Creating benchmark graph data
AssigneeInternational Business Machines Corporation
InventorsSongyun Duan, Anastasios Kementsietsidis, Kavitha Srinivas, Octavian Udrea
Life of this patent
Estimating the term…
Abstract
According to an aspect of the present principles, a method is provided for generating resource description framework benchmarks. The method includes deriving a resultant benchmark dataset with a user specified size and a user specified coherence from and with respect to an input dataset of a given size and a given coherence by determining which triples of subject-property-object to add to the input dataset or remove from the input dataset to derive the resultant benchmark dataset.
Claims3
A non-transitory computer readable storage medium comprising a computer readable program, wherein the computer readable program when executed on a computer causes the computer to perform the following:
- deriving a resultant benchmark dataset with a user specified size and a user specified coherence from and with respect to an input dataset of a given size and a given coherence;
- wherein said deriving step comprises:
- computing respective coins, each representing a respective value by which the coherence of the input dataset increases or decreases with respect to deriving the resultant benchmark dataset, when respectively adding to or removing from the input dataset all triples of subject-property-object with subjects that are instances of types in a plurality of type sets and with properties equal to a particular set of properties, each of the respective coins being computed with respect to all of the types in only a respective one of the plurality of type sets and with respect to only a respective one of the properties in the particular set of properties;
- determining an amount of the triples, as represented by the respective coins, to be added or removed from the input dataset to derive the resultant benchmark dataset;
- and
- adding or removing the determined amount of the triples from the input dataset so as to derive the resultant benchmark dataset.
A system for deriving a resultant benchmark dataset from and with respect to an input dataset of a given size and a given coherence, the system comprising:
- a coin calculator for computing respective coins, each representing a respective value by which the coherence of the input dataset increases or decreases with respect to deriving the resultant benchmark dataset, when respectively adding or removing from the input dataset all triples of subject-property-object with subjects that are instances of types in a plurality of type sets and with properties equal to a particular set of properties, each of the respective coins being computed with respect to all of the types in only a respective one of the plurality of type sets and with respect to only a respective one of the properties in the particular set of properties;
- a triple amount determiner for determining an amount of the triples, as represented by the respective coins, to be added or removed from the input dataset to derive the resultant benchmark dataset with a user specified size and a user specified coherence;
- and
- a triple modifier for adding or removing the determined amount of the triples from the input dataset so as to derive the resultant benchmark dataset,
- wherein the coin calculator, the triple amount determiner, and the triple modifier are implemented by a hardware processor and a memory device.
Dependencies are inferred from the claim text.