There are three sets of entities: Players, Teams and Games. Team may consist from one or two Players and are formed voluntary for each particular Game, i.e.
players A,B,C,D can form 11 Teams because Team AB is the same as Team BA.
That being said Teams may contain only unique set of players - BA is a duplication of AB.
The most obvious way to form the Teams and Games relationship is as many to many, but is this a way to go? The real question is how to model those restrictions in robust and scalable way, so they can handle, let's say, a team not of only 1-2 players but also a team of 1-20 players without (much) augmentation?
Here is an abstract interface I'm thinking of -
Team.find_or_create_by(player_ids: [p1.id, p2.id]) # find p1 and p2 team
Team.find_by(player_id: p1) # find all teams that p1 has participated in
p.s. I do not think that question title is really good and probably this is well known problem which has an established name, so if one can point it out I'll be really glad to swap the title.
Related
I wish to create a variable depicting similarity with respect to race to team members. In other words, I want to know of the people an individual shares a manager with, what percentage of the team is of the same race?
The variables I currently have are participant id, participant race, manager id, manager race, and team size. I know the racial breakdown of the teams, but I need the percentage of similar others in a team for each participant (in one column, not across columns split by race).
First calculate the number of participants of the same race for each participants, then divide that in the team size to get the ratio.
In the following code I assume you have the team ID although you didn't mention that in your post:
aggregate out=* mode=addvariables /break team_id participant_race /NRaceInTeam=n.
compute PRaceInTeam=NRaceInTeam/Team_Size.
I denormalising a OLTP database for use in a DWH.
At the moment I am denormalising studygroups.
Each studygroup has a key pointing towards 1 project.
Each project has a key pointing towards 1 department.
Each department has a key pointing towards 1 university.
Each universityhas a key pointing to 1 city.
Now I know that you are supposed to denormalize the sh*t out your OLTP but in this dwh department will be a dimension on its own. This goes for university also. Would it suffise to add a key from studygroup pointing at department or is it wiser to denormalize as far as you can and add all attributes from the department and all attributes from its M:1 related tables to the dimension studygroup? Even when department and university will be dimensions by themselves?
In other words: how far/deep do you go when denormalizing?
The key concept behind a dimensional model is:
Keep your fact tables in 3NF (third normal form);
De-normalize your dimensions into 2NF (second normal form)
So ideally, the only joins you should have in your model are the joins between fact tables and relevant dimensions.
As part of this philosophy:
Avoid "snow flake" designs, where dimensions contain keys to other dimensions. It's always possible to come up with a data model that allows the same functionality as the snow flakes, without violating 3NF/2NF rule;
Never have any direct joins between 2 separate dimensions (i.e, department and study group) directly. All relations among dimensions must be resolved via fact tables;
Never have any direct joins between 2 separate fact tables. Any relations among fact tables must be resolved via shared dimensions.
Finally, consider that dimensional design, besides optimization of the data for querying, serves a second important purpose: it's a semantic model of the business (or whatever else it represents). So, when making decisions about combining data elements into dimensions and facts, consider their "logical affinity" - they should make intuitive sense to the end users. If you have hard times explaining to a BI analyst the meaning of your dimension or fact table, most likely you've made a modeling mistake.
For example, in your case you should consider logical relations between universities, departments, study groups, etc. It's very likely that University/Department form a natural hierarchy. If so, they should belong to the same dimension. Study group, on the other hand, might not - let's assume, it's possible to form study groups across multiple universities and/or multiple departments. Such Many:Many relations are clear indication that they should be resolved via fact tables. In addition, relations between universities and departments are stable (rarely change), while study groups are formed and dissolved very often, and thus should be modeled separately.
In general, if you see 1:1 or 1:M relations between dimensional elements, it's often an indication that they should be de-normalized into the same table (again, only if their combination makes logical sense). If the relations are M:M, most likely they belong to different tables (you can force them into the same table, but often such tables look like Frankenstein creatures).
