EP2695098A1

EP2695098A1 - Method of parameterizing rules for broadcasting personal data

Info

Publication number: EP2695098A1
Application number: EP12708870.6A
Authority: EP
Inventors: David Pergament; Armen Aghasaryan; Jean Gabriel GANASCIA
Original assignee: Alcatel Lucent SAS
Current assignee: Alcatel Lucent SAS
Priority date: 2011-04-05
Filing date: 2012-03-16
Publication date: 2014-02-12
Also published as: FR2973906B1; FR2973906A1; JP2014515855A; CN103562929B; US20140026184A1; CN103562929A; WO2012136462A1; JP5864720B2; KR20140002025A; KR101519401B1

Abstract

The invention relates to a method of parameterizing rules for broadcasting personal data of a user (U) of a social network in relation to a target contact (CC). The method consists in retrieving behavioural data of the target contact. As a function of these behavioural data retrieved and of a sensitivities profile predefined by the user, an evaluation score is allocated to the target contact regarding the danger that is represented by propagating the personal data of the user. As a function of the score allocated to the target contact, a recommendation for parameterizing the rules for broadcasting their personal data is emitted for the user's purposes.

Description

METHOD FOR SETTING DATA BROADCASTING RULES

PERSONAL

The present invention relates to the field of social networks and the dissemination of personal data within these social networks.

More particularly, the invention relates to a method of setting the rules for the distribution of personal data of a user of a social network. The invention also relates to a system for setting up the distribution rules of personal data of a user of a social network, to an application server and to a computer program product.

[0003] Social networking sites allow millions of users around the world to open an account, create a profile and publish information or personal data about their privacy. Each user of a social network then creates his own network in which he accepts relationships, still referred to as contacts in the rest of the description, with other users. These contacts can be grouped according to their nature. For example, a user may have contacts belonging to the group of family members, or to the group of his close friends, or to the group of his more distant friends, or to the group of his co-workers. The user can also accept into his contact network unknown persons who asked him to be part of his network. Each user has the opportunity to control the visibility of his personal data by other users of the social network, whether his contacts or not. Thus, a user can decide to share only certain personal data with some contacts of his network. A social network therefore allows its users to enter personal data relating to privacy and to interact with other users. The information likely to be made available to the network mainly concerns the civil status, studies or profession, or centers of interest. This information then makes it possible to find users sharing the same interests. The use of social networks then extends to a simple sharing of personal data, relating to privacy, through photographs, links, or texts for example. But these social networks can also be used to form public groups to publicize institutions, companies or causes. variety. Interactions between members of such groups include the sharing of correspondence and multimedia documents in particular. In this case, all data published in these public groups are, unlike the profile, public and can be viewed by any user without the need to have an account on the social network in question . Since these data are public, they can be used by anyone, without the consent of the owner, for advertising, phishing or identity theft for example.

[0004] Moreover, some users, especially the younger ones, want to meet as many people as they like and who share the same interests. That's why they spread their personal data without any restrictions. Their personal data can then be disseminated by contacts of their network, then by contacts of their contacts who no longer belong to their network, and so on. Similarly, even a close contact may use his or her profile for business purposes, or even a contact, who is unfamiliar with the workings of social networks, may not correctly set his or her distribution rules so well. that his profile is public and that he finds himself a propagator without knowing it. In these cases, the users are no longer masters of their data, which may be widely disseminated, and which can then be reused without their consent. In particular, users' personal data is used by advertisers to send targeted advertisements. Social networks can also legally resell information about their members, not just their profile but also their consumer behavior, in order to better target advertising. Some companies also collect personal data, publicly available, to collect information about their employees. Recruiters can also collect information and use it for the selection of their candidates. Political or governmental organizations may also collect information and supplement their files. There are so-called "reputation" websites that allow any user to obtain third-party descriptions by searching for and collecting publicly available information on the web. Finally, because of the spread of their personal data, users also incur a high risk of identity theft. Other users are more cautious and do not wish to insert their personal data lest they be used without their consent or stolen.

