Method and system for clickpath funnel analysis
AssigneeMicrosoft Corporation
InventorsTeresa Mah, Hendricus D.J. Hoek, Ying Li
Life of this patent
Estimating the term…
Abstract
A system and method for extracting funnels from at least one input clickstream (CS) representing an ordered path of successively viewed web pages. The invention stores the CS in one or more tree structures in one scan, then traverses each tree structure to identify funnels satisfying input criteria, if any. The input criteria includes depth criteria, width criteria, starting page criteria, and end page criteria.
Claims31
A method for extracting at least one funnel from at least one input clickstream (CS), said CS representing an ordered path of web pages P1 to PN successively viewed by a user, said method comprising:
- storing each ordered path within the CS in one of up to N tree structures, each of said tree structures having a root node Pi corresponding to one of pages P1 to PN and having successive child nodes corresponding to the successive pages after Pi of Pi+1 to Pi+X, said X representing a depth criterion;
- and
- extracting a list of stored paths from each tree structure from the root node Pi to each end node to represent a set of funnels corresponding to the CS.
A method for analysis of funnels in at least one input clickstream (CS), said method comprising:
- storing one or more paths within the CS satisfying a first input criterion in one or more tree structures;
- and
- analyzing each tree structure to identify any of the paths that satisfy a second input criterion, said second input criterion representing a width criterion.
A method for extracting at least one funnel from one or more input clickstreams, each of said clickstreams representing an ordered path of web pages successively viewed by a user, said method comprising:
- creating a tree structure for storing the ordered path for each of the clickstreams, each tree structure having a root node corresponding to the first page of the clickstream and having a child node corresponding to each of the successive pages in the clickstream;
- for each input starting page, searching each tree structure to identify any stored path that starts at a node associated with the input starting page and satisfies an input depth requirement, said input depth requirement representing a desired number of pages in each path;
- storing each identified path in a temporary tree;
- and
- recursing through the temporary tree to identify any path that satisfies an input width requirement, said input width requirement representing a retention rate.
A computer-readable medium having computer-executable components for extracting at least one funnel from at least one input clickstream (CS) representing an ordered path of successively viewed pages P1 to PN of a user, said components comprising:
- a repository component for storing each ordered path within the CS in one of up to N tree structures, each of said tree structures having a root node Pi corresponding to one of pages P1 to PN and having successive child nodes corresponding to the successive pages after Pi of Pi+1 to Pi+X, said X representing a depth criterion;
- a support component for incrementing a counter associated with each node in each ordered path as the ordered path is stored;
- a funnel component for extracting a list of stored paths from each tree structure from the root node Pi to each end node to represent a set of funnels corresponding to the CS;
- and
- a criteria component for analyzing each stored path in each tree structure using the counters to identify the stored paths that satisfy one or more input criteria.
A computer-readable medium having stored thereon a data structure for a particular node in a tree structure storing at least one click path from one or more input clickstreams representing an ordered path of successively viewed web pages P1 to PN of a user, said tree structure having a root node, each particular node associated with one of the viewed web pages, said data structure comprising:
- a first field storing a page name representing a name of the viewed web page associated with the particular node;
- and
- a second field storing a support value representing a frequency of appearance for a particular path starting with the root node of the tree structure and including the particular node.
A method for extracting at least one funnel from at least one input clickstream (CS) representing an ordered path of successively viewed pages P1 to PN, said method comprising:
- reading through CS from P1 to PX, wherein X is less than or equal to N and represents an input depth;
- creating a first tree with a root node associated with page P1, and with successive child nodes associated with pages P2 to PX, wherein PX represents a child node with parent PX−1;
- incrementing a counter associated with each node in the first tree as the node is created;
- creating a second tree with a root node associated with page P2, and with successive child nodes associated with pages P3 to PX+1;
- incrementing a counter associated with each node in the second tree as the node is created;
- creating additional trees rooted at each page PN−X+2 to PN for all subpaths in CS starting with pages PN−X+2 to PN;
- storing the subpaths that start at each page and ending at PN in the respective tree so that new trees are created only when the trees or nodes have not already been created;
- and
- running through all paths in each tree to extract and output only paths that satisfy input depth and input width criteria.
Dependencies are inferred from the claim text.