NL2016943B1 - Collection method for a user-oriented audience measurement, from hits transferred to a node for site-oriented audience measurement, using hit registrations created by the node. - Google Patents

Abstract

Collection technique for a user-oriented audience measurement of consumption messages, called hits, that are transferred to a node for site-oriented audience measurement (5) by markers performed by terminals (2) in panel households (1). The method comprises the following steps: obtaining a collection of first registrations created by at least one intermediate element through which the hits pass, each first registration being a separate hit and including an identification of the terminal that sent the hit; obtaining a collection of second registrations created by the node, each second registration being a separate hit and including at least one hit parameter; and assignment to each second registration of a terminal identification, derived either from a unique first registration or from a plurality of first registrations comprising the same terminal identification.

Description

1, FIELD OF THE INVENTION

The field of the invention is that of audience measurement techniques.

More precisely, the invention relates to a collection technique for a user-oriented audience measurement of consumption messages, called hits, that are transferred to a node for site-oriented audience measurement by markers performed by terminals in panel households. For example, the marker is located in a web page that is executed by a browser, which is itself executed by the terminal. In a variant, the marker is in a third-party application, which itself is executed by the terminal.

In particular, but not exclusively, the invention applies to the user-oriented audience measurement of streamed content (e.g., a TV program) transmitted over the Internet.

More generally, the proposed technique can apply to:

• any type of information network;

• any type of content (in particular, but not exclusive, audio and / or video content: streamed or downloaded content, live content or deferred content, missed TV ('catch up TV'), video on demand (VOD), web radio, online available content (content available on internet sites, ie references in web pages that can be consulted with a browser executed by the terminal), content available on third-party applications performed by the terminal, etc.);

• any type of terminal (fixed or mobile, personal or shared): computer, smartphone, tablet, connected TV, etc.

2. TECHNICAL BACKGROUND

Nowadays there are two main measures (also called 'Web Analytics' tools) to perform a public measurement (in a broad sense) for a website, on the Internet: site-oriented audience measurement (in English 'site-centric') and user-oriented audience measurement ( in English 'user-centric').

2.1 Site-centric measurement

This is a measurement produced by the website on request that itself marks the web pages (HTML pages) that one wants to measure. The main purpose of the site-centric measurement is to measure the traffic volumes. This allows fine and precise monitoring: the behavior of the internet user on the site, the most consulted content, the most frequently used services and the internet return of the visits. The site-centric measurement also applies to measuring the effectiveness of online marketing activities: clicks, conversion rates, sales, return on investment, etc.

The operation of the "site-centric" measurement is based on marking each page of a site with a marker (also called "marker block" or "tag"), for example a Javascript code. As soon as a page is requested by a web browser via a terminal connected to the Internet (computer, smartphones, tablets, connected TVs, etc.), the marker placed on the page sends a request (also called 'hit' or 'consumption message') to the measurement server. This way the traffic on the pages is tracked. The site-centric measurement makes it possible to analyze the number of terminals (identified by their internet browsers) that have consulted a site during a certain period. The browsers are identified using cookies or with the IP address / User Agent pairing when measurement with a cookie is not possible. For a given browser and terminal, a cookie is placed on the terminal by the server of the audience measurement company (Web Analytics system) when a marker is downloaded from the server of the audience measurement company for the first time. This cookie is common to all sites marked by the public measurement company, because it is a "third-party cookie" (in English "third party cookie") that is managed by this company.

Web site marking technology (originally designed to measure website visits, as set out above) has subsequently been extended to third-party applications (also known as 'mobile applications' or 'publisher applications') that can be downloaded from online application stores (eg markers for iOS , Android or Windows). For example, it is suggested to insert a marker ("tag") in the mobile application. For this purpose, there are marker libraries for the various platforms for mobile applications. Mobile applications are tools that facilitate the search and consultation of all types of content for users. A mobile application that is connected essentially behaves like a browser, because information is read on the Internet for each operation or data access. Unlike a browser, a mobile application can generally take into account the functionality of the terminal on which it is running, thereby offering the user more efficiency.

The marking technique for web pages and applications can be summarized as follows. When a publisher marks his website to allow an internet audience measurement company to measure its audience, he integrates a marker into his HTML pages (the marker is then a call to a Javascript routine that is on the server of the audience measurement company) or in its third-party applications (the marker then consists of a library written in native code and is integrated in the third-party application). When each page or third-party application is loaded, a flag identifier (unique and persistent) is generated (if that has not been done before). In the case of a page (from a publisher's site) that can be consulted with the web browser of the terminal, then the tag identification is placed on the terminal by the server of the public measurement company (the cookie is in the generally common to all sites marked by the public measurement company). In the case of a third-party application proposed by a publisher in a third-party application store, the marker identification is generated by the marker and stored in a storage space of the terminal specific to each application, if it is not possible to uniquely identify the terminal to use. This is generally different per application.

In addition to measuring visits to HTML pages and third-party applications, certain markers can also measure the audience of audio and / or video content that is displayed by content playback devices (also known as "players" or "players"). For this purpose these markers are linked to the content players to know their operating status. During the presentation of content (including at the start) the marker sends periodic requests (also called "hits" or "consumption messages") to the internet audience measurement server to transfer the operating status of the player (player). These hits convey different parameters: the identity of the content, the position in the content (the time index of the playback), the status of the player (pause, playback ...), the marker identification (which allows sessions from the same terminal during the same day can be doubled) as well as a temporary session identification which is renewed with every new start of the player. In nominal mode the hits are sent every 60 seconds. The marking technique for web pages and applications makes it possible to measure the consumption (number and duration) of content.

A disadvantage of the marking technique is that it is limited to "site-centric" measurement and does not take into account the concept of a panel (this concept is specific to the "user-centric" measurement). With the "site-centric" measurement it is not possible to identify the internet users separately who are connecting to the site (they do not know who is behind the terminal connected to the Internet). It therefore does not make it possible to have a qualified audience. That is why the "user-centric" measurement (detailed below) complements the "site-centric" measurement.

