Visitor session classification based on clickstreams

Patent metadata

Field	Value
Publication number	US-10115121-B2
Application number	US-201314103056-A
Country	US
Kind code	B2
Filing date	Dec 11, 2013
Priority date	Dec 11, 2013
Publication date	Oct 30, 2018
Grant date	Oct 30, 2018

Abstract

Example systems and methods of classifying web visitor sessions based on clickstreams are presented. In one example, a plurality of web pages of a website is organized into a plurality of web page categories. A clickstream of each visitor to visit the plurality of web page categories of the website are divided into a plurality of visitor sessions. A mathematical distance between each of the plurality of visitor sessions is determined using a visitation metric based on the web page categories. Each of the visitor sessions is classified into a target group or a non-target group based on the mathematical distance between each of the visitor sessions and on an identification of at least one of the visitor sessions with an event corresponding to the target group.

First claim

What is claimed is:

1. A method of classifying web visitor sessions based on clickstreams, the method comprising:
utilizing a visitor session classification system to:
organize, utilizing a webpage categorization component of the visitor session classification system, web pages of a single website into web page categories;
divide, utilizing a visitor session determination component, clickstreams of users visiting the website into visitor sessions;
generate a matrix of weights corresponding to the web page categories relative to web page interactions to provide weightings for the web page categories, wherein each weight is applied based on a type of web page interaction, the matrix including a first weight for a first web page category relative to a first type of web page interaction;
determine, by at least one processor of a machine running an intersession distance component, an extent of similarity between a first visitor session of a first user and a second visitor session of a second user by comparing a first set of web page interactions, including the first type of web page interaction, by the first user corresponding with the first web page category to a second set of web page interactions by the second user corresponding with the first web page category, wherein the extent of similarity is determined based at least in part on applying the first weight to the first web page category relative to the first type of web page interaction;
classify, utilizing a visitor session classification component, the second visitor session into one of a target group or a non-target group based on the extent of similarity between the first visitor session and the second visitor session where the first visitor session is identified as corresponding to the target group based on an interaction with the website by the first user; and

directing targeted information related to the website to the second user when the second visitor session is classified into the target group based on the extent of similarity between the first visitor session and the second visitor session.

2. The method of claim 1 , wherein the interaction with the website comprises a purchase of at least one product or service via the website, and wherein the target group comprises a target market segment for the at least one product or service.

3. The method of claim 1 , wherein the web page categories comprise at least one of a home page category, a product page category, a user account page category, an online shopping cart page category, a search page category, and a help page category.

4. The method of claim 1 , wherein the dividing of the clickstreams of the users to visit the website into the visitor sessions comprises:
detecting, for each user gaps in the clickstream of the user that exceed a predetermined length of time; and
dividing, for each user the clickstream into visitor sessions according to the detected gaps.

5. The method of claim 1 , wherein the interaction by the first user with the first web page category includes a visitation count of webpages of the first web page category during the first visitor session.

6. The method of claim 1 , wherein the first set of web page interactions by the first user with the first web page category includes a visitation duration for webpages of the first web page category during the first visitor session.

7. The method of claim 1 , wherein the determining of the extent of similarity between the first and second visitor sessions comprises:
determining a raw value for visitation metrics of the first set of web page interactions of the first visitor session; and
computing a score for the visitation metrics based at least on the raw values of the visitation metrics for the first web page category;
determining a raw value for visitation metrics of the second set of web page interactions of the second visitor session; and
computing a score for the visitation metrics of the second visitor session based at least on the raw values of the visitation metrics for the first web page categories.

8. The method of claim 7 , wherein the scores for visitation metrics of the first and second visitor sessions comprise term frequency-inverse document frequency scores.

9. The method of claim 7 , wherein the determining of the extent of similarity between the first and second visitor sessions further comprises linearly scaling the scores for the visitation metrics of the first and second visitor sessions.

10. The method of claim 1 , wherein generating the matrix of weights includes employing large margin nearest neighbor classification.

11. The method of claim 1 , wherein generating the matrix of weights includes employing differential weights depending on a size of the target group relative to the non-target group.

12. The method of claim 1 , wherein the classifying the first visitor session into the target group or the non-target group employs a k- nearest neighbor algorithm.

13. A non-transitory computer-readable storage medium comprising instructions that, when executed by at least one processor of a machine, cause the machine to perform operations comprising:
organizing web pages of a single commercial website into web page categories:
dividing a clickstream of a first user visiting the commercial website into visitor sessions;
training a module of a visitor session classification system to apply a matrix of weights corresponding to the web page categories relative to web page interactions to provide weightings for the web page categories, wherein each weight is applied based on a type of web page interaction, the matrix including a first weight for a first web page category relative to a first type of web page interaction;
determining, using the trained module, an extent of similarity between a first visitor session of the first user and a second visitor session of the second user by comparing a first set of web page interactions, including the first type of web page interaction, by the first visitor corresponding with the first web page category to a second set of web page interactions by the second visitor corresponding with the first web page category, wherein the extent of similarity is determined based at least in part on applying the first weight to the first web page category relative to the first type of web page interaction; and
classifying, utilizing the visitor session classification system, the second visitor session into a target market group based on the extent of similarity between the first visitor session and the second visitor session where the first visitor session is identified as corresponding with the target group based on a transaction via the commercial website associated with the target group, the classification of the second visitor session into the target group resulting in targeted information related to the commercial website being sent to the second user.

14. A system comprising:
at least one processor; and
memory comprising instructions that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:
running a visitor session classification system to perform:
organizing web pages of a single website into web page categories;
dividing a clickstream of each user to visit the website into a plurality of visitor sessions;
generating, utilizing a category weight determination component, a matrix of weights corresponding to the web page categories relative to web page interactions to provide weightings for the web page categories, wherein each weight is applied based on a type of web page interaction, the matrix including a first weight for a first web page

Assignees

• Adobe Systems Inc

Inventors

• Pai Deepak
• Sharang Abhijit
• Macha Yadagiri Meghanath
• Agrawal Shradha

Classifications

• H04L67/22

  Electricity · mapped topic

• G06F17/3089

  Physics · mapped topic

• G06Q30/0204 Primary

  Market segmentation · CPC title

• H04L67/146

  Markers for unambiguous identification of a particular session, e.g. session cookie or URL-encoding · CPC title

• H04L67/02

  based on web technology, e.g. hypertext transfer protocol [HTTP] · CPC title