I am importing and linking datasets from multiple data sources into a single graph model. I might have one dataset for customer orders and a different dataset for customer issues coming from different systems. Each of those systems may have a different name or ID for the customer so we need to do some master data mapping between the different IDs.
I want to keep both representations of the customer in the model for lineage reasons and to allow querying by either name while returning both orders and complaints.
Any recommendations on how to model this? I can start with the basic facts:
merge(c1:Customer1 {ID:'abc'})-[:HAS_ORDER]-(o1:Order {ID:'123'})
merge(c1)-[:HAS_ORDER]-(o2:Order {ID:'234'})
merge(c2:Customer2 {ID:'xyz'})-[:HAS_ISSUE]-(i1:Issue {ID:'678'})
merge(c2)-[:HAS_ISSUE]-(i2:Issue {ID:'789'})
merge (c1)-[:SAME_AS]-(c2)
I could just handle it in my queries, like this:
match(c1:Customer1 {ID:'abc'})-[r1:SAME_AS*0..1]-(c2)-[r2]-(n1)
return c1.ID as source,type(r2) as relationship, n1.ID as target
which correctly returns
source relationship target
"abc" "SAME_AS" "xyz"
"abc" "HAS_ORDER" "234"
"abc" "HAS_ORDER" "123"
"abc" "HAS_ISSUE" "678"
"abc" "HAS_ISSUE" "789"
The problem is it assumes the sameas relationship is always with the starting node. If I want to find issues by the same customer as orders I would need a different query . I'd rather not have to explicitly think about same as in my queries (and force other people to).
A different approach would be to create new relationships between each customer and all the objects related to the paired customer...basically if customer abc is same as customer xyz which has issue 789 then customer abc also has issue 789 so create that relationship. I'd need to mark it as a derived relationship so I can keep track of what the base facts are.
Are there other approaches I should consider?
Related
I'm creating a relational database of a store and its stock of products.
In the brief, it says "products can be returned under agreed terms e.g. expiry date or manufacturers error", based on this I created a weak entity "Terms" with product_ID as the foreign key and errors & expiry as two attributes.
My logic was that the terms only exist if the product exists, therefore it is a weak attribute as every product has terms, but you wouldn't have terms not associated with a product.
Looking at it though, the "Terms" table would basically be Product ID (1) ---> Errors (No) ---> Expiry (01/01/23), and now I'm starting to think those two attributes should be attributes of the product table and not a separate entity, mainly because "Terms" doesn't have a partial/discriminator key that could be used as a composite primary.
Does anyone have any thoughts about which way is correct?
I think this answer really comes down to the trade-offs in terms of performance.
To make sure I understand your question correctly - you basically have two tables:
The main product table
A "lookup" table that just has Product_ID (FK), Errors, and Expiry as the columns
If this is the case, you have two options:
Just add Errors and Expiry as columns to the primary product table
Keep the two tables separated as you have them, and just JOIN that data when needed.
Option 1 has the benefit of keeping all the data in one table, assuming that "Expiry" and "Errors" are unique to the product_ID; if they're not, you may end up duplicating data, and it's better to keep these fields in your separate table to have a 1:Many relationship. The other drawback would be that if your main Product table is beefy, you've slowed down the query even further by adding these columns.
Option 2 can circumvent the two shortcomings of Option 1 - by keeping this data separate, your Product table is much lighter, and if you have a 1:many relationship, you don't duplicate data (saving you more memory overall!). The drawback with Option 2 is that your EDR gets a bit more complicated - you have one more table to keep track of.
Based on these, I recommend keeping your separate "lookup" table - the benefits of separating this data out will help you in the long run - but ultimately you'll need to weight the pros and cons since I don't know the extent of your project.
[Assuming there is a one to many relationship between an individual and an address, and assuming there is a one to many relationship between an agency and an address.]
Given the following table structure:
Wouldn't you want to merge the two address tables together and instead of using a foreign key within each one use a tie table?
Like this:
Are they both valid for normalization or only one?
Depends what you want to do.
In your second example with the tie tables, if I want to do a mailshot to my customers then my query has to go out to the agency tie table to exclude any agency addresses.
Of course you could have an address type column to differentiate but then you have a more complex query for your insert statement.
So although "address" is a global idea, sometimes it is easier to have it segregated by context.
