AU2006200041B2 - System and method for selecting alternative advertising inventory in place of sold out advertising inventory - Google Patents

Description

P/00/011 Regulation 3.2

AUSTRALIA

Patents Act 1990 COMPLETE SPECIFICATION STANDARD PATENT

(ORIGINAL)

Name of Applicant(s): Actual Inventor(s): Yahoo, Inc., of 701 First Avenue, Sunnyvale, California 94089, United States of America Kian-Tat LIM Address for Service: DAVIES COLLISON CAVE, Patent Trademark Attorneys, Level 10, 10 Barrack Street, Sydney, 2000, New South Wales, Australia Phone: 02 9262 2611 Fax 02 9262 1080 Attorney Code: CA Invention Title: System and method for selecting alternative advertising inventory in place of sold out advertising inventory The following statement is a full description of this invention, including the best method of performing it known to us:- SYSTEM AND METHOD FOR SELECTING ALTERNATIV ADVERTISING INVENTORY IN PLACE OF SOLD OUT ADVERTISING INVENTORY BACKGROUND OF THE INVENTION Use of the Internet by the general public is certainly gaining popularity.

More and more people are getting access to the Internet and the vast amount of information that it provides. Along with the rapid increase in the number of Internet users, advertising on the Internet has consequently become an important priority for many advertisers.

As a result, for web portals and ISPs, a significant amount of revenue can be generated from displaying advertisers' ad banners on displayed websites or web pages.

For example, for a preeminent portal such as Yahoo! which is visited daily by hundreds of thousands, if not millions, of users, considerable revenue can be made by displaying an advertiser's ads on its websites or web pages.

Generally, the advertisers pay a fee for each ad viewed .by web users.

Contracts to show ads are normally signed several weeks/months before ads get delivered. The duration of contracts ranges from one day to multiple years. Typically,.

there are several types of contracts, including regular contracts, exclusive contracts and infinite contracts. For regular contracts, the advertisers purchase a designated number of ad views on a chosen space (web page). For exclusive contracts, they purchase all the ad views on a chosen space. For infinite contracts, they purch ase all the leftover ad views on a chosen space after other regular contracts related to that space have been fulfilled.

In order to maximize their effects, ads ideally need to be placed in strategic locations, both physically and temporally, for maximum exposure to the targeted audience. Identify'ing strategic locations, therefore, is critical in planning any advertising efforts. Demographics are often used in the pr-ocess of determining strategic locations for advertising purposes. Hence, it is important to derive demographic data on a system-wide basis for web portals or ISPs which host a large number of websites or web pages. Thus, it would be desirable to develop a method and system that is capable of deriving demographic data from Internet user'behavior on a collective basis for web portals such as Yahoo!.

Furthermore, it would be beneficial if the collective demographic data can be broken down and used to identify particular websites, or web pages based on demographic groups or other characteristics, which may suit the respective purposes of specific ads. Hence, it would be desirable to develop a method and system that is capable of providing demographic data so as to allow more effective advertising strategies to be formulated for websites and web pages.

In addition, like advertising conducted through more traditional medium, such as TV or printed publications, advertising on the Internet is similarly subject to.

physical limitation. For obvious reasons, it is a natural and often most selected choice for advertisers to request ad views on the home page of a web portal or ISP. However, since there is a finite amount of physical space on a web page, demand for ad space or ad views on popular web pages often exceeds supply. Thus, a significant number of ads do not always get placed on the most desired web pages.

Ideally, in order to maximize revenue, the excess demand need to be diverted to other ad space or ad views which may be available on other web pages.

However, before advertisers can be convinced to place their ads on these other web pages, they need to be reasonably assured that their ads will similarly generate comparable beneficial results if placed on those web pages. Thus, it would be desirable to develop a method and system that is capable of providing demographic data derived from Internet user behavior relating to web pages so as to allow excess advertising demand to be channeled to other equally effective web pagei.

