08448380779 Call Girls In Civil Lines Women Seeking Men
USEMP Project Presentation ICT 2015
1. USEMP is partly funded by the
European Community’s Seventh
Framework Programme
User Empowerment for Enhanced
Online Presence Management
usemp
project
@usemp usemp
http://www.usemp-project.eu
2.
3. Legal:
– Profile transparency
– Data licencing
Business:
- Personal data value
network evolutions
Technical:
– USEMP disclosure scoring
& data value scoring framework
– USEMP personal assistant
Framework for the user
5. • Current and upcoming legal framework
• Legal implications for information-driven applications?
Prohibited forms of profiling and profile transparency
• Profiling in the DPD: automated individual decisions (art. 15) and right of
access (art. 12a)
• Profiling in proposed GDPR (art. 20, 14(ga)(gb))
– Other relevant DP requirements
• DPD/GDPR, ePrivacy Dir
• Data minimisation, purpose limitation, sensitive data, access & consent
withdrawal, security
• Legal ground for legitimate data processing
– Performance of a contract (art. 7b DPD)
– Consent (sensitive data: art. 8 DPD)
6. • Profile transparency
tools
• Data protection by
design and by default
(art. 23 GDPR)
• Data protection by
design will be legal
requirement
• USEMP tools as
example
7. • USEMP personal
assistant framework
for the user
• USEMP disclosure
scoring & data value
scoring framework
8. A F
A.1.1 F.1.1 F.3.1
Dimensions (e.g.
demographics, health factors
and condition, etc.)
0101 1101 1001
A.1 A.6 F.1 F.3
A.1.5
Attributes (e.g. sex,
age, smoking, etc.)
Values (e.g. male,
20-25 years old,
smoker, etc.)
OSN user
Inference algorithms
9. A F
A.1 A.6 F.1 F.3A.5
Disclosure scoring & data value scoring framework
• Value / Predicted Value
• Confidence of Prediction
• Level of Sensitivity
• Support
• Reach of disclosure/visibility
• Level of control
• Privacy Score
• Reach of Disclosure
• Monetary Value Indicators
Inference algorithms
0101 1101 1001
volunteered + observed data
(URLs, likes, posts, interactions)
10. A F
A.1 A.6 F.1 F.3A.5
Disclosure scoring & Data value scoring framework
• Privacy Score
• Reach of Disclosure
• Monetary Value Indicators
Inference algorithms
0101 1101 1001
Indicators related to Monetary Value
1. User influence I is based on following parameters :
• Number of objects (i.e., picture/video/post) that a user has created
• Number of first-hop & second-hop friends
• Total number of first hop & second hop friends that had an action on
each object (i.e., picture/video/post)
• Type of action (i.e., share, like, comment) of user j on the object i
2. a measure of the importance of an object M
Data value V =I∙M
11. • Value / Predicted Value
• Confidence of Prediction
• Level of Sensitivity
• Support
• Reach of disclosure/visibility
• Level of control
• Indicators related to Monetary Value
(audience metrics, OSN interaction metrics)
A F
A.1 A.6 F.1 F.3A.5
disclosure scoring & data
value scoring framework
• Privacy Score
• Reach of Disclosure
• Monetary Value Indicators
Inference algorithms
0101 1101 1001
personal assistance
framework/applications
volunteered + observed data
(URLs, likes, posts, interactions)
Audience Management
Disclosure Score Control
Privacy Policies
Data Value Visualization
Framework
Groups Communities
Rules Data-driven Social
Inferences are made using a variey of mechanisms: multimedia information extraction techniques and inference techniques and OSN presence data in order to fill (or predict) the word value.We use a user's OSN presence data in order to predict the values of the user's attributes. This involves both utilizing explicitly provided information and also producing a number of inferences. Each filled value is characterized by a number of scores, e.g. Expressing eg. The perceived sensitivity of a piece of information, the confidence that value is true for atrtributes, …
The framework enables the following two kinds of user awareness (1) navigation through the level of the hierarchy and understanding the scores for some particular value affect or are affected by the levels above and below it
(2) focus on specific aspects of the factors that are related to perceived privacy/disclosure
The scores that characterize each value are the following
(1) conficence of prediction: it represents how confident we are that the corresponding value is true and is typically computed by the inference algorithms along with the produced inference.
(2) level of sensitivity: it represents how sensitive the user thinks that this particular piece of information is. Perceived sensitivity <> legal notion of sensitive data. In our framework we provide the level playing field for users to develop their own privacy preferences.
(3) support: This field is not really a score, but rather it provides a link to the OSN presence data that support the particular value
(4) reach of disclosure: this score attempts to quantify the set of people to which the relevant information is accessible and consists of three sub-scores
(5) level of control: This score represents the ability of a user to control the disclosure of data about him/her
In addition to privacy scores, a set of personal data value indicators are required so that the end-users can be informed about the value of the data they are sharing. The proposed personal value indicators are based on the activities of the end user in the OSN environment and his/her OSN social graph.
Indicators related to Monetary Value are
1. User influence I is based on following parameters :
number of objects (i.e., picture/video/post) that a user has created
number of first-hop & second-hop friends
total number of first hop & second hop friends that had an action on each object (i.e., picture/video/post)
type of action (i.e., share, like, comment) of user j on the object i
2. a measure of the importance of an object M
Data value V =I∙M
Finally, the personal assistance framework is developed so that end-users can get insights on their data,how they are profiled, and the inferences that are made upon their data.This framework is then vizualised in following ways to the end users
Here you see how posted images on OSN could be used to profile the user as someone with possible health problems based on his/her drinking behavior
Here the user gets insights on how pictures are defined and which part of their privacy profile could be affected.