System and method for scalable semantic stream processing
First Claim
Patent Images
1. A computer processor system for processing data, comprising:
- a communications receiver configured to receive a plurality of streams of data from a plurality of respective sensors, wherein the sensors include;
at least one video sensor, at least one satellite sensor, at least one acoustic sensor, and at least one document source;
one or more computer processors configured to;
identify features of the streams that match known patterns based on the streams of data from the plurality of sensors, using methods specific to each type of the sensors, and process the identified features into infon data structures comprising a relation and at least two parameters on which the relation works;
form semantic binary frames (sb-frames) based on the infon data structures, each sb-frame comprising a segment of data from a stream and semantic information that is universal to the streams of data received from each type of sensor that includes natural language expressions describing the segment of data based on an identified purpose of the computer processor system, and each sb-frame being formed using at least one infon data structure in a respective stream of data, a previous infon data structure in the stream of data, and a next infon data structure in the stream of data;
form fiber groups based on the sb-frames, each fiber group comprising a plurality of sb-frames related by logical natural language expressions that relate the segment of data to other segments of data in the stream;
identify situations of interest as control groups based on the fiber groups; and
receive the sb-frames and the control groups as input, and process the sb-frames to identify a presence of situations of interest and produce a situated deduction.
2 Assignments
0 Petitions
Accused Products
Abstract
A system for collaborative analysis from different processes on different data sources. The system uses a unique approach to lightweight temporary data structures in order to allow communication of interim results among processes, and construction of semantically appropriate reports. The data structures are generated in near real time and their lightweight nature supports massive scaling, including many diverse streaming inputs.
-
Citations
17 Claims
-
1. A computer processor system for processing data, comprising:
-
a communications receiver configured to receive a plurality of streams of data from a plurality of respective sensors, wherein the sensors include;
at least one video sensor, at least one satellite sensor, at least one acoustic sensor, and at least one document source;one or more computer processors configured to; identify features of the streams that match known patterns based on the streams of data from the plurality of sensors, using methods specific to each type of the sensors, and process the identified features into infon data structures comprising a relation and at least two parameters on which the relation works; form semantic binary frames (sb-frames) based on the infon data structures, each sb-frame comprising a segment of data from a stream and semantic information that is universal to the streams of data received from each type of sensor that includes natural language expressions describing the segment of data based on an identified purpose of the computer processor system, and each sb-frame being formed using at least one infon data structure in a respective stream of data, a previous infon data structure in the stream of data, and a next infon data structure in the stream of data; form fiber groups based on the sb-frames, each fiber group comprising a plurality of sb-frames related by logical natural language expressions that relate the segment of data to other segments of data in the stream; identify situations of interest as control groups based on the fiber groups; and receive the sb-frames and the control groups as input, and process the sb-frames to identify a presence of situations of interest and produce a situated deduction. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
-
-
16. A method of processing data, comprising:
-
receiving, using a communications receiver, a plurality of streams of data from a plurality of respective sensors, wherein the sensors include;
at least one video sensor, at least one satellite sensor, at least one acoustic sensor, and at least one document source;identifying features of the streams that match known patterns based on the streams of data from the plurality of sensors, using methods specific to each type of sensor, and processing the identified features into infon data structures comprising a relation and at least two parameters on which the relation works; forming semantic binary frames (sb-frames) based on the infon data structures, each sb-frame comprising a segment of data from a stream and semantic information that is universal to the streams of data received from each type of sensor that includes natural language expressions describing the segment of data based on an identified purpose of the computer processor system, and each sb-frame being formed using at least one infon data structure in a respective stream of data, a previous infon data structure in the stream of data, and a next infon data structure in the stream of data; form fiber groups based on the sb-frames, each fiber group comprising a plurality of sb-frames related by logical natural language expressions that relate the segment of data to other segments of data in the stream; identifying control groups based on the fiber groups characterizing situations of interest; synthesizing sb-frames that conform to the control groups to form a series of facts as one or more situated deductions.
-
-
17. A device with one or more non-transitory computer-readable mediums storing a program that, when executed, causes a computer processor to:
-
receive, using a communications receiver, a plurality of streams of data from a plurality of respective sensors, wherein the sensors include;
at least one video sensor, at least one satellite sensor, at least one acoustic sensor, and at least one document source;identify features of the streams that match known patterns based on the streams of data from the plurality of sensors, using methods specific to each type of sensor, and process the identified features into infon data structure comprising a relation and at least two parameters on which the relation works; form semantic binary frames (sb-frames) based on the infon data structures, each sb-frame comprising a segment of data from a stream and semantic information that is universal to the streams of data received from each type of sensor that includes natural language expressions describing the segment of data based on an identified purpose of the computer processor system, and each sb-frame being formed using at least one infon data structure in a respective stream of data, a previous infon data structure in the stream of data, and a next infon data structure in the stream of data; form fiber groups based on the sb-frames, each fiber group comprising a plurality of sb-frames related by logical natural language expressions that relate the segment of data to other segments of data in the stream; identify control groups based on the fiber groups characterizing situations of interest; synthesize sb-frames that conform to the control groups to form a series of facts as one or more situated deductions.
-
Specification