By way of background, the Yahoo! web pages are generally organized in a tree structure. Fig. 1 is a simplified network diagram showing the structure of the Yahoo! network. At the top of the tree is the entire Yahoo! network. Under this node are various nodes, such as Yahoo! Shopping, Yahoo! Sports, Yahoo! Yellow Pages, Yahoo! Search, etc. Under each of these nodes, there may be a variety of descendant nodes, each of which may have a variety of additional descendant nodes. For example, under the Yahoo! Sports node are the NFL, NHL, NBA, etc. nodes, and under the NBA node are Standing, Statistics, Games, etc. nodes, and so on. The search result pages are also included as part of the tree. For instance, under the Yahoo! Search node are all the result pages from search words that are entered on the Yahoo! front page.

SUMMARY OF THE INVENTION The present invention relates to a system and method for collecting and deriving Internet user behavioral data. More specifically, the present invention relates to a system and method for collecting and deriving historical and demographic data based on Internet user behavior so as to allow alternative advertising inventory to be selected in place of sold out advertising inventory.

In an exemplary embodiment, the present invention includes a number of ad records. An ad record is generated for each ad appearing on a viewed web page.

Preferably, an additional record creation routine first creates one or more additional ad records based on one or more of the original ad records. This is performed to allow the computation of historical and demographic data for trees of web pages (all nodes descendant from a single node) and even entire web sites.

A filtering routine then processes all the ad records to create a first group of records and a second group of records. The first group of records contains only registered user records, while the second group of records contains records for all users, registered and unregistered. For a registered user, a record includes a P cookie, a L cookie, a B cookie, and a Space ID; whereas, for an unregistered user, a record does not contain any P cookie or L cookie and only includes a B cookie and a Space ID. The first group of records is created by examining whether a record has a P cookie and/or L cookie; each record also contains the associated B cookie and Space ID. The second group of records is created by extracting from each user's (registered or unregistered) record the associated B cookie and Space ID. The two groups of records are thus, namely, a first group containing only registered user records with each record additionally having a P cookie and a L cookie, and a second group containing records for all users (registered and unregistered).

The B cookie provides identification information about the user's particular browser. The P cookie provides demographic information, sich as age, sex, occupation, etc., about a registered user. The L cookie provides a user name of a registered user. Finally, the Space ID provides identification information about the web page for which the record is generated.

Each group of records is then processed in a different manner. For the group of registered user records, a first sorting routine sorts these records based on the L cookie, by user name. An indexing routine then creates an index for each of these records using the P cookie. Since the P cookie includes demographic information, the index to be created essentially represents a demographic profile. In a preferred embodiment, the index is represented by a bit map and each bit of the bit map represents a demographic characteristic.

A second sorting routine then sorts the first group of records based on the Space ED. By sorting the Space ED, the records are then grouped by web pages. Hefice, demographic information from records relating to the same web page can be obtained.

A third sorting routine also sorts the first group of records based on the indices associated with these records. Since the indices represent dem ographic profiles, records having the same demographic profile are then grouped together. Collective Space ID information can then be examined from these grouped records to derive information on the web pages most frquently viewed by users having the same demographic profile.

This information can then be used to identify alternative advertising inventory amongst the web pages.

The second group of records for all users are processed differently. A first tallying routine first calculates the respective total number of visits to each of a number of web pages identified firom the Space I]Ds of the second group of records.

A second tallying routine then determines a number of affinity relationships. One afffinity relationship is determined for each pair of different web pages amongst the web pages identified from the Space IDs of the second group of records.

This second tallying routine also maintains the respective cumulative totals for each of the determined affinity relationships. By using the respective total number of visits and the respective cumulative totals, information on the llkelihoodof a user visiting one web page and also visiting another web page can be determined. Such information can similarly be used to identify alternative advertising inventory.

Reference to the remaining portions oif the specification, including the drawings and claims, will realize other features and advantages of the present invention.

Further features and. advantages of the present invention, as well as the structure and operation of various embodiments of the present invention, are described in detail below with respect to accompanying drawings, like reference numbers indicate identical or functionally similar elements.