2.2 "User-centric" measurement

This measurement is based on following the navigation (surfing) on it

Internet from a panel (sample of internet users) that is representative of a population of internet users (for example Dutch internet users). The navigation of the panel is measured and an exploratory study is carried out in parallel to extrapolate their navigation to the internet population studied. By accurately identifying the panel member, the "user-centric" measurement can be used to analyze the number of unique visitors who have visited a site within a certain period. A unique visitor to the panel corresponds to a qualified and identified individual. Extrapolated, a panel member corresponds to thousands of unique visitors. The "user centric" measurement gives agencies, publishers and marketing professionals accurate and reliable reports about the audience of internet sites and internet use. This information can be used for strategic plans, competition investigations or media planning. For example, the results are updated every month.

A first known technique for 'user-centric' audience measurement is based on digital tattooing (in English 'audio watermarking') of an audio channel (also called 'audio component') of broadcast content (for example TV channels) or comparing their audio fingerprint ( in English 'audio fingerprinting') with reference fingerprints. This first known technique forms the basis for the audimeters that are used for reference research into daily TV viewing behavior. In the case of a TV set, the audimeter (often a mini-PC) captures the audio channel of the displayed channel (electrically or acoustically) in order to be able to process the signal and transfer the audience data to the server from the public measurement company. A disadvantage of this first known technique is that in the case of portable PCs, smartphones or tablets it is difficult to imagine permanently connecting an audimeter (i.e. an external device) to these by their nature mobile devices. Another disadvantage of this first known technique is that on certain types of terminals (often Apple terminals (registered trademark)) it is very difficult to capture the audio channel in digital format, directly at the level of the audio circuits.

A second known technique for "user-centric" audience measurement is based on the analysis of network traffic between the terminal and the Internet. This second known technique consists of utilizing network traces left behind at the level of a router installed in the home of the panel members or at the level of an HTTP proxy through which the terminals of the panel members must pass. For example, in the context of measuring internet traffic in a household recruited to participate in a panel, a solution that is used by the public measurement company is to connect a router to the box (ie the router modem, also called 'network connection equipment', supplied by the internet provider) and to ask the household to connect all terminals to it in order to be able to log the internet traffic that passes through it. The main disadvantage of this second known technique is that the network traffic may have no connection whatsoever with the content that is actually displayed on the terminal. For example, when the user pauses the player (player), the audiovisual stream is further stored in buffer memory, with no guarantee that it will later be decoded and displayed. So there is an inaccuracy in the measurement of streamed content that may be buffered in the terminals and never consumed. Another disadvantage is that this solution is limited to the home where the router is located.

A third known technique for "user-centric" audience measurement is based on a measurement program (also called "measurement software" or "measurement application") that is built into the terminal in order to be able to transfer and analyze the connection data (surf data) in real time. The "measurement application" was developed by the public measurement company and installed by the panel member on its terminal to be able to measure its use. The application works as a background task and, in a special implementation, installs a local proxy (also called "process proxy") on the terminal. All network connections (Wi-Fi, 3G / 4G, ...) pass through the local proxy and it can record all HTTP requests sent to the target servers as it passes. The main disadvantage of this third known technique is that it requires a very consistent R&D effort to maintain a measurement program on all operating systems (OS) and all internet browsers of the terminals, taking into account the developments of the large parties on the web, and that just as well on a computer (PC), tablet, smartphone, etc. On the other hand, when audiovisual content (stream) is protected with a technical security measure (or DRM, "Digital Rights Management" in English), the measurement program can generally not measure their consumption.

2.3 Use of the hits of a "site-centric" measurement in the context of a "user-centric" measurement

As explained earlier, the "site-centric" measurement is based on the sending of hits by the markers that are integrated in web pages or third-party applications. It has been proposed to use these hits in the context of a "user-centric" measurement. For example, as in the second known "user-centric" audience measurement technique presented above, a router installed in the household of panel members records (hits) transmitted by the terminals of that household. The thus logged hits can then be used by the public measurement company to know which HTML page or part of a third-party application has been visited by the panel member. But above all, in the case of hits transmitted by markers cooperating with players, they can be utilized to effect a measurement of video consumption. The knowledge of internet traces (HTTP requests) alone is not sufficient to accurately know the duration of consumption of streamed content, because the content may be placed in buffer memory and may be delayed. The marker, on the other hand, is a means to accurately find out which part of the content has been decoded and presented. In the context of measuring TV on internet displays (PC, tablet, mobile), one can use this technology to measure the audience of television channels (live, delayed or missed) in a panel of viewers. The condition is of course that the sites and the applications of the channels use this marking technology.

For several years it has been observed that internet parties are increasingly migrating their servers to secure connections via the HTTPS protocol. By setting up this type of connection, the internet user is assured on the one hand that he connects to the correct site and not to a hijacked site (server authentication) and on the other hand that the communication is encrypted and cannot be intercepted (confidentiality). The HTTPS protocol was originally used by banking or e-commerce sites, but is now largely used by all sites that belong to social networks and that store or exchange personal, potentially sensitive information. When an HTML page is loaded via HTTPS (that is, via a secure connection according to the HTTPS protocol), the internet browser requires that all parts of this page (images, Javascript, ...) be loaded via HTTPS from the point of view of the security, in order to prevent malicious code from entering through a secure website. Nowadays, most browsers still accept that items are loaded via the HTTP protocol in a page that is loaded via the HTTPS protocol. Only a warning icon will be displayed for the address bar, but it is likely that loading will be blocked in future versions of the browsers. Like all components, a marker from a public measurement company must also be loaded via HTTPS. Likewise, the hits sent by these markers must be sent via HTTPS.

