Determining engagement scores for sub-categories in a digital domain by a computing system

Abstract

In general, techniques are described to determine engagement scores representative of a level of engagement in a digital domain for a particular sub-category within the common category of entities on a social media platform. In accordance with these techniques, a computing system is configured to receive, from one or more client devices, messages composed by one or more users of the one or more client devices. Each of the messages includes a respective identifier, and each respective identifier is associated with a common category of entities. The computing system is further configured to determine, based on the messages, an engagement score that represents a level of engagement for a particular sub-category within the common category of entities. The computing system is further configured to output, for display at a display device operatively connected to the computing system, a visual representation of the engagement score.

1 . (canceled) 
     
     
         2 . A system, comprising:
 at least one processor; and   at least one non-transitory computer-readable storage medium storing instructions that are executable by the at least one processor to:
 receive a specification of a common category of entities, the common category of entities associated with a common brand market; 
 receive a specification of a first entity as associated with the common category of entities and as participating in the common brand market; 
 receive, from one or more client devices, a set of messages composed by one or more users of the one or more client devices accessing a social media platform; 
 identify a first subset of messages of the set of messages as referring to the common category of entities; 
 identify, without further user input, a second entity associated with the common category of entities; 
 calculate, for the first entity and based on the first subset of messages, a first engagement score that represents a level of engagement in a digital domain for the first entity; 
 calculate, for the second entity and based on the first subset of social media messages, a second engagement score that represents a level of engagement in a digital domain for the second entity; and 
 transmit, to a display device communicably coupled to the system, an indication of the first engagement score and the second engagement score. 
   
     
     
         3 . The system of  claim 2 , wherein the instructions are further executable by the at least one processor to identify the second entity based on a database having stored thereon a specification of the common brand market and a corresponding set of entities, wherein the set of entities includes both the first entity and the second entity. 
     
     
         4 . The system of  claim 2 , wherein the instructions to identify the second entity further include instructions to:
 parse the first subset of messages to identify one or more words as a potential entity;   when a threshold number of messages of the first subset of messages includes the identified one or more words, deem the potential entity as the second entity.   
     
     
         5 . The system of  claim 4 , wherein the instructions to parse the first subset of messages further include instructions to compare text in the first subset of messages against a dictionary. 
     
     
         6 . The system of  claim 2 , wherein the instructions to calculate the first engagement score further include instructions to calculate the first engagement score based on how many messages of the first subset of messages include a learned reference to the first entity. 
     
     
         7 . The system of  claim 2 , wherein the instructions are further executable by the at least one processor to identify the first subset of messages by applying a neural network to the set of messages. 
     
     
         8 . The system of  claim 7 , wherein the instructions to apply the neural network to the social media messages further include instructions to provide, as input to the neural network, one or more feature values, the one or more feature values including one or more of the specification of the common category of entities, and the specification of the first entity. 
     
     
         9 . The system of  claim 7 , wherein the instructions to apply the neural network to the social media messages further include instructions to provide, as input to the neural network, one or more feature values, the one or more feature values including one or more of:
 a count of likes associated with each message of the set of messages;   a count of shares associated with each message of the set of messages;   a count of occurrence of a predetermined word in each message of the set of messages; or   a count of occurrence of a predetermined phrase in each message of the set of messages.   
     
     
         10 . The system of  claim 7 , wherein the instructions to apply the neural network to the social media messages further include instructions to provide, as input to the neural network, one or more feature values, and wherein the neural network includes:
 a discretization layer to convert each feature value into a corresponding range of values;   a sparse calibration layer to, for each feature value:
 normalize the corresponding range of values to generate a set of normalized feature values; and 
 apply a feature value type-specific bias value to each normalized feature value of the set of normalized feature values. 
   
     
     
         11 . The system of  claim 10 , wherein the instructions further include instructions to train the sparse calibration layer using observed engagement information for the set of users in the digital domain, and by employing backpropagation and gradient descent. 
     
     
         12 . A method, comprising:
 receiving a specification of a common category of entities, the common category of entities associated with a common brand market;   receiving a specification of a first entity as associated with the common category of entities and as participating in the common brand market;   receiving, from one or more client devices, a set of messages composed by one or more users of the one or more client devices accessing a social media platform;   identifying a first subset of messages of the set of messages as referring to the common category of entities;   identifying, without further user input, a second entity associated with the common category of entities;   calculating, for the first entity and based on the first subset of messages, a first engagement score that represents a level of engagement in a digital domain for the first entity;   calculating, for the second entity and based on the first subset of social media messages, a second engagement score that represents a level of engagement in a digital domain for the second entity; and   transmitting, to a display device, an indication of the first engagement score and the second engagement score.   
     
