US 12,008,027 B2
Optimization for real-time, parallel execution of models for extracting high-value information from data streams
Luis F. Stevens, San Jose, CA (US); Vincent Schiavone, Berwyn, PA (US); Charles H. Leu, San Jose, CA (US); Shirin Hashemi, San Francisco, CA (US); and Mo Malakiman, San Jose, CA (US)
Assigned to Target Brands, Inc., Minneapolis, MN (US)
Filed by Target Brands, Inc., Minneapolis, MN (US)
Filed on Jun. 30, 2020, as Appl. No. 16/917,466.
Application 16/917,466 is a continuation of application No. 15/360,935, filed on Nov. 23, 2016, granted, now 10,698,935.
Application 16/917,466 is a continuation of application No. 14/214,490, filed on Mar. 14, 2014, granted, now 9,600,550, issued on Mar. 21, 2017.
Application 15/360,935 is a continuation in part of application No. 14/688,865, filed on Apr. 16, 2015, granted, now 10,599,697, issued on Mar. 24, 2020.
Application 15/360,935 is a continuation in part of application No. 14/214,410, filed on Mar. 14, 2014, granted, now 9,477,733, issued on Oct. 25, 2016.
Claims priority of provisional application 62/259,021, filed on Nov. 23, 2015.
Claims priority of provisional application 62/259,023, filed on Nov. 23, 2015.
Claims priority of provisional application 62/259,024, filed on Nov. 23, 2015.
Claims priority of provisional application 62/259,026, filed on Nov. 23, 2015.
Claims priority of provisional application 62/264,845, filed on Dec. 8, 2015.
Claims priority of provisional application 61/802,353, filed on Mar. 15, 2013.
Claims priority of provisional application 61/980,525, filed on Apr. 16, 2014.
Prior Publication US 2021/0279265 A1, Sep. 9, 2021
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 16/35 (2019.01); G06F 11/34 (2006.01); G06F 16/245 (2019.01); G06F 16/2455 (2019.01); G06F 16/90 (2019.01); G06F 16/901 (2019.01); G06F 16/95 (2019.01); G06F 16/9535 (2019.01); G06Q 30/02 (2023.01); G06Q 30/0201 (2023.01); G06Q 50/00 (2012.01)
CPC G06F 16/35 (2019.01) [G06F 11/3409 (2013.01); G06F 16/24568 (2019.01); G06F 16/9024 (2019.01); G06F 16/9535 (2019.01); G06Q 30/0201 (2013.01); G06Q 50/01 (2013.01)] 19 Claims
OG exemplary drawing
 
1. A method for identification of high-value information in data streams, comprising:
at a computer system including a plurality of processors and memory storing programs for execution by the processors:
receiving a plurality of filter graphs, each filter graph including a plurality of filter nodes interrelated by a plurality of graph edges, each filter node representing a classification model;
performing a continuous monitoring process for a data stream that includes a plurality of data packets from a plurality of sources, including:
without user intervention, in response to receiving the data stream with the plurality of data packets, distributing the plurality of data packets to inputs of the plurality of filter graphs; and
identifying whether each of the plurality of data packets includes high-value information related to a particular concept with regard to the respective filter graphs, based on parallel execution of the filter nodes included in the respective filter graphs, by applying predefined criteria associated with the particular concept with respect to classification models of corresponding filter nodes to text content and author information associated with the plurality of data packets;
wherein applying the predefined criteria to the text content and the author information of each data packet of the plurality of data packets includes:
executing one or more textual filters on the text content of the data packets and one or more author filters on the author information of the data packets in accordance with a first of the classification models; and
determining whether to tag each of the plurality of data packets with an identifier of the first classification model or to tag each of the plurality of data packets as rejected with the identifier of the first classification model based on whether each of the plurality of data packets is accepted by the first classification model;
upon determining that at least one of the plurality of data packets includes high-value information related to the particular concept, generating statistical information related to the high-value information; and
upon determining that the statistical information meets or exceeds a predetermined threshold, triggering an alarm, including sending a notification to a designated recipient indicating that the predetermined threshold has been met or exceeded.