You can get much better help by making your question more specific - draw your dimensional model, post it, and ask for specific issues/challenges you have. For general concepts, books from Kimball and Inmon are your best friends.
I am in the process of designing this E-R diagram for a shop of which I have shown part of below (the rest is not relevant). See the link please:
E-R diagram
The issue that I have is that the shop only sells two items, Socks and Shoes.
Have I correctly detailed this in my diagram? I'm not sure if my cardinalities and/or my design is correct. A customer has to buy at least one of these items for the order to exist (but has the liberty to buy any number).
The Shoe and Sock entities would have their respective ID attribute, and I am planning to translate to a relational schema like this:
(I forgot to add to my diagram the ORDER_CONTAINS relationship to have an attribute called "Quantity". )
Table: Order_Contains
ORDER_ID | SHOEID | SOCKID | QTY
primary key | FK, could be null |FK, could be null | INT
This clearly won't work since the Qty would be meaningless. Is there a way I can reduce the products to just two products and make all this work?
Having two one-to-many relationships combined into one with nullable fields is a poor design. How would you record an order containing both shoes and socks - a row per shoe with SOCKID set to NULL and vice-versa for socks, or would you combine rows? In the former case the meaning of QTY is clear though it depends on the contents of SHOEID/SOCKID fields, but what would the QTY mean in the latter case? How would you deal with rows where both SHOEID and SOCKID are NULL and the QTY is positive? Keep in mind Murphy's law of databases - if it can be recorded it will be. Worse, your primary key (ORDER_ID) will prevent you from recording more than one row, so a customer couldn't buy more than one (pair of) socks or shoes.
A better design would be to have two separate relations:
Order_Socks (ORDER_ID PK/FK, SOCKID PK/FK, QTY)
Order_Shoes (ORDER_ID PK/FK, SHOEID PK/FK, QTY)
With this, there's only one way to record the contents of an order and it's unambiguous.
You have not explained very well the context here. I'll try to explain from what I understand, and give you some hints.
Do your shop only and always (forever) sell 2 products? Do the details of these products (color, model, weight, width, etc...) need to be persisted in the database? If yes, then we have two entities in the model, SOCKS and SHOES. Each entity has its own properties. A purchase or a order is usually seen as an event on the ERD. If your customers always buys (or order) socks with shoes, then there will always be a link between three entities:
CLIENTS --- SHOES --- SOCKS
This connection / association / relationship is an event, and this would be the purchase (or order).
If a customer can buy separate shoes and socks, then socks and shoes are subtypes of a super entity, called PRODUCTS, and a purchase is an event between CUSTOMERS and PRODUCTS. Here in this case we have a partitioning relationship.
If however, your customers buy separate products, and your store will not sell forever only 2 products, and details of the products are not always the same and will not be saved as columns in a table, then the case is another.
Shoes and socks are considered products, as well as other items that can be considered in future. Thus, we have records/rows in a PRODUCTS table.
When a customer places an order (or a purchase), he (she) is acquiring products. There is a strong link between customers and products here, again usually an event, which would be the purchase (or a order).
I do not know if you do it, but before thinking of start a diagram, type the problem context in a paper or a document. Show all details present in the situation.
The entities are seen when they have properties. If you need to save the name of a customer, the customer's eye color, the customer's e-mail, and so on, then you will have certainly a CUSTOMER entity.
If you see entities relate in some way, then you have a relationship, and you should ask yourself what kind of relationship these entities form. In your case of products and customers, we have a purchasing relationship there between. The established relationship is a purchase (or an order, you call it). One customer can buy various products, and one product (not on the same shelf, is the type, model) can be purchased for several customers, thus, we have a Many-To-Many relationship.
The relationship created changes according to the context. Whatever, we'll invent something crazy here as examples. Say we have customers and products. Say you want to persist a situation where customers lick Products (something really crazy, just for you to see how the context says the relationship).
