reasonable to affirm that:
A Recipe is classified by zero-one-or-many RecipeCategoriesA RecipeCategory classifies zero-one-or-many Recipes
Such situation indicates, yes, that Recipe and RecipeCategory are involved in a many-to-many (M:N) relationship, which implies the existance of an associative entity type, that I am going to call RecipeCategorization.
Ingredient and Ingredient Category
Then, let us deal with Ingredient and IngredientCategory. In this case, we can affirm that:
An Ingredient is grouped by zero-one-or-many IngredientCategoriesAn IngredientCategory groups zero-one-or-many Ingredients
This means, yes, that Recipe and RecipeCategory are connected via another M:N relationship, which entails the existance of another associative entity type, that I denominated IngredientCategorization.
Recipe Category and Ingredient Category
As discussed above, one can observe that the concrete occurrences of RecipeCategory are meant to be (directly) associated with the specific instances of Recipe, and not with the occurrences of Ingredient. In the same way, the concrete instances of IngredientCategory are meant to be (directly) connected with the specific occurrences of Ingredient, and not with the instances of Recipe. Therefore, RecipeCategory and IngredientCategory are different entity types, and demand their own rspective individual considerations.
Recipe and Ingredient
Finally, we can assume that:
A Recipe includes one-to-many IngredientsAn Ingredient is included in zero-one-or-many Recipes
Thus, there is another M:N relationship, this time between Recipe and Ingredient, which signifies the existence of other associative entity type, that I am going to entitle RecipeListing.
#Illustrative logical model
Then, from the aforementioned analyzis and consequent formulations, I created the IDEF1X† logical model shown in Figure 1:
As demonstrated, each entity type is depicted in its corresponding individual box, and is displayed directly related to (i) its own attributes, contained in the respective box, and to (ii) the entity types that apply, by way of the relationship lines.
Of course, there are other indirect relationships that should be derived via the direct connections established here.
#Logical and physical elements
Once we have analized and defined the pertinent things of significance, it is time to determine how to manage and implement them by means of relations (or tables, if created in a certain SQL database management system), which are composed of domains (or columns) and tuples (or rows).
As relations are abstract resources, Dr. E. F. Codd —the originator of the Relational Paradigm— envisioned the utility of representing them in tabular form, so that, e.g., the users and implementers of a RDB can approach them in a more familiar way. In this respect, even though a relational table has a concrete shape, it is still a logical element of a given database, and its components, e.g., columns, rows, constraint declarations are logical as well.
In this regard, it is very important and of vast pragmatical value to differentiate logical from physical elements. For example, in file systems, a physical record can be made up of zero, one or more fields. In the case of a RDB, the logical elements can be served by one or more physical units, e.g., indexes, records, pages, extents, etc.
Thus, in accordance with the points detailed above, a table —being a logical level component— does not have fields.
#Expository (derived) SQL-DDL structure
That being said, and based on the logical model previously presented, both RecipeCategory and IngredientCategory (and the rest of the identified entity types too) require an individual base table, as exemplified in the following DDL structure:
-- You should determine which are the most fitting
-- data types and sizes for all your table columns
-- depending on your business context characteristics.
-- Also, you should make accurate tests to define the most
-- convenient physical implementation settings; e.g.,
-- a good INDEXing strategy.
-- As one would expect, you are free to make use of
-- your preferred (or required) naming conventions.
CREATE TABLE RecipeCategory -- Lookup table.
(
RecipeCategoryCode CHAR(2) NOT NULL, -- This column can store, e.g.: ‘O’ for ‘Omnivorous’; ‘VT’ for ‘Vegetarian’; ‘VG’ for ‘Vegan’; etc.
Name CHAR(30) NOT NULL,
Description CHAR(60) NOT NULL,
Etcetera CHAR(30) NOT NULL,
CreatedDateTime DATETIME NOT NULL,
CONSTRAINT RecipeCategory_PK PRIMARY KEY (RecipeCategoryCode),
CONSTRAINT RecipeCategory_AK1 UNIQUE (Name), -- ALTERNATE KEY.
