WO2010017647A1

WO2010017647A1 - Pull advertising method and system based on pull technology

Info

Publication number: WO2010017647A1
Application number: PCT/CA2009/001144
Authority: WO
Inventors: Line Tousignant
Original assignee: 9198-74 2 Quebec Inc.
Priority date: 2008-08-15
Filing date: 2009-08-17
Publication date: 2010-02-18

Abstract

A system for allowing a user of an online personal space to select an ad to be dispayed on the personal space of the user from a plurality of available ads, the system comprising a first database containing a user profile of the user, a second data base containing the available ads and advertisement target user profile of the available ads, an input, an output and a processor operatively connected to the first and second databases, the input and the output. The processor is so configured so as to rank the available ads based on a correlation between the user profile and the advertisement target user profile, present to the user the ranked ads in accordance with their ranking through the output, prompt the user to select a presented ad through the input and display the selected ad on the personal space of the user.

Description

PULL ADVERTISING METHOD AND SYSTEM BASED ON PULL TECHNOLOGY

TECHNICAL FIELD

[0001] The present invention relates to pull advertising method and system based on pull technology.

SUMMARY

[0002] The present invention relates to a method and system offering social networkers, bloggers or anyone with a personal website the possibility to add ads on their personal page(s) and to receive compensation based on various results measurements while providing constant refining of advertisers' advertising strategies and target customers' definition by using advertising metrics, statistic models and network metrics.

[0003] More specifically, in accordance with the present invention, there is provided a computer implemented method for allowing a user of an online personal space to select an ad to be displayed on the personal space of the user from a plurality of available ads, the method comprising: a. ranking available ads based on a correlation between a user profile of the user and advertisement target user profile of the available ads; b. presenting to the user the ranked ads in accordance with their ranking; c. prompting the user to select a presented ad; and d. displaying the selected ad on the personal space of the user.

[0004] In accordance with the present invention, the method further comprises the optional step of: aa. prompting the user to provide at least one selection criteria; and wherein step a. of ranking the available ads is based on a correlation between the user profile, the at least one selection criteria and the advertisement target user profile.

[0005] In accordance with the present invention, there is further provided a system for allowing a user of an online personal space to select an ad to be displayed on the personal space of the user from a plurality of available ads, the system comprising: a first database containing a user profile of the user; a second database containing the available ads and advertisement target user profile of the available ads; an input; an output; a processor operatively connected to the first and second databases, the input and the output, wherein the processor is so configured so as to; a. rank the available ads based on a correlation between the user profile and the advertisement target user profile; b. present to the user the ranked ads in accordance with their ranking through the output; c. prompt the user to select a presented ad through the input; and d. display the selected ad on the personal space of the user.

[0006] In accordance with the present invention, the processor of the system is further configured so as to optionally: aa. prompt the user to provide at least one selection criteria through the input; and wherein the processor is configured so as to rank the available ads based on a correlation between the user profile, the at least one selection criteria and the advertisement target user profile.

BRIEF DESCRIPTION OF THE FIGURES

[0007] Embodiments of the invention will be described by way of example only with reference to the accompanying drawings, in which:

[0008] Figure 1 is a schematic view of computing devices connected to a pull advertising system through a network; and

[0009] Figure 2 is a flow diagram depicting the pull advertising process according to an illustrative embodiment of the present invention.

DETAILED DESCRIPTION

[0010] Generally stated, the non-limitative illustrative embodiment of the present invention provides a pull advertising method and system based on a pull technology for use with online social networks, blogs and other websites where users have a "personal space". The pull advertising model offers social networkers, bloggers or anyone with a personal website the possibility to add an ad on their personal page(s) (e.g. web page) and to receive compensation for, for example, the number of monthly impressions generated on their page(s), blog(s) or website.

[0011] The pull advertising model reverses the process of traditional advertising and builds advertisers' target profile and strategies from the endorsement of users. Unlike the traditional push advertising model where ads are selected by the system according to various metrics and "pushed" onto a user's page whenever that page is accessed, the pull advertising model allows a user to select the ads he or she wishes to have displayed on his or her page.

[0012] Using advertising metrics, statistic models and network metrics, the pull advertising system provides advertisers with a constant refining of their advertisement target user profiles and advertisement campaign objectives and strategies. The pull advertising model allows for word of mouth projections and can take advantage of various consumers' network behavior to reach advertisers' specific advertising campaign objectives.

[0013] It should be noted that throughout the disclosure references to the term "ad" refers to an advertisement that may be of different formats, for example text, static or animated image, video, etc., and that the ad may include active links or mouseover activated functions.

[0014] Referring to Figure 1 , a user using a personal computer 12, laptop computer 14, personal assistant device 16, or any other such computing device, on which runs a user interface in the form of a communication software such as, for example, a web browser, may access an online social network, blog or other website located on the web server 32 of the pull advertising system 30 or on a third party web server 40 via an Internet connection 20 such as, for example, Ethernet (broadband, high-speed), wireless WiFi, cable Internet, satellite connection, cellular or satellite network, etc.