It is therefore very important to define rules for the distribution of personal data, so that social network users remain masters of their own personal data, relating to their privacy.

[0007] There are currently systems that provide social network users with services to alert them to how their data should be protected. One of these systems has been the subject of a patent application US201 1/0029566. The system described in this document analyzes whether a user's personal data is visible to each of his contacts. It then analyzes the degree of sensitivity of the data. Thus, the more sensitive a data is, the more it will be necessary to preserve it and avoid its diffusion. For this, the system distinguishes well-defined attribute fields, for example, the date of birth, the telephone number, the personal address, the trade, and so on. It also relies on the nature of the user's relationship with each of his or her contacts, that is, he or she takes relationships into account differently, depending on whether the contacts belong to a group identified as being family, or a group of close friends, or a group of distant friends or a group of colleagues for example. Then the system gives the user the possibility of setting globally the level of requirement that he has in terms of confidentiality of his personal data relating to his private life. For this, the user chooses to give access or not to certain attribute fields, depending on the nature of the relationship he has with each of his groups of contacts, that is to say according to trust in each group of contacts.

However, the existing systems are based only on the user data according to its level of requirement in terms of confidentiality. These systems do not make it possible to refine the rules for distributing personal data according to the behavior of a contact and the ability of said contact to propagate data.

The invention therefore aims to remedy at least one of the disadvantages of the prior art. In particular, the invention aims to make it possible to evaluate the danger that represent a potential contact of a social network user, to propagate personal data deemed sensitive by the user.

For this purpose, the subject of the invention is a method for setting up rules for broadcasting personal data of a user of a social network with respect to a target contact, said personal data being classified by categories, said method comprising the steps of:

define a profile of user sensitivities by prioritizing said categories of personal data and assigning a weighting to behavioral factors, according to a degree of importance that the user gives them in relation to a broadcast public,

recovering behavioral data from said target contact,

estimating a score for each behavioral factor of said target contact, from said recovered behavioral data, each behavior factor being noted for each hierarchical category of personal data in said user's sensitivity profile,

aggregating the estimated scores by taking into account the weighting attributed to each of the behavioral factors of the sensitivity profile, in order to obtain an overall score attributed to the target contact for all categories of personal data and for each of them ,

- Issue a parameterization recommendation for said user to set rules for the distribution of his personal data vis-à-vis said target contact based on the overall scores.

Thus, the method allows to assign an evaluation score to the target contact, and to establish a recommendation of parameterization for the user, according to the assessment of the danger that represents the contact target to propagate data.

[0012] According to other optional features of the method:

the calculation of the notes is further refined by a collaborative exchange of notes with at least one common contact between said user and said target contact,

- issuing a recommendation consists in issuing an alert message proposing to block access to a category of personal data for the said target contact, provided that the overall score obtained for said category of personal data is less than a predetermined threshold value,

the threshold value is modified according to a decision of said user to follow or not said recommendation issued,

the recovery of the behavioral data of said target contact is achieved by means of contacts common to said user and said target contact, and by means of publicly available data,

the sensitivity profile of said user is automatically modified according to requests made by said user for a display of the scores assigned to the target contact.

The invention also relates to a system for setting up personal data distribution rules of a social network user vis-à-vis a target contact, said personal data being classified by categories, characterized in that that said system comprises:

an input means enabling said user to define a sensitivity profile, by ranking said categories of personal data and weighting behavioral factors, according to a degree of importance that the user gives them in relation to 'a public broadcast,

a request module adapted to recover behavioral data from said target contact,

a calculation module capable of estimating and assigning a score to predetermined behavioral factors of said target contact, from said recovered behavioral data, each behavior factor being noted for each category of personal data hierarchical in said profile of sensitivities of said user,

an aggregation module capable of aggregating the estimated scores by taking into account the weighting attributed to each of said behavioral factors in said sensitivity profile, in order to obtain an overall score attributed to said target contact for all the categories of data personal hierarchy in the profile of sensitivities, and for each of them, a recommendation module able to issue a recommendation for setting the rules for the distribution of personal data for said user with respect to said target contact based on the overall scores obtained.

