README.md

---
dataset_info:
  features:
  - name: ResponseId
    dtype: string
  - name: locus_score
    dtype: float64
  - name: locus
    dtype: string
  - name: age
    dtype: string
  - name: gender
    dtype: string
  - name: gender_text
    dtype: float64
  - name: assigned_birth
    dtype: string
  - name: assigned_birth_text
    dtype: float64
  - name: religion
    dtype: string
  - name: religion_text
    dtype: string
  - name: political_aff
    dtype: string
  - name: political_aff_text
    dtype: float64
  - name: race_ethnicity
    dtype: string
  - name: prim_language
    dtype: string
  - name: first_language
    dtype: string
  - name: first_language_text
    dtype: string
  - name: highest_education
    dtype: string
  - name: employment_status
    dtype: string
  - name: employment_status_text
    dtype: float64
  - name: current_profession
    dtype: string
  - name: current_profession_text
    dtype: string
  - name: income
    dtype: string
  - name: marital_status
    dtype: string
  - name: marital_status_text
    dtype: string
  - name: family_status
    dtype: string
  - name: family_status_text
    dtype: string
  - name: disability_binary
    dtype: string
  - name: conditional_disability
    dtype: string
  - name: vaccination
    dtype: string
  - name: question
    dtype: string
  - name: response
    dtype: string
  - name: Creative Imagination
    dtype: float64
  - name: Responsibility
    dtype: float64
  - name: Intellectual Curiosity
    dtype: float64
  - name: Depression
    dtype: float64
  - name: Emotional Volatility
    dtype: float64
  - name: Trust
    dtype: float64
  - name: Productiveness
    dtype: float64
  - name: Conscientiousness
    dtype: float64
  - name: Anxiety
    dtype: float64
  - name: Respectfulness
    dtype: float64
  - name: Compassion
    dtype: float64
  - name: Energy Level
    dtype: float64
  - name: Negative Emotionality
    dtype: float64
  - name: Aesthetic Sensitivity
    dtype: float64
  - name: Assertiveness
    dtype: float64
  - name: Agreeableness
    dtype: float64
  - name: Extraversion
    dtype: float64
  - name: Organization
    dtype: float64
  - name: Sociability
    dtype: float64
  - name: Open-Mindedness
    dtype: float64
  - name: topic
    dtype: string
  - name: media_path
    dtype: string
  - name: mental_health
    dtype: string
  - name: sexual_health
    dtype: string
  - name: copd
    dtype: string
  - name: chronic_disease
    dtype: string
  - name: hiv
    dtype: string
  - name: nutrition
    dtype: string
  - name: substance_abuse
    dtype: string
  - name: media_type
    dtype: string
  - name: target_population
    dtype: string
  - name: behavior_change
    dtype: string
  splits:
  - name: train
    num_bytes: 34943710
    num_examples: 36109
  download_size: 2719587
  dataset_size: 34943710
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
license: cc
language:
- en
pretty_name: PHORECAST
---
# The PHORECAST Dataset

<!-- Provide a quick summary of the dataset. -->

Repository: https://github.com/rifaaQ/PHORECAST 

Paper: https://arxiv.org/abs/2510.02535 

## Dataset Details

PHORECAST (Public Health Outreach REceptivity and CAmpaign Signal Tracking) is a multimodal dataset curated to enable fine-grained prediction of both individual-level behavioral responses and community-wide engagement patterns to health messaging. 
The dataset maps the characteristics of diverse individuals onto their reactions from interacting with health marketing content.

### Dataset Description

Each participant:

1. Profiled Background – demographics, Big Five traits, locus of control, baseline health opinions.

2. Reviewed Campaigns – free-text and Likert-scale reactions to five curated campaigns.

Campaigns span seven categories:Nutrition & Diabetes, Vaccination / HIV / AIDS, Mental Health, Substance Abuse, Sexual Practices, COPD / Smoking, Chronic Diseases (e.g., Heart Disease, Cystic Fibrosis, Arthritis) and are annotated with target behavior, target population, and message type (Informative, Persuasive-Efficacy, Persuasive-Threat).

