Spam filtering in multimodal mobile communication

1. A method for filtering textual multimodal messages in the form of short messages at a mobile device, comprising:

receiving an electronic message at the mobile device, the electronic message being a short message;

enriching, via processing circuitry of the mobile device, the electronic message;

extracting, via the processing circuitry, one or more features from the electronic message;

generating, via the processing circuitry, according to an algorithm, dendritic cell signals based on the one or more features extracted from the electronic message;

subjecting, via the processing circuitry, the dendritic cell signals and the electronic message to a dendritic cell algorithm including one or more dendritic cells;

determining, via processing circuitry, a maturity of the one or more dendritic cells; and

classifying, via processing circuitry, the electronic message as spam based upon the maturity of the one or more dendritic cells,

wherein the one or more features includes one or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent,

wherein the one or more features extracted from the electronic message for said generating dendritic cell signals include a feature vector representing the electronic message,

wherein said generating dendritic cell signals according to the algorithm includes:

processing of the feature vector representing the electronic message as an input to each of a Naïve Bays (NB) algorithm and a Support Vector Machine (SVM) algorithm to generate a first decision on the classification of the electronic message with a first confidence level for the Naïve Bays (NB) algorithm and a second decision on the classification of the electronic message with a second confidence level for the Support Vector Machine (SVM) algorithm, and

performing a comparison of the first and second decisions on the classification of the electronic message to determine whether or not the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm matches the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm,

wherein when the first decision on the classification of the electronic message does not match the second decision on the classification of the electronic message, the generated dendritic cell signal is a Danger signal according to an average confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm,

wherein when the first decision on the classification of the electronic message matches the second decision on the classification of the electronic message and the first and second decisions on the classification indicate “spam,” then the generated dendritic cell signal is a PAMP signal according to a first maximum confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm, and

wherein when the first decision on the classification of the electronic message matches the second decision on the classification of the electronic message and the first and second decisions on the classification indicate “non-spam,” then the generated dendritic cell signal is a Safe signal according to a second maximum confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm.

2. The method of claim 1 , wherein the enriching includes adding additional information to the electronic message before extracting the one or more features from the content of the electronic message.

3. The method of claim 2 , wherein adding additional information to the electronic message includes adding semantic information to the content of the electronic message.

4. The method of claim 1 , wherein the one or more features includes two or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent.

5. The method of claim 1 , wherein the one or more features includes three or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent.

6. The method of claim 5 , wherein the electronic message is an instant message.

7. A mobile device configured to filter textual multimodal messages in the form of short messages at a mobile device, comprising:

a memory; and

a processor operatively coupled to the memory and configured to:

receive an electronic message, the electronic message being a short message;

enrich the electronic message;

extract one or more features from the electronic message;

generate dendritic cell signals based on the one or more features extracted from the electronic message according to an algorithm;

subject, via the processing circuitry, the dendritic cell signals and the electronic message to a dendritic cell algorithm including one or more dendritic cells;

determine a maturity of the one or more dendritic cells; and

classify the electronic message as spam based upon the maturity of the one or more dendritic cells,

wherein the one or more features includes one or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent,

wherein the one or more features extracted from the electronic message for the generation of the dendritic cell signals include a feature vector representing the electronic message,

wherein said generating dendritic cell signals according to the algorithm includes:

processing of the feature vector representing the electronic message as an input to each of a Naïve Bays (NB) algorithm and a Support Vector Machine (SVM) algorithm to generate a first decision on the classification of the electronic message with a first confidence level for the Naïve Bays (NB) algorithm and a second decision on the classification of the electronic message with a second confidence level for the Support Vector Machine (SVM) algorithm, and

performing a comparison of the first and second decisions on the classification of the electronic message to determine whether or not the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm matches the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm,

wherein when the first decision on the classification of the electronic message does not match the second decision on the classification of the electronic message, the generated dendritic cell signal is a Danger signal according to an average confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm,

wherein when the first decision on the classification of the electronic message matches the second decision on the classification of the electronic message and the first and second decisions on the classification indicate “spam,” then the generated dendritic cell signal is a PAMP signal according to a first maximum confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm, and

wherein when the first decision on the classification of the electronic message matches the second decision on the classification of the electronic message and the first and second decisions on the classification indicate “non-spam,” then the generated dendritic cell signal is a Safe signal according to a second maximum confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm.

8. The apparatus of claim 7 , wherein the processing circuitry is configured to add additional information to the electronic message before extracting the one or more features from the content of the electronic message.

9. The apparatus of claim 8 , wherein the processing circuitry is configured to add semantic information to the electronic message.

10. The apparatus of claim 7 , wherein the one or more features includes two or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent.

11. The apparatus of claim 7 , wherein the one or more features includes three or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent.

12. The apparatus of claim 11 , wherein the electronic message is an instant message.

13. A non-transitory computer readable medium storing instructions which when executed by a computer of a mobile device cause the computer to execute a process for filtering textual multimodal messages in the form of short messages at a mobile device, the process comprising:

receiving an electronic message at the mobile device, the electronic message being a short message;

enriching the electronic message using the mobile device;

extracting one or more features from the electronic message using the mobile device;

generating dendritic cell signals based on the one or more features extracted from the electronic message using the mobile device according to an algorithm;

subjecting the dendritic cell signals and the electronic message to a dendritic cell algorithm including one or more dendritic cells using the mobile device;

determining a maturity of the one or more dendritic cells using the mobile device; and

classifying the electronic message as spam based upon the maturity of the one or more dendritic cells using the mobile device,

wherein the one or more features includes one or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent,

wherein the one or more features extracted from the electronic message for said generating dendritic cell signals include a feature vector representing the electronic message,

wherein said generating dendritic cell signals according to the algorithm includes:

processing of the feature vector representing the electronic message as an input to each of a Naïve Bays (NB) algorithm and a Support Vector Machine (SVM) algorithm to generate a first decision on the classification of the electronic message with a first confidence level for the Naïve Bays (NB) algorithm and a second decision on the classification of the electronic message with a second confidence level for the Support Vector Machine (SVM) algorithm, and

performing a comparison of the first and second decisions on the classification of the electronic message to determine whether or not the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm matches the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm,

wherein when the first decision on the classification of the electronic message does not match the second decision on the classification of the electronic message, the generated dendritic cell signal is a Danger signal according to an average confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm,

wherein when the first decision on the classification of the electronic message matches the second decision on the classification of the electronic message and the first and second decisions on the classification indicate “spam,” then the generated dendritic cell signal is a PAMP signal according to a first maximum confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm, and

wherein when the first decision on the classification of the electronic message matches the second decision on the classification of the electronic message and the first and second decisions on the classification indicate “non-spam,” then the generated dendritic cell signal is a Safe signal according to a second maximum confidence level of the first decision on the classification of the electronic message for the Naïve Bays (NB) algorithm and the second decision on the classification of the electronic message for the Support Vector Machine (SVM) algorithm.

14. The non-transitory computer readable medium of claim 13 , wherein the enriching includes adding additional information to the electronic message before extracting the one or more features from the content of the electronic message.

15. The non-transitory computer readable medium of claim 14 , wherein adding additional information to the electronic message includes adding semantic information to the content of the electronic message.

16. The non-transitory computer readable medium of claim 13 , wherein the one or more features includes two or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent.

17. The non-transitory computer readable medium of claim 13 , wherein the one or more features includes three or more of existence of JavaScript, characteristics of a recipient address field, characteristics of a subject field, and a domain from which the electronic message was sent.