According to other optional features of the system:

the system further comprises a learning module (80) able on the one hand to modify parameterization decision rules, according to a decision by said user to follow or not said recommendation issued and on the other hand, to modify the sensitivity profile of said user according to a request from said user for a display of the scores assigned to the target contact,

the system further comprises a filtering module adapted to establish a correspondence between the categories of hierarchical personal data of said profile of sensitivities of the user and the behavioral data of said target contact retrieved by said request module.

The invention further relates to an application server comprising at least one microprocessor and a memory for implementing the parameterization method as described above.

The invention finally relates to a computer program product intended to be loaded into a memory of an application server, the computer program product comprising portions of software code implementing the method such as described above, when the program is executed by an application server processor.

Thus, the invention improves the confidentiality and / or secure control of the dissemination of personal data about a user without requiring encryption of personal data. Consequently, the invention constitutes a simple and effective alternative that does not require the use of encryption algorithms requiring significant software and hardware resources (in particular in terms of processor and memory) to avoid an anarchic distribution of personal data. It is therefore well adapted to the context of social networks. Other advantages and features of the invention will appear on reading the following description given by way of illustrative and nonlimiting example, with reference to the appended figures which represent:

• Figure 1, a simplified diagram of a social network on which users meet,

FIG. 2, a diagram of a system for setting up the distribution rules for the personal data of a social network user vis-à-vis a target contact,

FIG. 3, a diagram of a graphic interface for displaying the scores estimated by the system of FIG. 2, for a selected target contact,

FIG. 4, a flowchart representing the process steps implemented by the system of FIG. 2.

In the following description, the term "user" refers to a social network user who opened an account, created his profile to publish personal data and created a network of contacts including different groups of contacts. A target contact is defined as another user of said social network who wishes to integrate the user's network of contacts, or that the user plans to integrate or that the user has already integrated into his network of contacts.

Figure 1 shows a network, through which users U, C, CC connect their respective computers 1, 2, 3 to a remote server RS social network. A user U then encounters C and CC contacts of the social network. He may want to integrate a CC target contact into his contact network. In this case, the user connects via an IT telecommunication network to a remote parameter server SP arranged to implement the parameterization method according to the invention.

The system shown in Figure 2 helps the user to set his own data dissemination rules, based on an assessment of the danger that the target contact to propagate said data. For this, the system analyzes behavioral data of the target contact. FIG. 2 is described in parallel with FIG. 4 to clarify the role of each functional module of the system in the parameterization process. In a first step 300, the user previously defines his profile PROF sensibilities in terms of dissemination of personal data, relating to his private life. To do this, an input means 10, for example in the form of a graphical interface that is displayed on the screen of his computer, allows the user to define this PROF profile. Thus, for each predetermined group of data, the user ranks the categories of personal data that he considers more or less relevant and to which he gives more or less importance vis-à-vis a broadcast. The categories of data deemed important, or sensitive, are the categories of data that the user does not want to propagate on a global telecommunication network such as the web.

A first group that is taken into account to achieve this profile, called "themes", includes all the topics addressed by the user, according to thematic categories. Thus, within this group, the user can be vigilant about the dissemination of his personal data concerning thematic categories on his family or politics and gives them a high degree of importance. On the other hand, it may give less or no importance to a category relating to sport, for example. In this case, the user classifies the thematic categories in order of importance, in a drop-down menu for example. Thus, in the example, the user places the subject on the family first, the subject on politics second, while he places the subject on the sport last.