Returning to the aforementioned solution for 'user-centric' measurement based on the utilization of registered (logged) hits by a router present in a household of panel members (that is the second known technique presented above), it is determined that the router that is placed for every hit in the middle of an encrypted HTTPS connection (established between a terminal in the household and the server of the public measurement company), only logs the domain name to which the hit is sent, but does not register any of the parameters included in this hit . For example, for hits transmitted by markers that interact with playback devices (players), the router does not record the status of the player, the name of the content, and the time index of playback. The logs (recordings of network traces corresponding to hits) can therefore not be used to achieve a "user-centric" measurement (of video consumption in the aforementioned example).

Similarly, a disadvantage of the third known technique is that if the 'measurement application' and the application in use (of which one wants to measure internet traffic) are each in their own sandbox, the local proxy of the 'measurement application' after having an HTTPS -connection has no longer access to the details of the hits (same problem with a router or an external proxy). The hits sent via HTTPS (for example hits of the "eStat streaming" type) cannot be used. A sandbox is a mechanism with which (a) program (s) can be executed with fewer risks for the operating systems.

In other words, in the context of a 'user-centric' measurement, it is currently possible to use the parameters in the hits that are sent over a non-secure connection (for example via the HTTP protocol), but not the parameters in the hits that are sent over a secure connection (for example via the HTTPS protocol). It is therefore not possible to use all hits sent by the terminals in a household of panel members and it is therefore not possible to achieve a public measurement in all cases.

3. OBJECTS OF THE INVENTION

In at least one embodiment, the invention has the particular object of overcoming these various disadvantages of the prior art.

More precisely, in at least one embodiment of the invention, it is an object to provide a technique with which it is possible to collect (and thus make usable) all parameters in the context of a 'user-centric' measurement (user-oriented audience measurement) are included in the hits sent by the terminal or the terminals of a panel household, hits that are transmitted either via an unsecured connection (e.g. via the HTTP protocol) or via a secure connection (e.g. via the HTTPS protocol).

At least one embodiment of the invention also has the purpose of providing such a technique that requires no modification of the existing markers nor of the existing terminals.

Another object of at least one embodiment of the invention is to provide such a technique that is easy to use and inexpensive.

4, EXPLANATION OF THE INVENTION

In a particular embodiment of the invention, a collection technique is proposed, for user-oriented audience measurement, of consumption messages, called hits, that are transmitted to a node for site-oriented audience measurement by markers performed by terminals in panel households. The method comprises the following steps:

- obtaining a collection of first registrations, created by at least one intermediate element through which the hits pass, each first registration being a separate hit and comprising an identification of the terminal that sent the hit;

- obtaining a collection of second registrations created by the node, each second registration being a separate hit and comprising at least one hit parameter, and

- assignment to every second registration of a terminal identification, derived either from a unique first registration or from several first registrations comprising the same terminal identification.

The principle of the proposed solution therefore consists of assigning a terminal identification to every second registration (also called "second log") created by the site-oriented audience measurement node (also called "site-oriented measurement node"). This makes it possible to associate a terminal identification with (the) known parameter (s) in every second registration created by the node. This terminal identification (for example the MAC address of the terminal) makes it possible to know information that is required for a user-oriented measurement, such as, for example, the identity of the person (s) registered at the terminal. and / or the type of terminal (for example 'shared' or 'individual', 'iOS tablet' or 'Android smartphone' ...).

It is thus possible to collect (and make usable in the context of a user-oriented audience measurement) all parameters in the hits sent by the terminal (s) of a panel household, even when these hits are transmitted via a secure connection (and the network equipment therefore does not have access to these parameters). In the context of a user-oriented audience measurement, the user of the terminal is generally asked (at the start of a user session) to log in by selecting his first name from a list of persons in the household. This information (registration) is linked to the identification of the terminal (for example the MAC address). You can therefore use the terminal identification as an association key between:

• on the one hand the second hit registrations of the terminal (second logs of hits created by the node), whereby a terminal identification is assigned to each of these second registrations, and • on the other hand the content of the notification (identity of the person), to inform the public. qualify with socio-demographic data.

According to a special feature, said hit parameter belongs to the group comprising: information about the status of a content player, an identification of the content, a time index of the playback of content and an identification of the visualization session.

In this way, the proposed solution applies to a collection for a user-oriented audience measurement of audio and / or video content displayed by content players ("players").

In a first particular implementation, the set of first registrations comprises a plurality of subsets, each comprising the first registrations created by a separate intermediate element, which is a router installed in a panel household to which each terminal of said panel household is connected to access an information network.

According to a particular feature of this first particular implementation, the allocation step comprises the following steps, for a given panel household comprising a given router:

- obtaining a first subset, comprising the first registrations created by the given router, with a network address of the given household therein;

- obtaining a second subset, comprising the second registrations created by the node, with said network address of the given household;

- determining the first sessions in which groups are based, starting from the first subset and thanks to the terminal identification included in each first registration, first registrations associated with a consumption of the same content by the same terminal;

- determining the second sessions in which groups are based, starting from the second subset and thanks to a session identification included in each second registration, second registrations associated with a consumption of the same content by the same terminal;

- obtaining pairs of first and second sessions, by assigning a first session to every second session, whereby the start times of the second sessions are compared with the start times of the first sessions, and

- for each couple of first and second sessions, assignment to every second registration of the second session of said couple of terminal identification that is common to all first registrations of the first session of said couple.

In a second particular implementation, the set of first registrations comprises a plurality of subsets each comprising first registrations created by a separate intermediate element, which is an internal proxy in one of the terminals.

According to a particular feature of this second particular implementation, the mapping step comprises the following steps, for a given terminal comprising a given internal proxy:

- obtaining a first subset, comprising the first registrations, created by the given internal proxy, with a network address of the given terminal therein;

- obtaining a second subset, comprising the second registrations created by the node, with said network address of the given terminal;

- determining the first sessions in which the groups are grouped, starting from the first subset and thanks to the terminal identification in each first registration, first registrations associated with a consumption of the same content by the given terminal;

- determining the second sessions in which groups are based, starting from the second subset and thanks to a session identification in every second registration, second registrations associated with a consumption of the same content by the given terminal;

- obtaining pairs of first and second sessions, by assigning a first session to every second session, whereby the start times of the second sessions are compared with the start times of the first sessions, and

- for each couple of first and second sessions, assignment, to every second registration of the second session of the said couple, of the terminal identification that is common to all first registrations of the first session of the said couple.

