Concepts

Every graph pattern contains at least one node pattern. The simplest graph pattern is a single, empty node pattern:

The empty node pattern matches every node in a property graph. In order to obtain a reference to the nodes matched, a variable needs to be declared in the node pattern:

With this reference, node properties can be accessed:

Adding a label expression to the node pattern means only nodes with labels that match will be returned. The following matches nodes that have the Stop label:

The following more complex label expression matches all nodes that are either a TrainStation and a BusStation or StationGroup:

A map of property names and values can be used to match on node properties based on equality with the specified values. The following matches nodes that have their mode property equal to Rail:

More general predicates can be expressed with a WHERE clause. The following matches nodes whose name property starts with Preston:

See the node patterns reference section for more details.

The simplest possible relationship pattern is a pair of dashes:

This pattern matches a relationship with any direction and does not filter on any relationship type or property. Unlike a node pattern, a relationship pattern cannot be used in a MATCH clause without node patterns at both ends. See path patterns for more details.

In order to obtain a reference to the relationships matched by the pattern, a relationship variable needs to be declared in the pattern by adding the variable name in square brackets in between the dashes:

To match a specific direction, add < or > to the left or right hand side respectively:

To match on a relationship type, add the type name after a colon:

Similar to node patterns, a map of property names and values can be added to filter on properties of the relationship based on equality with the specified values:

A WHERE clause can be used for more general predicates:

See the relationship patterns reference section for more details.

Any valid path starts and ends with a node, with relationships between each node (if there is more than one node). Path patterns have the same restrictions, and for all valid path patterns the following are true:

These are all valid path patterns:

These are invalid path patterns:

This section contains an example of matching a path pattern to paths in a property graph.

It uses the following graph:

To recreate the graph, run the following query against an empty Neo4j database:

The graph contains a number of train Stations and Stops. A Stop represents the arrival and departure of a train that CALLS_AT a Station. Each Stop forms part of a sequence of Stops connected by relationships with the type NEXT, representing the order of calling points made by a train service.

The graph shows three chains of Stops that represent different train services. Each of these services calls at the Station with the name Denmark Hill.

To return all Stops that call at the Station Denmark Hill, the following motif is used (the term motif is used to describe the pattern looked for in the graph):

In this case, three paths in the graph match the structure of the motif (plus the predicate anchoring to the Station Denmark Hill):

In order to return the name of each Stop that calls at a Station, declare a variable in the Stop node pattern. The results will then have a row containing the departs value of each Stop for each match shown above:

An equijoin is an operation on paths that requires more than one of the nodes or relationships of the paths to be the same. The equality between the nodes or relationships is specified by declaring the same variable in multiple node patterns or relationship patterns. An equijoin allows cycles to be specified in a path pattern. See graph patterns for more complex patterns.

This section uses the following graph:

To recreate the graph, run the following query against an empty Neo4j database:

To illustrate how equijoins work, we will use the problem of finding a round trip between two Stations.

In this example scenario, a passenger starts their outbound journey at London Euston Station and ends at Coventry Station. The return journey will be the reverse order of those Stations.

The graph has three different services, two of which would compose the desired round trip, and a third which would send the passenger to Birmingham International.

The solution is the following path with a cycle:

If unique properties exist on the node where the cycle "join" occurs in the path, then it is possible to repeat the node pattern with a predicate matching on the unique property. The following motif demonstrates how that can be achieved, repeating a Station node pattern with the name London Euston:

The path pattern equivalent is:

There may be cases where a unique predicate is not available. In this case, an equijoin can be used to define the desired cycle by using a repeated node variable. In the current example, if you declare the same node variable n in both the first and last node patterns, then the node patterns must match the same node:

Putting this path pattern with an equijoin in a query, the times of the outbound and return journeys can be returned:

This section uses the example of Stations and Stops used in the previous section, but with an additional property distance added to the NEXT relationships:

As the name suggests, this property represents the distance between two Stops. To return the total distance for each service connecting a pair of Stations, a variable referencing each of the relationships traversed is needed. Similarly, to extract the departs and arrives properties of each Stop, variables referencing each of the nodes traversed is required. In this example of matching services between Denmark Hill and Clapham Junction, the variables l and m are declared to match the Stops and r is declared to match the relationships. The variable origin only matches the first Stop in the path:

Variables that are declared inside quantified path patterns are known as group variables. They are so called because, when referred outside of the quantified path pattern, they are lists of the nodes or relationships they are bound to in the match. To understand how to think about the way group variables are bound to nodes or relationships, it helps to expand the quantified path pattern, and observe how the different variables match to the elements of the overall matched path. Here the three different expansions for each value in the range given by the quantifier {1,3}:

The subscript of each variable indicates which instance of the path pattern repetition they belong to. The following diagram shows the variable bindings of the path pattern with three repetitions, which matches the service that departs Denmark Hill at 17:07. It traces the node or relationship that each indexed variable is bound to. Note that the index increases from right to left as the path starts at Denmark Hill:

For this matched path, the group variables have the following bindings:

The second solution is the following path:

The following table shows the bindings for both matches, including the variable origin. In contrast to the group variables, origin is a singleton variable due to being declared outside the quantification. Singleton variables bind at most to one node or relationship.