A second group, called "object types", groups the types of content, according to different categories that define the means by which data is published. These categories of content types vary from one social network to another. The most common in social networks are for example photos, videos, status, events, or groups. In his profile of sensitivities, the user defines what categories of object types are more or less important for him. Thus, it can give more importance to photos than status. In this case too, it classifies each type of object, according to the importance that it grants him. Moreover, in the definition of its sensitivity profile, the user takes into account, in addition, another group of data, referred to as "behavioral factors". This group groups different categories of behaviors that can have a target contact with respect to privacy. These different categories of behavior are, for example, the fact of easily propagating data that does not belong to the target contact, or the manner in which the target contact disseminates the data, especially if feelings are expressed during the broadcast, or the fact not to set privacy rules when the target contact creates their profile in a social network. Thus, the user can give more importance to a propensity factor to propagate, which assesses the dangers of a target contact to propagate personal data. Other factors take into account the popularity of the target contact, the way in which it propagates the data, if the target contact quotes other contacts during the data dissemination, etc. These factors are detailed below in relation to the calculation module. The user then assigns a weighting, or importance score, for example between 0 and 1, such as 0.4, the lowest score being judged to be less important than the highest score. Thus, according to the degree of eligibility that he gives to each of the behaviors that a target contact may have, the user assigns them a weighting.

The user therefore defines his sensitivity profile by prioritizing the categories of personal data and weighting behavioral factors, according to the importance that the user attaches to said categories of personal data and to said behavioral factors vis-à-vis public broadcasting.

In a variant, the user can also associate a theme with an object type. For example, he can define that the data concerning the subject on the family in the object type "photos" is sensitive, while the same subject is not in the object type "status" for example. In this case too, we can attribute to this association a weighting between 0 and 1.

The sensitivity profile thus defined by the user is advantageously recorded in a storage means 11. This storage means may be remote and is for example in the form of a database. The user then selects, in step 310, a target contact CC1 for which he wishes to evaluate the danger that it represents in terms of data dissemination. This selection of the target contact can be done via a graphical interface that appears on the screen of his computer. This graphic interface is referenced 60 in FIGS. 2 and 3. The selection of the target contact then triggers the operation of a request module 20.

This request module 20 is used to recover, in step 320, DC behavioral data relating to the selected target contact, vis-à-vis which the user wants to set the distribution rules of his personal data. For this, the module 20 is divided into two entities 21 and 24. The first entity 21 collects publicly available data on the web. Thus, a first collector 22 research on the web if there is available information on the behavior of the target contact, relating to compliance with privacy rules. This collector can for example check if the target contact owns a website and if the parameters of this website, in terms of respect for personal data, are high or low. Another collector 23 makes it possible to retrieve information from social networks for which the target contact is a member but for which he has not set rules concerning respect for privacy and the dissemination of his personal data. This collector 23 can also retrieve information from social networks, and more particularly from public profiles, that is to say non-parametric profiles of users of these networks with which the target contact has interacted. The second entity 24 retrieves behavioral data on the target contact from the user's contact network. Thus, a first collector 25 makes it possible to recover data on the target contact directly on its profile, and visible to the user. In this case, the user must then be in a specific relationship with the target contact, that is to say that he has already integrated it into his network of contacts. Another collector 26 consists in collecting data on the target contact from information retrieved from the profiles of the common contacts between the user and the target contact. In this case, the user and the target contact do not need to be in direct contact. It is the information held by the common contacts that will serve. Thus, for example, this collector 26 can access comments that the target contact has given on topics held by the common contacts. Finally, another collector 27 can retrieve notes the user's contacts, to qualify the target contact for protection and privacy. In this case, the calculated and visible scores in the profiles of the common contacts are for example obtained with this same system. The data thus recovered are transmitted to a filtering module 30. The sensitivity profile recorded in the storage means 1 1 is also transmitted to the filtering module. This filtering module 30 makes it possible, from the behavioral data DC recovered by the different data collectors 22, 23, 25, 26, 27 of the request module 20, and data relating to the profile of the user's sensitivities. mapping between the user-prioritized data categories and the DC behavioral data of the selected CC1 target contact. Thus, all the data for which a correspondence could not be established are not retained for the subsequent step of rating notes. The data for which a correspondence could be established are retained and transmitted to the input of the next functional module 40. This filtering module 30 is optional, it facilitates subsequent estimates by eliminating all the data for which no match could be established. To make its analyzes, to establish its correspondences and to carry out its filtering, the module 30 is based advantageously on techniques of semantic analysis. A calculation module 40 then makes it possible, in step 330, to estimate and assign a note N _F / di to predetermined behavioral factors of the target contact CC1. For this, the calculation module 40 is based on the data transmitted by the filtering module 30. A behavior factor is associated with each category of personal data prioritized in the sensitivity profile and, for each of these associations, it is assigned to it a note N _F / di so for each theme and each type of object selected by the user in its sensitivities profile, a note is estimated and assigned to the propensity factor of the target contact spread the data, and so continued for each behavioral factor.