According to a special feature of the first and / or the second particular implementation, the acquisition of the pairs of first and second sessions is accomplished by also comparing the end times of the second sessions with the end times of the first sessions.

According to a particular feature of the first and / or the second particular implementation, if, for a given first session, the comparison of the end times indicates that the first given session must be assigned to a different second session than the session indicated by the comparison of the start times, then the first given session is simultaneously assigned to the second session indicated by the comparison of the start times, thus forming a first session couple, and to said second session indicated by the comparison of the end times, thus a second forming session couple.

In a third particular implementation, the intermediate element is an external proxy outside the terminals, through which the network traffic passes from the terminals, to gain access to an information network.

According to a special feature of this third particular implementation, for a given hit transferred by a given terminal:

- the first registration regarding the given hit includes information specific to the given terminal, at least at the time of the transfer of the given hit, inserted by the proxy;

- the second registration concerning the given hit comprises the said information specifically for the given terminal, at least at the time of the transfer of the given hit, inserted by the node, and

- the terminal identification included in the first registration regarding the given hit is assigned to the second registration regarding the given hit, using at least the said information specific to the given terminal as an association key.

According to a special feature of this third particular implementation, said information specific to the given terminal comprises an outbound port number for the proxy through which the hit goes.

In another embodiment of the invention, a computer program product is proposed that includes the program instruction code for deploying the aforementioned method (in any of the various embodiments) when said program is run on a computer.

In another embodiment of the invention, a storage medium is proposed that is computer readable and non-transient, on which is stored a computer program that includes an instruction set executable by a computer for deploying the aforementioned method (in any of the different embodiments).

In another embodiment of the invention, a collection device is proposed, for user-oriented audience measurement, of consumption messages, called hits, that are transmitted to a node for site-oriented audience measurement by markers performed by terminals in panel households. The collection device is configured and adapted for:

- obtaining a collection of first registrations created by at least one intermediate element through which the hits pass, each first registration being a separate hit and including an identification of the terminal that sent the hit;

- obtaining a collection of second registrations created by the node, each second registration being a separate hit and comprising at least one hit parameter, and

- assigning to every second registration of a terminal identification, derived either from a unique first registration or from several first registrations comprising the same terminal identification.

The collection device advantageously comprises means for inserting the steps it accomplishes in the collection method as previously described, in any of the various embodiments.

5. LIST OF FIGURES

Other features and advantages of the invention appear upon reading the following description, given as an indicative and non-limitative example, and the accompanying drawings, in which:

figure 1 illustrates the general principle of a collection method according to a first special embodiment of the invention;

figure 2 presents an overview of a system that makes it possible to use a collecting method according to the first special embodiment of the invention;

figure 3 presents an organization chart of a collection method according to the first special embodiment of the invention;

- figures 4 to 7 illustrate the acquisition step of couples of first and second sessions shown in figure 3;

figure 8 presents an overview of a system that makes it possible to use a collecting method according to a second special embodiment of the invention;

figure 9 presents an organization chart of a collection method according to the second special embodiment of the invention;

figure 10 presents the structure of a collecting device according to a special embodiment of the invention, and

- figure 11 presents an overview of a system that makes it possible to use a collecting method according to a third special embodiment of the invention;

6. DETAILED DESCRIPTION

It is to be noted that the second and third implementations of the preceding paragraph ("Explanation of the invention") correspond to the third and second embodiments of this paragraph, respectively.

In all figures of the present document identical elements and steps are indicated with the same reference number.

Now, in conjunction with Figures 1 to 7, a collection method according to a first particular embodiment of the invention is presented.

As illustrated in Figures 1 and 2, a panel household 1 is considered, comprising at least one terminal 2, 2a, 2b (computer, smartphone, tablet, connected TV, etc.) connected to an information network (often the Internet) via a box 4 (also called 'router modem' or 'network connection equipment') provided by an internet provider. Only from the point of view of simplification is the box not shown in Figure 2.

It is assumed that each terminal participates in the deployment of a site-oriented ("site-centric") measurement. To this end, he executes markers (integrated in web pages or third-party applications) that transfer hits (consumption messages) to a node for site-oriented audience measurement 5. In the case that they are transferred by markers that work together with players, these hits usually include included as parameters are the status of the player, the name of the content, the time index of the playback and an identification of the visualization session (the exploitation of which will be worked out below).

For example, each marker is located in a web page that is executed by a browser, which is itself executed by the terminal. In a variant, each marker is in a third-party application, which itself is executed by the terminal.

It is also assumed that people want to use these hits (including the parameters included in this) to use a user-oriented ("user-centric") measurement. To this end, in this first particular embodiment illustrated in Figures 1 and 2, the panel household 1 is equipped with, inter alia, a router 3. Each terminal of the household is connected to the router 3, which itself is connected to the box 4. The router 3 can thus log the ongoing internet traffic from each terminal 2, 2a, 2b of the household.

In this first particular embodiment, the proposed technique makes it possible to collect the hits and the parameters therein, regardless of the nature (secured or not) of the connection 6 established between the terminal 2 and the node 5.

In the case that the connection between the terminal 2 and the node 5 is secured (for example according to the HTTPS communication protocol), where the router 3 is placed in the middle of this HTTPS connection for each hit, then the router creates a registration (also log) 'called) which is no different than a network trace comprising the domain name to which the hit was sent and an identification of the terminal that sent the hit. Each terminal connected to the router is often identified by a unique connection identifier (for example, its MAC address or its local IP address on the local network (LAN) of the household), allowing the router to log the internet traffic from each terminal separately . In Fig. 2, the logs associated with the hits transmitted by the terminal A (reference 2a) are indicated by 21a and the logs associated with the hits transmitted by the terminal B (reference 2b) are designated 21b.

