Analysis apparatus, analysis method and recording medium for recording analysis program

• US 8,219,673 B2
• Filed: 03/26/2009
• Issued: 07/10/2012
• Est. Priority Date: 03/28/2008
• Status: Expired due to Fees

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Uyama, Masashi, Kohda, Youji, Watanabe, Satoru
• Primary Examiner(s): Elchanti, Hussein
• Assistant Examiner(s): Anwari, Maceeh
• Application Number:
  US12/411,930
• Publication Number:
  US 20090248863A1
• Time in Patent Office: 1,202 Days
• Field of Search: 709/224
• US Class Current: 709/224
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

An analysis apparatus includes a history analysis part analyzing the trend based on the access log and deriving the latent trait to an access destination, a principal components analysis part analyzing principal components of the frequency distribution of the access, a classification part classifying the correlation of the principal components into plural classes, an event recording part recording the events to induce to the access destination, a first generating part generating a latent trait principal component table; a second generating part generating a class construction ratio table stored the class as the result of classification, a third generating part generating an event classification table stored the class and the contents of the event, and an analysis result recording part recording the latent trait principal component table, the class construction ratio table and the event classification table, as an event feature model of the effect on the execution of the event.

Claims

1. An analysis apparatus incorporated in a computer and configured to access and analyze the contents recorded in a history recording part that records access records to a plurality of access destinations accessed by user terminals via a communication network, the analysis apparatus comprising:

a history analysis part analyzing a trend of an access history for each unit period and deriving a latent trait indicating a latent possibility of access to a preset target access destination for each session from the contents recorded in the history recording part that records, for each session, an access history indicating an access or non-access to each of the plurality of access destinations and the preset target access destination, the preset target access destination being a subset destination of the plurality of access destinations indicating a level of completion relating to an access of one of the plurality of access destinations;

a principal component analysis part analyzing principal components of a frequency distribution of the latent trait derived by the history analysis part for each unit period;

a classification part classifying a correlation of the principal components for each unit period into a plurality of classes based on the analysis of the principal components;

an event recording part recording, for each unit period, contents of one or more events executed to induce an access to the preset target access destination;

a first generating part deriving a distribution of a latent trait for each principal component and generating a latent trait principal component table that stores a correspondence between each principal component and each latent trait;

a second generating part generating a class construction ratio table that stores a correspondence between each class and a composition ratio of each principal component;

a third generating part generating an event classification table that stores a correspondence between each class and the contents of each event; and

an analysis result recording part recording the latent trait principal component table generated by the first generating part, the class construction ratio table generated by the second generating part and the event classification table generated by the third generating part, as an event feature model of an effect on execution of one of the one or more events.

2. The analysis apparatus according to claim 1, further comprising:

an acceptance part accepting the input of the contents of an event scheduled to be executed; and

a latent trait prediction part deriving the prediction of the distribution of the latent traits upon execution of a scheduled event based on the contents of the scheduled event accepted and the event feature model recorded in the analysis result recording part.

3. The analysis apparatus according to claim 2,

wherein the latent trait prediction part

derives the predicted class from the contents of the scheduled event based on the event classification table,

derives the principal component construction ratio predicted from the predicted class based on the class construction ratio table, and

derives the prediction of the distribution of the latent traits from the predicted principal component construction ratio based on the latent trait principal component table.

4. The analysis apparatus according to claim 2, further comprising:

an access behavior model recording part recording the latent trait derived by the history analysis part as an access behavior model corresponding to the session associated with the access history;

an execution result analysis part deriving the frequency distribution of the latent traits in an execution result accepted in the acceptance part based on an execution result and the access behavior model recorded in the access behavior model recording part;

a prediction-achievement comparison part comparing the distribution of the latent traits predicted by the latent trait prediction part with the distribution of the latent traits derived by the execution result analysis part; and

a protective measure support part classifying the difference in the distribution of the latent traits into one of a plurality of preset types based on the difference between the prediction and the achievement of the latent traits constituting the result of comparison by the prediction-achievement comparison part.

5. The analysis apparatus according to claim 4,

wherein the history analysis part

selects the access destination having the correlation of not less than a given value with the target access destination with respect to the access or non-access from the access history for each unit period recorded in the history recording part,

sets each session in correspondence with the access or non-access to the selected access destination, and

derives the latent traits for each session based on the result of analysis of the correspondence between each session and the access or non-access to the selected access destination according to the item response theory.

6. The analysis apparatus according to claim 5,

wherein the classification part

normalizes the score of each principal component based on the result of analysis of the principal components for each unit period in accordance with the result of dividing the score of the principal component by the size coefficient providing the total absolute value of the scores of the principal components, and

classifies the correlation of the normalized scores of the principal components.

7. The analysis apparatus according to claim 4,

wherein the classification part

normalizes the score of each principal component based on the result of analysis of the principal components for each unit period in accordance with the result of dividing the score of the principal component by the size coefficient providing the total absolute value of the scores of the principal components, and

classifies the correlation of the normalized scores of the principal components.

