Note that: Raw data is stored in .csv format. The file naming convention contains information about sensor type, sensor id, and sensor location, separated by “_”. For example: in MITes_01_RawCorrectedData_Trunk.RAW_DATA.csv, MITes is the type of sensor used, Trunk is the sensor location, which means Torso. The same convention applies to Study 2 directories. Annotation files are stored in several different formats, including .xlsx, .xml, and .csv, with the same contents, just for the sake of users’ convenience. Annotator1Stereotypy.annotation or Annotator2Stereotypy.annotation are offline annotations, Phone.annotation is online annotation. Files with pattern Annotation*Intervals.xlsx (containing the same information with *.annotation.* files) are currently used in the preprocessing script for Study 2 data due to legacy reasons, but should be changed to use files *.annotation.* in the future. Study 1 has only one online and one offline annotation while Study 2 has two online and one offline annotations. Experiments in the paper only used Annotator1 offline annotations, but one can refer to our previous Ubicomp’ 09 paper [1] for discussion about online and offline annotations. preprocessedDataAndLabels.mat is the intermediate file used by the scripts to store all the preprocessed data and annotations. featureVectorAndLabels.mat is the intermediate file used by the scripts to store all the computed feature sets along with the labels. Users can refer to this file directly using their own classifiers (Detailed description of the format can be found in file preExp.m in folder src/MATLAB). 3.2.1 Raw data format The raw data .csv file has four columns representing unix timestamp, raw x value, raw y value, and raw z value, respectively. There is no header row. The raw value is an integer in the range of 0 - 1023 for Wockets and 0 - 512 for Mites, with a dynamic range ±2g for MITes sensors and ±4g for Wockets sensors. 1274781775000,475,436,435 1274781775009,474,437,436 1274781775017,474,438,433 … 1274782392242,501,469,415 1274782392253,501,468,416 1274782392263,504,466,418 3.2.2 Annotation file format For .xml file, Annotation data format (*.annotation.xml) Annotated activity labels are encoded in a human and computer readable XML file. This is stored in the Annotation subdirectory inside the session folder. Some datasets may be labeled by more than one annotator using more than one category or set. Each file uses the XML format. At the top of the file you may find some comments assigned automatically during generation of this document by a source application. The first XML node defines the name/session for the dataset, along with the annotator and other relevant details about this document: The following properties may be defined: DATASET: A short name of the protocol to be used in the visualization software ANNOTATOR: Identification of the person who performed the annotation EMAIL: Annotator’s contact information DESCRIPTION: A clear, concise description of the protocol/annotation set METHOD: Indication of how the annotation was done. Options are direct_observation_paper: start/stop times were written down on paper as activities were performed direct_observation_mobile start/stop times were entered directly into a computer using custom annotation software in real-time video: annotations were coded from a video recording audio: annotations were coded from an audio recording NOTES. Any special notes should be included here. In particular, a full description of how the intensity values (if used) were determined should be included here. The main content of the the annotation.xml file comprises time-stamped label data. Each ANNOTATION node can be assigned a Globally Unique Identifier (GUID) to allow synchronization between multiple annotators. The LABEL node can also be assigned a Globally Unique Identifier (GUID) which allows lookup in a database of labels to find the specific definition of the label in use at the time this document was generated searching for an item The START_DT and STOP_DT date-time nodes receive values in the form [YEAR]-[MO]-[DY] [HR]:[MI]:[SE].[MSe]. 2011-01-25 11:42:54.396 2011-01-25 11:43:02.396 The RATINGS NODES are optional timestamps used to allow annotators to express how certain they feel the observed label represented the activity they observed. The SETNAME and GUID can be used to map meaningful labels onto ratings values. A METARATING is also available for applications where a second rating values is desired, such as intensity. A final PROPERTIES node contains tracking information about the annotation, such as the annotation set of origin and the time of creation/modification An example AnnotationData file can be found here: annotation.xml. Although activity labels are experiment-specific, researchers would be encouraged to reuse the annotation schema of others whenever possible in order to facilitate data sharing. Examples from prior experiments would be featured on the data sharing website.