Secondly, your customer data would usually be changing much more than your agency data. There may also be organisational and legal requirements around storage of personal data that make it better to separate the two.
e.g. in a health records system I want to be able to easily extract / restrict client data and to keep my configuration or commissioning data separate.
Thus in all the client systems I have used, the model tends to be the first one you describe rather than the second.
So I have 3 main entities. Airports, Customers, and Vendor.
Each of these will have multiple contacts I need to relate to each.
So they way I set it up currently.
I have the following tables..
Airport
Customer
Vendor
I then have one Contacts table and a xref for Airport, Customer, Vendor...
I am questioning that and was thinking a contacts table for each ..
Airport and AirportContacts
Customer and CustomerContacts
Vendor and VendorContacts
Any drawbacks to either of these designs?
To me, the deciding factor is duplication of entities vs "one version of the truth". If a single real-world person can be a contact for more than one of the other entities, then you don't want to store that single person in multiple contact tables, because then you have to maintain any changes to his/her properties in multiple places.
If you put the same "Joe Smith" in both AirportContacts and VendorContacts, then one day when you look and see his city is "Denver" in one table and "Boston" in another table, which one will you consider to be the truth?
But as someone mentioned in comments, if a contact can only be associated with one of the three other entities ("types" as you call them), then putting them in separate tables makes the most sense.
And there's yet a third scenario. Say "Joe Smith" can be a contact for both Airports and Vendors. But say that he has some properties, like his gender and age, which are the same regardless of which "type" he is being considered, but there might be some properties, like phone number, or position/job title, which could depend on the "type". Maybe he uses one phone in his capacity as an Airport Vendor, and a different phone as a Vendor Contact. Moreover, maybe there are some properties that apply to one type of contact that don't apply to the others. In these cases, I would look at some kind of hybrid approach where you keep common properties in a single Contact table, and "Type"-specific properties in their own type-related tables. These type-related tables would be bridge tables that have FKs back to the Contact table and to the main entity table of the "Type" they are related to (Vendor, Customer or Airport).
What I have so far ... Dont mind some of the data types.. just placed quick placeholders..
So I know the convention for naming M-M relationship tables in SQL is to have something like so:
For tables User and Data the relationship table would be called
UserData
User_Data
or something similar (from here)
What happens then if you need to have multiple relationships between User and Data, representing each in its own table? I have a site I'm working on where I have two primary items and multiple independent M-M relationships between them. I know I could just use a single relationship table and have a field which determines the relationship type, but I'm not sure whether this is a good solution. Assuming I don't go that route, what naming convention should I follow to work around my original problem?
To make it more clear, say my site is an auction site (it isn't but the principle is similar). I have registered users and I have items, a user does not have to be registered to post an item but they do need to be to do anything else. I have table User which has info on registered users and Items which has info on posted items. Now a user can bid on an item, but they can also report a item (spam, etc.), both of these are M-M relationships. All that happens when either event occurs is that an email is generated, in my scenario I have no reason to keep track of the actual "report" or "bid" other than to know who bid/reported on what.
I think you should name tables after their function. Lets say we have Cars and People tables. Car has owners and car has assigned drivers. Driver can have more than one car. One of the tables you could call CarsDrivers, second CarsOwners.
EDIT
In your situation I think you should have two tables: AuctionsBids and AuctionsReports. I believe that report requires additional dictinary (spam, illegal item,...) and bid requires other parameters like price, bid date. So having two tables is justified. You will propably be more often accessing bids than reports. Sending email will be slightly more complicated then when this data is stored in one table, but it is not really a big problem.
I don't really see this as a true M-M mapping table. Those usually are JUST a mapping. From your example most of these will have additional information as well. For example, a table of bids, which would have a User and an Item, will probably have info on what the bid was, when it was placed, etc. I would call this table... wait for it... Bids.
For reporting items you might want what was offensive about it, when it was placed, etc. Call this table OffenseReports or something.
You can name tables whatever you want. I would just name them something that makes sense. I think the convention of naming them Table1Table2 is just because sometimes the relationships don't make alot of sense to an outside observer.
There's no official or unofficial convention on relations or tables names. You can name them as you want, the way you like.
If you have multiple user_data relationships with the same keys that makes absolutely no sense. If you have different keys, name the relation in a descriptive way like: stores_products_manufacturers or stores_products_paymentMethods
I think you're only confused because the join tables are currently simple. Once you add more information, I think it will be obvious that you should append a functional suffix. For example:
Table User
UserID
EmailAddress
Table Item
ItemID
ItemDescription
Table UserItem_SpamReport
UserID
ItemID
ReportDate
Table UserItem_Post
UserID -- can be (NULL, -1, '', ...)