CONSTRAINT RecipeCategory_AK2 UNIQUE (Description) -- ALTERNATE KEY.
);
CREATE TABLE Recipe
(
RecipeNumber INT NOT NULL,
Name CHAR(30) NOT NULL,
Description CHAR(60) NOT NULL,
Etcetera CHAR(30) NOT NULL,
CreatedDateTime DATETIME NOT NULL,
CONSTRAINT Recipe_PK PRIMARY KEY (RecipeNumber),
CONSTRAINT Recipe_AK1 UNIQUE (Name), -- ALTERNATE KEY.
CONSTRAINT Recipe_AK2 UNIQUE (Description) -- ALTERNATE KEY.
);
CREATE TABLE RecipeCategorization -- Associative table.
(
RecipeNumber INT NOT NULL,
RecipeCategoryCode CHAR(2) NOT NULL, -- Retains meaningful and readable values.
Etcetera CHAR(30) NOT NULL,
ClassifiedDateTime DATETIME NOT NULL,
CONSTRAINT RecipeCategorization_PK PRIMARY KEY (RecipeNumber, RecipeCategoryCode), -- Composite PK.
CONSTRAINT RecipeCategorization_to_Recipe_FK FOREIGN KEY (RecipeNumber)
REFERENCES Recipe (RecipeNumber),
CONSTRAINT RecipeCategorization_to_RecipeCategory_FK FOREIGN KEY (RecipeCategoryCode)
REFERENCES RecipeCategory (RecipeCategoryCode)
);
CREATE TABLE IngredientCategory -- Lookup table.
(
IngredientCategoryNumber INT NOT NULL,
Name CHAR(30) NOT NULL,
Description CHAR(60) NOT NULL,
Etcetera CHAR(30) NOT NULL,
CreatedDateTime DATETIME NOT NULL,
CONSTRAINT IngredientCategory_PK PRIMARY KEY (IngredientCategoryNumber),
CONSTRAINT IngredientCategory_AK1 UNIQUE (Name), -- ALTERNATE KEY.
CONSTRAINT IngredientCategory_AK2 UNIQUE (Description) -- ALTERNATE KEY.
);
CREATE TABLE Ingredient
(
IngredientNumber INT NOT NULL,
Name CHAR(30) NOT NULL,
Description CHAR(60) NOT NULL,
Etcetera CHAR(30) NOT NULL,
CreatedDateTime DATETIME NOT NULL,
CONSTRAINT Ingredient_PK PRIMARY KEY (IngredientNumber),
CONSTRAINT Ingredient_AK1 UNIQUE (Name), -- ALTERNATE KEY.
CONSTRAINT Ingredient_AK2 UNIQUE (Description) -- ALTERNATE KEY.
);
CREATE TABLE IngredientCategorization -- Associative table.
(
IngredientNumber INT NOT NULL,
IngredientCategoryNumber INT NOT NULL,
Etcetera CHAR(30) NOT NULL,
GroupedDateTime DATETIME NOT NULL,
CONSTRAINT IngredientCategorization_PK PRIMARY KEY (IngredientNumber, IngredientCategoryNumber), -- Composite PK.
CONSTRAINT IngredientCategorization_to_Ingredient_FK FOREIGN KEY (IngredientNumber)
REFERENCES Ingredient (IngredientNumber),
CONSTRAINT IngredientCategorization_to_IngredientCategory_FK FOREIGN KEY (IngredientCategoryNumber)
REFERENCES IngredientCategory (IngredientCategoryNumber)
);
CREATE TABLE IngredientListing -- Associative table.
(
RecipeNumber INT NOT NULL,
IngredientNumber INT NOT NULL,
Etcetera CHAR(30) NOT NULL,
IncludedDateTime DATETIME NOT NULL,
CONSTRAINT IngredientListing_PK PRIMARY KEY (RecipeNumber, IngredientNumber), -- Composite PK.
