Cloud4all is building key parts of an infrastructure that is planned to be used in the real world after the funded life of the project. This means that we are doing a great effort in implementing a production-quality platform that can serve real users in real environments. However, Cloud4all is a research project, and as such we are moving forward the state of the art in several fields:
- Context-aware adaptation of accessibility features.
- A multi-platform, multi-toolkit architecture
- Automatic generation of metadata
- A federation of AT databases
- Getting the best fit: matchmaking
- Impact on Standards
Unlike other projects in the fields of context awareness and UI adaptation, our main contribution is to use dynamic context changes –brightness, noise- as triggers to change the application settings that are most suitable for the present context conditions. These changes may react to a short-term situation and may modify the interaction user-device, going beyond the change of brightness in response to light level, or changing volume in response to ambient volume, as this context information can be transmitted through the whole architecture to end with a new software deployed in the user device, e.g. the system may decide to employ a screen reader under unfavourable conditions, thus changing the access mode rather than improving the usability in the current one.
The approach undertaken to save network and battery time is not unworthy to note at this point, as it is not widely seen on other Context-Aware related projects. This is achieved thanks to the separation of responsibilities in several components: these components are implemented in the same philosophy of Inversion of Control as the rest of the Architecture inheriting its innovative approach. Moreover, the Minimatchmaker is an innovative component that transfers part of the processing to the user device making decisions and action quicker regarding changes to the application settings (much as the basal ganglia provide faster responses to some stimuli in the human body). It also saves network load, user expenses and computing costs for the cloud components.
Learn more about context-awareness in D103.1. Rule sets for the automatic adaptation of the user profile
The Cloud4all architecture represents an unprecedented effort to combine cutting edge, cloud-based technologies with the production-oriented techniques of developing software that can reliably scale and grow with time. Unlike traditional research projects, the Cloud4all architecture has been designed and developed from the beginning to be used by real people in production. With this and the emphasis on community-based sustainability, this architecture has the potential to carry on Cloud4all’s efforts and goals well beyond the funded period of the project.
Few authoring tools incorporate metadata generation, particularly the ability to determine metadata automatically. This is a problem because without metadata it can be difficult for users to find relevant content and know whether or not a particular resource is suitable for their personal needs and preferences. And without accessibility metadata, it is very difficult for Cloud4all/GPII to appropriately match user needs with the appropriate content alternatives. The best place to generate metadata is not after the fact, but while the materials are being authored in the first place. Cloud4All is developing metadata tools that can be incorporated into any authoring environment. Where accessibility metadata is typically complex and confusing for most non-expert authors, the Cloud4All metadata components provide an integrated, seamless, and easy to use means of automatically generating and refining accessibility information. This will allow authoring tools to easily add accessibility metadata creation to the authoring process, increasing the availability of metadata “in the wild” and improving the odds that the Cloud4All architecture will be able to find resources that match the needs and preferences of the individual user.
Based on the experience of the EASTIN network, Cloud4all introduces the following key innovations in the field of federating databases:
- A mechanism to identify the records, coming from different distributed databases, that represent the same product, in order to store the information into a single record in the Unified Listing and assign a Global Unique Sync Number to the original sources (to be used for future synchronizations);
- A mechanism for bi-directional federation of the resources: every modification made in one of the federated databases is propagated to the other databases. For example if a product record is updated in a database, a feedback notice is provided to all the other databases that include information on that product.
The proposed methodology can contribute to reducing the time needed to maintain the data, both by sharing the effort among different information providers and by providing companies with a more convenient mechanism to update their data across databases.
Matchmaking is part of the overall personalisation framework of Cloud4all/GPII algorithms responsible for mapping between user needs and preferences and configurations of ICT products in a certain context of use. Cloud4all statistical matchmaking will use machine learning algorithms on a wider base than other similar approaches. Cloud4all will not only adapt existing user interfaces, but also decide which application should be used to support the user’s preferences, for example a certain type of screen reader. In some scenarios, Cloud4all will even choose the channel of communication, for example moving from audio to text in high-noise environments. This requires the machine learning algorithms to be adaptive to the current scenario and capable of dealing with a large variety of user preferences, contexts and queries.
In the end, the communication and effort needed to create a good standard pays off with consensus and adoption. In this area standardization is key. Therefore much needs to and is being invested in creating a solid, stable, and internationally adopted framework of standards. A trend however is occurring in accessibility standards. Initially accessibility standards were focused on requirements; what had to be done to products to make them more accessible. Next, features or aspects in other mainstream standards appeared along with standards specifically designed to create technologies for people with disabilities. More recently though, standards are being created that will not only enable new forms of accessibility but can also enhance mainstream usability. The mainstream aspect can help in adoption (where the standards do not threaten the ability of large companies to dominate a market) and can allow mainstream and assistive technology manufacturers to provide access through their products in less expensive and broader fashions.