[0015] Further to the web server 32, the pull advertising system 30 includes an advertisement server 34, a user database 36 and an advertisement database 38, all of which will be detailed further below.

[0016] The pull advertising process is initiated when a user, connected to, for example, a website on either the pull advertising system web server 32 or a third party web server 40, elects to "add an ad" on one or more of its page(s). Upon this user's action, the pull advertising system web server 32 or third party web server 40 connects to the advertisement server 34 which provides the user with a selection of ads to choose from. Suggested ads are selected accordingly to the level of correlation (ranking) between the user and available ads data such as the profile, the communication behavior and the advertisement selection behavior of the user (stored in the user database 36), and the advertisement target user profiles and advertisement campaign objectives of the ads (stored in the advertisement database 38). Upon selection of a desired ad by a user, related data is used to update the user advertisement selection behavior which continuously improves the relevance of the selection of ads suggested to a user. The pull advertising process as well as the user advertisement selection behavior update will be further detailed below.

[0017] It is to be understood that the presence of the pull advertising system web server 32 is optional; in the absence of which the pull advertising process can simply be initiated remotely from personal pages located on third party web servers 40, connecting directly to the advertisement server 34 of the pull advertising system 30.

User data

User profile

[0018] The user profile is a collection of information aggregated about a user and is stored within the user database 36. It is to be understood that there is a user profile for each of the users. The information may include general characteristics about the user, for example:

- age;

- sex;

- location;

- profession;

- income;

- causes supported (which can be a good indicator of brands that the user will or will not want to support);

- activities;

- interests; and

- lifestyle.

[0019] It may also include information about the user's actions and exposure related to currently running or past advertising campaigns, such as:

- products bought or explored on advertisers' websites; - a list of "pushed" advertisements that the user has seen including the frequency of exposure, the total number of times seen and the user's actions (e.g. clicking on the ad, moving the mouse over it) associated with the advertisements; and

- sites visited when leaving the website (i.e. through clickstream or browsing analysis).

[0020] It is to be understood that that the user profile is not limited to this information and that additional information may be added or information omitted.

[0021] The attributes that are stored in the user profile may be represented in one of many ways, for example:

- with a single value possible for each attribute, for example a single integer that describes the income in dollars (e.g. $45000);

- a representation that includes an uncertainty, for example the income in dollars could be represented as $45000 plus or minus $5000;

- a representation that includes a lower and an upper limit; for example $42000-$47500;

- a representation that describes membership in one or more predefined categories, for example $40000-$50000;

- a representation that describes membership in one or more predefined categories

- a representation that includes a probabilistic distribution over the potential values, for example income: expected value $45000, standard deviation $45000;

- a list of keywords that describe such attributes as interests or hobbies;

- a list of keywords with a weight attached to each one that describe such attributes as hobbies where some might be more important than others; and - another representation known in the art.

[0022] It is to be understood that the collected user information and its use may be subject to any and all Privacy Act provisions or any other applicable privacy guidelines.

User communication behavior

[0023] The user communication behavior, which is stored within the user database 36, is the user's past and present behavior on the Web, social networking sites and blog sites. It may include, for example:

- the content of, and history of, any publicly accessible web page which is identifiable as belonging to that user, such as their personal web page, their personal page on social network sites, any blogs that are identifiable as being written by that user, etc.;

- connections between this user and other users via such actions as adding a friend on a social network; joining a group on a social networking site; sending a message through a social networking site; adding a blog to the "blogroll" on a blogging site; mentioning the friend by name in a public communication such as a "wall" message, etc.;

- topics, products, services and brands discussed on the user's blog and/or personal page or in the user's communications (i.e. through the use of a word syntax recognition or named entity recognition technology);

- the user's opinion on the products, services and brands, or rating of ads, as specifically entered by the user as part of a user interface the pull advertising process (which will be detailed further below);

- the role of a user within its community on a social network. The user may be an influencer, or communicator, part of a community sharing common interests; and - the user's opinion on topics, products, services and brands discussed in their blog, on their personal page (through the use of named entity recognition and sentiment analysis technology).

[0024] It is to be understood that the user communication behavior may be included within the user profile or be separate. It is also to be understood that the collected user behavior information and its use may be subject to any and all Privacy Act provisions or any other applicable privacy guidelines.

User advertisement selection behavior

[0025] The user advertisement selection behavior, which is stored within the user database 36, describes the user's actions when interacting with the advertisements displayed on the system. These interactions could occur as part of the "add an ad" process or when viewing ads on other user's personal pages.

[0026] The user advertisement selection behavior may include information about the user's actions when interacting with the "add an ad" process, for example:

- lists of advertisements, brands, products or services that the user has clicked on or not clicked on when choosing to "ad an ad";

- lists of categories or keywords that the user has chosen in order to narrow down or filter the list of ads available; and

- the amount of time that the user has spent in each category and watching or looking at each ad.

