Statistical message classifier
View Patent ↗A system and method are disclosed for improving a statistical message classifier. A message may be tested with a machine classifier, wherein the machine classifier is capable of making a classification on the message. In the event the message is classifiable by the machine classifier, the statistical message classifier is updated according to the reliable classification made by the machine classifier. The message may also be tested with a first classifier. In the event that the message is not classifiable by the first classifier, it is tested with a second classifier, wherein the second classifier is capable of making a second classification. In the event that the message is classifiable by the second classifier, the statistical message classifier is updated according to the second classification.
1. A method for processing a message, comprising:
maintaining whitelisted messages in memory, the whitelisted messages sent over a communication network to an intended recipient;
maintaining a table of message features in memory, wherein each message feature corresponds to a good count and a spam count, the good count based on a number of times the message feature is associated with a previously received message determined not to be unsolicited and the spam count based on a number of times the message feature is associated with a previously received message determined to be unsolicited;
maintaining a table of user classifications of previously received messages in memory;
receiving a message sent over the communication network;
executing instructions stored in memory, wherein execution of the instructions by a processor:
parses the message to identify a message feature,
determines the number of times the message feature has appeared in previously received whitelisted messages, wherein each of the previously received whitelisted messages is associated with an address to which the intended recipient has previously sent a message,
computes a score for the message feature identified in the received message based on:
the good count and spam count associated with the identified message feature,
a weighted version of the user classification of received messages associated with the identified message feature, and
the number of times the message feature has appeared in previously received whitelisted messages, and
processes the received message according to the computed score.
2. The method of claim 1 , wherein the feature includes a word used to determine that the message is a spam message.
3. The method of claim 1 , wherein each user classification indicates that the previously received message is junk or unjunk.
4. The method of claim 1 , wherein the score indicates that the received message is an unsolicited message, and processing the received message includes deleting the received message.
5. The method of claim 1 , wherein the score indicates that the received message is an unsolicited message, and processing the received message includes quarantining the received message.
6. The method of claim 1 , wherein the score indicates that the received message is not an unsolicited message, and processing the message includes delivering the message to an intended recipient.
7. The method of claim 6 , wherein the execution of instructions by the processor further updates a statistical classifier.
8. The method of claim 1 , wherein the previously received whitelisted messages are stored in an address book associated with the intended recipient of the message.
9. A system for processing a message, comprising:
memory for storing:
whitelisted messages sent to an intended recipient;
a table of message features, wherein each message feature corresponds to a good count and a spam count, the good count based on a number of times the message feature is associated with a previously received message determined not to be unsolicited and the spam count based on a number of times the message feature is associated with a previously received message determined to be unsolicited, and
a table of user classifications of previously received messages;
a processor to execute a program stored in memory, wherein execution of the program by the processor:
determines the number of times the message feature has appeared in previously received whitelisted messages, wherein each of the previously received whitelisted messages is associated with an address to which the intended recipient has previously sent a message,
computes a score for the message feature identified in the received message based on:
the good count and spam count associated with the identified message feature,
a weighted version of the user classification of received messages associated with the identified message feature, and
the number of times the message feature has appeared in previously received whitelisted messages, and
processes the received message according to the computed score.