Microsoft Technology Licensing, LLC

Peter Dahle Heen

Jan-Ove Almli Karlberg

Tor Kreutzer

Amund Kronen Johansen

Dag Steinnes Eidesen

Steffen Viken Valvåg

Jon Meling

Åge Andre Kvalnes

Age Kvalnes

Steffen Viken Valvag

Daniele Vettorel

Dag Brattli

Åge Kvalnes

Examples of the present disclosure describe systems and methods for ontology-based graph query optimization. In an example, ontology data relating to a graph or isolated collection may be collected. The ontology data may comprise uniqueness and topology information and may be used to reformulate a query in order to yield a query that is more performant than the original query when retrieving target information from a graph. In an example, reformulating a query may comprise reordering one or more parameters of the query relating to resources, relationships, and/or properties based on uniqueness information. In another example, the query may be reformulated by modifying the resource type to which the query is anchored based on the topology information. The reformulated query may then be executed to identify target information in the isolated collection, thereby identifying the same target information as the original query, but in a manner that is more performant.

Ontology-based graph query optimization

Controlling access to nodes in a relational graph at query time by using an approximate membership query (AMQ) filter and ordered queries based on historic grants or denials of access according to security context enables a more efficient querying of the relational graph while preserving access controls. Security contexts that grant or deny access to a node are stored in an associated AMQ filter and are queried according to the subject's security context in an order based on the frequency at which the security contexts have previously granted or denied access to nodes in the relational graph.

Access controlled graph query spanning

Approximate Membership Query (AMQ) Filters are used in conjunction with graph queries to a relational graph to provide change monitoring that span views associated with the queries. Each node from the relational graph spanned by a graph query and the index structure for the view are added as members to an AMQ filter. When a change is made to the relational graph, the changed nodes are queried against the AMQ filter. When a changed node is noted as a candidate member of the AMQ filter, the graph query may be rerun to update the view associated with the query. Otherwise, the graph query is not rerun, thus saving computing resources and improving the systems hosting and querying the relational graph.

Change monitoring spanning graph queries

Systems, methods, and computer readable devices embodying instructions are provided herein for reducing latencies and/or improving computational efficiency when traversing data stored in a relational graph by caching subgraphs and enabling the utilization thereof. More specifically, after a user performs a graph query, the resulting subgraphs of the graph query are cached in a reusable form. Subsequent graph queries are able to identify cached subgraphs based on the graph query. Further, the subsequent graph query is operable to integrate the cached subgraphs as part of the result of subsequent graph query, which may include a portion or the entire result of the subsequent graph query being composed from cached subgraphs, thereby improving the computational efficiency and performance of querying relational graphs, reducing the query execution cost required to traverse the relational graphs, and improving the functionality of the computing devices hosting the relational graphs and running the queries.

Caching of subgraphs and integration of cached subgraphs into graph query results

Traversing data stored in a relational graph by utilization of probabilistic characteristics associated with the graph nodes is disclosed. When a user submits a request with a graph query, an initial node associated with the graph query is identified. Further, the edge type associated the node is extracted from the graph query. When traversing the graph by following relevant edges from the initial node to new nodes, each new node is queried with the extracted edge type. If the query for the node is negative, then the edges for the particular node are not enumerated. However, if the query for the node is positive, then the edges for the particular node are enumerated for expanding the subgraph. This process continues until the subgraph is expanded to include all relevant nodes. Thus, the computational efficiency is improved by reducing the number of edges that must be traversed when performing graph queries.

Growing community of inventors

Tromsø, Norway

Peter Dahle Heen