[0027] It may also include information about user's interactions when interacting with ads on other user's personal pages, such as:

- whether or not the user clicked on the ad;

- whether, if the ad is a video, the user played the video;

- if the ad is a video, for how long the user played the video and whether or not the video finished before the user left the page or scrolled the video off the page; - whether or not the user moved the mouse over the ad and for how long the user left the mouse over the ad; and

- whether or not the user bought a product when they clicked on the ad and were sent to the advertiser's website.

[0028] It is to be understood that the user advertisement selection behavior may be included within the user profile or be separate.

Advertisement data

Advertisement target user profile

[0029] The advertisement target user profile, which is stored within the advertisement database 38, is a set of attributes that describe the socio- demographic attributes of people to which an advertiser wants to show its advertisements. The information may include target attributes which are general characteristics about the user, for example:

- age;

- sex;

- location;

- profession;

- income;

- causes supported;

- activities;

- interests; and

- lifestyles.

[0030] It is to be understood that that the advertisement target user profile is not limited to this information and that additional information may be added or information omitted. [0031] The attributes that are stored in the advertisement target user profile may be represented in one of many ways, for example:

- a number, for example an age in years;

- one or more categories, for example profession could be split up into several industry segments;

- one or more ranges, for example age could be split up into 12-17, 18-24, 18-49, 25-34, 35-44, 45-54, 55-64, 65+; and

- a location could be represented as a country, state or province, region, municipality and postal code, with each of these attributes having its own target value(s).

[0032] In addition, where appropriate, the advertiser may be given the option of allowing multiple criteria. This may be done in one of the following ways:

- allowing the advertiser to select multiple values, for example several locations; and

- allowing the advertiser to select a range of values, for example $30000- $50000 for income.

[0033] For each of these target attributes, the advertiser may optionally include a weight. In one embodiment, the weight could be a number between 0 and 10. A weight of 0 would represent that the advertiser did not want this attribute to influence the ranking in any way. As weights increase, the fact that a user matches the target attribute may have increasingly more effect on the ranking. A weight of 10 would represent that the advertiser never wanted anyone who didn't exactly match their criteria exactly to see the ad. This can be used, for example, to restrict certain advertisements such as those containing sexually explicit content to people who were aged 18 years or over, or to restrict alcohol advertising to those people who were aged 21 years or over. Advertisement campaign objectives

[0034] Advertisement campaign objectives, which are stored within the advertisement database 38, are composed of specific advertising campaign objectives such as, for example, building brand awareness vs. driving sales. Different objectives call for different strategies of frequency, scope and scale of a campaign.

Pull advertising process

[0035] Referring now to Figure 2, there is shown a flow diagram of an illustrative example of the pull advertising process 100 executed by the advertisement server 34 of the pull advertising system 30. Steps of the process 100 are indicated by blocks 101 to 130.

[0036] The process 100 starts at block 101 where a user accesses the pull advertising system 30 through the pull advertising system web server 32 or third party web server 40.

[0037] At block 102, the user profile of the user accessing the pull advertising system 30 is updated in the user database 36 with pertinent information about the user gathered from any of its personal pages.

[0038] As regard to the user communication behavior, it is updated every time a user accesses a social network, whether or not the user accesses the pull advertising system 30. Accordingly, patterns of communication and level of influence of users on their social networks is constantly measured and this information used to update their user profiles.

[0039] Then, at block 103, the user is presented with its personal pull advertising page, which provides the user with various options such as the option to add an ad on a personal page.

[0040] Optionally, the advertisement server 34 may communicate to a user, through, for example, its personal pull advertising page or email, that a campaign for a previously added ad is coming to end and invite the user to select a new ad. Furthermore, the advertisement server 34 may monitor results of ad campaigns and adjust to the terms and conditions (i.e. budget, impressions, frequency, etc) specified for each ad in the advertisement database 38. When approaching a campaign ending date or the maximum number of impressions desired by an advertiser for an ad, a notification is sent to users carrying this specific ad on their personal page(s) inviting them to select a new ad.

[0041] Then, at block 104, the process 100 verifies if the user has selected to add a new ad. If so, it proceeds to block 106, otherwise it returns to block 102.

[0042] Optionally, at block 106, the process 100 allows the user to enter one or more selection criteria, for example brands, product categories, service categories, keywords, etc., in order to provide information to the process 100 as to desired or preferred ads.

[0043] The process 100 then ranks the available ads, at block 108, based on a correlation between the user profile and the advertisement target user profiles of the available ads. Optionally, if one or more selection criteria were entered at block 106, the correlation further takes into account the one or more selection criteria. In an alternative embodiment, the ranking is also based on a correlation between the user advertisement selection behavior, its influence within its community or its belonging to one or more communities, and the advertisement target user profiles. In another alternative embodiment, the ranking is further based on a correlation between the user communication behavior and the advertisement campaign objectives of the available ads. The ranking schemes will be further detailed below.