But every hitlog created by the router does not include any of the parameters included in the hit. For example, for hits transmitted by markers that interact with playback devices (players), the router does not record the following parameters: the status of the player, the name of the content, the time index of playback, and the identification of the visualization session. The hit logs 21a, 21b created by the router are therefore not in themselves usable for effecting a "user-centric" measurement.

The general principle of the collection method according to the first embodiment consists of looking up the hit parameters directly in the records (logs) 22 of the site-centric measurement node 5. To this end, it is possible to find the public IP address (network address) of the household as an association key, because the measurement node logs the (public) source IP address of the hits. In other words, the log created by the node 5 and the log created by the router 3 each include the public IP address of the router that sent this hit for each hit.

For example, the router 3 can consult a web service 8 such as "http://www.whatismyip.com/" to know its public IP address, that is the address of the box 4 of the household 1 (also "IP address of called the household). This consultation is symbolized by the arrow with reference 9 in figure 2.

As soon as the IP address is known and registered by the router 3, the router sends the address to a collection device 20 (this transfer is symbolized by the arrow with reference 10 in Figure 2). Thus, the collecting device 20 can subsequently retrieve the logs 22 of the measuring node 5 which relate to the hits (with their parameters) sent from this IP address. This retrieval is symbolized in Figure 1 with the arrow with reference 7. Incidentally, the collecting device 20 receives the hit logs 21a, 21b created by the router 3.

But all terminals in a household have the same public IP address (that of the box 4 of the household), as a result of which the collecting device 20 cannot directly assign the hit logs 21a, 21b created by the router 3 to the hit logs 22 created by the measurement node 5 In other words, the use of the public IP address of the household as an association key is not sufficient. It does, however, make it possible to obtain a subset of hit logs created by the router 3 and a subset of hit logs created by the measurement node 5.

But it does not make it possible to assign a hitlog created by the router 3 (this log comprising an identification of the terminal that sent the hit, for example its MAC address) to every hitlog created by the measurement node 5.

In other words, it does not allow the collecting device 20 to associate to every hit log created by the measurement node 5 the identification of the terminal that sent this hit. However, this is necessary in the context of a user-oriented audience measurement, on the one hand to know the identity of the person (s) registered at this terminal and on the other hand to know the type of terminal. In the context of a user-oriented audience measurement, the user of the terminal is generally asked to log in by selecting his first name from a list of persons in the household. This information (registration) is then linked to the traffic of the terminal (either in the router or at the back office level), in order to qualify the public with socio-demographic data.

In order to enable the collecting device 20 to associate with each hit log created by the measurement node 5 the identification of the terminal that sent the hit, a solution proposed in the first particular embodiment of the invention consists of reconstituting visualization sessions from the side of the router 3 and bring them closer to calculated visualization sessions on the part of the measuring node 5. This approach between visualization sessions (also called comsumption sessions) is achieved, for example, on the basis of the closest start and end times of sessions.

Figure 3 presents an organization chart of the collection method according to the first particular embodiment of the invention, including the aforementioned solution with approximation of visualization sessions (hereinafter also simply referred to as "sessions"). The collecting device 20 accomplishes this method for each panel household equipped with a router.

In a step 31, the collection device 20 obtains the first hit logs created by the router 3 containing the IP address of the household.

In a step 32, the collection device 20 obtains the second hit logs created by the measurement node 5, which also contains the IP address of the household.

In a step 33, the collecting device 20 determines the first sessions in which are grouped, starting from the result of step 31 and, thanks to the terminal identification included in each first hit log, first hit logs associated with a consumption of the same content by the same terminal;

In a step 34, the collecting device 20 determines the second sessions in which are grouped, starting from the result of step 32 and thanks to a session identification included in each second hit log, second hit logs associated with a consumption of the same content by the same terminal;

In a step 35, the collecting device 20 obtains pairs of first and second sessions, by assigning a first session to every second session, by comparing the start times of the second sessions with the start times of the first sessions, as well as (in a special bet) the end times of the second sessions with the end times of the first sessions.

In a step 36, for each pair of first and second sessions, the collecting device 20 assigns to each second hit log of the second session of the pair the terminal identification that is common to all first hit logs of the first session of the pair.

Thereby the intended goal can be achieved: knowing the identity of the user (s) and the type of terminal that has consumed the content to which the couple comprising this second session and the first session assigned to it relate.

Figures 4 to 7 illustrate the aforementioned acquisition step of couples of first and second sessions, represented by 35 in Figure 3.

Although there are several separate technologies of Web Analytics, each technology generally uses a concept of unique session identification to group all hits (and therefore all corresponding hit logs) in a single and unique visualization session.

For the hit logs created by the measurement node 5, one can therefore identify a visualization session thanks to the unique session identification carried by the logged hit parameters. Thus knows with certainty the start time and the duration of the session thanks to the times of the first and the last hit.

For the hit logs created by the router 3, the information for each HTTPS hit is limited to the time of transmission, because none of the parameters carried by the hit can be read by the router. However, thanks to the unique identification of the connection (for example the MAC address), the router has the option of separately logging the HTTPS hits sent by each terminal. Thus, the simultaneous consumption of the same video content by two terminals in the household leads to two separate sets of hit logs. Although there are several separate Web Analytics technologies, the majority of corresponding markers periodically send hits (for example, every 60 seconds in nominal mode for Médiamétrie's eStat streaming technology) to indicate that the player (player) is still on is playing. Hits can also be sent to indicate actions (for example pause / play / stop / skip / ...) on the video stream. Under these circumstances, the likely end of a session can be detected by the absence of a hit during the nominal period (60 s in the case of the "eStat streaming" marker).

For a given household with a known IP address, Figure 4 illustrates the approach of second sessions (grouping the second hit logs created by the measurement node 5) to the first sessions (grouping the first hit logs created by the router 3). In this example, the first sessions with references S _r / A, i and S _r / A, 2 (each grouping the first hit logs transmitted by the terminal A) are assigned to the second session with reference S _n , 3 and the second session respectively with reference S _n , 3. The first session with reference S _r / B, i (each grouping the first hit logs transmitted by the terminal B) is assigned to the second session with reference S _n , 2.

