The system pipeline

The overview of the proposed pipeline to automatically generate postcards based on food posts data. The pipeline consists of two core modules, i.e., (A) information extraction, and (B) postcard composition. We demonstrate the pipeline with the food post on Instagram as an example. The dotted line indicates that the textual data could be used to validate the nutrition information if they contain related information. The final output includes a postcard’s (1) backside and (2) front side designs based on the layout templates.
Sample results from our experiment

Various examples from participants' generated results. The participants provided qualitative feedback indicating that the postcard designs encouraged them to review their physical and mental well-being differently from conventional visualization methods. The feedback highlighted that the postcards facilitated non-judgmental self-reflection and helped them gain new perspectives on their food habits.
Evaluation Criteria
- Data: How the nutritional values and emotional status encoded in the postcards influenced their understanding of their food intake.
- Context: How the postcards helped them recall the context in which the food was consumed.
- Action: How the postcards influenced their actions or decisions related to food and health.
- Value: How the postcards helped them derive personal value or insights from their food posts.