Please refer to our paper to learn more about how public health experts collected the campaign database. 

![image](https://cdn-uploads.huggingface.co/production/uploads/65d7cfbf1b0043dbe0cb053b/4mKKzKMYUjKo7usIseiSP.png)


- **Curated by:** Researchers from the University of Maryland, College Park.
<!-- - **Funded by [optional]:** [More Information Needed] -->
<!-- - **Shared by [optional]:** [More Information Needed] -->
- **Language(s) (NLP):** English
- **License:** The dataset is available under the Creative Commons NonCommercial (CC BY-NC 4.0).

<!-- ### Dataset Sources [optional]

<!-- Provide the basic links for the dataset. -->

<!-- - **Repository:** [More Information Needed] -->
<!-- - **Paper [optional]:** [More Information Needed] -->
<!-- - **Demo [optional]:** [More Information Needed] -->

## Uses
The Dataset is provided for the purpose of research and educational use in the field of natural language processing, conversational agents, social science and related areas; and can be used to develop or evaluate artificial intelligence, including Large Vision Language Models (VLMs).

### Direct Use

Evaluate vision-language models, study variability in campaign receptivity, guide health message design.


### Out-of-Scope Use

The dataset should not be used for applications requiring verified factual accuracy, critical decision-making, or any malicious or unethical activities.


## Dataset Structure

Each row consists of an individual's reaction (both numerical and te to one public health campaign, alongside their demographic and personality information. 

## Dataset Creation

### Curation Rationale
The PHORECAST dataset aims to map real human profiles (demographics, personality, and locus of control) to their responses / reactions from interacting with various public health campaigns. The primary purpose is for academic research to study how different people interact with stimuli and simulate how and why different communities respond differently to visuals. The results will be used to build an AI simulator that can mimic real world communities. 

### Source Data

<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->

#### Data Collection and Processing

All collection and processing stages were done using Python. More information can be found in the paper and on our github.  

#### Who are the source data producers?

Correspondence to rqadri@umd.edu 

### Annotations [optional]

<!-- If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. -->

#### Annotation process

<!-- This section describes the annotation process such as annotation tools used in the process, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->

[More Information Needed]

#### Who are the annotators?

<!-- This section describes the people or systems who created the annotations. -->

[More Information Needed]

#### Personal and Sensitive Information

<!-- State whether the dataset contains data that might be considered personal, sensitive, or private (e.g., data that reveals addresses, uniquely identifiable names or aliases, racial or ethnic origins, sexual orientations, religious beliefs, political opinions, financial or health data, etc.). If efforts were made to anonymize the data, describe the anonymization process. -->

[More Information Needed]

## Bias, Risks, and Limitations

The dataset is primarily in English, limiting global applicability of our method. 
Sample Representation: While the dataset includes over 1,000 participants across diverse demographics, it is not fully representative of all populations. Certain groups (e.g., older adults, low-literacy populations, or non-English speakers) are underrepresented.
Contextual Biases: Responses are shaped by the cultural and temporal context in which the data were collected (e.g., during/after global health crises).

### Recommendations

<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.

## Citation [optional]

<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**

@misc{qadri2025phorecastenablingaiunderstanding,
      title={PHORECAST: Enabling AI Understanding of Public Health Outreach Across Populations}, 
      author={Rifaa Qadri and Anh Nhat Nhu and Swati Ramnath and Laura Yu Zheng and Raj Bhansali and Sylvette La Touche-Howard and Tracy Marie Zeeger and Tom Goldstein and Ming Lin},
      year={2025},
      eprint={2510.02535},
      archivePrefix={arXiv},
      primaryClass={cs.CY},
      url={https://arxiv.org/abs/2510.02535}, 
}

**APA:**

[More Information Needed]

<!-- ## Glossary [optional]

<!-- If relevant, include terms and calculations in this section that can help readers understand the dataset or dataset card. -->

[More Information Needed] -->

<!-- ## More Information [optional]

[More Information Needed] -->

<!-- ## Dataset Card Authors [optional]

[More Information Needed] -->

## Dataset Card Contact

[More Information Needed]