| ID | 4.3.5 |
| Title | Prediction |
| Expert | Niels Snoeck (TELIN) Christian Raeck (FOKUS) |
| Priority | mandatory |
| Description | The SPICE platform must be able to predict end-user, group, and service behaviour based on end-user, group, and service models learnt from past context (behaviour) information. The prediction result is then used to provide contextual recommendations.The SPICE platform should include a reasoner that attempts to predict contextual changes, such as a person's next location or activity. A measure of the quality of a prediction should be included as meta-information. |
| Rationale | Contextual recommendations are part of all three service scenarios, i.e. that the knowledge inference process must include this functionality. The scope of the SPICE platform encompasses mainly, but not exclusively: a) modality recommendations (related to the "Dynamic Desktop", mainly for end-users) b) content recommendations (mainly for end-users and groups) c) service recommendations out of a list of available services (either for end-users or services).Next virtual location and next user activity are examples of prediction of contextual changes. |
| Type | functional |
| Depends on | 4.2.7 - Learning and Recommending Algorithms 8.1.2 - Limited push-behaviour of Spice 8.1.3 - Non-biased recommendations and queries results 8.1.5 - No user location tracking w/o user's explicit consent 8.1.6 - Users' rating of consumed services |
| Child dependencies | 4.2.7 - Learning and Recommending Algorithms 7.3.2 - Adaptation Decisions 7.3.4 - Changing the Presentation Modality and Interactive Modality |
| Environment | |
| Other_info | |
| Category | enterprise;technical;user;device |
| Subcategory | |
| Subcategory2 | |
| Scenario_scene | unified.scene5 unified.scene7 unified.scene8 unified.scene14 |
| SPICE_value | (seamless) service adaptation;service matching |
| Demo | |
| Keywords | prediction;context;data;knowledge |
| Home |