Returning to the original goal, which was to return the sequence of depart times for the Stops and the total distance of each service, the final query exploits the compatibility of group variables with list comprehensions and list functions such as reduce():

This section uses the following graph:

To recreate it, run the following query against an empty Neo4j database:

One of the pitfalls of quantified path patterns is that, depending on the graph, they can end up matching very large numbers of paths, resulting in a slow query performance. This is especially true when searching for paths with a large maximum length or when the pattern is too general. However, by using inline predicates that specify precisely which nodes and relationships should be included in the results, unwanted results will be pruned as the graph is traversed.

Here are some examples of the types of constraints you can impose on quantified path pattern traversals:

The same example used in the shortest path section above is used here to illustrate the use of inline predicates. In that section, the shortest path in terms of number of hops was found. Here the example is developed to find the shortest path by physical distance and compared to the result from the shortestPath function.

The total distance from Hartlebury to Cheltenham Spa following the path yielded by shortestPath is given by the following query:

Whether this is the shortest path by distance can be checked by looking at each path between the two end Stations and returning the first result after ordering by distance:

This shows that there is a route with a shorter distance than the one with fewer Stations. For a small dataset, this query will be fast, but the execution time will increase exponentially with the graph size. For a real dataset, such as the entire rail network of the UK, it will be unacceptably long.

One approach to avoid the exponential explosion in paths is to put a finite upper bound to the quantified path pattern. This works fine where the solution is known to lie within some range of hops. But in cases where this is not known, one alternative would be to make the pattern more specific by, for example, adding node labels, or by specifying a relationship direction. Another alternative would be to add an inline predicate to the quantified path pattern.

In this example, an inline predicate can be added that exploits the geospatial property location of the Stations: for each pair of Stations on the path, the second Station will be closer to the endpoint (not always true, but is assumed here to keep the example simple). To compose the predicate, the point.distance() function is used to compare the distance of the left-hand and the right-hand Station to the destination Cheltenham Spa:

This query shows that there are only four paths solving the query (a number that remains constant even if the data from the rest of the UK railway network was included). Using inline predicates or making quantified path patterns more specific where possible can greatly improve query performance.

In addition to the single path patterns discussed so far, multiple path patterns can be combined in a comma-separated list to form a graph pattern. In a graph pattern, each path pattern is matched separately, and where node variables are repeated in the separate path patterns, the solutions are reduced via equijoins. If there are no equijoins between the path patterns, the result is a Cartesian product between the separate solutions.

The benefit of joining multiple path patterns in this way is that it allows the specification of more complex patterns than the linear paths allowed by a single path pattern. To illustrate this, another example drawn from the railway model will be used. In this example, a passenger is traveling from Starbeck to Huddersfield, changing trains at Leeds. To get to Leeds from Starbeck, the passenger can take a direct service that stops at all stations on the way. However, there is an opportunity to change at one of the stations (Harrogate) on the way to Leeds, and catch an express train, which may enable the passenger to catch an earlier train leaving from Leeds, reducing the overall journey time.

This section uses the following graph, showing the train services the passenger can use:

To recreate the graph, run the following query against an empty Neo4j database:

The solution to the problem assembles a set of path patterns matching the following three parts: the stopping service; the express service; and the final leg of the journey from Leeds to Huddersfield. Each changeover, from stopping to express service and from express to onward service, has to respect the fact that the arrival time of a previous leg has to be before the departure time of the next leg. This will be encoded in a single WHERE clause.

The following visualizes the three legs with different colors, and identifies the node variables used to create the equijoins and anchoring:

For the stopping service, it is assumed that the station the passenger needs to change at is unknown. As a result, the pattern needs to match a variable number of stops before and after the Stop b, the Stop that calls at the changeover station l. This is achieved by placing the quantified relationship -[:NEXT]->* on each side of node b. The ends of the path also needs to be anchored at a Stop departing from Starbeck at 11:11, as well as at a Stop calling at Leeds:

For the express service, the ends of the path are anchored at the stop b and Leeds station, which lds will be bound to by the first leg. Although in this particular case there are only two stops on the service, a more general pattern that can match any number of stops is used:

Note that as Cypher only allows a relationship to be traversed once in a given match for a graph pattern, the first and second legs are guaranteed to be different train services. (See relationship uniqueness for more details.) Similarly, another quantified relationship that bridges the stops calling at Leeds station and Huddersfield station is added:

The other node variables are for the WHERE clause or for returning data. Putting this together, the resulting query returns the earliest arrival time achieved by switching to an express service: