# aaq

**Repository Path**: mirrors_microsoft/aaq

## Basic Information

- **Project Name**: aaq
- **Description**: Code and Data artifact for Affective Air Quality: A Dataset of Gas Sensor Data during Emotional Experiences
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2025-09-25
- **Last Updated**: 2025-09-27

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Affective Air Quality (AAQ)
This repository hosts the official code and data artifact for the "Affective Air Quality" dataset. Details about the user study and data collection can be found in our [paper][paper-link]. The dataset released it's the first to include gas sensing data emitted by humans while experiencing different emotions using a wearable device. More specifically, the detection of volatile organic compounds, nitrogen dioxide, carbon monoxide and alcohol while watching 12 emotional videos (3 per valance-arousal quadrant). It contains anonymized user data from a psychophysiological experiment focused on understanding how video-elicited emotions impact chemical breath composition (and exuded body gases in general). All data collected is fully anonymized.

### Ethics
The study and prototype was approved by the Certified Review Board and Compliance at Microsoft. No changes were made to the study protocol after approval. 

## Repository Contents

1. [Participant Data](#1-participant-data): intake form that includes demographics, personality, self-reported mood and sleep, as well as session notes.
2. [Interviews](#2-exit-interview): transcripts of the audio recordings for all semi-structured exit interviews.
3. [Gas Sensor Data](#3-gas-sensor-data): includes sensor readings from participants exposed to emotional videos for 33 minutes, with some experiencing a mint fresh breath strip in the last 5 minutes, for analyzing reactions via wearable and desktop gas sensors. It features raw gas sensor data, including NO2, ethyl alcohol, VOC, and CO levels. 
4. [Emotional Ratings](#4-emotional-ratings): psychophysiological experiment outcomes, including participant responses to stimuli via valence, arousal, and familiarity ratings, enriched with detailed timestamps for video segments and associated gas sensor data for precise analysis, capturing both subjective ratings and objective timestamps across various study segments.

## 1. Participant Data
### Intake Survey
`Qualtrics_Intake.csv` compiles responses from participants to the initial intake form, which includes inquiries on: (1) the timing and nature of their last consumed beverage and meal, (2) sleep quality via the Richards-Campbell Sleep Questionnaire, (3) personality traits through the short Big Five personality test, (4) current and general anxiety levels measured by the short State-Trait Anxiety Inventory (STAI), (5) mood states assessed by the short form of the International Positive and Negative Affect Schedule (I-PANAS-SF) and (6) emotional regulation skills via the ERQ Questionnaire.

**Key Columns:**
* `ID`: Participant ID Number, a unique numerical identifier for each participant.
* Demographics:
	* `Q1`: Age.
	* `Q2`: Gender.
* Last meal:
	* `Q3`: Timing and content of last caffeine or taurine intake.
	* `Q4`: Timing and content of last meal.
* Richards-Campbell Sleep Questionnaire (RCSQ):
	* `Q5_1`: RCSQ question 1 - Sleep depth.
	* `Q6_1`: RCSQ question 2 - Time to fall asleep.
	* `Q7_1`: RCSQ question 3 - Time awake.
	* `Q8_1`: RCSQ question 4 - Woke up in night.
	* `Q9_1`: RCSQ question 5 - Sleep quality.
* Big Five Inventory (BFI) short version: `Q11_1` - `Q11_12`.
	* `Q11_1` - `Q11_12`: Self-perceptions across various dimensions.
* State-Trait Anxiety Inventory (STAI) short version: 
	* `Q12_1` - `Q12_10`: Current feelings of anxiety (state).
	* `Q13_1` - `Q13_10`: General feelings of anxiety (trait).
* Short Form International Positive and Negative Affect Schedule (I-PANAS-SF):
	* `Q14_1` - `Q14_10`: Present emotional state.
* Emotion Regulation Questionnaire (ERQ):
	* `Q15_1` - `Q15_10`: Emotional regulation.
   
### Participant Logs
`Participant_Logs.xlsx` contains a log of participants who took part in the experiment. It provides insights into each participant's session, notes, and details about the data collected. Key columns:
    
  - `Participant`: Participant ID number, a unique numerical identifier for each participant.
  - `Listerine`: Indicates if a Listerine strip was used in the participant's final video viewing.
  - `Status`: Status of the participant's session.
  - `Date`: Date of the session.
  - `Notes`: Observations or notes from the session (e.g., coughs).
  - `Data_Status`: Session data status.
  - `Sampling_Freq`: Precise sampling frequency for raw gas data.
  - `Breathing_Type`: Predominant breathing method/type (nose or mouth) during the session, as determined in the exit interview.

## 2. Exit Interview
The `Interview Transcripts` folder includes each participant's exit interview transcript as `<P>_InterviewTranscript.txt` where `<P>` denotes the participant ID (34 in total). Due to initial synchronization issues in our recording software, data from 11 sensor sessions were lost from [gas sensor data](#3-gas-sensor-data), but their exit interviews were kept. Participant IDs: 708048, 29054, 150054, 174036, 176572, 381770, 523122, 547253, 675300, 831769, 871253.

## 3. Gas Sensor Data
This folder presents timestamped sensor readings from 23 participants while watching approximately 33 minutes of different emotional videos (randomized). During the last approximately 5 minutes of the recording, half of the participants were randomly assigned to dissolve a Listerine (mint) fresh breath strip in their mouth. (See Emotional_Rating_Data for exact timestamps to segment the Gas Sensor Data.). 

