System, method, and service for segmenting a topic into chatter and subtopics

• US 20050256905A1
• Filed: 05/15/2004
• Published: 11/17/2005
• Est. Priority Date: 05/15/2004
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Tomkins, Andrew S., Gruhl, Daniel Frederick, Guha, Ramanathan Vaidhyanath
• Granted Patent: US 7,281,022 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing
• CPC Class Codes: Y10S 707/99948 Application of database or data structure, e.g. distributed, multimedia, or image

Abstract

A topic segmenting system segments a topic into chatter and subtopics. The system decomposes a conversation into topics, producing a time-based structure for topics and subtopics in the conversation. The system extracts a large number of topics at all levels of granularity. Some of the topics extracted correspond to broad topics and some correspond to “spiky” topics or subtopics. The system comprises a process for automatically detecting spiky regions of a topic. For each possible broad topic, the present system finds regions where coverage of the broad topic overlaps significantly with the spiky region of another topic. The system then removes the spiky subtopic from the conversation. Processing is repeated until all discernable topics have been identified and removed from the conversation, yielding random topics of little duration or intensity.

Claims

1. A method of segmenting a complex topic into a plurality of spiky subtopics, comprising:

receiving a conversation containing the complex topic;

identifying a spiky subtopic in the conversation;

identifying instances associated with the spiky subtopic;

removing the instances associated with the spiky subtopic;

recursively identifying additional spiky subtopics in the conversation and removing instances associated with the additional spiky subtopics until a predetermined chatter criterion for topic segmentation has been achieved; and

selectively outputting at least some of the plurality of spiky subtopics.

2. The method of claim 1, wherein the complex topic comprises chatter.

3. The method of claim 2, wherein recursively identifying additional spiky subtopics in the conversation comprises examining the at least some of the additional spiky subtopics for further decomposition.

4. The method of claim 2, wherein the conversation contains a blog stream.

5. The method of claim 4, wherein identifying the spiky subtopic in the conversation comprises identifying the spiky subtopic in the blog stream.

6. The method of claim 5, wherein the blog stream comprises a stream of articles.

7. The method of claim 6, wherein the blog stream of articles comprises data.

8. The method of claim 6, wherein identifying the spiky subtopic in the blog stream comprises searching for a predetermined keyword.

9. The method of claim 3, wherein examining the at least some of the additional spiky subtopics for further decomposition comprises decomposing the additional spiky subtopics into at least two topics and chatter.

10. The method of claim 9, wherein examining the at least some of the additional spiky subtopics for further decomposition comprises using a heuristic approximation approach.

11. A computer program product having a plurality of executable instruction codes for segmenting a complex topic into a plurality of spiky subtopics, comprising:

a first set of instruction codes for receiving a conversation containing the complex topic;

a second set of instruction codes for identifying a spiky subtopic in the conversation;

a third set of instruction codes for identifying instances associated with the spiky subtopic;

a fourth set of instruction codes for removing the instances associated with the spiky subtopic;

a fifth set of instruction codes for recursively identifying additional spiky subtopics in the conversation and removing instances associated with the additional spiky subtopics until a predetermined chatter criterion for topic segmentation has been achieved; and

a sixth set of instruction codes for selectively outputting at least some of the plurality of spiky subtopics.

12. The computer program product of claim 11, wherein the complex topic comprises chatter.

13. The computer program product of claim 12, wherein the fifth set of instruction codes examines the at least some of the additional spiky subtopics for further decomposition.

14. The computer program product of claim 12, wherein the conversation contains a blog stream.

15. The computer program product of claim 14, wherein the second set of instruction codes identifies the spiky subtopic in the blog stream.

16. The computer program product of claim 15, wherein the blog stream comprises a stream of articles.

17. The computer program product of claim 16, wherein the blog stream of articles comprises data.

18. The computer program product of claim 16, wherein the second set of instruction codes further searches for a predetermined keyword.

19. The computer program product of claim 13, wherein the fifth set of instruction codes decomposes the additional spiky subtopics into at least two topics and chatter.

20. The computer program product of claim 19, wherein the fifth set of instruction codes examines the at least some of the additional spiky subtopics for further decomposition by means of a heuristic approximation approach.

21. A system for segmenting a complex topic into a plurality of spiky subtopics, comprising:

a topic identifier receiving a conversation containing the complex topic, and identifies a spiky subtopic in the conversation;

a topic assigner, connected to the topic identifier, for identifying instances associated with the spiky subtopic;

a topic segmenter, connected to the topic assigner, for removing the instances associated with the spiky subtopic;

wherein the topic identifier, the topic assigner, and the topic segmenter recursively identify additional spiky subtopics in the conversation and remove instances associated with the additional spiky subtopics until a predetermined chatter criterion for topic segmentation has been achieved; and

wherein at least some of the plurality of spiky subtopics are selectively outputted.

22. The system of claim 21, wherein the complex topic comprises chatter.

23. The system of claim 22, wherein the conversation contains a blog stream.

24. The system of claim 23, wherein the blog stream comprises a stream of articles.

25. The system of claim 24, wherein the blog stream of articles comprises data.

26. A method of segmenting a topic into chatter and a plurality of subtopics, comprising:

specifying a plurality of input data from which a topic is segmented;

specifying a chatter criteria by which topic segmenting is deemed to be sufficient; and

invoking a topic segmenting utility program wherein the input data and the chatter criteria are made available to recursively identify additional spiky subtopics in the conversation and to remove instances associated with the additional spiky subtopics until a predetermined chatter criterion for topic segmentation has been achieved, and further to selectively output at least some of the plurality of spiky subtopics.

27. A service for segmenting a complex topic into a plurality of spiky subtopics, comprising:

a topic identifier receiving a conversation containing the complex topic, and identifies a spiky subtopic in the conversation;

a topic assigner, connected to the topic identifier, for identifying instances associated with the spiky subtopic;

a topic segmenter, connected to the topic assigner, for removing the instances associated with the spiky subtopic;

wherein the topic identifier, the topic assigner, and the topic segmenter recursively identify additional spiky subtopics in the conversation and remove instances associated with the additional spiky subtopics until a predetermined chatter criterion for topic segmentation has been achieved; and

wherein at least some of the plurality of spiky subtopics are selectively outputted.

28. A method of segmenting a topic into a chatter and a plurality of subtopics, comprising:

specifying a plurality of input data from which a topic is segmented;

specifying chatter criteria by which segmentation is deemed to be sufficient;

invoking a topic segmentation utility program wherein the input data and the chatter criteria are made available to the topic segmentation utility program; and

selectively outputting the topic and the plurality of subtopics for at least some of the input data.