Abstract
Context data library consists of data in different levels of
abstraction; raw data, features, contexts and annotatated target contexts. Raw
data is compacted by extracting features and/or context atoms, which
represent the data in an abstracted form. This paper discusses structure,
naming, and format for feature and context data. The representation is
compatible with annotation data format. Represented with the same format,
features and annotated target contexts can together be straightforwardly used
for evaluating the results of context recognition methods, and for visualising
the data at different levels of abstraction.
| Original language | English |
|---|---|
| Title of host publication | Benchmarks and a Database for Context Recognition: Pervasive 2004 Workshop Proceedings |
| Place of Publication | Zurich |
| Pages | 32-37 |
| Publication status | Published - 2004 |
| MoE publication type | A4 Article in a conference publication |
| Event | Benchmarks and a database for context recognition: held in conjuction with the 2nd International Conference on Pervasive Computing, PERVASIVE 2004 - Linz/Vienna, Austria Duration: 18 Apr 2004 → 23 Apr 2004 |
Conference
| Conference | Benchmarks and a database for context recognition: held in conjuction with the 2nd International Conference on Pervasive Computing, PERVASIVE 2004 |
|---|---|
| Country/Territory | Austria |
| City | Linz/Vienna |
| Period | 18/04/04 → 23/04/04 |