MULTIMEDIA SPAM DETERMINATION USING SPEECH CONVERSION

• US 20090147930A1
• Filed: 12/05/2007
• Published: 06/11/2009
• Est. Priority Date: 12/05/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Rice, Christopher Russell
• Granted Patent: US 8,144,841 B2
• Time in Patent Office: Days
• US Class Current: 379/88.120
• CPC Class Codes: H04L 65/1079 of unsolicited session attempts, e.g. SPIT
• CPC Class Codes: H04M 2201/40 using speech recognition speech recognition per se G10L15/00
• CPC Class Codes: H04M 2201/60 Medium conversion
• CPC Class Codes: H04M 3/436 Arrangements for screening incoming calls , i.e. evaluating the characteristics of a call before deciding whether to answer it based on the calling party profile H04M3/42059; based on location H04M3/42348; based on presence H04M3/42365; diversion H04M3/54
• CPC Class Codes: H04M 3/5335 Message type or catagory, e.g. priority, indication

Abstract

Determining unwanted multimedia message originated from a multimedia call. An interface receives a multimedia message from a telecommunication service operator, and multimedia message is stored on a storage area. A unified communication system manages communication message delivery to at least one of the following access points for the recipient, and the unified communication system includes a processor configured to execute computer-executable instructions for identifying a voice portion from content of the received multimedia message. An anti-spam engine is applied to analyze a textual representation of the identified voice portion, and a confidence ranking of the representation is generated in response to the analyzed representation. It is determined whether the multimedia message is an unwanted multimedia message as a function of the confidence ranking and the analyzed representation.

Claims

1. A method for determining an unwanted multimedia message originating from a multimedia call, said multimedia call being initiated by a caller for a recipient and received by a telecommunication network servicing the recipient, said telecommunication network being unable to complete the call to the recipient and prompting the caller to leave a multimedia message for the recipient, said method comprising:

receiving the multimedia message from the telecommunication network;

identifying a voice portion of the received multimedia message;

generating a textual representation of the identified voice portion;

applying an anti-spam engine to analyze the generated representation;

generating a confidence ranking of the generated representation in response to the analyzed representation; and

determining whether the multimedia message is the unwanted multimedia message as a function of the confidence ranking and the analyzed representation.

2. The method of claim 1, wherein the applying comprises applying a Bayesian filter to the generated representation, and wherein the generating comprises generating the confidence ranking of the generated representation after applying the Bayesian filter.

3. The method of claim 1, further comprising comparing the generated confidence ranking with a deletion threshold and removing the multimedia message if the confidence ranking exceeds a deletion threshold.

4. The method of claim 3, further comprising comparing the generated confidence ranking with a suspicion threshold and the deletion threshold, and generating a notification to the user when delivering the multimedia message to the user if the confidence ranking exceeds the suspicion threshold but is lower than the deletion threshold.

5. The method of claim 4, wherein the notification comprises at least one of the following:

an audio message indicating the multimedia message from the caller is considered as a unwanted message and a textual message indicating the multimedia message from the caller is considered as a unwanted message.

6. The method of claim 4, wherein the deletion threshold is determined by at least one of the following:

the user and a unified communication system, and wherein the suspicion threshold is determined by at least one of the following;

the user and the unified communication system.

7. The method of claim 1, wherein the multimedia message comprises at least one of the following:

a voice message and a video message.

8. The method of claim 1, wherein the generated representation includes at least one of the following:

a text transcript and an annotated extensible markup language (XML) file.

9. A system for determining an unwanted multimedia message originating from a multimedia call, said multimedia call being initiated by a caller for a recipient and received by a telecommunication service operator servicing the recipient, said telecommunication service operator being unable to complete the call to the recipient and prompting the caller to leave a multimedia message for the recipient, said system comprising:

an interface for receiving the multimedia message from the telecommunication service operator;

a storage area for storing the multimedia message;

a unified communication system for managing communication message delivery to at least one of the following access points for the recipient;

a mail access point, a voice access point, and a portable phone access point; and

wherein

said unified communication system includes a processor configured to execute computer-executable instructions for;

identifying a voice portion from content of the received multimedia message;

generating a textual representation of the identified voice portion;

applying an anti-spam engine to analyze the generated representation;

generating a confidence ranking of the generated representation in response to the analyzed representation; and

determining whether the multimedia message is the unwanted multimedia message as a function of the confidence ranking and the analyzed representation, and

wherein the interface provides a notification to the recipient at the access point in response to the determination.

10. The system of claim 9, wherein the processor is configured to apply a Bayesian filter to the generated representation, wherein the processor is configured to generate the confidence ranking of the generated representation after applying the Bayesian filter.

11. The system of claim 9, wherein the processor is further configured to compare the generated confidence ranking with a deletion threshold and remove the multimedia message if the confidence ranking exceeds a deletion threshold.

12. The system of claim 11, wherein the processor is configured to compare the generated confidence ranking with a suspicion threshold and the deletion threshold, and wherein the interface provides the notification to the user when delivering the multimedia message to the user if the confidence ranking exceeds the suspicion threshold but is lower than the deletion threshold.

13. The system of claim 9, wherein the interface provides the notification to the recipient as a function of the access point, wherein the notification comprises at least one of the following:

an audio message indicating the multimedia message from the caller is considered as a unwanted message and a textual message indicating the multimedia message from the caller is considered as a unwanted message.

14. The system of claim 12, wherein the deletion threshold is determined by at least one of the following:

the user and a unified communication system, and wherein the suspicion threshold is determined by at least one of the following;

the user and the unified communication system.

15. The system of claim 9, wherein the multimedia message comprises at least one of the following:

a voice message and a video message.

16. The system of claim 9, wherein the generated representation includes at least one of the following:

a text transcript and an annotated extensible markup language (XML) file.

17. A method for determining an unwanted multimedia message originating from a multimedia call, said multimedia call being initiated by a caller for a recipient and received by a telecommunication network operator servicing the recipient, said telecommunication network operator being unable to complete the call to the recipient and prompting the caller to leave a multimedia message for the recipient, said method comprising:

receiving the multimedia message from the telecommunication network operator, said multimedia message including at least one of the following;

a voice message and a video message;

identifying a voice portion of the received multimedia message;

generating a representation of the identified voice portion, said representation including at least one of the following;

a text transcript and an annotated extensible markup language (XML) file;

applying an anti-spam engine to analyze the generated representation, said anti-spam engine applying a Bayesian filter to the generated representation and generating a confidence ranking thereof in response to the analyzed representation after applying the Bayesian filter;

determining whether the multimedia message is the unwanted multimedia message as a function of the confidence ranking and the analyzed representation; and

delivering a notification indicative of the determining to a mail inbox of the recipient.

18. The method of claim 17, wherein delivering the notification comprises delivering at least one of the following:

an audio message indicating the multimedia message from the caller is considered as a unwanted message and a textual message indicating the multimedia message from the caller is considered as a unwanted message.

19. The method of claim 17, further comprising comparing the generated confidence ranking with a deletion threshold and removing the multimedia message if the confidence ranking exceeds a deletion threshold.

20. The method of claim 19, further comprising comparing the generated confidence ranking with a suspicion threshold and the deletion threshold, and generating a notification to the user when delivering the multimedia message to the user if the confidence ranking exceeds the suspicion threshold but is lower than the deletion threshold.