SPAM DETECTION FOR ONLINE SLIDE DECK PRESENTATIONS

• US 20160065605A1
• Filed: 11/19/2014
• Published: 03/03/2016
• Est. Priority Date: 08/29/2014
• Status: Abandoned Application

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: LinkedIn Corporation (Microsoft Corporation)
• Inventors: Yan, Baoshi, Guo, Jiaqi, Liu, Haishan, Amin, Mohammad Shafkat
• Application Number:
  US14/547,948
• Publication Number:
  US 20160065605A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 21/56 Computer malware detection or handling, e.g. anti-virus arrangements
• CPC Class Codes: G06F 3/04847 Interaction techniques to control parameter settings, e.g. interaction with sliders or dials
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: H04L 63/1425 Traffic logging, e.g. anomaly detection

Abstract

The disclosed systems and methods are directed to detecting spam in an electronic presentation and determining whether the electronic presentation should be moderated. The example systems and methods may employ one or more classifiers for classifying an electronic presentation and, should the electronic presentation fall within a predetermined classification, the electronic presentation may be analyzed further for the presence of spam. Further analysis of the electronic presentation may include invoking one or more filters to determine whether the electronic presentation includes words and/or phrases known to be associated with spam. Where the electronic presentation is determined to contain spam, the electronic presentation may be removed from a database of electronic presentations, excluded from search results, or flagged for moderation by a moderator.

Claims

1. A computer-implemented method comprising:

receiving an electronic presentation, the electronic presentation comprising a plurality of slides, wherein at least one slide contains content for viewing by a user;

extracting content from a slide selected from the plurality slides based on a determination that the selected slide contains content;

determining a plurality of features for each slide of the plurality of slides based on the content extracted from a corresponding slide;

assigning a classification to each slide based on the features determined for the corresponding slide, the assigned classification identifying the type of content contained within the corresponding slide;

applying a filter to each slide based on the features determined for the slide, the applied filter identifying whether the slide contains a predetermined plurality of alphanumeric characters;

determining whether each slide of the plurality of slides contains spam based on the applied filter and assigned classification to the slide;

adjusting the spam determination of each slide of the plurality of slides based on a location of a corresponding slide relative to the plurality of slides of the electronic presentation; and

determining whether the electronic presentation is spam based on the adjusted spam determination for each slide of the plurality of slides.

2. The computer-implemented method of claim 1, wherein assigning the classification to each slide is based on the application of a maximum entropy classifier to each slide.

3. The computer-implemented method of claim 1, wherein assigning the classification to each slide is based on a classification model trained for the classification.

4. The computer-implemented method of claim 1, further comprising selecting the filter from a plurality of filters to apply to each slide based on the classification assigned to the corresponding slide, wherein a first filter of the plurality of filters is associated with a first classification and a second filter of the plurality of filters is associated with a second classification.

5. The computer-implemented method of claim 1, further comprising modifying the electronic presentation based on the adjusted spam determination.

8. The computer-implemented method of claim 1, further comprising identifying the electronic presentation for moderation based on the adjusted spam determination.

6. The computer-implemented method of claim 6, wherein modifying the electronic presentation comprises removing the electronic presentation from being discoverable by a search query applied to a plurality of electronic presentations.

7. The computer-implemented method of claim 6, wherein modifying the electronic presentation comprises removing a slide selected from a plurality of slides based on the adjusted spam determination for the selected slide.

9. A system comprising:

a non-transitory, computer-readable medium storing computer-executable instructions; and

one or more processors in communication with the non-transitory, computer-readable medium that, having executed the computer-executable instructions, are configured to;

receive an electronic presentation, the electronic presentation containing a plurality of slides, wherein at least one slide contains content for viewing by a user;

for each slide of the plurality of slides, determine a plurality of features for a corresponding slide, the determined features based on content extracted from the corresponding slide;

assign at least one classification to each slide of the plurality of slides based on the features determined for the corresponding slide;

determine whether a filter is satisfied for each slide of the plurality of slides, the filter identifying whether a given slide includes a plurality of alphanumeric characters;

determine a spam value for each slide of the plurality of slides, the spam value based on the assigned classification for the corresponding slide, whether the filter was satisfied for the corresponding slide, and a location of the corresponding slide relative to the plurality of slides; and

determine an overall spam value for the electronic presentation, the overall spam value based on each spam value determined for each slide of the plurality of slides.

10. The system of claim 9, wherein the one or more processors are further configured to assign the classification to each slide based on the application of a maximum entropy classifier to each slide.

11. The system of claim 9, wherein the one or more processors are configured to assign the classification to each slide based on a classification model trained for the classification.

12. The system of claim 9, wherein the one or more processors are further configured to select the filter from a plurality of filters to apply to each slide based on the classification assigned to the corresponding slide, wherein a first filter of the plurality of filters is associated with a first classification and a second filter of the plurality of filters is associated with a second classification.

13. The system of claim 9, wherein the one or more processors are further configured to modify the electronic presentation based on the overall spam value.

14. The system of claim 13, wherein the one or more processors are configured to modify the electronic presentation by removing the electronic presentation from being discoverable by a search query applied to a plurality of electronic presentations.

15. The system of claim 13, wherein the one or more processors are configured to modify the electronic presentation by removing a slide selected from a plurality of slides based on the adjusted spam determination for the selected slide.

16. The system of claim 9, wherein the one or more processors are further configured to identify the electronic presentation for moderation based on the overall spam value.

17. A non-transitory, computer-readable medium storing computer-executable instructions thereon that, when executed by one or more processors, cause the one or more processors to perform a method, the method comprising:

receiving an electronic presentation, the electronic presentation comprising a plurality of slides, wherein at least one slide contains content for viewing by a user;

extracting content from a slide selected from the plurality slides based on a determination that the selected slide contains content;

determining a plurality of features for each slide of the plurality of slides based on the content extracted from a corresponding slide;

assigning a classification to each slide based on the features determined for the corresponding slide, the assigned classification identifying the type of content contained within the corresponding slide;

applying a filter to each slide based on the features determined for the slide, the applied filter identifying whether the slide contains a predetermined plurality of alphanumeric characters;

determining whether each slide of the plurality of slides contains spam based on the applied filter and assigned classification to the slide;

adjusting the spam determination of each slide of the plurality of slides based on a location of a corresponding slide relative to the plurality of slides of the electronic presentation; and

determining whether the electronic presentation is spam based on the adjusted spam determination for each slide of the plurality of slides.

18. The non-transitory, computer-readable medium of claim 17, wherein assigning the classification to each slide is based on the application of a maximum entropy classifier to each slide.

19. The non-transitory, computer-readable medium of claim 17, wherein the method further comprises modifying the electronic presentation based on the adjusted spam determination.

20. The non-transitory, computer-readable medium of claim 19, wherein modifying the electronic presentation comprises removing the electronic presentation from being discoverable by a search query applied to a plurality of electronic presentations.