     
         13 . The method of  claim 12 , the identifying the second entity further comprising:
 parsing the first subset of messages to identify one or more words as a potential entity;   when a threshold number of messages of the first subset of messages includes the identified one or more words, deeming the potential entity as the second entity.   
     
     
         14 . The method of  claim 13 , the parsing the first subset of messages further including comparing text in the first subset of messages against a dictionary. 
     
     
         15 . The method of  claim 12 , the calculating the first engagement score further including calculating the first engagement score based on how many messages of the first subset of messages include a learned reference to the first entity. 
     
     
         16 . The method of  claim 12 , the identifying the first subset of messages including applying a neural network to the set of messages, the method further including providing, as input to the neural network, one or more feature values, the one or more feature values including one or more of the specification of the common category of entities, and the specification of the first entity. 
     
     
         17 . The method of  claim 12 , the identifying the first subset of messages including applying a neural network to the set of messages, the method further including providing, as input to the neural network, one or more feature values, wherein the neural network includes:
 a discretization layer to convert each feature value into a corresponding range of values;   a sparse calibration layer to, for each feature value:
 normalize the corresponding range of values to generate a set of normalized feature values; and 
 apply a feature value type-specific bias value to each normalized feature value of the set of normalized feature values. 
   
     
     
         18 . The method of  claim 17 , further comprising training the sparse calibration layer using observed engagement information for the set of users in the digital domain, and by employing backpropagation and gradient descent. 
     
     
         19 . A non-transitory computer-readable storage medium storing instructions that, when executed, cause at least one processor of a computing system to:
 receive a specification of a common category of entities, the common category of entities associated with a common brand market;   receive a specification of a first entity as associated with the common category of entities and as participating in the common brand market;   receive, from one or more client devices, a set of messages composed by one or more users of the one or more client devices accessing a social media platform;   identify a first subset of messages of the set of messages as referring to the common category of entities;   identify, without further user input, a second entity associated with the common category of entities;   calculate, for the first entity and based on the first subset of messages, a first engagement score that represents a level of engagement in a digital domain for the first entity;   calculate, for the second entity and based on the first subset of social media messages, a second engagement score that represents a level of engagement in a digital domain for the second entity; and   transmit, to a display device communicably coupled to the computing system, an indication of the first engagement score and the second engagement score.   
     
     
         20 . The non-transitory computer-readable storage medium of  claim 19 , the instructions to identify the second entity further including instructions that when executed, further cause the at least one processor to:
 parse the first subset of messages to identify one or more words as a potential entity;   when a threshold number of messages of the first subset of messages includes the identified one or more words, deem the potential entity as the second entity.   
     
     
         21 . The non-transitory computer-readable storage medium of  claim 20 , the instructions to parse the first subset of messages further including instructions that when executed, further cause the at least one processor to compare text in the first subset of messages against a dictionary. 
     
     
         22 . The non-transitory computer-readable storage medium of  claim 19 , the instructions to calculate the first engagement score further including instructions that when executed, further cause the at least one processor to calculate the first engagement score based on how many messages of the first subset of messages include a learned reference to the first entity. 
     
     
         23 . The non-transitory computer-readable storage medium of  claim 19 , the instructions to identify the first subset of messages further including instructions that when executed, further cause the at least one processor to apply a neural network to the set of messages, the instructions further including instructions causing the at least one processor to provide, as input to the neural network, one or more feature values, the one or more feature values including one or more of the specification of the common category of entities, and the specification of the first entity. 
     
     
         24 . The non-transitory computer-readable storage medium of  claim 19 , the instructions to identify the first subset of messages further including instructions that when executed, further cause the at least one processor to apply a neural network to the set of messages, the instructions further including instructions causing the at least one processor to provide, as input to the neural network, one or more feature values, wherein the neural network includes:
 a discretization layer to convert each feature value into a corresponding range of values;   a sparse calibration layer to, for each feature value:
 normalize the corresponding range of values to generate a set of normalized feature values; and 
 apply a feature value type-specific bias value to each normalized feature value of the set of normalized feature values. 
   
     
     
         25 . The non-transitory computer-readable storage medium of  claim 19 , the instructions further including instructions causing the at least one processor to train the sparse calibration layer using observed engagement information for the set of users in the digital domain, and by employing backpropagation and gradient descent.