The files feature gas sensor data from both the wearable headset and standalone desktop prototypes for each participant with the following format: `<P>_Gas-Data_Raw.csv` where `<P>` denotes the participant ID. Both prototypes are equipped with Winsen metal oxide gas sensors (GM102B, GM302B, GM502B, GM702B), with specific sensor readings identifiable by column names:   `<prototype>_<sensor_type>` e.g: `desktop_GM702B` for desktop GM702B sensor data, or `headset_GM302B` for wearable GM203B data.

** Columns:**
- `Timestamp`: Timestamp of the reading.
- `<prototype>_GM102B`: Nitrogen dioxide (NO2) sensor reading.
- `<prototype>_GM302B`: Ethyl alcohol(C2H5CH) sensor reading.
- `<prototype>_GM502B`: Volatile Organic Compounds (VOC) sensor reading.
- `<prototype>_GM702B`: Carbon monoxide (CO) sensor reading.

The values for columns `<prototype>_<sensor_type> are raw values from 0-1024 from the Arduino Read, thus to calculate voltage response the following formula should be applied: 
##### V = (raw value / 1023) * Vref

## 4. Emotional Ratings

This folder contains the psychophysiological experiment outcomes, detailing the stimuli presented and participant reactions. It includes valence, arousal, and familiarity ratings, along with timestamps for all study segments (e.g., video start and end times), facilitating raw [gas sensor data](#3-gas-sensor-data)(`<P>_Gas-Data_Raw.csv`) segmentation. `<P>` is the participant number and `<DATE>` is the date (and hour) of data collection.
    
- **`target_emotion`**: Specifies the intended emotional response the film clip aims to elicit from the participant based on previous studies in Psychology. It includes the following categories: relaxation, disgust, sad, happy, relaxation.
  
- **`quadrant`**: Categorizes emotions based on Valence (positive or negative feelings) and Arousal (level of excitement or calmness). The quadrants are High Valence Low Arousal (HVLA), Low Valence High Arousal (LVHA), Low Valence Low Arousal (LVLA), and High Valence High Arousal (HVHA).

- **`film_clip`**: The identifier or name of the film clip shown to participants.

- **`movie_file`**: The file name of the movie clip used in the study.

- **`duration`**: The length of time (in seconds) the film clip was played.

- **`neutral_file`**: The file name of a neutral clip shown to participants, used as a control or baseline.

- **`neutral_duration`**: The length of time (in seconds) the neutral clip was played.

- **`paper_sources`**: References to academic papers or sources that discuss the film clip or the methodology of emotion elicitation.

- **`getBreath`**, **`holdTime`**, **`exhaleTime`**, **`secondHoldOffset`**: Parameters related to breathing exercises that participants were instructed to perform during the study, such as the time to inhale, hold breath, and exhale.

- **`trials.thisRepN`**, **`trials.thisTrialN`**, **`trials.thisN`**, **`trials.thisIndex`**: Technical columns related to the trial number, current repetition, and index of the trial within the experimental design.

- **`study_start`**, **`Welcome.started`**, **`key_resp.started`**, **`contact_us.started`**, **`privacy.started`**, **`terms.started`**, **`code.started`**, **`trademarks.started`**, **`year.started`**, **`text_4.started`**, etc.: Timestamps marking the start of different sections of the study or the moment specific screens were shown to participants.

- **`key_resp.keys`**, **`key_resp.rt`**, **`key_resp.duration`**: Responses from participants, including the keys pressed, response times, and duration of key presses during various parts of the study.

- **`contact_us.numClicks`**, **`privacy.numClicks`**, **`terms.numClicks`**, **`code.numClicks`**, **`trademarks.numClicks`**, **`year.numClicks`**, etc.: The number of times participants clicked on various sections of the study, such as contact information, privacy policy, terms of service, etc.

- **`v_score_e_tr.response`**, **`a_score_e_tr.response`**, **`f_score_e_tr.response`**: Valence, arousal, and familiarity ratings for the trial video. These ratings help assess the participant's emotional response to a neutral video before the main experiment.

- **`v_score_e.response`**, **`a_score_e.response`**, **`f_score_e.response`**: Valence, arousal, and familiarity ratings for the emotional video. These columns capture the participant's emotional response to the primary stimulus video.

- **`v_score_n.response`**, **`a_score_n.response`**, **`f_score_n.response`**: Valence, arousal, and familiarity ratings for a neutral video shown after the emotional video, used to measure baseline emotional state.

- **`participant`**, **`session`**, **`date`**, **`expName`**, **`psychopyVersion`**, **`frameRate`**: Metadata about the experiment, including participant ID, session number, date of experiment, name of the experiment, version of PsychoPy software used, and the frame rate of the display.

## Code of Conduct

This project has adopted the [Microsoft Open Source Code of Conduct][oss-conduct-link].

## Privacy Statement

Your privacy is important to us. This privacy statement explains the personal data Microsoft processes, how Microsoft processes it, and for what purposes. Please read the [Microsoft privacy statement][privacy-link] for more information.

[privacy-link]: http://go.microsoft.com/fwlink/?LinkId=521839
[oss-conduct-link]: https://opensource.microsoft.com/codeofconduct/
[paper-link]: https://arxiv.org/abs/2509.15774