ItemID
PostDate
Table UserItem_Bid
UserId
ItemId
BidDate
BidAmount
Then the relation will have a Role. For instance a stock has 2 companies associated: an issuer and a buyer. The relationship is defined by the role the parent and child play to each other.
You could either put each role in a separate table that you name with the role (IE Stock_Issuer, Stock_Buyer etc, both have a relationship one - many to company - stock)
The stock example is pretty fixed, so two tables would be fine. When there are multiple types of relations possible and you can't foresee them now, normalizing it into a relationtype column would seem the better option.
This also depends on the quality of the developers having to work with your model. The column approach is a bit more abstract... but if they don't get it maybe they'd better stay away from databases altogether..
Both will work fine I guess.
Good luck, GJ
GJ
Consider a database with tables Products and Employees. There is a new requirement to model current product managers, being the sole employee responsible for a product, noting that some products are simple or mature enough to require no product manager. That is, each product can have zero or one product manager.
Approach 1: alter table Product to add a new NULLable column product_manager_employee_ID so that a product with no product manager is modelled by the NULL value.
Approach 2: create a new table ProductManagers with non-NULLable columns product_ID and employee_ID, with a unique constraint on product_ID, so that a product with no product manager is modelled by the absence of a row in this table.
There are other approaches but these are the two I seem to encounter most often.
Assuming these are both legitimate design choices (as I'm inclined to believe) and merely represent differing styles, do they have names? I prefer approach 2 and find it hard to convey the difference in style to someone who prefers approach 1 without employing an actual example (as I have done here!) I'd would be nice if I could say, "I'm prefer the inclination-towards-6NF (or whatever) style myself."
Assuming one of these approaches is in fact an anti-pattern (as I merely suspect may be the case for approach 1 by modelling a relationship between two entities as an attribute of one of those entities) does this anti-pattern have a name?
Well the first is nothing more than a one-to-many relationship (one employee to many products). This is sometimes referred to as a O:M relationship (zero to many) because it's optional (not every product has a product manager). Also not every employee is a product manager so its optional on the other side too.
The second is a join table, usually used for a many-to-many relationship. But since one side is only one-to-one (each product is only in the table once) it's really just a convoluted one-to-many relationship.
Personally I prefer the first one but neither is wrong (or bad).
The second would be used for two reasons that come to mind.
You envision the possibility that a product will have more than one manager; or
You want to track the history of who the product manager is for a product. You do this with, say a current_flag column set to 'Y' (or similar) where only one at a time can be current. This is actually a pretty common pattern in database-centric applications.
It looks to me like the two model different behaviour. In the first example, you can have one product manager per product and one employee can be product manager for more than one product (one to many). The second appears to allow for more than one product manager per product (many to many). This would suggest the two solutions are equally valid in different situations and which one you use would depend on the business rule.
There is a flaw in the first approach. Imagine for a second, that the business requirements have changed and now you need to be able to set 2 Product Manager to a product. What will you do? Add another column to the table Product? Yuck. This obviously violates 1NF then.
Another option the second approach gives is an ability to store some attributes for a certain Product Manager <-> Product relation. Like, if you have two Product Manager for a product, then you can set one of them as a primary...
Or, for example, an employee can have a phone number, but as a product manager he/she can have another phone number... This also goes to the special table then.
Approach 1)
Slows down the use of the Product table with the additional Product Manager field (maybe not for all databases but for some).
Linking from the Product table to the Employee table is simple.
Approach 2)
Existing queries using the Product table are not affected.
Increases the size of your database. You've now duplicated the Product ID column to another table as well as added unique constraints and indexes to that table.
Linking from the Product table to the Employee table is more cumbersome and costly as you have to ink to the intermediate table first.
How often must you link between the two tables?
How many other queries use the Product table?
How many records in the Product table?
in the particular case you give, i think the main motivation for two tables is avoiding nulls for missing data and that's how i would characterise the two approaches.
there's a discussion of the pros and cons on wikipedia.
i am pretty sure that, given c date's dislike of this, he defines relational theory so that only the multiple table solution is "valid". for example, you could call the single table approach "poorly typed" (since the type of null is unclear - see quote on p4).