Encoding system, method, and recording medium for time grams

Patent metadata

Field	Value
Publication number	US-10049140-B2
Application number	US-201514839730-A
Country	US
Kind code	B2
Filing date	Aug 28, 2015
Priority date	Aug 28, 2015
Publication date	Aug 14, 2018
Grant date	Aug 14, 2018

Abstract

An encoding system for encoding an event time series, the system including an inter-arrival time computing device configured to compute an inter-arrival time between a plurality of input events and computes a sequence of events, a transformation device configured to transform the inter-arrival time between the plurality of input events into discrete time symbols, and a mapping device configured to map the input events and the discrete time symbols using a dictionary to output a time gram.

First claim

What is claimed is:

1. An encoding system including a processor and a memory for encoding an event time series, the system comprising:
a processor, and
a memory, the memory storing instructions to cause the processor to:
compute, using said processor, an inter-arrival time between each of a plurality of input events and computes a sequence of events;
transform, using said processor, each of the inter-arrival times between the plurality of input events into discrete time symbols by quantizing the time symbols between two events, each discrete time symbol representing one inter-arrival time of the computed inter-arrival times between two of the input events; and
map, using said processor, the input events and the discrete time symbols using a dictionary to output a time gram,

wherein the mapping introduces symbol sets and grammar rules to group the plurality of input events together that have a predetermined threshold inter-arrival time and also maps a sequence of events with the inter-arrival times for recognized out of sequence patterns between the inter-arrival times between the plurality of input events.

2. The encoding system according to claim 1 , wherein the dictionary comprises at least one of rules, patterns, and grammars.

3. The encoding system according to claim 1 , wherein the dictionary dynamically evolves based on previously mapped time grams by the mapping.

4. The encoding detection system according to claim 1 , wherein the dictionary comprises a combination of pattern mining algorithms including at least one of sequitur, Ngrams, and LZ.

5. The encoding system according to claim 4 , wherein a regex engine is used in the pattern mining algorithms.

6. The encoding system according to claim 4 , wherein the dictionary comprises multiple data combined-data sets based on frequently-mined matching rules.

7. The encoding system according to claim 1 , wherein the memory further stores instructions to cause the processor to recursively apply the mapping such that the output time gram is used as an input into a second event sequence.

8. The encoding system according to claim 1 , wherein the time gram is further annotated to capture domain specific semantics.

9. The encoding system according to claim 1 , wherein the transforming includes lossless or lossy compression such that the original time series is recoverable after the mapping outputs the time gram.

10. The encoding system according to claim 1 , wherein the memory further stores instructions to cause the processor to recognize, using said processor, patterns in the time gram output by the mapping.

11. The encoding system according to claim 10 , wherein the memory further stores instructions to cause the processor to store patterns, using said memory, in the time gram output by the mapping recognized by the pattern recognition.

12. The encoding system according to claim 11 , wherein the memory further stores instructions to cause the processor to store analyze, using said processor, the stored patterns in a storage device and score a probability that an event will occur next based on the stored patterns.

13. The encoding system according to claim 12 , memory further stores instructions to cause the processor to learn and predict, using said processor, based on the probability output by the scoring, the an inter-arrival time until a next event of the recognized pattern will occur.

14. The encoding system according to claim 13 , wherein the memory further stores instructions to cause the processor to transform predicts the next event in real-time.

15. The encoding system according to claim 1 , wherein the quantization is based on at least one of a history, a context, and semantics such that the quantization by the transforming dynamically updates.

16. A method for encoding an event time series, the method comprising:
transforming, using a processor, each of the inter-arrival times between the plurality of input events into discrete time symbols by quantizing the time symbols between two events each discrete time symbol representing one inter-arrival time of the computed inter-arrival times between two of the input events; and
mapping the plurality of input events and the discrete time symbols using a dictionary to output a time gram,
wherein the main introduces symbol sets and grammar rules to group the plurality of input events together that have a predetermined threshold inter-arrival time and also maps a sequence of events with the inter-arrival times for recognized out of sequence patterns between the inter-arrival times between the plurality of input events.

17. A non-transitory computer-readable recording medium recording an encoding program for encoding an event time series, the program causing a computer to perform:
transforming, using a processor, each of the inter-arrival times between the plurality of input events into discrete time symbols by quantizing the time symbols between two events, each discrete time symbol representing one inter-arrival time of the computed inter-arrival times between two of the input events; and
mapping the plurality of input events and the discrete time symbols using a dictionary to output a time gram,
wherein the mapping introduces symbol sets and grammar rules to group the plurality of input events together that have a predetermined threshold inter-arrival time and also maps a sequence of events with the inter-arrival times for recognized out of sequence patterns between the inter-arrival times between the plurality of input events.

Assignees

• IBM

Inventors

• Halim Nagui
• Parthasarathy Srinivasan
• Pavuluri Venkata N
• Sow Daby Mousse
• Turaga Deepak Srinivas

Classifications

• G06F16/2477 Primary

  Temporal data queries · CPC title

• G06F17/30551 Primary

  Physics · mapped topic