As regards the estimation of the propensity factor score to propagate data, the calculation module 40 counts, from the data supplied to it, the number of times the target contact has commented or tagged for example objects, such as photo or video links or status, do not not belonging. The more the target contact often does, the higher the score assigned to the factor. For example, when dealing with the propagation of a status, the intensity of the propagation is measured by taking into account the number of times that the target contact propagated the object, the number of times that other users also propagated the object, and the number of users who saw the object without spreading it. Thus, when the target contact has sent three comments on a status for example, the note of the propensity factor to propagate the type of object "status" will be higher than when it sends only one comment. Similarly, when the target contact clicks a "like" button under a type of object published by him or one of his contacts, this allows his contacts to know what he likes. Thus, if several contacts press the "like" button for a particular status for example, then this status will be highly propagated and the propensity factor score to propagate will be high.

[0034] The popularity factor represents the popularity of the target contact in comparison with reference measurements. These reference measurements can for example be defined as the average behavior of the user's contacts. In particular, the rating given to this popularity factor is based on the number of contacts the target contact has in their relationship network, on the percentage of people in an "event" object that the target contact has created, or on the number of times an object type is propagated.

[0035] The sensitivity factor represents the degree of neutrality of a sentence. The measure of the degree of neutrality can be achieved by conventional techniques of extracting emotions, such as the detection of smileys for example, which are stylized drawings of faces used to express emotions. The analysis of the degree of neutrality of all the terms of a sentence can also be carried out using statistical dictionaries, such as, for example, the dictionary "SentiWordnet" (registered mark) which can be consulted at the internet address. http://sentiwornet.isti.cnr.it. The aggregation of the score assigned to each of the terms of a sentence gives the note of the sentence. The more extreme the note is, that is, close to 0 or 1, the more sensitive the sentence is. A score of 0.5 means that the target contact remains neutral in the propagation of his messages and does not transmit his feelings. This factor is important because it reveals the quality of the spread when personal feelings are spread.

The exposure factor makes it possible to deduce if the target contact has configured its personal data distribution parameters in a private or public sense. It helps the user to know if he can interact with the target contact without risk.

To assign a score to the diffusion factor, the calculation module is based on the number of times that the data of the target contact speak of third parties. For that, it bases for example on the contents of the messages evoking thirds as well as the photos which are marked, or tagged, with its contacts. In this case, the calculation module analyzes the percentage of contacts involved, the number of times they are cited, etc. The proximity factor represents the proximity of the target contact to the user. Finally, the propensity factor to facilitate diffusion makes it possible to know if the target contact facilitated access to already propagated data.