[0044] At block 110, the process 100 presents to the user the ranked ads in accordance with their ranking. [0045] Then, at block 112, the process 100 verifies if the user found an ad of interest among the presented ads. If so, it proceeds to block 114, otherwise it proceeds to block 120.

[0046] At block 114, the user selects the ad of interest from the presented ads. When selecting an ad, a user may optionally be invited to write a few words expressing an opinion about the service/product/brand or rate the ad. This information may also be used to update the user profile. Furthermore, friends of the user may also be given the possibility to write their own comments/opinions about the service/product/brand selected by a friend. The selected ad and, optionally, the user's and user's friends' comments then appear on the user's personal page until the user changes it, or until the advertising campaign comes to an end; whichever comes first.

[0047] In an alternative embodiment, the pull advertising system 30 may give the user the opportunity to select more than one ad for a same personal page or specific advertisement positions on a personal page (i.e. in the case where a personal page contains multiple areas dedicated to ad placement). In a further alternative embodiment, the pull advertising system 30 may actually allow the user to select multiple ads. In such a case, the selected ads are displayed in an alternate fashion on the personal page of the user. The display frequency of each selected ad may then be specified, for example, by the user (subject to some advertiser or system limitations) and/or determined by the pull advertising system 30 based on various parameters, such as advertiser requirements, and/or network metrics.

[0048] Once an ad is selected, at block 116, the process 100 updates the user advertisement selection behavior based on selected ads and non-selected ads. The user advertisement selection behavior update may also be initiated whenever a user interacts with an ad on a personal page. Data collected by the user advertisement selection behavior for a plurality of users also highly benefits push advertising strategies. A clearer understanding of users' profiles and more information on their social networking behavior helps advertisers to increase push advertising strategies efficiency. The pull advertising model provides marketers with "samples" of customers and interested prospects behavior in a natural environment, instead of in a context of traditional focus groups, where several factors can influence their response. Actually, other than usual advertising campaign metrics; several other market research reports can be extracted from the pull advertising system 30.

[0049] At block 118, the user selects on which of its personal page(s) the selected ad is to appear. It is to be understood that in an alternative embodiment the user may not be allowed to select which of its personal page(s) the ad is to appear on or may be limited only to specific personal pages by the pull advertising system 30 depending on remote web site policies. The process 100 then returns to block 102 where the user has the possibility to add another ad.

[0050] At block 120, in the case where a user cannot find an ad of interest among the presented ads, the process 100 offers the user the possibility to make a special request. If the user wishes to make a special request, the process 100 proceeds to block 122. If not, the process 100 proceeds to block 130 where the user advertisement selection behavior is updated based the non-selected ads and then proceeds back to block 102.

[0051] At block 122, the process 100 displays a special request form for the user to fill.

[0052] Then, at block 124, the process 100 verifies if the requested service, product or brand is available in the advertisement database 38 but was simply ranked so low as to be placed near the end of the list of presented ads. If so, the ad is retrieved and proposed to the user (unless, as mentioned previously, the advertiser has requested an automatic rejection of specific websites or specific types of content, in which case the ad will not be proposed to the user) and the process 100 proceeds to block 116, which was previously detailed. [0053] If not, the process 100 proceeds to block 126 where it notifies the sales force of the pull advertising system 30 that there is a special request pending and, finally, at block 128, it notifies the user that the requested service, product or brand is not available and that the sales force was notified. While the special request remains pending, the pull advertising system 30 may send the user notifications on his special request's progress and, optionally, propose similar ads based on the preferences of other users with similar profiles. The process 100 proceeds to block 130 where the user advertisement selection behavior is updated based the non-selected ads and then proceeds back to block 102.

Ranking schemes

User profile vs advertisement target user profile

[0054] The correlation between the user profile and an advertisement target user profile can be determined using, for example, the cosine similarity measure (CSM). The CSM includes the following steps:

1. filtering out ads which the advertiser wants to exclude (for example, if the advertiser had age > 21 and the person was 16);

2. converting the advertisement target user profile to a feature vector;

3. multiplying the advertisement target user profile attribute vector by the attribute weights, if specified by the advertiser;

4. converting the user profile to a feature vector;

5. calculating the cosine similarity between the advertisement target user profile feature vector and the user profile feature vector;

6. converting the cosine similarity to a percentage; and

7. filtering out ads which do not match the one or more user selection criteria (optional).

[0055] Other criteria may also influence the suggested ranking, such as:

- Advertising campaign's budget. The selection of ads have to respect the budget determined by each advertiser and/or the number of impressions desired; therefore, although the ranking prioritizes the degree of relevance, it also considers other factors and may reject an ad based on the fact that the desired number of impressions would be already close to 100%. Such additional filters are necessary to avoid annoyance among users, who could otherwise be forced to go through the same process again very shortly.

