SCALABLE TREND DETECTION IN A PERSONALIZED SEARCH CONTEXT

• US 20160063071A1
• Filed: 08/27/2014
• Published: 03/03/2016
• Est. Priority Date: 08/27/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Guy, Ido, Hashavit, Anat, Steier, Tal, Kutiel, Gilad, Levin, Roy
• Granted Patent: US 10,210,214 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24568 Data stream processing; Continuous queries
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/248 Presentation of query results
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9538 Presentation of query results

Abstract

A computer implemented method, a computerized system and a computer program product for detecting scalable trends in a personalized search context. The computer implemented method comprising obtaining search results matching a search query, wherein the search query is a query defined by a user, wherein the search results comprise occurrences of events, wherein each occurrence is associated with a timestamp. The method further comprises calculating, by a processor, in response to obtaining the search results, trend scores for the events based on the occurrences comprised by the search results. The method further comprises providing an output to the user based on the trend scores.

Claims

1. A computer-implemented method comprising:

obtaining search results matching a search query, wherein the search query is a query defined by a user, wherein the search results comprise occurrences of events, wherein each occurrence is associated with a timestamp;

in response to obtaining the search results, calculating, by a processor, trend scores for the events based on the occurrences comprised by the search results; and

providing an output to the user based on the trend scores.

2. The computer implemented method of claim 1, wherein said calculating is performed in a linear time with respect to a number of the occurrences comprised by the search results.

3. The computer implemented method of claim 1, wherein said calculating comprises recursively calculating the trend scores for the events based on the occurrences comprised by the search results.

4. The computer implemented method of claim 3, wherein said recursive calculating is based on the following recursive formula:

trendScore(t,i)=β

·

trendScore(t,i−

1)+eventScore−

serviceRate(t,i),

wherein t is an event, wherein i is a time, wherein trendScore(t, i) is a trend score of the event at time i based on the occurrences at and before time i, wherein eventScore is a score of the event at time i based on occurrences of the event at time i, wherein serviceRate is weighted average of an arrival rate of the event t at time i, wherein β

is a constant.

5. The computer implemented method of claim 4, wherein serviceRate is calculated based on the following formula:

serviceRate(t, i)=α

·

serviceRate(t, i−

1)+(1−

α

)·

eventScore, wherein α

is a constant.

6. The computer implemented method of claim 3, wherein said recursive calculating is based on the following recursive formula:

trendScore(t,i)=β

·

trendScore(t,i−

d)+eventScore−

serviceRate(t,i)·

(GeometricSum(α

,d−

1)+1),

wherein t is an event, wherein i is a time, wherein a first occurrence of the event t is at time i, wherein a second occurrence of the event t is at time i−

d, wherein there are no additional occurrences of the event t between the first occurrence and the second occurrence, wherein trendScore(t, i_i) is a trend score of the event t at time i based on the occurrences at and before time i, wherein eventScore is a score of the event at time i based on occurrences of the event at time i, wherein serviceRate(t, i) is weighted average of an arrival rate of the event t at time i, wherein β

is a constant, wherein GeometricSum(α

, d−

1) is a geometric sum with a ratio α and

a first term d−

1.

7. The computer implemented method of claim 6, wherein serviceRate is calculated based on the following formula:

serviceRate(t,i)=α

^d·

serviceRate(t, i−

d)+(1−

α

)·

eventScore, wherein α

is a constant.

8. The computer implemented method of claim 1, wherein said calculating is performed within a timeframe of less than one minute.

9. The computer-implemented method of claim 1, wherein said calculating is performed by a single pass on the occurrences comprised by the search results.

10. The computer implemented method of claim 1 further comprises filtering the occurrences of the events based on a timeframe.

11. A computerized apparatus having a processor, the processor being adapted to perform the steps of:

obtaining search results matching a search query, wherein the search query is a query defined by a user, wherein the search results comprise occurrences of events, wherein each occurrence is associated with a timestamp;

in response to obtaining the search results, calculating trend scores for the events based on the occurrences comprised by the search results; and

providing an output to the user based on the trend scores.

12. The computerized apparatus of claim 11, wherein said calculating is performed in a linear time with respect to a number of the occurrences comprised by the search results.

13. The computerized apparatus of claim 11, wherein said calculating comprises recursively calculating the trend scores for the events based on the occurrences comprised by the search results.

14. The computerized apparatus of claim 13, wherein said recursive calculating is based on the following recursive formula:

trendScore(t,i)=β

·

trendScore(t,i−

1)+eventScore−

serviceRate(t,i),

wherein t is an event, wherein i is a time, wherein trendScore(t, i) is a trend score of the event at time i based on the occurrences at and before time i, wherein eventScore is a score of the event at time i based on occurrences of the event at time i, wherein serviceRate is weighted average of an arrival rate of the event t at time i, wherein β

is a constant.

15. The computerized apparatus of claim 14, wherein serviceRate is calculated based on the following formula:

serviceRate(t, i)=α

·

serviceRate(t, i−

1)+(1−

α

)·

eventScore, wherein α

is a constant.

16. The computerized apparatus of claim 13, wherein said recursive calculating is based on the following recursive formula:

trendScore(t,i)=β

·

trendScore(t,i−

d)+eventScore−

serviceRate(t,i)·

(GeometricSum(α

,d−

1)+1),

wherein t is an event, wherein i is a time, wherein a first occurrence of the event t is at time i, wherein a second occurrence of the event t is at time i−

d, wherein there are no additional occurrences of the event t between the first occurrence and the second occurrence, wherein trendScore(t, i_i) is a trend score of the event t at time i based on the occurrences at and before time i, wherein eventScore is a score of the event at time i based on occurrences of the event at time i, wherein serviceRate(t, i) is weighted average of an arrival rate of the event t at time i, wherein β

is a constant, wherein GeometricSum(a, d−

1) is a geometric sum with a ratio α and

a first term d−

1.

17. The computerized apparatus of claim 16, wherein serviceRate is calculated based on the following formula:

serviceRate(t, i)=α

^d·

serviceRate(t, i−

d)+(1−

α

)·

eventScore, wherein α

is a constant.

18. The computerized apparatus of claim 11, wherein said calculating is performed within a timeframe of less than one minute.

19. The computerized apparatus of claim 11, wherein said calculating is performed by a single pass on the occurrences comprised by the search results.

20. A computer program product comprising a computer readable storage medium retaining program instructions, which program instructions when read by a processor, cause the processor to perform a method comprising:

obtaining search results matching a search query, wherein the search query is a query defined by a user, wherein the search results comprise occurrences of events, wherein each occurrence is associated with a timestamp;

in response to obtaining the search results, calculating trend scores for the events based on the occurrences comprised by the search results; and

providing an output to the user based on the trend scores.