Normalizing ingested signals

Abstract

The present invention extends to methods, systems, and computer program products for normalizing ingested signals. In general, different types of raw signals including source data in different pluralities of data dimensions and including other characteristics are ingested. Per raw signal, a transdimensionality transform is applied to recode and normalize the source data into a normalized signal that includes normalized data in a common reduced plurality of dimensions including a time dimension, a location dimension, and a context dimension. Normalization can include inferring signal annotations from the source data and using the annotations and/or the other characteristics to derive time, location, and context dimensions. Derivation can include computing a probability of a real-world event and including the probability in the context dimension. An real-world event is detection from the normalized data in the time, location, and context dimensions and an entity is notified of the real-world event.

1 . A method comprising:
 ingesting a raw signal including source data in a plurality of data dimensions;   applying a transdimensionality transform to the raw signal recoding and normalizing the source data into a normalized signal that includes normalized data in a common reduced plurality of dimensions including a time dimension, a location dimension, and a context dimension, comprising:
 inferring a signal annotation from the source data and other signal characteristics of the raw signal; and 
 deriving the time dimension, the location dimension, and the context dimension from a combination of the other signal characteristics and the signal annotation, the deriving including at least:
 computing a single source probability value for the raw signal, from a plurality of the other signal characteristics of the raw signal, that at least approximates a probability that the raw signal actually indicates an occurrence of a real-world event type; and 
 inserting the probability into the context dimension of the normalized data; 
 
   detecting a real-world event of the real-world event type from the normalized data in the time dimension, the location dimension, and the context dimension including at least the single source probability value; and   notifying an entity about the real-world event.   
     
     
         2 . The method of  claim 1 , further comprising accessing the transdimensionality transform defined and structured in a normalization dimensional model. 
     
     
         3 . The method of  claim 1 , further comprising:
 ingesting another raw signal including other source data in a further plurality of data dimensions; and   applying the transdimensionality transform to the other raw signal recoding and normalizing the other source data into another normalized signal that includes other normalized data in the common reduced plurality of dimensions including the time dimension, the location dimension, and the context dimension; and   wherein detecting the real-world event of the real-world event type comprises detecting the real-world event from the other normalized data in the time dimension, the location dimension, and the context dimension included in the other normalized signal.   
     
     
         4 . The method of  claim 3 , wherein ingesting a raw signal comprises ingesting the raw signal from a social media network source; and
 wherein ingesting another raw signal comprises ingesting the other raw signal from a source other than the social media network source.   
     
     
         5 . The method of  claim 4 , wherein ingesting the other raw signal from a source other than the social media network source comprises ingesting the other raw signal from one of: a camera feed, a listening device feed, weather data, IoT device data, crowd sourced traffic information, a 911 call, satellite data, air quality sensor data, smart city sensor data, or public radio communication. 
     
     
         6 . The method of  claim 1 , wherein deriving the time dimension, the location dimension, and the context dimension comprises:
 computing probability details indicating a probabilistic model used to calculate the probability; and   including the probability details in the context dimension; and   wherein detecting the real-world event of the real-world event type comprises detecting the real-world event from the probability and the probability details.   
     
     
         7 . The method of  claim 1 , wherein computing a probability value comprises computing a probability value that at least approximates a probability of a real-world event type selected from among: a fire, police presence, an accident, a natural disaster, weather, a shooter, a concert, or a protest;
 wherein detecting a real-world event of the real-world event type comprises detecting the real-world event of the real-world event type selected from among: the fire, the police presence, the accident, the natural disaster, the weather, the shooter, the concert, or the protest; and   wherein notifying an entity about the real-world event comprises notifying one of: a person, a business entity, or a governmental agency.   
     