8. The analysis apparatus according to claim 2,

wherein the history analysis part

selects the access destination having the correlation of not less than a given value with the target access destination with respect to the access or non-access from the access history for each unit period recorded in the history recording part,

sets each session in correspondence with the access or non-access to the selected access destination, and

derives the latent traits for each session based on the result of analysis of the correspondence between each session and the access or non-access to the selected access destination according to the item response theory.

9. The analysis apparatus according to claim 8,

wherein the classification part

normalizes the score of each principal component based on the result of analysis of the principal components for each unit period in accordance with the result of dividing the score of the principal component by the size coefficient providing the total absolute value of the scores of the principal components, and

classifies the correlation of the normalized scores of the principal components.

10. The analysis apparatus according to claim 2,

wherein the classification part

normalizes the score of each principal component based on the result of analysis of the principal components for each unit period in accordance with the result of dividing the score of the principal component by the size coefficient providing the total absolute value of the scores of the principal components, and

classifies the correlation of the normalized scores of the principal components.

11. The analysis apparatus according to claim 1,

wherein the history analysis part

selects the access destination having the correlation of not less than a given value with the target access destination with respect to the access or non-access from the access history for each unit period recorded in the history recording part,

sets each session in correspondence with the access or non-access to the selected access destination, and

derives the latent traits for each session based on the result of analysis of the correspondence between each session and the access or non-access to the selected access destination according to the item response theory.

12. The analysis apparatus according to claim 11,

wherein the classification part

normalizes the score of each principal component based on the result of analysis of the principal components for each unit period in accordance with the result of dividing the score of the principal component by the size coefficient providing the total absolute value of the scores of the principal components, and

classifies the correlation of the normalized scores of the principal components.

13. The analysis apparatus according to claim 1,

wherein the classification part

normalizes the score of each principal component based on the result of analysis of the principal components for each unit period in accordance with the result of dividing the score of the principal component by the size coefficient providing the total absolute value of the scores of the principal components, and

classifies the correlation of the normalized scores of the principal components.

14. An analysis method for causing a computer to function as an analysis apparatus which is configured to access and analyze contents recorded in a history recording log file that records access records to a plurality of access destinations accessed by user terminals via a communication network, the analysis method for causing the computer to execute:

analyzing a trend of an access history for each unit period and deriving the latent trait indicating a latent possibility of access to a preset target access destination for each session from the contents recorded in a history recording part that records, for each session, an access history indicating an access or non-access to each of the plurality of access destinations and the preset target access destination, the preset target access destination being a subset destination of the plurality of access destinations indicating a level of completion relating to an access of one of the plurality of access destinations;

analyzing principal components of a frequency distribution of the latent trait derived by the analyzing for each unit period;

deriving a distribution of a latent trait for each principal component and generating a latent trait principal component table that stores a correspondence between each principal component and each latent trait;

classifying a correlation of the principal components for each unit period into a plurality of classes based on the analysis result of the principal components executed in the analyzing;

recording, for each unit period, contents of one or more events executed to induce an access to the preset target access destination;

generating a class construction ratio table that stores a correspondence between each class and a composition ratio of each principal component;

generating an event classification table that stores a correspondence between each class and the contents of each event; and

recording the latent trait principal component table generated by the first generating, the class construction ratio table generated by the second generating and the event classification table generated by the third generating, as an event feature model of an effect on execution of one of the one or more events.

15. A non-transitory computer-readable recording medium storing a computer program for an analysis apparatus which is configured to access and analyze contents recorded in a history recording log file that records access records to a plurality of access destinations accessed by user terminals via a communication network, the computer program making the computer to execute:

analyzing a trend of an access history for each unit period and deriving the latent trait indicating a latent possibility of access to a preset target access destination for each session from the contents recorded in a history recording log file that records, for each session, an access history indicating an access or non-access to each of the plurality of access destinations and the preset target access destination, the preset target access destination being a subset destination of the plurality of access destinations indicating a level of completion relating to an access of one of the plurality of access destinations;

analyzing the principal components of a frequency distribution of the latent trait derived by the analyzing for each unit period;

deriving a distribution of a latent trait for each principal component and generating a latent trait principal component table that stores a correspondence between each principal component and each latent trait based on the analysis result of the principal components executed in the analyzing;

classifying a correlation of the principal components for each unit period into a plurality of classes based on the analysis result of the principal components executed in the analyzing;

generating a class construction ratio table that stores a correspondence between each class and a composition ratio of each principal component;

generating an event classification table that stores a correspondence between each class and the contents of each event; and

recording the latent trait principal component table generated by the first generating, the class construction ratio table generated by the second generating and the event classification table generated by the third generating, as an event feature model of an effect on execution of one of the one or more events.