- Campaign running dates.

As for the budget, advertising campaign dates influence the suggested selection. Based on, for example, a campaign ending date shorter than one week, the ranking would reject an ad for the same reasons mentioned above. Although the material for an advertising campaign should be eliminated from the advertisement database 38 when a campaign is over; there could be some exceptions. For example, such an exception would occur when an advertiser is running a "flighting" advertising campaign. A flighting advertising campaign is when a campaign period of activity is followed by a period with no advertising and then followed by a second period of activity.

- New products.

Products still unknown from consumers may also influence the ranking. As it is desirable to offer users the possibility to discover new products introduced to the market, new products labeled under the "discovery" category, and also labeled under their appropriate category, may be suggested to users when presenting potential products of interest to them, as determined throughout the advertisement target user profile analysis. The discovery category may be very valuable for early adopters who are interested in the discovery and promotion of "the latest".

- Advertisers' requests. Requests of an automatic rejection of specific websites or specific types of content, in order to avoid the risk of being associated with inappropriate content, can also influence the ranking.

- Popularity of an ad.

Popularity of an ad does not influence the ad selection proposal process but it could influence the user's selection. The ranking may suggest a combination of frequently selected ads as well as less frequently selected ads. Pricing for advertisers and compensation for users may be determined using a demand and supply model. Therefore, the ranking may then indicate to the user that the selection of a frequently selected ad could result in a lower compensation (per click or any other compensation metrics) than for a less frequently selected one, thus inciting users to select less frequently selected ads and creating an equilibrium for the ads selection. Cost to advertisers may then also be adjusted according to the number of users having selected a specific ad.

User profile and user advertisement selection behavior vs advertisement target user profile

[0056] In an alternative embodiment, the ranking is also based on a correlation between the user advertisement selection behavior and the advertisement target user profiles. This might be implemented by adding extra features to the user profile and advertisement target user profile feature vectors that are compared using the CSM.

[0057] The user advertisement selection behavior may be in the form of, for example, a user statistics model, which contain statistics modeling a user's behavior to particular categories of advertisement. For example, the user statistics model may model which product/service category a user is likely to click on in the pull advertising process 100. This model may contain a probability for each product/service category. When nothing is known about a user, the probabilities would all be equal. When a user clicked on a product/service category in the pull advertising process 100, the probability for the product/service category that the user clicked on may be increased, and the probabilities for all of the other product/service categories may be slightly decreased. This may be implemented as an update of the user database 36. Over time, the user statistics model builds a user-specific model of the product/service category that a user is likely to select.

[0058] Similar models may be built for other characteristics of the user, for example:

- the product/service category that the user is likely to be interested in;

- the brands that the user is likely to be interested in;

- characteristics of the ads that the user is likely to be interested in (for example, that the user prefers videos to static images);

- the product type that the user is likely to be interested in (for example, within the product category "electronics" the user is more interested in personal music players than in projectors);

- the kind of behavior exhibited (for example, the user is likely to browse for a lot of ads rather than search for a specific ad); and

- other characteristics known in the art.

[0059] The user statistics model may be built from the following signals:

- the product/service category that a user selected in the pull advertising process 100;

- the products/services advertised in the ads that a user selected in the pull advertising process 100;

- the brands advertised in the ads that a user selected in the pull advertising process 100;

- the product/service category, products advertised or brands advertised in the ads that a user chose to add to their page in the pull advertising process 100; - if applicable, the one or more selection criteria used by the user during the pull advertising process 100;

- whether or not the user clicked on an ad which was presented to them on another user's personal page; and

- other signals known in the art.

[0060] It may also be possible to obtain signals about what a user wasn't interested in. For example, if the user was presented with five ads ranked in order of relevance and the user chose ad number 3, then that may be taken as an indication that the user didn't like ads number 1 or 2 and the user statistics model may be updated to decrease the probabilities associated with these ads, whilst increasing the probabilities associated with ad number 3.

[0061] The process to update the statistics may be implemented in real-time by performing an update of the statistics model of the user advertisement selection behavior after each user action in the pull advertising process 100. The process to update the statistics may also be done less frequently and take into account the final choice of the user. For example, if the user was presented with five ads and the user watched ads number 1 and 2 before watching ad number 3 and choosing to add it to their page, then this may be interpreted as a signal that the user didn't like ads number 1 and 2 but liked ad number 3, even though the user had clicked on ads number 1 and 2.

