Process and framework for facilitating data sharing using a distributed hypergraph

• US 9,996,567 B2
• Filed: 05/29/2015
• Issued: 06/12/2018
• Est. Priority Date: 05/30/2014
• Status: Active Grant

• Current Assignee: Georgetown University
• Original Assignee: Georgetown University
• Inventors: Smart, J. Cory
• Primary Examiner(s): Hershley, Mark E
• Application Number:
  US14/724,921
• Publication Number:
  US 20150347480A1
• Time in Patent Office: 1,110 Days
• Field of Search: 707743
• CPC Class Codes: G06F 16/2237 Vectors, bitmaps or matrices
• CPC Class Codes: G06F 16/2246 Trees, e.g. B+trees
• CPC Class Codes: G06F 16/2264 Multidimensional index structures
• CPC Class Codes: G06F 16/26 Visual data mining; Browsing structured data

Abstract

A knowledge model “overlay” for organizing and analyzing large, dynamic data volumes is implemented as a hypergraph that manifests as a result of a distributed theory-driven data source transformation process. This process maps exponentially growing data into an asymptotically limited information space. Within this space, real-world entities (e.g. people, computers, cities, Earth) and their complex interdependencies (e.g. social networks, connectivity, causal relationships) are represented as attributed hypergraph elements (i.e. both hypervertices and hyperedges). Attributes are represented as state vectors affixed to any hypergraph element. Transformation of source system data into this overlay structure is accomplished with minimal data movement and replication using a universal “pointer” like mechanism that is managed in a decentralized fashion by the respective transformation components. Access to the knowledge overlay is performed via a hypergraph communication protocol encapsulated within a common hypergraph API and accompanying hypergraph toolkit.

Claims

1. A process for organizing data from multiple heterogeneous data sources in a knowledge hypergraph, the knowledge hypergraph being accessible by multiple unrelated analytics operating on one or more analytics engines, the process comprising:

receiving a first data set from a first data source at a first of multiple hypergraph servers;

graphing by the first of multiple hypergraph servers first data from the first data set in a knowledge hypergraph in accordance with a knowledge model (M), the knowledge hypergraph being defined by hypergraph elements including hypervertices and hyperedges, wherein attributes of the first data are represented in the hypergraph by first state vectors affixed to one or more hypergraph elements;

receiving a second data set from a second data source at a second of multiple hypergraph servers;

graphing by the second of multiple hypergraph servers second data from the second data set in the hypergraph, wherein attributes of the second data are represented in the hypergraph by second state vectors affixed to one or more hypergraph elements;

further wherein at least one of the first state vectors and at least one of the second state vectors is affixed to the same hypergraph element;

receiving at the first hypergraph server from a first of the one or more analytics engines a search request initiated by a first analytic and directed to the hypergraph; and

querying by the first hypergraph server the hypergraph, wherein query results include attributes from both the first data and the second data;

wherein the hypergraph exists as a distributed entity across the multiple hypergraph servers and further wherein the multiple hypergraph servers communicate peer-to-peer regarding the knowledge hypergraph using a defined hypergraph transfer protocol (HGTP).

2. The process according to claim 1, wherein the first data set and the second data set are received from the first and second hypergraph servers via a common hypergraph application programming interface.

3. The process according to claim 2, wherein the search request is submitted to the first hypergraph server and the query results are returned to the first analytic via the common hypergraph application programming interface.

4. The process according to claim 2, wherein the first and second hypergraph servers create new hypergraph elements in the knowledge graph responsive to a create primitive received from an associated adapter, the create primitive being defined by the common hypergraph application programming interface.

5. The process according to claim 4, wherein each of the created hypergraph elements includes a universally unique identifier (UUID) and further wherein the UUID includes embedded routing information to a persisted location of the created hypergraph element.

6. The process according to claim 5, wherein the persisted location of the created hypergraph element is one of the first source and the second source.

7. A process for requesting information from a distributed knowledge graph, the process comprising:

formulating, at a first adapter, a query directed to a first hypergraph element of the distributed knowledge graph using a common hypergraph application programming interface for processing by a first hypergraph server associated with the first adapter, the first hypergraph server being programmed to receive queries in a format of the common hypergraph application programming interface and further being programmed to communicate with other hypergraph servers using a defined hypergraph transfer protocol (HGTP);

packaging, by the first hypergraph server, the query into a HGTP query message and forwarding the query message to a second hypergraph server in accordance with first location instructions for the first hypergraph element at a second adapter;

receiving, by the second hypergraph server, the packaged query message, un-packaging the packaged query message and relaying the query to the second adapter;

retrieving results to the query from the associated first hypergraph element by the second adapter and returning the results to the second hypergraph server using the common hypergraph application programming interface; and

packaging, by the second hypergraph server, the results into a HGTP results message and forwarding the results message to the first hypergraph server in accordance with second location instructions for the first adapter associated therewith.

8. The process according to claim 7, wherein the query includes a request for attributes associated with the first hypergraph element.

9. The process according to claim 8, wherein formulating the query includes providing a universally unique identifier (UUID) for first hypergraph element.

10. The process according to claim 9, wherein formulating the query includes providing a universally unique identifier (UUID), an identifier of the method and parameters of the method for first hypergraph element.

11. The process according to claim 7, wherein the query includes a request for invoking a method associated with the first hypergraph element, and further wherein retrieving results to the query from the associated first hypergraph element by the second adapter includes executing the method associated with the first hypergraph element and retrieving the results therefrom.

12. The process according to claim 7, wherein forwarding the packaged query message includes determining, by the first hypergraph server, the first location instructions for the first hypergraph element using the received UUID.

13. The process according to 7, wherein the packaged query message includes location instructions for the first adapter.