BRIE DESCRIPTON OF HEDRAWINGS Fig. 1 is a simplified network diagram showing the structure of the Yahoo! network; and Figs. 2A-E are simplified flow diagrams ilustrating the operation of an exemplary embodiment of the present invention.

DESCRIPTION OF THE SPECIFIC EMB3ODIENTS The present invention will now be described. Figs. 2A-E are simplified flow diagrams illustrating the opertion of an embodiment of the present invention.

Referring to Fig. 2A, the adlogs 20 are collectively an inventory which provides a record of all the ads which have been displayed to users during a predetermined time period.

The predetermined time period can vary based on a number of factors, such as processing needs, system storage constraints, etc. Every time a user views a particular web page, information relating to all the ads appearing on that particular web page is recorded in the adlogs 20. Preferably, an ad record is generated for each ad displayed on each viewed web page. JFor example, if a viewed web page displays ten (10) ads, then ten (10) ad records, one for each ad, are generated in the adlogs 20. In an exemplary embodiment, each entry or ad record in the adlogs 20 contains various information about the user viewing the ad, including, for example, a B cookie, a P cookie, a L cookie, and a Space

ID.

The B cookie contains identifying information about the browser used by a particular user. Such information typically include, for example, a serial number assigned to a browser.

The P cookie contains certain demographic information about a user, such as gender, date of birth, zip code, country, occupation, industry, and interests, etc. The information represented by the P cookie is generally obtained from the user at the time the usrsigns upor registers with aweb portal or ISP. Hence, the Pcoolde is usually only applicable to a registered user. 'Similarly, the L cookie contains the user name of a registered user.

Finally, the Space II) contains identification information which indicates the specific web page the user has visited. In other words, the Space ID provides information about what web page has been viewed by the user.

In an exemplary embodiment, an additional record creation process 22 examines each ad record in the adlogs 20 to determine if additional ad records need to be created. This is done to compute data for trees of web pages and/or entire web sites so as to more accurately reflect the distribution of the demographic data: As noted above, same web portals, such as Yahoo!, organize their web pages in a tree structure with nodes. In a preferred embodiment, for each ad record relating to a particular web page (or node), an additional, identical record with the proper Space ID is generated for each and every node that is an ancestor of that particular node, to the extent that such identical record(s) is not already present in the adlogs. In other words, additional ad records may be generated for each and every node that is above that particular node.

The foregoing can be illustrated by an example. As shown in Fig. 1, when a user directly visits the web page for the San Francisco 49ers 16, by getting to that web page by entering the appropriate URL, ad records for ads displayed on that web page are generated in the adlogs 20. In addition, for each of those ad records, an additional ad record is also generated respectively for each of the nodes above, namely, the NFL web page 14, the Yahool Sports web page 12, and the Yahoo! Network web page 10. Since each of these web pages are separate and distinct from each other, their respective ad records necessarily reflect their own corresponding Space IDs but otherwise contain the same demographic information as the originating records. No additional ad records would be created, however, if the user had arrived at the San Francisco 49ers web page 16 via the less direct route of successively clicking on the appropriate hyperlinks beginning from the Yahoo! network page 10. This is because the appropriate records would have been added to the adlogs already by virtue of the user visiting each web page starting from the Yahoo! network page 10. The newly created additional ad records are then added to the original ad records. By adding these additional ad records, information on trees of web pages and entire web sites can be collected and made available for subsequent use.

All the ad records are then filtered by a filtering process 24 using different criteria to generate two groups of records. More specifically, a first group of records is generated for registered users, and a second group of records is generated for all users, registered or otherwise. Since only records of registered users would contain a L cookie, this is achieved by examining whether a L cookie is present in an record. Alternatively, a P cookie can also be used to generate the first group of records. Hence, after the filtering process 24, respective records for the two groups of users, namely, the registered users and all users, are identified and grouped together.