More accurately, as illustrated in Figure 5, for a first session that groups the first hit logs created by the router (in this example with the reference S _r / B, i), the collector 20 calculates the time distance between the start time of this first session and the start time of each of the nearby second sessions (grouping the second hit logs created by the measurement node) and retaining the second session for which the time distance is smallest. In the illustrated example, of the time distances D1, D2 and D3, the smallest is D2, so the first session with reference S _r / B, i is assigned to the second session with reference S _n , 2.

As illustrated in Figure 6, the same comparison is made with the end time to verify that the end time of the second session preserved (S _n , 2 in this example) also corresponds to the shortest of the time distances between the end time of the first session and the end times of the nearby second sessions. In the illustrated example, it is verified because of the time distances D1 ', D2' and D3 'is the smallest D2'.

Figure 7 illustrates the case where a terminal (in this example terminal A) connects two visualization sessions in a very short time, shorter than the period of transmission of hits by the marker. The two first sessions with reference S _r / A, i and S _r / A, 2 in figures 4 to 6 are aggregated into a single first session with reference S _r / A, i in figure 7. This creates an inconsistency for the allocation: by comparing the start times of the session, the first session with reference S ' _r / A, i will be assigned to the second session with reference S _n , i; but the comparison of the end times of the session will lead to assignment to the second session with reference S _n , 3. The proposed solution in this case consists of assigning the first session with reference S ' _r / A, i (and therefore terminal A) to two second sessions with references S _n , i and S _n , 3. In other words, two session couples are thus formed: (S ' _r / A, i, S _n , i) and (S' _r / A, i, Sn, 3). Thus, after the completion of step 36 using each of these pairs, the intended goal is achieved: knowing the identity of the user (s) and the type of terminal that the second sessions with reference S _n , i and S _n , 3 consumed, logged by the measurement node.

Now, in conjunction with Figures 8 and 9, a collection method according to a second particular embodiment of the invention is presented.

As illustrated in Figure 8, this second embodiment differs from the first (described above) in that the network equipment with which the first hit logs can be obtained (i.e. logging the internet traffic from terminals of households of panel members) is not a router (for every household) is but a proxy 82.

In Figure 8, of the logs created by the proxy 82, the logs associated with the hits transmitted by the terminal A (reference 2a) are designated 81a and the logs associated with the hits transmitted through the terminal B (reference 2B) are designated 81b.

The terminals of the same household go through the same proxy, either using separate input ports, individually assigned to each terminal, or using a common port for all household terminals (or the panel), but with separate login and password to enable them. distinguished (in the case of a proxy with authentication). In all cases, the proxy is therefore able to log the terminals of a household separately. In other words, each hit log created by the proxy 82 includes an identification of the terminal that sent the hit.

In case the connection between the terminal 2 and the node 5 is secured (for example according to the HTTPS communication protocol), where the proxy 82 is placed in the middle of this HTTPS connection for each hit, the proxy creates a hit log that is no different then a network trace comprising the domain name to which the hit was sent and an identification of the terminal that sent the hit. But this hitlog created by the proxy does not include any of the parameters included in the hit.

The source IP address of the logged hits on the measurement node is always that of the proxy 82. A different association key must be used, such as, for example, the outgoing proxy port from the connection to the measurement node (also called 'source port'), which means different terminals that come out of the proxy can be distinguished.

It is to be noted that in order to apply the same allocation mechanism to the proxy 82 as to the router 3 (in the first embodiment), it must be borne in mind that the proxy masks all terminals behind the same proxy IP address, just as all the terminals of a household are masked after the same IP address of the household. The biggest difference is that thousands of terminals are hidden behind the proxy, while only a few are hidden behind the router. Thus, the allocation based on the probable start / end of the sessions has little chance of functioning effectively if there are thousands of terminals.

The outbound port of the proxy is information specific to the terminal, at least during the transfer of the hit. More accurately, in a first implementation, the information is only specific to the terminal in a temporary manner. In other words, the outgoing ports of the proxy are dynamically reassigned. An outbound port of the proxy is only assigned (and therefore only specific) for a given terminal for a limited period of time. In a variant, the specific information is invariable in time (all hits sent by a given terminal use the same outbound port of the proxy, which is specific to the given terminal). For this variant it is necessary to have a larger number of outgoing ports of the proxy and / or to handle a smaller number of terminals.

In the second embodiment, for each hit, the proxy 82 must retrieve and log the outgoing port number of the proxy. In other words, for each hit, the first log of this hit created by the proxy 82 includes the outbound port number of the proxy through which the hit goes to reach the measurement node.

The measurement node does the same. He must retrieve and log the proxy outbound port number. In other words, for each hit, the second log of the hit created by the measurement node 5 includes the outbound port number of the proxy through which the hit goes to reach the measurement node.

Thus, the collecting device 20 can bring the first hit logs 81a, 81b created by the proxy and the second hit logs 22 created by the measurement node closer together. In the event that the outbound ports of the proxy are dynamically reassigned, this approach is not only based on the outbound port number of the proxy (included in both the first and second hit logs), but also at the time of transfer of each hit (both the first and the second hit logs are time stamped).

Figure 9 presents an organization chart of a collection method according to the second particular embodiment of the invention.

In a step 91, the gatherer 80 obtains the first hit logs created by the proxy 82 with in each case an outbound port number of the proxy (information inserted by the proxy and specifically for the terminal that transmitted the hit) and time stamp information (substantially corresponding to the time of transfer of the hit between the proxy and the node).

In a step 92, the collecting device 80 obtains the second hit logs, created by the measurement node 5, which also contains in each case an outbound port number of the proxy (information inserted by the measurement node and specifically for the terminal that transmitted the hit) and time stamp information (substantially corresponding with the moment of transfer of the hit between the proxy and the node).

