UpStory: the Uppsala Storytelling dataset

<p>The&nbsp;<strong>UpStory dataset</strong> is an anonymized child-child interaction dataset, with an experimental manipulation for the level of rapport. It contains data pertaining to pairs of classmates (ages 8-10) playing a storytelling game in a naturalistic setting; pairs are selected to either promote close and friendly interactions (high-rapport condition), or promote distant interactions between acquaintances (low-rapport condition). Due to the experimental design, most children participated in two pairs: one high-rapport and one low-rapport.</p> <p>A copy of this text is included in the ZIP file.</p> <h2>Dataset Contents</h2> <p>The dataset contains data for 35 pairs. Each pair is given an ID starting with <code>P</code> (high-rapport condition) or <code>N</code> (low-rapport condition), followed by the academic year (2 or 3), and 2 additional digits. E.g.: N251 is a low-rapport pair from year 2; P318 is a high-rapport pair from year 3. Similarly, each child is given a 2-digit ID. E.g.: child 17 participated in pairs P245 and N255.</p> <p>Each pair played between 1 and 5 rounds of the game. Each round is provided as an individual sample, with its own associated time series as CSV files. In total, 106 rounds are provided.</p> <h3>Child Information</h3> <p>The top-level CSV file <code>child-info.csv</code> offers child-level information, including the following items:</p> <ul> <li><code>child_id</code>: the child's ID (a random 2-digit unique identifier).</li> <li><code>gender</code>: <code>boy</code> or <code>girl</code>.</li> <li><code>year</code>: academic year the child belonged to (<code>2</code> or <code>3</code>).</li> <li><code>age</code>: the child's age in years at the beginning of the data collection effort. Either an exact value (<code>9</code> or <code>10</code>), or a range (<code>8-9</code>).</li> </ul> <h3>Pair Information</h3> <p>The top-level CSV file <code>pair-info.csv</code> offers pair-level information, including the following items:</p> <ul> <li><code>pair_id</code>: the pair ID, as described above.</li> <li><code>condition</code>: the experimental condition this pair belonged to (<code>low_rapport</code> or <code>high_rapport</code>).</li> <li><code>distance</code>: the distance between the two participants in their year's friendship network (integer in range <code>2 &lt;= n &lt;=56</code> for Year 2 pairs, and <code>2 &lt;= n &lt;= 20</code> for Year 3 pairs).</li> <li><code>year</code>: academic year the children belonged to (<code>2</code> or <code>3</code>).</li> <li><code>rounds</code>: number of game rounds the pair played (<code>1 &lt;= n &lt;= 5</code>).</li> <li><code>child_1</code>: first child in the pair (lower ID; 2 digits).</li> <li><code>child_2</code>: second child in the pair (higher ID; 2 digits).</li> </ul> <h3>Sources</h3> <p>The dataset contains time-series data extracted from two different video sources, each one overviewing the play area from one side: the <code>left-camera</code> and <code>right-camera</code>. Each video source has its own top-level folder, with data extracted from that source inside it.</p> <p>In each source folder, you will find CSV files named <code>&lt;source&gt;-&lt;pair_id&gt;-round-&lt;round_number&gt;-&lt;face|pose&gt;.csv</code> (e.g., <code>left-camera-N249-round-1-face.csv</code>). There is a separate file for each round of the game; each pair typically played ~3 rounds (min: 1, max: 5). As the names suggest, <code>face</code> files contain information related to head pose and facial expression, while <code>pose</code> files contain information related to full body pose.</p> <h3>Face Data</h3> <p>Face data was extracted with OpenFace, and contains most information that is produced by the tool. See the OpenFace documentation for more details. Time series are given at 25Hz; entries are indexed by <code>frame</code> (0-indexed) and <code>child_id</code>. Included data:</p> <ul> <li><code>confidence</code> and <code>success</code> indicators.</li> <li>Per-eye gaze 3D vectors.</li> <li>Joint gaze angle.</li> <li>Eye landmark information in 2D (frame position in pixels) and 3D (estimated distances).</li> <li>Head position and rotation information in 3D.</li> <li>Face keypoint locations in 2D (frame position in pixels) and 3D (estimated distances).</li> <li>AU presence estimates for 18 AUs (binary variables: 0 or 1).</li> <li>AU intensity estimates for 17 AUs (continuous variables: 0 to 5).</li> </ul> <h3>Pose Data</h3> <p>Pose data was extracted with OpenPose. Time series are given at 25Hz; entries are indexed by <code>frame</code> (0-indexed), <code>child_id,</code> and <code>joint</code> (named body part that the row refers to). Data provided per row:</p> <ul> <li><code>x</code>: horizontal position in the frame, in pixels, left-to-right (float; range 0-width).</li> <li><code>y</code>: vertical position in the frame, in pixels, top-to-bottom (float; range 0-height).</li> <li><code>confidence</code>: OpenPose's reported prediction confidence (float; range 0-1).</li> </ul>