US 12,386,904 B2
Systems and methods for measuring collected content significance
Gregory Bala, Morgan Hill, CA (US); Sebastian Brinkmann, San Francisco, CA (US); Hendrik Bartel, San Francisco, CA (US); James P. Hawley, Oakland, CA (US); Philip Kim, San Fransisco, CA (US); Yang Ruan, Oakland, CA (US); Mark Strehlow, San Fransisco, CA (US); and Faithlyn A. Tulloch, San Fransisco, CA (US)
Assigned to Truvalue Labs, Inc., San Francisco, CA (US)
Appl. No. 16/328,673
Filed by TruValue Labs, Inc., San Francisco, CA (US)
PCT Filed Aug. 29, 2017, PCT No. PCT/US2017/049221
§ 371(c)(1), (2) Date Feb. 26, 2019,
PCT Pub. No. WO2018/044955, PCT Pub. Date Mar. 8, 2018.
Claims priority of provisional application 62/380,854, filed on Aug. 29, 2016.
Prior Publication US 2019/0205341 A1, Jul. 4, 2019
Int. Cl. G06F 15/16 (2006.01); G06F 16/00 (2019.01); G06F 16/2457 (2019.01); G06F 16/2458 (2019.01); G06F 16/906 (2019.01); G06F 16/908 (2019.01); G06F 16/9536 (2019.01); G06F 16/9537 (2019.01)
CPC G06F 16/906 (2019.01) [G06F 16/00 (2019.01); G06F 16/24578 (2019.01); G06F 16/2477 (2019.01); G06F 16/908 (2019.01); G06F 16/9536 (2019.01); G06F 16/9537 (2019.01)] 13 Claims
OG exemplary drawing
 
1. A method of identifying meta-events, the method comprising:
receiving event items over a given time period, wherein each event item:
is keyed to a point in time within the given time period;
includes a content; and
is associated with a sentiment rating;
analyzing the event items to determine one or more predetermined areas of interests;
measuring 1) a number of event items within a particular area of interest within the given time period, 2) a precision of the event items within the particular area of interest within the given time period, wherein the precision is a first positive function of a numerical proximity of the sentiment rating of each event item within the particular area of interest, 3) an accuracy of the event items within the particular area of interest within the given time period, wherein the accuracy is a second positive function of a numerical measure of a textual similarity of the content of each event item within the particular area of interest, and 4) a magnitude of the event items outside the particular area of interest within the given time period, wherein:
the magnitude is a derivative of a change in a particular characteristic of the event items outside of the particular area of interest; and
the magnitude is computed as:
M(c1, . . . cn)=fM(max(s(t(c1)), . . . ,s(t(cn)))−min(s(t(c1)), . . . ,s(t(cn))))
where t(ci) is a time stamp of ith content event within a time sampling window, s(t) is a content assessment score at time t, and fM is a positively directed function;
calculating a first significance score for the particular area of interest within the given time period based on the number of event items within the particular area of interest within the given time period, the precision of the event items, the accuracy of the event items, and the magnitude of the event items;
wherein the first significance score is calculated according to at least one of the following formulas:
SI≡the first significance score in the particular area of interest≡wMM(c1, . . . cn)+wIV(c1, . . . cn)+wPP(c1, . . . cn)+wAA(c1, . . . cn),
or
SI≡the first significance score in the particular area of interest≡[wMM(c1, . . . cn)][wIV(c1, . . . cn)][wPP(c1, . . . cn)][wAA(c1, . . . cn)]
where, in the above:
M(c1, . . . cn)≡the magnitude of the event items;
V(c1, . . . cn)≡the number of event items;
P(c1, . . . cn)≡the precision of the event items; and
A(c1, . . . cn)≡the accuracy of the event items;
calculating a historical significance score for the particular area of interest over an entire duration for which event items are present;
calculating a historical significance signal by aggregating the historical significance score over the entire duration for which event items are present;
comparing the first significance score and the historical significance score;
determining that a meta-event has occurred when the first significance score exceeds the historical significance score by a threshold measure, wherein the threshold measure is determined based on a statistical value of the historical significance signal; and
providing a notification or alert to a user of a client device that the meta-event has occurred.