View Source

Target release	Mid-17 or later
User Story	Link to related JIRA epic or feature - please copy URL
Document status
Date	1 June 2017
Document owner	Marios Meimaris
Designer	Marios Meimaris Kostis Pristouris George Papastefanatos
Developers	Marios Meimaris Kostis Pristouris Stavros Maroulis
Stakeholders	Marios Meimaris George Papastefanatos Kostis Pristouris Abi James Benjamin Wulff
QA	Lead tester
Work Package deliverable	Link to Epic in JIRA for the related work package deliverable

Summary

To provide the analytics module for the skills recognition and profiling of users based on their activity and preferences
To provide a model (i.e., ontology for modeling personal skills and competencies) and implementation for the User Profiling Module of SlideWiki
To implement a supervised classification method for enrichment of user profiles from data collected by the Student Information System
To reuse and extend open-source tools (Weak, Apache Mahout, Apache Spark MLib)

Background and goals

User Profiling is applied on two levels, static and dynamic. Thus, SlideWiki will follow a hybrid model for User Profiling.

Static data for user profiling include the user-related data that the user has provided through the creation and updating of his/her profile. These include:

Demographic data
- Name
- Age
- Education
Educational and Research Interests
Affiliation
Participation in Groups

Dynamic data for user profiling include extracted and inferred data from user activity. These include:

Deck consumption statistics
- Average time spent on reading/consuming decks
- Number of presentations given
- Usage of external sources, multimedia etc.
- Themes and Domains of created decks (e.g. through semantic annotation and/or tagging)
Content creation
- Average time spent on creating/editing content
- Number of decks the user is participating as an owner/editor
- Statistics on usage of the user's created decks from other users (i.e., reachability/popularity of created content)
- Themes and Domains of created decks (e.g. through semantic annotation and/or tagging)

Assumptions

The user profiling module should interact with the activity tracking components in order to ensure that the needed data is produced (e.g. tracking of time spent on decks)