Some factors analyze only the behavioral data not belonging to the target contact, such as propensity factor to propagate, others only take into account the behavioral data belonging to the target contact, such as the exposure factor, and others combine the two, such as the sensitivity factor, for example. In a variant, it may be interesting to calculate the score of certain factors taking into account the totality of all the behavioral data, and not related to thematic categories or object types, for example to calculate the score of the factor of proximity between the user and the target contact.

The notes N _{F d} i thus estimated are then transferred to an aggregator module 50. This module makes it possible to calculate (step 340) a global evaluation score NGdi associated with the target contact CC1, for all the categories of data. personal hierarchy in the profile of sensitivities, and also for each of these categories of personal data. This overall rating NG _d i reflects the behavior of the target contact CC1 selected, vis-à-vis the protection of personal data, that is to say, it assesses the dangerousness of the target contact to propagate data the user's personal details. This aggregator module 50 in a variant embodiment can be confused with the calculation module 40. It calculates an overall score by aggregating all the scores estimated by the module calculator 40 for each behavioral factor associated with a category of personal data. The aggregation takes into account the weighting of each behavioral factor as defined in the user's sensitivity profile. The higher the weighting of behavioral factors in the sensitivity profile, the more sensitive they are to the user, and the more they affect the value of the overall score. The calculation is therefore weighted according to the importance given to the various behavioral factors by the user.

In an alternative embodiment the notes can also be estimated collaboratively. Indeed, two users in contact, who have a relationship of great confidence and who share a lot of data can exchange the notes they have estimated for the same target contact and combine them to further refine their estimates. Therefore, optionally, and with the agreement of his / her contact (s), the user retrieves the score assigned to the target contact by his / her contact (s) and checks whether the information is relevant to him / her. . It can for example take into account the number of contacts that took into account this collaborative calculation, or the added value of this note, to include it in a significant way in its estimate. In return, the user sends the note he estimated to his (her) contact (s). This recovery of the notes for a collaborative calculation is performed by the collector 27 of the request module 20 as described above.

When the notes have been allocated for the target contact, they are advantageously recorded in a storage means 51. This storage means may for example be a database. This database also stores the context in which the estimates were made. For example, context can include contacts that contributed to rating estimates. Thus, the notes for each target contact of a user are stored and no longer need to be recalculated each time. On the other hand this database can allow to resume a note when it is necessary to recalculate. This may for example be the case when the user integrates into his network new contacts common to the target contact.

The notes thus obtained are advantageously displayed, for example through the graphical interface 60, which is displayed on the computer screen of the user. This interface 60 is the one that has previously been used to select the CC1 target contact. It is represented diagrammatically in FIG. 3. It makes it possible to display to the user the scores attributed to the target contact that he plans to integrate into his contact network. As soon as the user selects a target contact CC1, in a selection menu 61, the operation of the system is triggered by this interface. The user can use this interface after receiving an invitation from a target contact unknown to him, or if he wants to obtain more information about a person who is already in his network of contacts. This makes it possible to better regulate its distribution parameters of its personal data. The scores obtained are transmitted by the aggregator module 50 and are displayed on the interface 60. A first field 62 displays the overall score NG obtained for the target contact CC1 for all categories of personal data. In the example of FIG. 3, the overall score NG assigned to the target contact CC1 is equal to 0.35. Other fields 63a, 63b, 63c display the overall scores NG _d i obtained according to the themes and types of objects. Thus, in the example of FIG. 3, the field 63a displays a note equal to 0.4 for the type of object "photo". The field 63b displays a score of 0.1 for the subject relating to the family Fam and the field 63c displays a note equal to 0.7 for the object type "event" EV. These three notes therefore mean that the selected target contact CC1 tends to propagate the data about the subject very broadly on the family, it also broadcasts the photos, but it diffuse less the type of object "event". These fields 63 display in particular the notes in an order corresponding to the preferences of the user, that is to say according to the themes and types of objects that are most relevant to him. The results are also displayed according to their values. This interface also allows the user to view all the notes that have been estimated and not only the most relevant, thanks to drop-down menus.