In addition to identifying these two groups of records, records for the two groups are further sorted based on the L cookie and the B cookie, respectively. The first group is sorted by the L cookie and the second group is sorted by the B cookie. Referring to Figs. 2B and 2E, the two groups of records are then stored separately, for example, in databases 26, 40. The records for the two groups are treated differently, as will be further described below.

Processing of the first group of records, records of the registered users, is described next. The records for the registered users are sorted based on the L cookie.

As mentioned above, the L coolde contains information on the user name for a registered user. Thus, by sorting the L cookie, all the records and information relating to a particular registered user are grouped together.

Referring to Fig. 2B, after the records of the registered users are sorted, each of these records, in particular, the P cookie, is then processed by an indexing process 28 to create an index for that particular record. An index is a bit map which represents the demographic data in each rocord. Each bit in the bit map is design~ated a particular demographic characteristic. For example, the first bit of the bit map may represent the sex, male, of a user. Similarly, other bits can be used to represent demographic characteristics such as between age 30-39, annual income exceeding $100,000, engineer as occupation, etc. By selecting between the values and"' for each bit, the presence or absence of a particular demographic characteristic in each record is indicated.

In an alternative embodiment, additional bit(s) in a bit map are created based on the web page (obtained via the Space ID of each record) the user has visited.

These bits reflect user history and interests, which are valuable additions to the demographic characteristics of the user.

At this point having created the respective indices for these records, different processes can be used to process these records and their associated indices. For example, the records can be sorted by either Space ED or bit map and then processed accordingly, as will be further described below.

In one process 30, as shown in Fig. 2Q, according to an exemplary embodiment, these records ame sorted by Space ED. As mentioned before, Space ID provides the identification information used to. indicate the specifid web page the user has visited. By sorting the Space ED, all the records originating f-rm the same web page are grouped together. Practically, this means that a the demographic information relating to each web page is collected and available for subsequent use.

Once sorted by Space ID, the records are then summ arized at 32. During this summarizing process 32, information from the records are collected and stored.

More specifically, for each Space ED, the respective bit maps or index information for all records having that Space ID is now available and stored in a database 34 for future access. In other words, demographic information for all users who have visited a particular web page is accessible from the database 34.

In addition, as part of the summarizing process 32, for each Space IID, the bit count for each bit of the bit map is calculated By calculating the bit count for each bit, the total number of users having a specific demographic -characteristic who have visited that particular web page can be determined. For example, the total number of males who have viewed a particular web page can be determined.

Additional demographic evaluations within a specified demographic group can further be performed by using the bit maps. Continuing with the above example, using records identified as having a male user, the bit representing the age between 20-29 can further be selected to determine the total number of males within that particular age group. Additional refinements can be made using other bits of the bit map as well.

Hence, by examining the individual bits, or combinations thereof, in the bit map, various demographic profiles can be determined for a particular web page.

Tn another process 36, as shown in Fig. 2D, according to an exemplary embodiment, the records for the registered users are sorted by bit map. By sorting the records based on their bit maps, records with identical bit maps are grouped together.

Since each bit of a bit map represents a demographic characteristic and the bit map represents a demographic profile, users with identical demographic profiles are then grouped together.

Additional information can be obtained from these records which have been grouped based on demographic profiles. This information is then formatted and stored in a database 38 for subsequent use. For example, for each group of records having a specific demographic profile, Space ED information can be extracted from each record within the group. By examining the collective Space ED information, the web pages most frequently visited by users having that specific demographic profile can be identified. Thus, information on web pages that are fr-equently visited by the respective demographic groups is available. Using such information, advertisers can be advised appropriately regarding the placement of their ads. Ads can then be more strategically tailored and positioned to maximnize their exposure and efficacy on the targeted demographic groups.