CONSTRAINT IngredientListing_to_Recipe_FK FOREIGN KEY (RecipeNumber)
REFERENCES Recipe (RecipeNumber),
CONSTRAINT IngredientListing_to_Ingredient_FK FOREIGN KEY (IngredientNumber)
REFERENCES Ingredient (IngredientNumber)
);
--
--
With such structure, you prevent ambiguities and all their logical and pragmatical repercussions. You avoid mixing the representation of multiple entity types in a single table, and avoid mixing the meaning and intention of each of their attributes (via columns), which permits restricting more easily their corresponding domains of values and the subsequent references via FOREIGN KEY (FK) constraints.
The structure, hence, aids by itself substantially in reflecting the business context under consideration with precision, remaining consistent with the aspects delimited in the analyzis.
Practical considerations regarding the Recipe Category table
Since the RecipeCategory table (1) will fulfill a lookup role, and (2) it will store only a few rows, I consider that defining it with a PRIMARY KEY (PK) that is meaningful and, at the same time, physically light and narrow, i.e., of type CHAR(2) or maybe CHAR(3), would be of great help. So it might keep, e.g., the rows that follow:
+-——————————————————-+-——————————-+-—————————————————————————-+-————-+-————-+
| RecipeCategoryCode | Name | Description | Etc… | Cre… |
+-——————————————————-+-——————————-+-—————————————————————————-+-————-+-————-+
| O | Omnivorous | Category of recipes that… | … | … |
+--------------------+------------+---------------------------+------+------+
| VT | Vegetarian | Category of recipes that… | … | … |
+--------------------+------------+---------------------------+------+------+
| VG | Vegan | Category of recipes that… | … | … |
+--------------------+------------+---------------------------+------+------+
| F | Foo | Category of recipes that… | … | … |
+--------------------+------------+---------------------------+------+------+
| B | Bar | Category of recipes that… | … | … |
+--------------------+------------+---------------------------+------+------+
In this manner, when such PK migrates to the RecipeCategorization table as a FK constraint declaration, its values will retain their meaning and intention, making the result sets much more readable than, say, an INT value, which definitely can assist in the interpretation of the information retrieved via, e.g., derived tables (those obtained by virtue of a SELECT statement or via a VIEW definition).
All this remains in agreement with the spirit of the relational model seminal paper†† published in 1970 by Dr. Codd, where he included the following relevant note:
Naturally, as with any data put into and retrieved from a computer system, the user will normally make far more effective use of the data if he is aware of its meaning.
Relational databases and application programs
The extract of your question referred to bellow has to do, among other things, with the connection between (1) the database under discussion and (2) the application programs (apps) that will work with it:
I tested the ideas in the related Q & A Relational tables with same content, custom solution. It's easier to handle data on the frontend side and housekeeping, but data integrity could be an issue if the database grows.
It is opportune to point out that the data, in its essence and of its nature, is a highly valuable organizational asset; as a consequence, it should be administered as such. This fundamental resource tends to outlive apps, app development platforms and programming paradigms.
With that, a RDB should be an independent (self-protective, self-describing, etc.) software component that is capable of being accessed by multiple apps and —using OOP parlance— it should not be “coupled” with the code of any of these apps.
So, you should shape and implement the relevant processes and the user interface with the pertinent app development tools, and manage the data by means of the instruments provided by the relational theory, the modeling platforms and the SQL platform of choice. In this way, all the software components will work harmonically; they will be independent but, at the same time, well connected.
#Endnote and reference
† Integration Definition for Information Modeling (IDEF1X) is a highly recommendable data modeling technique that was established as a standard in december 1993 by the United States National Institute of Standards and Technology (NIST). It is solidly based on (a) some of the early relational model works authored by Dr. E. F. Codd; on (b) the Entity-Relationship view, developed by Dr. P. P. Chen; and also on (c) the Logical Database Design Technique, created by Robert G. Brown.
†† Codd, E. F. (June 1970). A Relational Model of Data for Large Shared Data Banks, Communications of the ACM, Volume 13 Issue 6 (pp. 377-387). New York, NY, USA.