[0062] Once a statistics model of the user advertisement selection behavior had been obtained for a user, this model may be used to influence the ranking of the advertisements. If the CSM was used as the ranking scheme as described above, then the information from the statistics model of the user advertisement selection behavior would be added to the user's feature vector ranked by the CSM. Further information about the product/service (for example, product/service category or type) would be added to the target feature vector. A CSM calculated on these augmented feature vectors would then provide a score that was affected by the user's product preferences. [0063] If information was added to the statistics model of the user advertisement selection behavior for which there was no corresponding information in the advertisement target user profile (for example, if the statistics model of the user advertisement selection behavior included a keyword model but there were no keywords attached to the advertisement target user profile), then the advertisement target user profile could be augmented with information from the user profiles of users that selected that ad. For example, in order to produce a keyword model for a advertisement target user profile, the keywords of all users in the statistics model of the user advertisement selection behavior who had chosen to add the ad to their personal page may be combined and this combined set of keywords used as the keyword model for the advertisement target user profile.

User profile and user advertisement selection behavior vs advertisement target user profile with user communication behavior vs advertisement campaign objectives

[0064] In a further alternative embodiment, the ranking is further based on a correlation between the user communication behavior and the advertisement campaign objectives of the available ads.

[0065] The goal of influencing the ranking in this manner is in order to optimize the execution of the advertising campaign to meet the advertiser campaign objectives as rapidly and efficiently as possible. This is done by considering both the user and the one ore more community in which it finds itself. As a result, two users with identical user profiles, but with different user communication behavior, could be presented with a very different ad ranking.

[0066] There exist many ways of analyzing social networks. These techniques are collectively known as social network metrics. In general, these social network metrics are abstract graph-theory based calculations that are implemented by representing the social network as an abstract graph. An abstract graph includes one or more types of node, which represent entities such as users. Each entity, such as a user, has an associated node. An abstract graph also includes one or more types of edges, which represent relationships between two entities. Each relationship between two entities has an associated edge. For example, friendship between two users, each of which is represented as a user node, might be represented as a friendship edge between the two users. Each edge may further be directed (so that it goes in a single direction) or undirected (so that there is no direction associated with it). Optionally, an edge may have a weight associated with it in order to represent the degree of association. Optionally still, a node or an edge may have other information associated with it.

[0067] It is possible to identify many types of relationships that might be associated with one or more types of edges:

- a friendship edge between user A and user B might be added if user A had asked user B to be their friend;

- a communication edge between user A and user B might be added if user A had sent a message to user B, and the weight on this edge might represent the amount or frequency of communication between the users;

- an "in group" edge between user A and group B might be added if user A had elected to join group B; and

- other types of edges will be apparent to one skilled in the art. [0068] Such relationships may be used for the following purposes:

- Tracking the number of times that a user has seen each ad (i.e. measured by frequency quintiles). A user that has been under-exposed to an ad might not have seen it often enough for it to remember the product/service or brand later, which means that the ad has not reached optimal visibility. On the other hand, a user that has been over-exposed to an ad might become annoyed by the ad and form a negative opinion of the product/service or brand being advertised; that is when an ad reaches saturation. Accordingly, money spent by the advertiser on further showing the ad has a decreasing return on investment.

- When an advertising campaign has reached saturation among a certain group of users, the ranking may consider other groups of users (where the group could be a set of users who were closely connected in the social network as given by the user communication behavior) that has similar user profiles to those that have reached saturation, and continue the campaign targeting these users.

At the end of an advertising campaign or at certain points during the campaign, the advertiser may be presented with the characteristics of users that liked their product/service and how the users differed from the characteristics in the advertisement target user profile. This allows the advertiser to gain market intelligence about their product/service and potentially refine their campaign or future campaigns strategies and segmentation.

Choosing the execution strategy of the advertising campaign based upon the user communication behavior. For example, the pull advertising system 30 constantly updates the links between users and updates the correlation between a user profile and an advertisement target user profile based on the number of links, as well as on the direction of these links, between the user and its community. An advertiser running a promotional campaign with short term goals would require high word of mouth velocity (that is the speed at which users will "spread the word" within their network). Consequently, the ranking system would heavily weight users representing nodes with high outdegree. On the contrary, for advertising campaigns aimed at building brand awareness (long term goals), the ranking would privilege users with a high influence or with a high indegree and high eigenvector centrality. In this way the pull advertising system 30 is able to customize the ranking of the ads in order to optimize the advertisement campaign objectives.

The user communication behavior includes the context of the social network of the user. This information can be used to calculate social network metrics, which include but are not limited to the following well- known metrics: o Indegree: the number of links to a node. This value may be used to determine the level of influence that this user has on their friends in the social network. o Outdegree: the number of nodes the user is connected to. A high outdegree score indicates a user who is extroverted and who initiates a lot of contacts with other members. Such users may highly contribute to word of mouth. o Eigenvector Centrality:

The eigenvector centrality measures the influence of a node within a network against all other nodes. This measure assigns relative scores based on the principle that connections to a high score contribute more to the scored of the node in question than same number of connections to low scoring nodes. Users with a high eigenvector centrality may be good starting points for campaigns as they are globally very influential; it may also be possible to charge a higher rate to place ads on the personal page of a user who has high eigenvector centrality's page. o Betweenness:

This measures a user's ability to bridge together different networks that would not otherwise connect. If this user is also popular and influential, he or she can become a key in widening the scope of an advertising campaign by reaching users who would otherwise be difficult to target. The betweeness allows for projections of word of mouth velocity. o Clustering Coefficient:

It measures the "cliquishness" of a group of friends. A high value indicates a group that is highly influenced from within the group, but not easily influenced from outside the group. A group with a high clustering coefficient would most likely easily result in a viral effect within the group. However, influencing them from outside would be much more difficult and it would be necessary to target members with a high betweenness score from other parts of the social network.