Processing of the second group of records, records of all the users, is next described. Referring to Fig. 2E, the records for all the users are sorted based on the .B cookie and stored in database 40. As mentioned above, the B cookie provides information used to identify a user's particular browser. Consequently, by sorting the B cookie, all the behavioral data originating from the same browser is grouped together. It is recognized that in certain sitiations, such as where an Internet-enabled computer is generally accessible to the public or where such computer is used by various family members, multiple users may use the same browser at various times thereby producing results which are representative of many users. From a statistical perspective, so long as the same browser is consistently used by the same general group of users, the behavioral data collected from that browser remains usefuL The records for all the users are processed in a different manner. It should be noted that the respective processing of the first group of records and the second group of records is independent of each other. As noted above, the records for all the users are sorted by the B cookie. Sorting by the B cookie is preferred since unregistered users have not previously provided any registration or demographic information, therefore, records for the unregistered users do not contain any P cookies. These records, however, include the Space IDs since Space ID information is captured during a browsing session, regardless of the status of a user.

The sorted records of all the users go through an affinity and tallying process 42. More specifically, for each B cookie, Li., each browser, a cumulative total is tallied and kept for each individual unique Space ID, an individual tally for each of the web pages visited by that browser is maintained. The collective cumulative totals for various unique Space IDs from all the browsers are then combined and stored for subsequent use.

Furthermore, various types of information are derived using the collective Space ID information collected from each browser. For example, an affinity relationship is determined for every pair of different Space IDs and the corresponding affinity count for that pair of Space IDs is incremented. By determining affinity relationships amongst all the web pages visited by the same browser, information can be obtained to predict the tendency and usage behavior of the user(s) using that browser in terms of the user(s) viewing one web page in connection with another. Information on the affinity relationships amongst the web pages visited by all the browsers is then collected. It should be understood that an affinity relationship can be made to correlate more than a pair of different web pages. Such relationship can involve, for eximple, three or more different web pages. A person with ordinary skills in the art will know of ways to implement such affinity relationship.

The foregoing can be illustrated by way of an example. Assume that web pages A, B and C are visited by a particular browser. The respective cumulative totals for keeping track of the number of visits to web pages A, B and C are all incremented by one In addition, since the browser visited three web pages A, B and C, the affinity relationships A-B, A-C, and B-C are created and the corresponding affinity counts are incremented.

After the records of all the users are processed, there is a cumulative total for each of the web pages collectively visited by the browsers and there is an affinity count for each pair of different web pages, both of which are stored, for example, in a database 44. By using the cumulative totals and the affinity counts, advertisers can further be advised appropriately regarding the placement of their ads.

Using the above example, assume that after processing the records of all the users, it is determined that the cumulative totals for web pages A, B and C are 100, 200 and 300, respectively and that the affinity counts for affinity relationships A-B, A-C and B-C are 50, 100 and 150, respectively. With the foregoing numbers, it can be fuirther calculated that the percentage of users visiting web page A who also visited web page B is 500/a and that the percentage of users visiting web page B who also visited web page A is 25%. The percentages are calculated as follows. Since the cumulative total for web page A is 100 and the affinity count for A-B is 50, that means that 100 users have visited web page A, and that out of those 100 users visiting web page A, only 50 of them also visited web page B, therefore, resulting in the 50% figure. Similarly, since the cumulative total for web page B is 200 and the affinity count for A-B is 50, that means that out of the 200 users visiting web page B, only 50 of them also visited web page A, resulting in the 25% figure.

Thus, with these percentages, advertisers can be appropriately informed of their alternatives in the event that their first choice is sold out. For example, using the figure arrived above, an advertiser can see that if ad views for web page A are all sold out, an ad intended for an audience viewing web page is likely to have a chance of getting viewed by that same audience if that ad is alternatively placed on web page B. That is because 50%1 of the users viewing web page A usually visit web page B also. Likewise, using the 25% figure, it can be seen that an ad intended for an audience viewing web page B is only likely to be viewed by 25% of that same audience if that ad is placed on web page A. Hence, by using the derived information, advertisers can be advised to place ads on alternative web pages which maximize the intended exposure to P:\WPDOCS\AMDSpiflcions\1 264373 .dc.3 Juuy 2006 -11the targeted audience. Consequently, excess demand is channeled to other possible alternatives. By selling more ad views on their web pages, web portals and ISPs can thus generate more revenue.