In a step 93, the gatherer 80 assigns a first hit log (and thus the terminal identification therein) to every second hit log, using the outbound port number of the proxy as an association key (in combination with the timestamp information in the hit logs, in the case of a dynamic re-allocation of outgoing ports of the proxy). Thereby the intended goal can be achieved: associating the parameters included in the second hit log with the terminal that transmitted the hit, whereby the identity of the user (s) and the type of terminal that consumed the content can be known.

Now, in conjunction with Figure 11, a collection method is presented according to a third particular embodiment of the invention.

This third embodiment differs from the first (described above with figures 1 to 7) in that the intermediate element that makes it possible to obtain the first hit logs (i.e. log the internet traffic from the terminals of households of panel members) is not the router 3 of the household, but an internal proxy within each terminal, hereafter called 'local proxy'.

More precisely, each terminal 2a ", 2b" of the household comprises:

• an application 2aA, 2as whose internet traffic is to be measured. This application is, for example, an internet browser that executes web pages with markers therein or a third-party application with markers therein. The markers, when executed, transfer hits (consumption messages) to the node for site-oriented audience measurement 5;

• a "measurement application" 2bA, 2bs (see definition above), which functions as a background task and installs a local proxy 2ca, 2cb (also called "process proxy") on the terminal.

In the case that the connection between a terminal 2a ', 2b' (more precisely an application 2aA, 2aB whose traffic is to be measured) and the node 5 is protected (for example according to the HTTPS communication protocol), where the local proxy 2ca, 2cb for every hit is placed in the middle of this HTTPS connection, the proxy creates a registration (also called 'log') that is no different than a network trace including the domain name to which the hit was sent and an identification of the terminal that sent the hit ( for example his MAC address). In Figure 11, the logs associated with the hits transmitted by the terminal A (reference 2a ") are indicated by 21a" and the logs associated with the hits transmitted by terminal B (reference 2b ") are indicated by 21b".

But each hitlog created by the local proxy 2ca, 2cb does not include any of the parameters included in the hit. For example, for hits transmitted by markers that interact with playback devices (players), the local proxy does not record the following parameters: the status of the player, the name of the content, the time index of the playback, and the identification of the visualization session. The hit logs 21a, 21b created by the local proxies 2ca, 2cb are therefore not in themselves usable for effecting a "user-centric" measurement.

As in the first embodiment, the proposed solution consists of looking up the hit parameters directly in the records (logs) 22 of the site-centric measurement node 5. But in the third embodiment the public IP addresses (network addresses) of the terminals are used for this as an association key, because the measurement node logs the (public) source IP address of the hits. In other words, the log created by the node 5 and the log created by the local proxy 2ca, 2cb each include the public IP address of the terminal that sent that hit for each hit.

For example, the "measuring application" 2bA, 2bs of each terminal 2a ", 2b" can consult a web service 8 such as "http://www.whatismyip.com/" in order to know the public IP address of this terminal. This consultation is symbolized by the arrows with references 9a 'and 9b' in Figure 11. The public IP address of a terminal results from an address translation, starting from a private IP address in the private network 110 to which the terminal belongs (local network behind) the box 4 of the household 1, mobile network ...).

As soon as the public IP address of the terminal is known and registered by the 'measuring application' 2bA, 2bB, the latter transfers the address to the collecting device 20 '(this transfer is symbolized by the arrows with references 10a' and 10b 'in the figure 11). Thus, the collecting device 20 can subsequently retrieve the logs 22 of the measuring node 5 which relate to the hits (with their parameters) sent from this IP address. Incidentally, the collecting device 20 "receives the hit logs 21a", 21b "created by the local proxies 2ca, 2cb.

But multiple terminals in the same private network 110 may have the same public IP address, as a result of which the collecting device cannot directly assign the hit logs 21a ', 21b' created by the local proxies 2ca, 2cb to the hit logs 22 created by the measurement node 5. In other words , the use of the public IP address of the terminal as an association key is not sufficient. It makes it possible to obtain a subset of hitlogs created by the local proxies 2ca, 2cb and a subset of hitlogs created by the measurement node 5.

But it does not make it possible to assign a hitlog created by the local proxies 2ca, 2cb (this log comprising an identification of the terminal that sent the hit, for example its MAC address) to every hitlog created by the measurement node 5.

In other words, it does not make it possible for the collecting device 20 ’to associate to every hit log created by the measurement node 5 the identification of the terminal that sent this hit. But, as explained above, this is necessary in the context of a user-oriented audience measurement, on the one hand to know the identity of the person (s) registered at this terminal and on the other hand to know the type of terminal.

In order to allow the collecting device 20 'to associate with each hit log created by the measurement node 5 the identification of the terminal that sent the hit, a solution proposed in the third particular embodiment of the invention is to reconstitute visualization sessions of the the local proxies 2ca, 2cb and to bring them closer to calculated visualization sessions on the part of the measuring node 5. This approach between visualization sessions (also called comsumption sessions) is accomplished, for example, on the basis of the closest start and end times of sessions (as explained above with figures 4 to 7).

It must be underlined that the greater the number of terminals in the private network (and therefore the greater the likelihood that multiple terminals of this private network have the same public IP address), the greater the risk of assignment errors (errors in bringing visualization sessions closer together) ) is.

Figure 10 presents the structure of a collection device 100 according to a special embodiment of the invention. This collecting device also has the reference 20 in Figure 2 (first embodiment described with Figures 1 to 7), 80 in Figure 8 (second embodiment described with Figures 8 and 9) and 20 'in Figure 11 (third embodiment described with Figure 11 ).

It comprises a directly accessible memory 103 (for example a RAM memory), a processing unit 102, equipped with for example a processor and controlled by a computer program stored in a read-out memory 101 (for example a ROM memory or a hard disk). For example, during the initialization, the instruction code of the computer program is loaded into the directly accessible memory 103 before it is executed by the processor of the processor 102.