[0069] Using these metrics in combination depending upon the advertiser campaign objectives presents several advantages. It helps in evaluating the "quality" of a user or of a group's members. Mixing these metrics with frequency of impression statistics, for example frequency quintiles, and with advertisement target user profiles results in very precise strategies and highly efficient advertising campaigns.

[0070] At the end of an advertising campaign, or at certain point during the campaign, the advertiser may be presented with the characteristics of users that liked their product/service and how the users differed from the characteristics in the advertisement target user profile. This allows the advertiser to gain market intelligence about their product/service and potentially refine their campaign or future campaigns strategies and segmentation.

[0071] Although the CSM was used to illustrate the various ranking schemes described above, other techniques, measures or procedures may be used. For example, in an alternative embodiment, the degree of correlation may be generated by a recommendation engine. A recommendation engine will be familiar to one skilled in the art as a process that determines a score that indicates, for a given user and ad, whether a user will like the ad or not. The score will normally be lower if the user is less likely to like the advertisement, and higher if the user is more likely to like the advertisement. [0072] This score may be based upon several sources of information, for example:

- information known about a user (from the user profile and other sources);

- information known about the ad (from the advertisement target user profile and other sources);

- the degree of correlation between the user profile (possibly augmented by the user communication behavior and user advertisement selection behavior) and the advertisement target user profile as provided by a simpler correlation measure, such as the CSM;

- information gained from the preferences of other similar users, where the similarity may be determined by the user communication behavior. For example a user's social network friends may be considered to be similar users; and

- other information known in the art.

[0073] It will be familiar to one skilled in the art that a recommendation engine provides scores that are in general superior to the CSM, as these scores not only measure the degree of correlation between the user profile and the advertisement target user profile but also take into account the preferences of users. Two users may prefer different ads even though the basic information in their respective user profiles is very similar.

Other functionalities

[0074] Private personalized reports informing users of campaigns' dates and of the number of impressions generated by their respective personal page(s) can be produced at anytime and are updated daily. Users are compensated accordingly to, for example, the number of impressions generated by their personal page(s) and may have the option to exchange cumulated points for money or products or to give them to a friend or a cause of their choice. Of course, a user's visits to its own personal page will not be considered in the number of impressions and advertisers are also be protected against fraud in the case, for example, of an automated system that would "hit" users' personal page(s).

[0075] When it comes to visits from friends to a user's personal page, the pull advertising system 30 may be set so as to retire advertisements once they have reached a level of frequency saturation among a given group. This feature protects advertisers from users' friends who would intentionally generate repetitive hits on a user's personal page. An exception to this process could occur if the ad is still generating word of mouth or actions from visitors. Appropriate adjustments can be made accordingly to results measurements and advertisers' campaign objectives.

[0076] Several metrics can be used to develop reports for advertisers, including, for example:

- impressions;

- frequency;

- click through rate;

- word of mouth activity;

- user's actions;

- visitor's actions;

- prospects profile and lifestyle definition and refinement; and

- network metrics allowing for behavior analysis, and for campaign's strategies refinement and redefinition.

[0077] It is to be understood that although throughout the disclosure reference is made to separate servers 32 and 34 as well as separate databases 36 and 38, these may be implemented on one or more physical device and/or may be combined. It is to be further understood that the user 36 and advertisement 38 databases may equally be implemented by a data structure within a computer memory. [0078] Although the present invention has been described by way of particular embodiments and examples thereof, it should be noted that it will be apparent to persons skilled in the art that modifications may be applied to the present particular embodiment without departing from the scope of the present invention.

Claims

WHAT IS CLAIMED IS:

1. A computer implemented method for allowing a user of an online personal space to select an ad to be displayed on the personal space of the user from a plurality of available ads, the method comprising: a. ranking available ads based on a correlation between a user profile of the user and advertisement target user profiles of the available ads; b. presenting to the user the ranked ads in accordance with their ranking; c. prompting the user to select a presented ad; and d. displaying the selected ad on the personal space of the user.

2. A method according to claim 1 , wherein the online personal space is selected from a group consisting of a blog and a web page on a social network.

3. A method according to claim 1 , wherein the advertisement target user profile includes weighted attributes and wherein the correlation between the user profile and the advertisement target user profile takes into consideration the weight of the attributes.

4. A method according to claim 1 , further comprising the step of: aa. prompting the user to provide at least one selection criteria; and wherein step a. of ranking the available ads is based on a correlation between the user profile, the at least one selection criteria and the advertisement target user profile.

