Profile Initialization Brainstorming
As we recently mentioned in the SP2 Call from May 31st, we will have to think about Profile initialization problems. This page serves the purpose of collecting ideas, existing solutions and impacts on the project. Please use a new headline for your items.
Requirements for User Profiles/Data Sets
- For development, generated profiles are sufficient.
- For learning and validation, real profiles are required.
- The statistical matchmaker requires a vast dataset, so manual creation of user profiles is not sufficient.
GUIDE User Initialization Application
GUIDE User Initialization Application (UIA) is a simple step-by-step configuration of a general interface. In each step, different types of contents and different contexts of interaction are presented, so the user can test different components and parameters, and the system learns the user characteristics, from his impairments to his preferences. By making every user go throug this application once, GUIDE can offer each one the most appropriate interface and (multimodal) way of interaction. From that moment on, every time the user interacts with a GUIDE application, the system is initially adapted to him: User Initialization application - Prototype. Characteristics:
- Initial user enrolment, playful, not mandatory
- Games which track the user performance while playing and use this information as an input to user profile refinements
- specific games for attention, vision and motor skills provide relevant information for the user profile
- games are designed to motivate and reward users for repeated playing and individual improvements
- games support rehabilitation and overall fitness of the target user groups
Web-4-All Preference Wizard Software
Web4All provides a system for easily configuring multi-user public access computer workstations. The preferences used by the system based on IMS ACCLIP Specification and thereby on the ISO 24751. The preferences document is created using a preference 'discovery aid' program. A user can create different Web-4-All instances for different contexts.
PreferAble Web Interface
PreferAble is preference editor, developed as part of the CulturAll project. It is part of a set of open source Web services of TransformAble. "PreferAble provides an easy-to-use Web interface which enables the user to edit and store their preferences. PreferAble walks the user through a series of questions about how the application should appear and behave. This allows the user to configure preferences including screen enhancements such as larger type and higher contrast colour schemes, preferred language, control preferences, and required alternatives to multimedia content such as captions and audio descriptions. Project TransformAble"
- Demo: A Demo of PreferAble has been integrated into the SAKI Learning Environment. Project TransformAble
Minutes and presentations from Cloud4all Telcos
- Activity 102.1: File:A102 1 - Implementation Plan.pptx
More information about ID102.1 can be found at: https://sites.google.com/site/cloud4allmanage/home/sp1/id102-1
- File:A101 2 - Profile Initialization.ppt
- Colin, Jess, and Joanna's UI Options Overview Presentation
- File:Cloud4all telco 13.11.12.ppt
Profile Data from Other Sources
Some profile data are available in other projects or on the Web.
- The Guide project has some profiles focused on specific disabilities, e.g. wet macular degeneration, Parkinson's disease, etc, which can be used for simulation.
- The MyUI project makes a distinction between the impairment itself (e.g. visual impairment) and the interpretation of these data (e.g. large font size). The number of profiles in the project is small.
- The VERITAS project uses profiles that strongly focus on the modeling of (motor) impairments for the purpose of simulations, with no data on adaptations.
- The VICON project has some OWL files. The complete ontology contains user model data (e.g. Vicon:UserModelArthritis, Vicon:UserModelFieldOfVision, UserModelHearingAidWithProduct and Vicon:UserModelHearing500Hz).
Some data sets are also available elsewhere, but these are unrelated to accessibility or UI adaptation:
- Cyberemotions http://www.cyberemotions.eu/data.html : European Project on Opinion Mining (some datasets may be available from them for experiments )
- Infochimps: Data sets, databases, statistics for free download and for sale.
- Kevin Chai: (links to) datasets.
- Polarity v2.0 (from LingPipe): http://www.cs.cornell.edu/people/pabo/movie-review-data/review_polarity.tar.gz
- Subjectivity v1.0 (from LingPipe): http://www.cs.cornell.edu/people/pabo/movie-review-data/rotten_imdb.tar.gz
These data sets would require a lot of time for reformatting. They might be useful after the first evaluation phase, i.e. as input for the "persona" generator.