It is to be understood that while the present invention as described herein is used in connection with identifying alternative advertising inventory amongst web pages, the present invention can be easily implemented for other areas of application such as identifying web pages which fit specified demographic profiles so as to facilitate searching of relevant web pages. For example, assume that a user is particularly interested in a specific web page amongst the results returned from a search. Using the present invention, other relevant web pages, similar to the one that the user is interested in, can be identified and shown to the user.

A person of ordinary skill in the art will know of other ways and methods to apply the present invention.

The present invention as described herein can be implemented using both hardware and/or software, or a combination thereof. In a preferred embodiment, the databases that are used to store the adlogs and other processed records are implemented using sorted flat files techniques. The various processes such as filtering, sorting, creating the index or bit map, etc.

are preferably implemented using computer software such as etc. A person of ordinary skill in the art will know of other ways, methods and techniques to implement the present invention.

It is understood that the examples and embodiments described herein are for illustrative purposes only and that various modifications or changes in light thereof will be suggested to persons skilled in the art and are to be included within the spirit and purview of this application and scope of the appended claims. All publications, patents, and patent applications cited herein are hereby incorporated by reference for all purposes in their entirety.

Throughout this specification and the claims which follow, unless the context requires otherwise, the word "comprise", and variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.

The reference to any prior art in this specification is not, and should not be taken as, an acknowledgement or any form of suggestion that that prior art forms part of the common general knowledge in Australia.

Claims (5)

1. A system for collecting and deriving historical and demographic data based on user behavior, comprising: a first plurality of user records, each user record having web page identification information and browser identification information; a first tallying mechanism configured to calculate respective total number of visits to each of a plurality of web pages using said web page identification information from said first plurality of user records; and a second tallying mechanism configured to determine a plurality of affinity relationships and keep track of respective cumulative totals for each of said plurality of affinity relationships, wherein one affinity relationship is determined for each pair of different web pages amongst said plurality of web pages.

2. The system according to claim 1, wherein said respective total number of visits and said respective cumulative totals for each of said plurality of affinity relationships are capable of being used to identify alternative advertising space.

3. The system according to claim 1, wherein said respective total number of visits to each of said plurality of web pages and said respective cumulative totals for each of said plurality of affinity relationships are capable of being used to identify alternative advertising space.

4. The system according to claim 1, wherein said respective total number of visits to each of said plurality of web pages and said respective cumulative totals for each of said plurality of affinity relationships are capable of being used to identify one or more relevant web pages fitting specified search criteria or user preference. A method for collecting and deriving historical and demographic data based on user behavior, comprising steps of: retrieving a plurality of user records, each user record having web page identification information and browser identification information; P:\WPDOCSVADf\SpariclionsI264373 I.do.5 JmuAy 2006

13- calculating respective total number of visits to each of a plurality of web pages using said web page identification information from said plurality of user records; identifying a plurality of groups of web pages based on said browser identification of said plurality of user records; determining a plurality of affinity relationships, wherein one affinity relationship is determined for each pair of different web pages amongst each of said plurality of groups of web pages; and keeping track of respective cumulative totals for each of said plurality of affinity relationships. 6. The method according to claim 5, further comprising step of: deriving alternative advertising inventory using said respective total number of visits and said respective cumulative totals. 7. The method according to claim 5, further comprising step of: identifying one ore more relevant web pages fitting specified search criteria or user preference using said respective total number of visits and said respective cumulative totals. 8. A system for collecting and deriving historical and demographic data based on user behaviour, substantially as herein described with reference to the accompanying figures. 9. A method for collecting and deriving historical and demographic data based on user behaviour, substantially as herein described with reference to the accompanying figures. DATED this 6th day of January, 2006 YAHOO, INC. By Their Patent Attorneys DAVIES COLLISON CAVE