5. A method according to claim 4, wherein the at least one selection criteria is selected from a group consisting of a brand, a product category, a service category and a keyword.

6. A method according to claim 1 , wherein step a. of ranking the available ads is further based on a correlation between a user advertisement selection behavior of the user and the target profile.

7. A method according to claim 6, further comprising the step of: e. updating the user advertisement selection behavior based on at least one of selected ads, viewed ads and non-selected ads.

8. A method according to claim 6, wherein step a. of ranking the available ads is further based on a correlation between a user communication behavior of the user and advertisement campaign objectives of the available ads.

9. A method according to claim 8, wherein the user communication behavior includes a measure of outdegree related to the number of outward links from the user and a measure of indegree related to the number of inward links to the user.

10. A method according to claim 9, wherein the correlation between the user communication behavior and the advertisement favors users having a high outdegree for advertisement campaign objectives with short term goals and favor users having a high indegree for advertisement campaign objectives with long term goals.

11. A method according to claim 8, further comprising the step of: cc. prompting the user to enter a comment or rating of the subject of the selected ad; and wherein the user communication behavior is updated based on the comment or rating.

12. A method according to claim 1 , wherein step a. of ranking the available ads further includes matching the advertisement target user profile with the user profile and excluding ads having a target profile not compatible with the user profile.

13. A method according to claim 1 , wherein the correlation is determined using a cosine similarity measure or a recommendation engine.

14. A method according to claim 1 , wherein in step c. the user is prompted to select a plurality of presented ads and wherein in step d. the selected ads are displayed in an alternate fashion on the personal space of the user.

15. A method according to claim 13, wherein the frequency of display of each selected ad is determined by the user, by advertiser requirements or by network metrics.

16. A system for allowing a user of an online personal space to select an ad to be displayed on the personal space of the user from a plurality of available ads, the system comprising: a first database containing a user profile of the user; a second database containing the available ads and advertisement target user profile of the available ads; an input; an output; a processor operatively connected to the first and second databases, the input and the output, wherein the processor is so configured so as to; a. rank the available ads based on a correlation between the user profile and the advertisement target user profile; b. present to the user the ranked ads in accordance with their ranking through the output; c. prompt the user to select a presented ad through the input; and d. display the selected ad on the personal space of the user.

17. A system according to claim 16, wherein the online personal space is selected from a group consisting of a blog and a web page on a social network.

18. A system according to claim 16, wherein the advertisement target user profile includes weighted attributes and wherein the correlation between the user profile and the advertisement target user profile takes into consideration the weight of the attributes.

19. A system according to claim 16, wherein the processor is further configured so as to: aa. prompt the user to provide at least one selection criteria through the input; and wherein the processor is configured so as to rank the available ads based on a correlation between the user profile, the at least one selection criteria and the advertisement target user profile.

20. A system according to claim 19, wherein the at least one selection criteria is selected from a group consisting of a brand, a product category, a service category and a keyword.

21. A system according to claim 16, wherein the first database further contains a user advertisement selection behavior of the user and wherein the processor is further configured so as to rank the available ads based on a correlation between the user advertisement selection behavior and the target profile.

22. A system according to claim 21 , wherein the processor is further configured so as to: e. update the user advertisement selection behavior based on at least one of selected ads, viewed ads and non-selected adds.

23.A system according to claim 21, wherein the first database further contains a user communication behavior of the user and the second database further contains advertisement campaign objectives of the available ads, and wherein the processor is further configured so as to rank the available ads based on a correlation between the user communication behavior and the advertisement campaign objectives.

24.A system according to claim 23, wherein the user communication behavior includes a measure of outdegree related to the number of outward links from the user and a measure of indegree related to the number of inward links to the user.

25.A system to claim 24, wherein the correlation between the user communication behavior and the advertisement favors users having a high outdegree for advertisement campaign objectives with short term goals and favor users having a high indegree for advertisement campaign objectives with long term goals.

26.A system according to claim 23, wherein the processor is further configured so as to: cc. prompt the user to enter a comment or rating of the subject of the selected ad; and wherein the user communication behavior is updated based on the comment or rating.

27. A system according to claim 16, wherein the processor is further configured so as to match the advertisement target user profile with the user profile and exclude ads having a target profile not compatible with the user profile.

28.A system according to claim 16, wherein the correlation is determined using a cosine similarity measure or a recommendation engine.

29. A system according to claim 16, wherein the processor is further configured so as to prompt the user to select a plurality of presented ads display the selected ads in an alternate fashion on the personal space of the user.

30. A system according to claim 29, wherein the frequency of display of each selected ad is determined by the user, by advertiser requirements or by network metrics.

31. A system according to claim 16, wherein the first and second databases are implemented on the same device.

32. A system according to claim 31 , wherein the processor is incorporated into the device in which are implemented the first and second databases.