This figure 10 only illustrates a particular one of several possible forms, whereby the collecting device realizes the various illustrated steps of the first or the second embodiment. The proposed technique can be realized on any reprogrammable calculator (a computer, a DSP processor or a microcontroller) that executes a program that includes an instruction sequence, or on a function-bound calculator (e.g., a set of logical ports such as an FPGA or an ASIC, or any other hardware module). In the case of an implementation on a reprogrammable calculator, the corresponding program (i.e., the instruction set) can be stored on a removable or non-removable storage medium (such as, for example, a floppy disk, a CD-ROM, or a DVD-ROM), the storage medium can be read in whole or in part by a computer or a processor.

Claims (14)

CONCLUSIONS A collection technique for a user-oriented audience measurement of consumption messages, called hits, that are transferred to a site-oriented audience measurement node (5) by markers performed by terminals (2) in panel households (1), characterized in that it comprises the following steps : - obtaining (31; 91) a set of first registrations, created by at least one intermediate element (3; 82) through which the hits pass, each first registration being a separate hit and including an identification of the terminal that sent the hit; - obtaining (32; 92) a set of second registrations created by the node, each second registration being a separate hit and comprising at least one hit parameter, and - assignment (33-35; 93) to every second registration of a terminal identification, derived either from a unique first registration or from several first registrations comprising the same terminal identification. A method according to claim 1, wherein said at least one hit parameter belongs to the group comprising: information about the status of a content player, an identification of the content, a time index of playing content and an identification of the visualization session. A method according to any one of claims 1 and 2, wherein the set of first registrations comprises a plurality of subsets, each comprising the first registrations created by a separate intermediate element, which is a router (3) installed in a panel household and connected to is each terminal (2a, 2b) of said panel household to gain access to an information network. The method of claim 3, wherein the allocation step comprises the following steps, for a given panel household comprising a given router: - obtaining (31) a first subset, comprising the first registrations created by the given router, with a network address of the given household therein; - obtaining (32) a second subset, comprising the second registrations created by the node, with said network address of the given household therein; - determining (33) the first sessions in which groups are based, starting from the first subset and thanks to the terminal identification included in each first registration, first registrations associated with a consumption of the same content by the same terminal; - determining (34) the second sessions in which groups are based, starting from the second subset and thanks to a session identification included in each second registration, second registrations associated with a consumption of the same content by the same terminal; - obtaining (35) pairs of first and second sessions, by assigning a first session to every second session, comparing the start times of the second sessions with the start times of the first sessions, and - for each pair of first and second sessions, assignment (36) to each second registration of the second session of said couple, of the terminal identification that is common to all first registrations of the first session of said couple. The method of any one of claims 1 and 2, wherein the set of first registrations comprises a plurality of subsets, each comprising first registrations created by a separate intermediate element, which is an internal proxy in one of the terminals (2a, 2b). The method of claim 5, wherein the mapping step comprises the following steps, for a given terminal comprising a given internal proxy: - obtaining a first subset, comprising the first registrations, created by the given internal proxy, with a network address of the given terminal therein; - obtaining a second subset, comprising the second registrations created by the node, with said network address of the given terminal; - determining the first sessions in which groups are based, starting from the first subset and thanks to the terminal identification included in each first registration, first registrations associated with a consumption of the same content by the given terminal; - determining the second sessions in which groups are based, starting from the second subset and thanks to a session identification included in each second registration, second registrations associated with a consumption of the same content by the given terminal; - obtaining pairs of first and second sessions, by assigning a first session to every second session, whereby the start times of the second sessions are compared with the start times of the first sessions, and - for each couple of first and second sessions, assignment, to every second registration of the second session of the said couple, of the terminal identification that is common to all first registrations of the first session of the said couple. Method according to claim 4 or 6, wherein the acquisition of the pairs of first and second sessions is achieved by also comparing the end times of the second sessions with the end times of the first sessions. A method according to claim 7, wherein, for a given first session, if the comparison of the end times indicates that the first given session must be assigned to a different second session than the session indicated by the comparison of the start times, the the first given session is simultaneously assigned to the second session indicated by the comparison of the start times, thus forming a first session couple, and to said second session indicated by the comparison of the end times, thus forming a second session couple. The method according to any of claims 1 and 2, wherein the intermediate element is an external proxy (82) outside the terminal, through which the network traffic of the terminals (2a, 2b) goes to gain access to an information network. The method of claim 9, wherein, for a given hit transmitted by a given terminal: - the first registration concerning the given hit information comprises specifically for the given terminal, at least at the time of the transfer of the given hit, inserted by the proxy; - the second registration concerning the given hit comprises the said information specifically for the given terminal, at least at the time of the transfer of the given hit, inserted by the node, and - the terminal identification included in the first registration concerning the given hit is assigned to the second registration regarding the given hit, using at least the said information specific to the given terminal as an association key. The method of claim 10, wherein said information specific to the given terminal comprises an outbound port number for the proxy through which the hit goes. A computer program product, comprising program instruction code for implementing the method according to at least one of claims 1 to 11, when said program is executed on a computer. A storage medium, computer-readable and non-transient, that stores a computer program product according to claim 12. A collection device (80) for a user-oriented audience measurement of consumption messages, called hits, that are transmitted to a node for site-oriented audience measurement (5) by markers performed by terminals (2) in panel households (1), characterized in that the collection device is configured and adapted for: - obtaining a collection of first registrations created by at least one intermediate element through which the hits pass, each first registration being a separate hit and including an identification of the terminal that sent the hit; - obtaining a collection of second registrations created by the node, each second registration being a separate hit and comprising at least one hit parameter, and - assigning to each second registration a terminal identification, derived either from a unique first registration or from several first registrations comprising the same terminal identification. 1/10

Figure 1 2/10 HOUSEHOLD, BACKOFFICE / WEB

3/10

9 4/10

Figure 4 5/10

Figure 5 6/10

Figure 6 7/10

T'g / JS 8/10 HOUSEHOLD,> BACKOFFICE / WEB

Figure 8 9/10

10/10 PRIVATE NETWORK, BACKOFFICE / WEB