On the other hand, the user may want to know how the rating was assigned. Therefore, by selecting a note, for example the score of 0.4 assigned for the photo object in Figure 3, two other windows 64, 65 appear. A first window 64 displays the public data Dl belonging to the target contact and obtained directly on the profile of the target contact or on other public sites. Thus, the window can for example display the percentage of photos that the target contact has tagged, 78% in the example of Figure 3, and the percentage, 23% in the example of Figure 3, of common contacts CCom tagged . This window can also display a status for example to reveal some of the data that allowed the generation of the note. A second window 65 displays the behavioral data of the target contact not belonging to the target contact and obtained through the CCI common contacts with which it has interaction. These two windows are an example of display. The data can be displayed in another way, for example in several windows, each window being linked to a collector 22, 23, 25, 26, 27 of the request module 20.

The notes thus obtained are both transmitted to a recommendation module 70 and to a learning module 80. The user's browsing history in the display interface 60 of the notes is advantageously transmitted. to the learning module Ap 80. Thus, the data of the display history make it possible to better understand and apprehend the sensitivities of the user. Indeed, if the user often asks to display information on a particular theme that was not considered important in his profile 1 1 of sensitivities, its importance will then be raised and updated in its profile 1 1 of PROF sensitivities , so that this data is displayed in the first the following times.

The operation of the recommendation module 70 in turn is triggered by the interface 60, when the user requests to display recommendations to configure its settings options rules for the dissemination of his personal data. This module 70 thus makes it possible to establish a recommendation strategy by comparing the scores assigned to the target contact with predefined threshold values in decision rules, contained in a storage means 81 such as a database for example. This database 81 contains basic decision rules that can be applied by default. Such a rule can for example consist in saying that if the note NG _d i obtained, for a particular type of object, is lower than a threshold value If, for example 0.75, then the target contact can not have access to the data of this type of object. Otherwise, there may be access. The decision rules, stored in the database 81, are transmitted to the recommendation module 70 and, depending on the notes transmitted to it, it issues one or more recommendation (s) REC1 (di), REC2 (di) to destination of the user (steps 350, 351, 352). So, in one example, the note NGdi obtained for the photo object is 0.4 and below a predefined threshold value Si of 0.75 for this object (step 350). In this case, the recommendation module 70 issues a recommendation REC1 (step 351) to say that access to the photo object CC1 should not be given. On the other hand, if the score obtained for the event object is, for example, 0.7, and greater than the predefined threshold value Si of 0.6 for this object for example, then the recommendation module 70 issues a recommendation REC2 for the object. an event object consisting in saying that the user can give access to this object to the target contact CC1 (step 352). The recommendations issued in terms of setting up the rules for distributing personal data with respect to the target contact are then displayed in another window 91 of another graphical interface 90 which is displayed on the screen of the user. 'user. The user can then follow these recommendations (step 360) and, in this case, his own distribution rules vis-à-vis the target contact, stored in a database type of storage means 92, will be automatically updated (step 370). He can also refuse the recommendation. In both cases, the learning module 80 is informed of the decision of the user (step 380) and updates (step 390) the decision rules contained in the database 81, so that next time the behavior of the system better corresponds to the wishes of the user. For example, if a recommendation is to prevent the target contact from accessing the object type "photo" and still giving the user access, the threshold value Si of the note for this type of object is lowered in the corresponding decision rule.

In another embodiment, the two storage means 1 and 81 respectively of the user's sensitivity profile and decision rules can be collected in a single database.

The figures and their descriptions made above illustrate the invention rather than limiting it.

Although some figures show different functional entities / modules as distinct blocks, this does not exclude in any way embodiments of the invention in which a single entity / module performs several functions, or several entities / modules perform only one function. The functions of the various elements illustrated in the Figures, in particular the functional blocks marked as "processing modules" or "processors", may be realized by the use of dedicated hardware as hardware able to run a computer program in association with a appropriate computer program. When the function is performed by a processor, it can be by a single dedicated processor, or by a single shared processor, or by a plurality of individual processors some of which can be shared. The mentioned or illustrated databases can be centralized or distributed. Thus, the Figures should be considered as a very schematic illustration of the invention.

