Update README.md
Browse files
README.md
CHANGED
|
@@ -97,6 +97,32 @@ This structure enables consistent representation across recipes, improves filter
|
|
| 97 |
|
| 98 |
---
|
| 99 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 100 |
## Exploratory Data Analysis (EDA)
|
| 101 |
|
| 102 |
### Sample Entry
|
|
|
|
| 97 |
|
| 98 |
---
|
| 99 |
|
| 100 |
+
## Risks and Bias
|
| 101 |
+
|
| 102 |
+
The dataset inherits several biases from the original Food.com source:
|
| 103 |
+
|
| 104 |
+
- **Regional bias**: Western and American cuisines dominate, leading to under-representation of other cuisines.
|
| 105 |
+
- **Popularity bias**: Highly rated and frequently interacted recipes are over-represented, which may skew recommendation models.
|
| 106 |
+
- **Nutritional estimation bias**: Nutrition values are approximate and may vary from real-world measurements.
|
| 107 |
+
|
| 108 |
+
These biases may affect fairness and diversity in downstream models.
|
| 109 |
+
|
| 110 |
+
---
|
| 111 |
+
|
| 112 |
+
## Intended Uses and Limitations
|
| 113 |
+
|
| 114 |
+
**Intended Uses**
|
| 115 |
+
- Ingredient-based recipe recommendation research
|
| 116 |
+
- Cold-start personalization and ranking model development
|
| 117 |
+
- Educational and academic applications
|
| 118 |
+
|
| 119 |
+
**Not Intended For**
|
| 120 |
+
- Medical or nutritional decision-making
|
| 121 |
+
- High-stakes personalization without bias mitigation
|
| 122 |
+
- Commercial deployment without further validation
|
| 123 |
+
|
| 124 |
+
---
|
| 125 |
+
|
| 126 |
## Exploratory Data Analysis (EDA)
|
| 127 |
|
| 128 |
### Sample Entry
|