There would be an intimate connection between customers and products entities (really close... I think...). In this case, the relationship represents a history of customers licking products. This would generate an EVENT. In this event you could put properties such as the date, the amount of times a customer licked a proper product, the weather, the time, the traffic light color on the street, etc., only what you need to persist according to your context, your needs.
Remember that for N-N relationships created, we need to see if new entities (out of relationship) will emerge. This usually happens when you are decomposing the conceptual model to the logical model. Probably, product orders will generate not one but two entities: The ORDER and the products of orders. It is within the products of orders that you place the list of products ordered from each customer, and the quantity.
I would like to present various materials to study ERD, but unfortunately they are all in Portuguese. I hope I have helped you in some way. If you want to be more specific about your problem, I think I can really help you best. Anything, please ask.
I'm trying to model contractor relationships in Neo4J and I'm struggling with how to conceptualize subcontracts. I have nodes for Government Agencies (label: Agency) and Contractors (label:Company). Each of these have geospatial Office nodes with the HAS_OFFICE relationship. I'm thinking of creating a node that represents a Government Contract (label: Contract).
Here's what I'm struggling with: A Contract has a Government Agency (I'm thinking this is a "HAS CONTRACT" relationship) and one or more prime contractor(s) (I'm thinking this is a "PRIME" relationship). Here's where it gets complicated. Each of those primes contractors can have subcontractors under the prime contract only. Graphically, this is:
(Agency) -[HAS_CONTRACT]-> (Contract) -[PRIME]-> (Company 1) -[SUB]-> (Company 2)
The problem I'm struggling with is that the [SUB] relationship is only for certain contracts -- not all. For example:
Agency 1 -HAS-> Contract ABC -P-> Company 1 -S-> Company 2
Agency 1 -HAS-> Contract ABC -P-> Company 3 -S-> Company 4
Agency 2 -HAS-> Contract XYZ -P-> Company 1
Agency 2 -HAS-> Contract XYZ -P-> Company 4 -S-> Company 2
I want some way to search on that so I can ask cypher questions like "Find ways Agency 2 can put money on contract with Company 2." It should come back with the XYZ contract through Company 4, and NOT the XYZ contract through Company 1.
It seems like maybe storing and filtering on data within the relationship would work, but I'm struggling with how. Can I say Prima and Sub relationships have a property, "contract_id" that must match Contract['id']? If so, how?
Edit: I don't want to have to specify the contract name for the query. Based on #MarkM's reply, I'm thinking something like:
MATCH (a:Agency)-[:HAS]-(c:Contract)-[:PRIME {contract_id:c.id}]
-(p:Company)-[:SUB {contract_id:c.id}]-(s:Company)
RETURN s
I'd also like to be able to use things like shortestPath to find the shortest path between an agency and a contractor that follows a single contract ID.
I'd create the subcontractor by having two relationships, one to the contractor and one to the contract.
(:Agency)-[:ISSUES]->(con:Contract)-[:PRIMARY]->(contractor:Company)
(con:Contract)-[:SECONDARY]->(subContractor:Company)<-[:SUBCONTRACTS]-(contractor:Company)
Perhaps you can mode your use-case as a graph-gist, which is a good way of documenting and discussing modeling issues.
This seems pretty simple; I apologize if I've misunderstood the question.
If you want subcontractors you can simply query:
MATCH (a:Agency)-[:HAS]-(:Contract)-[:PRIME]-(p:Company)-[:SUB]-(s:Company) RETURN s
This will return all companies that are subcontractors. The query matches the whole pattern. So if you want XYZ contract subcontractors you simply give it the parameter:
MATCH (a:Agency)-[:HAS]-(:Contract {contractID: XYZ})-[:PRIME]-(p:Company)-[:SUB]-(s:Company) RETURN s
You'll only get company 2.
EDIT: based on your edit:
"Find ways Agency 2 can put money on contract with Company 2"
This seems to require some domain-specific knowledge which I don't have. I assume Agency 2 can only put money on subcontractors but not primes?? I might help if you reword so we know exactly what your trying to get from the graph. From my reading it looks like you want all companies that are subcontractors under Company 2's contracts. Is that right?