Claims

claims

1. A method of setting rules for the distribution of personal data of a user of a social network with respect to a target contact, said personal data being classified by categories, characterized in that said method comprises the following steps :

defining (300) a user's sensitivity profile (PROF) by prioritizing said categories of personal data and assigning a weighting to behavioral factors, according to a degree of importance which the user gives them vis-à-vis with respect to a public broadcast,

recovering (320) behavioral data (DC) from said target contact,

estimating (330) a note (N _{F /} cn) for each behavioral factor of said target contact, from said recovered behavioral data, each behavior factor being noted for each hierarchical category of personal data in said sensitivity profile of said user,

- aggregate (340) the estimated scores taking into account the weighting attributed to each of the behavioral factors of the sensitivity profile, to obtain a global score (NGdi) attributed to the target contact for all categories of personal data and for each of them,

- issuing (350, 351, 352) a parameterization recommendation for said user to set rules for the dissemination of his personal data vis-à-vis said target contact based on global ratings.

2. The method of claim 1, wherein the calculation of the notes is further refined by a collaborative exchange of notes with at least one common contact between said user and said target contact.

3. Method according to one of claims 1 to 2, wherein the issuing of a recommendation is to issue an alert message proposing to block access to a category of personal data for said target contact, since the overall score (NGdi) obtained for said category of personal data is less than a predetermined threshold value (Si).

4. Method according to claim 3, wherein the threshold value (Si) is modified according to a decision of said user to follow or not said recommendation issued.

5. Method according to one of the preceding claims, wherein the recovery of the behavioral data of said target contact is achieved through contacts common to said user and said target contact, and through publicly available data.

6. Method according to one of the preceding claims, wherein the sensitivity profile of said user is automatically modified based on requests made by said user for a display of the ratings assigned to the target contact.

7. System for setting up rules for the distribution of personal data of a social network user vis-à-vis a target contact, said personal data being classified by categories, characterized in that said system comprises:

an input means (10) enabling said user to define a sensitivity profile, by prioritizing said categories of personal data and by weighting behavioral factors, according to a degree of importance that the user gives them vis-à-vis -vis of a public distribution, - a request module (20) able to recover behavioral data of said target contact,

a calculation module (40) capable of estimating and assigning a score (NF / _Û ) to predetermined behavioral factors of said target contact, from said recovered behavioral data, each behavior factor being noted for each hierarchical category of personal data in said profile of sensitivities of said user,

an aggregation module (50) capable of aggregating the estimated scores by taking into account the weighting attributed to each of said behavioral factors in said sensitivity profile, in order to obtain an overall score (NGdi) attributed to said target contact for all categories of personal data prioritized in the profile of sensitivities, and for each of them,

- A recommendation module (70) able to issue a recommendation setting the rules for the dissemination of personal data for said user with respect to said target contact based on the overall scores obtained.

8. System according to claim 7, further comprising a learning module (80) able firstly to modify parameterization decision rules, according to a decision of said user to follow or not said recommendation issued and d. on the other hand, to modify the sensitivity profile of the user based on a request of said user for a display of the scores assigned to the target contact.

9. System according to one of claims 7 to 8, further comprising a filtering module (30) adapted to establish a correspondence between the hierarchical personal data categories of said profile of sensitivities of the user and the behavioral data of said target contact retrieved by said request module.

Application server (SP) comprising at least one microprocessor and a memory for implementing the method according to one of claims 1 to 6.

11. A computer program product intended to be loaded into a memory of an application server (SP), the computer program product comprising portions of software code implementing the method according to one of the claims. 1 to 6 when the program is executed by an application server processor.