Automatic contacts sorting
View Patent ↗To automatically sort multiple contacts of a user, in some examples, a system may be configured to monitor physiological signals, which reflect the emotional responses, of the user during communications between the user and his/her contacts and, further, to classify the contacts into multiple contact groups that may be sorted by the emotional responses of the user.
1. A method for categorizing contacts, comprising:
generating a quality of experience (QoE) vector space for each of one or more contacts associated with a user;
monitoring multiple communications between the user and each of the one or more contacts;
collecting one or more physiological signals from the user during each of the multiple monitored communications;
classifying each of the multiple monitored communications into multiple predetermined classifications;
updating the QoE vector space for each of the one or more contacts based on the collected physiological signals;
categorizing the one or more contacts into multiple contact groups in accordance with the updated QoE vector spaces, the categorizing including:
randomly selecting multiple QoE vector spaces, from the updated QoE vector spaces, as multiple centers,
computing a distance value between each unselected QoE vector space and each of the multiple centers,
clustering each unselected QoE vector space to one of the multiple centers that corresponds to a minimum one of the computed distance values to form the multiple contact groups,
calculating a new center for each of the multiple contact groups,
calculating a mean square value for each of the multiple contact groups,
reverting to the computing of the distance value if the calculated mean square value is larger than a predetermined mean square value, and
validating the categorizing if the calculated mean square value is less than or equal to the predetermined mean square value; and
sorting the categorized contact groups based on the updated QoE vector spaces.
2. The method of claim 1 , wherein the generated QoE vector space for each of the one or more contacts associated with the user includes multiple vectors in respective multiple dimensions, each of which corresponds to one of the multiple predetermined classifications.
3. The method of claim 2 , wherein each of the multiple vectors includes one or more elements, each of which corresponds to one of the one or more physiological signals.
4. The method of claim 1 , wherein the multiple monitored communications include at least one of telephone conversation, a text message exchange, an email exchange, or a video chat.
5. The method of claim 1 , further comprising identifying at least one of the one or more contacts based on at least one of an account name, an IP address, a phone number, a user name, a facial image, and a voice of the user.
6. The method of claim 1 , wherein the multiple predetermined classifications include at least work, study, travel, and entertainment.
7. The method of claim 1 , wherein each of the multiple predetermined classifications is associated with one or more topical terms.
8. The method of claim 7 , wherein the classifying includes extracting one or more key words from each of the multiple communications.
9. The method of claim 8 , wherein the classifying further includes calculating a semantic relatedness value between the one or more topical terms and the one or more extracted key words.
10. The method of claim 1 , wherein the one or more physiological signals include at least one of blood pressure, breath frequency, pulse, voice, facial expression, or brain activities.
11. A system, comprising:
a QoE detector configured to collect one or more physiological signals from a user during each of multiple communications between the user and multiple contacts;
a communication monitor configured to monitor the multiple communications;
a vector generator configured to generate a QoE vector space for each of the multiple contacts; and
a processor coupled to a memory storing executable components, the processor operable to execute or facilitate execution of one or more of the executable components, the executable components comprising:
a classifier configured to classify the multiple monitored communications in accordance with multiple predetermined classifications;
an update manager configured to update the QoE vector space for each of the multiple contacts based on the collected one or more physiological signals;
a categorizer configured to:
categorize the multiple contacts into multiple friends groups in accordance with the updated QoE vector spaces,
randomly select multiple QoE vector spaces, from the updated QoE vector spaces, as multiple centers,
compute a distance value between each of the unselected QoE vector spaces and each of the multiple centers,
cluster each of the unselected QoE vector spaces to one of the multiple centers that corresponds to a minimum one of the computed distance values to form the multiple contact groups,
calculate a new center for each of the multiple contact groups,
calculate a mean square value for each of the multiple contact groups,
revert to compute the distance value if the mean square value is larger than a predetermined mean square value, and
end if the mean square value is less than or equal to the predetermined mean square value; and
a sorter configured to sort the contact groups based on the updated vector spaces.
12. The system of claim 11 , wherein the QoE detector includes a camera, a microphone, a sphygmomanometer, an electroencephalography monitor, a heart rate monitor, or a combination thereof.
13. The system of claim 11 , wherein the QoE vector space includes multiple vectors in respective multiple dimensions, each of which corresponds to one of the multiple predetermined topics.
14. The system of claim 11 , wherein each of the multiple predetermined classifications is associated with one or more topical terms.
15. The system of claim 14 , wherein classifier is further configured to extract one or more key words from each of the multiple communications.
16. The system of claim 15 , wherein the classifier is further configured to calculate a semantic relatedness value between the one or more topical terms and the one or more extracted key words.
17. A non-transitory computer-readable medium that stores executable-instructions that, when executed, cause one or more processors to perform operations comprising:
monitoring multiple communications between a user and one or more contacts;
collecting one or more physiological signals from the users during each of the multiple monitored communications;
generating a QoE vector space for each of the one or more contacts based on the one or more physiological signals;
categorizing the one or more contacts into multiple contact groups in accordance with the QoE vector spaces, the categorizing including:
randomly selecting multiple QoE vector spaces, from the updated QoE vector spaces, as multiple centers,
computing a distance value between each of the unselected QoE vector spaces and each of the multiple centers,
clustering each of the unselected QoE vector spaces to one of the multiple centers that corresponds to a minimum one of the computed distance values to form the multiple contact groups,
calculating a new center for each of the multiple contact groups,
calculating a mean square value for each of the multiple contact groups,
reverting to the computing of the distance value if the calculated mean square value is larger than a predetermined mean square value, and
validating the categorizing if the mean square value is less than or equal to the predetermined mean square value; and
sorting the categorized contact groups based on the QoE vector spaces.
18. The non-transitory computer-readable medium of claim 17 , further comprising classifying the multiple communications in accordance with multiple predetermined classifications.
19. The non-transitory computer-readable medium of claim 18 , further comprising:
associating one or more topical terms with each of the multiple predetermined topics;
extracting one or more key words from each of the multiple communications; and
calculating a semantic relatedness value between the one or more topical terms and the one or more extracted key words.