If that's what you want, again you just give Neo the path:
MATCH (a:Agency: {AgencyID: 2)-[:HAS]
-(c:Contract)-[:PRIME]-(:Company)-[:SUB]-(s:Company: {companyID: 2)
RETURN c, s
This will give you a list of all contracts under XYZ for which Company 2 is a subcontractor. With the current example, it will one row: [c:Contract XYZ, s:Company 2]. If Agency 2 had more contracts under which Company 2 subcontracted, you would get more rows.
You can't do this: [:PRIME {contract_id:c.id}] [:SUB {contract_id:c.id}] because Prime and Sub relationships shouldn't have contract_id properties. They don't need them — the very fact that they are connected to a contract is enough.
One thing that might make this a little more complicated is if the subcontractors also have subcontractors, but that's not evident.
Okay take 2:
So the problem isn't captured well in the original example data — sorry for missing it. A better example is:
Agency 1 -HAS-> Contract ABC -P-> Company 1 -S-> Company 2
Agency 1 -HAS-> Contract XYZ -P-> Company 1 -S-> Company 3
Now if I ask
MATCH (a:Agency)-[:HAS_CONTRACT]-(ABC:Contract {id:ABC})-[:PRIME]
-(c:Company)-[:SUBS]-(c2) RETURN c2
I'll get both Company 2 and 3 even though only 2 is on ABC, Right?
The problem here is the data model not the query. There's no way to distinguish a company's subs because they are all connected directly to the company node. You could put a property on the sub relationship with the prime ID, but a better way that really captures the information is to add another contract node under company. Whether you label this as a different type depends on your situation.
Company1 then [:HAS] a contract which the subs are connected to. The contract can then point back to the prime contract with a relationship of something like [:PARENT] or [:PRIME] or maybe from the prime to the sub with a [:SUBCONTRACT] relationship
Now everything becomes much easier. You can find all subcontracts under a particular contract, all subcontracts a particular company [:HAS], etc. To find all subcontractors under a particular contract you could query something like this:
MATCH (contract:Contract { id:"ContractABC" })-[:PRIME]-(c:Company)
-[:HAS]->(subcontract:Contract)-[:PARENT]-(contract)
WITH c, subcontract
MATCH (subcontract)-[:SUBS]-(subcontractor:Company)
RETURN c, subcontractor
This should give you a list of all companies and their subcontractors under contract ABC. In this case Company 1, Company 2 (but not company 3).
Here's a console example: http://console.neo4j.org/?id=flhv8e
I've left the original [:SUB] relationships but you might not need them.
I am trying to understand the concept of ER modelling, but I do not yet succeed. I have designed the ER model about movie database, but I do not know wheather it is a good design and how to connect the entities:
between Actor and Film i want to say "actor can play in each film only once" and at the same time "many actors can play in many movies" -- is it 1 to 1 relation or many to many?
and HOW do we need to think about entities ans relations between them? relations to one user, one film, one actor, one director, or in general?
UPDATE: new question : should the relation between Director and Film be 1 to many or many to many? I want to say : "one director can have many films && many directors can have may films" ??
Think about it like this: There are many movies. There are many actors. It makes sense that you would only want to include each actor in a particular movie once, but otherwise you want to be able to "mix and match" the movies and actors to express the relationship.
Looking at your diagram, you don't seem to have any fields which express the relationship between Film and Actor - those lines need to match actual fields. Read up on foreign keys: http://en.wikipedia.org/wiki/Foreign_key
The relationship between Actor and Movie that you want is actually many-to-many. You can express this with a "join table" (you'd need to add this to your diagram).
Something like this would work:
FilmActor
-------
uidFilm
uidActor
And put a unique constraint on those two fields together so it can't be duplicated (i.e. the same Actor can't appear in a Film twice)