     
         8 . A computer system comprising:
 a processor;   system memory coupled to the processor and storing instructions configured to cause the processor to:
 ingest a raw signal including source data in a plurality of data dimensions; 
 apply a transdimensionality transform to the raw signal recoding and normalizing the source data into a normalized signal that includes normalized data in a common reduced plurality of dimensions including a time dimension, a location dimension, and a context dimension, comprising:
 infer a signal annotation from the source data and other signal characteristics of the raw signal; and 
 derive the time dimension, the location dimension, and the context dimension from both the other signal characteristics and the signal annotation, the deriving including at least:
 compute a single source probability value for the raw signal, from a plurality of the other signal characteristics of the raw signal, that at least approximates a probability that the raw signal actually indicates an occurrence of a real-world event type; and 
 insert the probability into the context dimension of the normalized data; 
 
 
 detect a real-world event of the real-world event type from the normalized data in the time dimension, the location dimension, and the context dimension including at least the single source probability value; and 
 notify an entity about the real-world event. 
   
     
     
         9 . The computer system of  claim 8 , further comprising instructions configured to access the transdimensionality transform defined and structured in a normalization dimensional model. 
     
     
         10 . The computer system of  claim 8 , further comprising instructions configured to:
 ingest another raw signal including other source data in a further plurality of data dimensions; and   apply the transdimensionality transform to the other raw signal recoding and normalizing the other source data into another normalized signal that includes other normalized data in the common reduced plurality of dimensions including the time dimension, the location dimension, and the context dimension; and   wherein instructions configured to detect the real-world event of the real-world event type comprises instructions configured to detect the real-world event from the other normalized data in the time dimension, the location dimension, and the context dimension included in the other normalized signal.   
     
     
         11 . The computer system of  claim 8 , wherein instructions configured to ingest a raw signal comprise instructions configured to ingest the raw signal from a social media network source; and
 wherein instructions configured to ingest another raw signal comprises instructions configured to ingest the other raw signal from a source other than the social media network source.   
     
     
         12 . The computer system of  claim 11 , wherein instructions configured to ingest the other raw signal from a source other than the social media network source comprise instructions configured to ingest the other raw signal from one of: a camera feed, a listening device feed, weather data, IoT device data, crowd sourced traffic information, a 911 call, satellite data, air quality sensor data, smart city sensor data, or public radio communication. 
     
     
         13 . The computer system of  claim 8 , wherein instructions configured to derive the time dimension, the location dimension, and the context dimension comprise instructions configured to:
 compute probability details indicating a probabilistic model used to calculate the probability; and   including the probability details in the context dimension; and   wherein instructions configured to detect the real-world event of the real-world event type comprise instructions configured to detect the real-world event from the probability and the probability details.   
     
     
         14 . The computer system of  claim 8 , wherein instructions configured to compute a probability value comprises instructions configured to compute a probability value that at least approximates a probability of a real-world event type selected from among: a fire, police presence, an accident, a natural disaster, weather, a shooter, a concert, or a protest;
 wherein instructions configured to detect a real-world event of the real-world event type comprise instructions configured to detect the real-world event of the real-world event type selected from among: the fire, the police presence, the accident, the natural disaster, the weather, the shooter, the concert, or the protest; and   wherein instructions configured to notify an entity about the real-world event comprise instructions configured to notify one of: a person, a business entity, or a governmental agency.   
     
     
         13 - 20 . (canceled) 
     
     
         21 . The method of  claim 1 , wherein the single source probability value is derived based on a consideration of one or more of a source, a type, an age, or a content of the normalized signal. 
     
     
         22 . The method of  claim 21 , further comprising:
 ingesting a second raw signal including second source data in a plurality of data dimensions;   applying a transdimensionality transform to the second raw signal to generate a second normalized signal;   deriving a second time dimension, a second location dimension, and a second context dimension for the second normalized signal; and   computing a second single source probability for the second raw signal from the second normalized signal, wherein the real-world event of the real-world event type is detected using both the single source probability value for the raw signal and the second single source probability for the second raw signal.   
     
     
         23 . The method of  claim 22 , wherein the normalized signal and the second normalized signal are derived from different raw signal source types. 
     
     
         24 . The method of  claim 23 , wherein the raw signal is one of a social signal, a web signal, or a streaming signal and the second raw signal is a different one of the social signal, the web signal, or the streaming signal. 
     
     
         25 . The method of  claim 1 , wherein the single source probability value is derived from signal characteristics within the context dimension of the normalized signal. 
     
     
         26 . The method of  claim 1 , further comprising updating the single source probability value based on a time-decay function.