Transcript WP2_Task2_4_NTUA

Design & Development of Social
Media/Mining Data Capturing Module
Savona, 10th April 2014
Overview
 Social Data
 Social Media Data Capturing Tool
 3-Step Approach
 Further Issues
 Data Capturing Tool Architecture
 Previous Experience of NTUA
Social Data
 Social Media


Relate people, facts, events and locations
Openness, availability and real-time nature
 Energy-related Social Data




Special events (i.e. number of people attending)
Data illustrating people’s behavior
Energy profiles of buildings
Energy behavioral changes
Social Media Data Capturing Tool
 Design and develop a social media tool to provide
targeted information from social media that can
significantly affect energy demand/behavior of endusers in the city;
 The tool will gather information, regarding the expected
participation of citizens in scheduled major events and
feed it to the central OPTIMUS DSS;
 Major events: concerts, festivals, football games,
special events due to elections, etc.
3-Step Approach
Heterogeneous captured data will be transformed into meaningful information for different stakeholders by
applying advanced analysis procedures.
Analysis & evaluation of Social Media
 Classification of the social media landscape under specific categories, depending on their authentication
schemes, semantic expressiveness, social network topology and evolution;
 Selection of the most popular social media networks (e.g. twitter, facebook, forums, etc), through an
evaluation under the scope of specific criteria and requirements regarding energy demand/behavior of
end-users in the city.
Information Representation
 Definition of a generic, platform agnostic information model for social media content and interactions, to
establish a high level abstraction for information communicated;
 Identification of metadata, data, content, and other social network entities (e.g. re-tweets, hashtags,
network connections, etc).
Social Media Monitor
 Development of a “Monitor” to constantly harvest social media for specific content, based on simple
filtering rules defined on the generic data model;
 Filtering examples: provide content within a given location, for the past 10 hours, content communicated
by a select user/user group, content annotated with specific tags, metadata or keywords, etc.
Tools and methodologies: Rapid Miner, OpenDover, i-Sieve, SAS® Sentiment Analysis
Further Issues
Security
 All the information derived from end-users through social media/web will not reveal at any case any kind
of personal users’ data (e.g. identity, location, etc);
 Personal-related data will be deleted after feeding the OPTIMUS DSS and all necessary safety
requirements will be fully met so as to secure that any personal data will not be accessible by any nonauthorized personnel.
Other Uses
 Feedback gathered from the population concerning their feelings, ease/disease, demands upon the
building and the facilities they use, will be analyzed, clustered, evaluated in order to have samples of
behavioral habits to assess and eventually modify/correct;
 Trends and cluster of typical behaviors will be created;
 The social media tool will also be used, not only as a source of input data but also to provide the endusers with suggestions on possible behavior modifications.
Data Capturing Tool Architecture
Previous Experience of NTUA
FITMAN: Future Internet Technologies for MANufacturing
The FITMAN “Unstructured and Social Data Analytics”
Specific Enabler (FITMAN-Anlzer) extracts
unstructured data from selected web resources
and social data from selected social networks and
turns such user-generated content to knowledge to
be used for the benefit of manufacturers.
Ms. Stella Androulaki: [email protected]
Dr. Haris Doukas: [email protected]
Mr. Vangelis Spiliotis: [email protected]
Mr. Vangelis Marinakis: [email protected]