Apparatus and method for identifying constituents in a social network
View Patent ↗A social analytic system collects signals from different social network accounts. Social metrics are derived for the accounts and the accounts classified as different types of constituents for a company or primary account based on the social metrics. The constituents may include any combination of advocates, detractors, influencers, spammers, employees, partners, and/or market. Some of the social metrics used for classifying the different types of constituents may include a volume of the signals, types of message interactions, number of unique messages, sentiment, number of subscribers, alignment of constituent and company messages, and/or average signal length.
1. A method for a model-based social analytic system including a database system for accessing and processing data collected from different social networks to identify different candidate accounts, the method comprising:
using the database system to access the different social networks via a network to extract and store different signals including different posted messages;
identifying with the database system the signals associated with a company account;
identifying with the database system the signals associated with a candidate account of the different candidate accounts, the candidate account being an account within the company account or regarding a topic associated with the company account;
identifying with the database system a number of the signals generated by the candidate account regarding the topic;
identifying with the database system a sentiment for the signals associated with the candidate account regarding the topic;
identifying with the database system the candidate account as an advocate account when the sentiment for the signals associated with the candidate account regarding the topic is positive and the number of the signals generated by the candidate account regarding the topic is above a first threshold;
identifying with the database system the candidate account as an detractor account when the sentiment for the signals associated with the candidate account regarding the topic is negative and the number of the signals generated by the candidate account regarding the topic is above the first threshold;
identifying with the database system the candidate account as an influencer account when the number of the signals generated by the candidate account regarding the topic is above a second threshold lower than the first threshold and the candidate account has a third threshold number of subscribers;
counting with the database system a number of original messages generated and posted by the candidate account for a time period;
counting with the database system a number of total messages sent by the candidate account for the time period; and
classifying with the database system the candidate account as one of the advocate account, the detractor account, and the influencer account based on a ratio of the number of original messages generated and posted by the candidate account for the time period, and the number of total messages sent by the candidate account for the time period.
2. The method of claim 1 , further comprising:
identifying different types of social interactions for the signals associated with the candidate account; and
classifying the candidate account based on the types of social interactions for the signals associated with the candidate account.
3. The method of claim 2 , wherein the different types of social interactions comprise posted messages and echoed messages.
4. The method of claim 1 , further comprising:
generating a first term vector from the signals associated with the company account;
generating a second term vector from the signals associated with the candidate account; and
detecting an alignment between the first term vector and the second term vector.
5. The method of claim 1 , further comprising:
identifying a ratio of the number of original signals generated by the candidate account to a number of the original signals and echoed signals sent by the candidate account;
classifying the candidate account as the detractor based on a first value for the ratio; and
classifying the candidate account as the advocate based on a second value for the ratio lower than the first value.
6. The method of claim 1 , further comprising:
identifying signal lengths for the signals associated with the candidate account; and
identifying the candidate account as the advocate account or a spam account based on the signal lengths.
7. The method of claim 1 , further comprising:
identifying a ratio of unique signals written by the candidate account to a combination of the unique signals and other signals repeated or forwarded by the candidate account;
increasing an advocate score when the ratio is above a first amount and increasing a spam score when the ratio is below a second amount lower than the first amount;
increasing the advocate score for the signals associated with the candidate account aligned with the signals associated with the company account;
increasing the spam score for the signals associated with the candidate account not aligned with the signals associated with the company account;
identifying an average length of the signals associated with the candidate account;
increasing the advocate score when the average length of the signals is above a first length;
increasing the spam score when the average length of the signals is above a second length less than the first length; and
identifying the candidate account as the advocate account or a spam account based on the advocate score and the spam score.
8. The method of claim 1 , further comprising ranking the company account with other company accounts based on a number of different types of constituents associated with the company account.
9. The method of claim 1 , further comprising:
ranking the advocate, detractor, or influencer account with other, detractor, or influencer accounts associated with the company account based on the number of signals, the sentiment, and the number of subscribers associated with the candidate account.
10. A database system to access and process data collected from different social networks to identify different candidate accounts, the database system comprising:
a processor coupled to a memory;
the processor configured to:
access the different social networks via a network to extract and store different signals including different posted messages;
identify the signals associated with a company account;
identify the signals associated with a candidate account of the different candidate accounts, the candidate account being an account within the company account or regarding a topic associated with the company account;
identify a number of the signals generated by the candidate account regarding the topic;
identify a sentiment for the signals associated with the candidate account regarding the topic;
identify the candidate account as an advocate account when the sentiment for the signals associated with the candidate account regarding the topic is positive and the number of the signals generated by the candidate account regarding the topic is above a first threshold;
identify the candidate account as an detractor account when the sentiment for the signals associated with the candidate account regarding the topic is negative and the number of the signals generated by the candidate account regarding the topic is above the first threshold;
identify the candidate account as an influencer account when the number of the signals generated by the candidate account regarding the topic is above a second threshold lower than the first threshold and the candidate account has a third threshold number of subscribers;
count a number of original messages generated and posted by the candidate account for a time period;
count a number of total messages sent by the candidate account for the time period; and
classify the candidate account as one of the advocate account, the detractor account, and the influencer account based on a ratio of the number of original messages generated and posted by the candidate account for the time period, and the number of total messages sent by the candidate account for the time period.
11. The database system of claim 10 , wherein the processor is further configured to:
identify different types of social interactions for the signals associated with the candidate account; and
classify the candidate account based on the types of social interactions for the signals associated with the candidate account.
12. The database system of claim 11 , wherein the different types of social interactions comprise posted messages and echoed messages.
13. The database system of claim 10 , wherein the processor is further configured to:
generate a first term vector from the signals associated with the company account;
generate a second term vector from the signals associated with the candidate account; and
detect an alignment between the first term vector and the second term vector.
14. The database system of claim 10 , wherein the processor is further configured to:
identify a ratio of the number of original signals generated by the candidate account to a number of the original signals and echoed signals sent by the candidate account;
classify the candidate account as the detractor based on a first value for the ratio; and
classify the candidate account as the advocate based on a second value for the ratio lower than the first value.
15. The database system of claim 10 , wherein the processor is further configured to:
identify signal lengths for the signals associated with the candidate account; and
identify the candidate account as the advocate account or a spam account based on the signal lengths.
16. The database system of claim 10 , wherein the processor is further configured to:
identify a ratio of unique signals written by the candidate account to a combination of the unique signals and other signals repeated or forwarded by the candidate account;
increase an advocate score when the ratio is above a first amount and increase a spam score when the ratio is below a second amount lower than the first amount;
increase the advocate score for the signals associated with the candidate account aligned with the signals associated with the company account;
increase the spam score for the signals associated with the candidate account not aligned with the signals associated with the company account;
identify an average length of the signals associated with the candidate account;
increase the advocate score when the average length of the signals is above a first length;
increase the spam score when the average length of the signals is above a second length less than the first length; and
identify the candidate account as the advocate account or a spam account based on the advocate score and the spam score.
17. The database system of claim 10 , wherein the processor is further configured to rank the company account with other company accounts based on a number of different types of constituents associated with the company account.
18. The database system of claim 10 , wherein the processor is further configured to:
rank the advocate, detractor, or influencer account with other, detractor, or influencer accounts associated with the company account based on the number of signals, the sentiment, and the number of subscribers associated with the candidate account.