Spam filtering and person profiles
Systems and methods to reduce false positives in spam filtering. In one aspect, a method includes automatically generating profiles for persons identified in messages, such as incoming and outgoing e-mails. Messages flagged as spam can be checked against the profile to identify false positives.
1. A method, comprising:
determining, by at least one processor, at least one false positive spam message from a first set of messages sent to a user and classified as spam, the determining the at least one false positive spam message comprising:
querying, over a network, a computing device that supports a social network of the user,
determining, based on social data received from the computing device, that a person associated with the at least one false positive spam message is in the social network of the user, the person selected from the group consisting of sender or an additional recipient of the false positive spam message,
computing a relevancy score for a sender of the false positive spam message, the relevancy score for the sender computed based on a presence of the person in the social network of the user, the presence of the person comprising information pertaining to interactions of the person with the social network, the interactions of the person including a number of endorsements for the person by others in the social network, and a number of posts of the person in the social network, and
comparing the relevancy score of the sender to a threshold; and
in response to determining the at least one false positive spam message, causing identification of the at least one false positive spam message for the user.
2. The method of claim 1 , wherein determining the at least one false positive spam message further comprises:
ranking a plurality of persons, including the sender, to determine a respective relevancy score for each of the persons, each of the persons having previously sent a message to the user; and
comparing the relevancy score of the sender to a threshold.
3. The method of claim 2 , wherein the relevancy score for each of the persons is based at least on a number of messages from the person that the user has filed.
4. The method of claim 1 , further comprising:
scanning a second set of messages to generate a plurality of profiles for persons, the persons including the sender, each of the persons associated with at least one of the second set of messages, and the profiles including a first profile comprising a first address;
wherein the determining the at least one false positive spam message further comprises determining that a message was previously received from or sent to the first address.
5. The method of claim 4 , further comprising determining a relevancy score for each of the first set of messages, and wherein the determining the at least one false positive spam message further comprises using the relevancy score of a message in the first set of messages.
6. The method of claim 5 , wherein determining the relevancy score for each of the first set of messages comprises determining the relevancy score based on at least one action that the user has taken with the respective message.
7. The method of claim 5 , wherein determining the relevancy score for each of the first set of messages comprises increasing a relevancy score when a message is moved by the user from an inbox to a folder.
8. The method of claim 1 , wherein the determining the at least one false positive spam message further comprises determining whether a body of a message includes an address of a second person, and comparing a relevancy score of the second person to a threshold.
9. The method of claim 1 , wherein determining whether the sender is in the social network of the user comprises determining whether the sender is within a predetermined level in the social network.
10. The method of claim 1 , further comprising storing a set of addresses, and wherein the determining the at least one false positive spam message further comprises comparing an address for the sender with the set of addresses.
11. The method of claim 1 , further comprising storing a set of addresses, and wherein the determining the at least one false positive spam message further comprises comparing recipient addresses of the first set of messages with the set of addresses to select the at least one false positive spam message when at least one recipient address is in the set of addresses.
12. A system, comprising:
at least one processor; and
memory storing instructions configured to instruct the at least one processor to:
determine at least one false positive spam message from a first set of messages sent to a user and classified as spam, the determining the at least one false positive spam message comprising:
querying, over a network, a computing device that supports a social network of the user,
determining, based on social data received from the computing device, whether a sender or an additional recipient of the at least one false positive spam message is in the social network of the user,
computing a relevancy score for the sender, the relevancy score for the sender computed based on a presence of the sender or the additional recipient in the social network of the user, the presence of the sender comprising information pertaining to interactions of the sender with the social network, the interactions of the person including a number of endorsements for the person by others in the social network, and a number of posts of the person in the social network; and
comparing the relevancy score of the sender to a threshold; and
in response to determining that the recipient is in the social network of the user, cause identification of the at least one false positive spam message.
13. The system of claim 12 , wherein the instructions are further configured to instruct the at least one processor to rank a plurality of persons, including the recipient, to determine a respective relevancy score for each of the persons, and wherein determining the at least one false positive spam message further comprises comparing the relevancy score for the recipient to a threshold.
14. The system of claim 12 , wherein the instructions are further configured to instruct the at least one processor to:
scan messages to generate a plurality of profiles for persons, the persons including the recipient, the profiles including a first profile storing a first address;
wherein the determining the at least one false positive spam message further comprises determining that the user previously sent a message to the first address.
15. The system of claim 12 , wherein the instructions are further configured to instruct the at least one processor to store the received social data in a person profile of the recipient.
16. The system of claim 12 , wherein the instructions are further configured to instruct the at least one processor to determine a relevancy score for each of the first set of messages, and wherein the determining the at least one false positive spam message further comprises using the relevancy score of a message in the first set of messages.
17. A non-transitory computer-readable storage medium storing computer-readable instructions, which when executed, cause a computing system to:
determine a false positive spam message from a first set of messages sent to a user and classified as spam, the determining the false positive spam message comprising:
querying, over a network, a computing device that supports a social network of the user,
determining, based on social data received from the computing device, that a sender or additional recipient of the false positive spam message is in a social network of the user,
computing a relevancy score for the sender, the relevancy score for the sender computed based on a presence of the sender or the additional recipient in the social network of the user, the presence of the sender comprising information pertaining to interactions of the sender with the social network, the interactions of the person including a number of endorsements for the person by others in the social network, and a number of posts of the person in the social network; and
comparing the relevancy score of the sender to a threshold; and
in response to determining the false positive spam message, cause identification of the false positive spam message.
18. The computer-readable storage medium of claim 17 , wherein the determining the false positive spam message further comprises determining whether a body of the false positive spam message includes an address of a first person other than the sender, and comparing a relevancy score of the first person to a threshold.
19. The computer-readable storage medium of claim 17 , wherein the computer-readable instructions further cause the computing system to store a set of addresses, and wherein the determining the false positive spam message further comprises comparing a recipient address of the false positive spam message to the set of addresses.
20. The method of claim 1 , wherein the interactions of the person comprise data selected from a group of data consisting of: a number of friends or connections of the person in the social network and/or other social networks; an average number of friends or connections each friend or connection of the person has in the social network and/or the other social networks; a degree of connectedness of the person on a social network graph; whether the person has a public photo in the social network and/or the other social networks; and how long the person has been a member of the social network and/or the other social networks.