gmanolache/BioInteract

---
dataset_info:
features:
- name: sourceTaxonName
dtype: string
- name: sourceTaxonRank
dtype: string
- name: targetTaxonName
dtype: string
- name: targetTaxonRank
dtype: string
- name: interactionTypeName
dtype: string
- name: sourceTaxonKingdomName
dtype: string
- name: sourceTaxonPhylumName
dtype: string
- name: sourceTaxonClassName
dtype: string
- name: sourceTaxonOrderName
dtype: string
- name: sourceTaxonFamilyName
dtype: string
- name: sourceTaxonGenusName
dtype: string
- name: targetTaxonKingdomName
dtype: string
- name: targetTaxonPhylumName
dtype: string
- name: targetTaxonClassName
dtype: string
- name: targetTaxonOrderName
dtype: string
- name: targetTaxonFamilyName
dtype: string
- name: targetTaxonGenusName
dtype: string
- name: year
dtype: string
- name: month
dtype: string
- name: day
dtype: string
- name: decimalLatitude
dtype: string
- name: decimalLongitude
dtype: string
- name: license
dtype: string
- name: referenceCitation
dtype: string
- name: sourceVernacularName
list: string
- name: targetVernacularName
list: string
- name: imageURL
dtype: string
splits:
- name: train
num_bytes: 153224188
num_examples: 256758
download_size: 10399847
dataset_size: 153224188
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
license: cc
task_categories:
- text-to-image
- image-to-text
- visual-question-answering
- zero-shot-image-classification
language:
- en
pretty_name: >-
Scalable Benchmark for Evaluating Vision-Language Models under Semantic
Variation
size_categories:
- 100K<n<1M
---

# BioInteract: Scalable Benchmark for Evaluating Vision-Language Models under Semantic Variation

<!-- Banner links -->
<div style="text-align:left;">
<a href="https://georgianagmanolache.github.io/biointeract/" target="_blank" style="display:inline-block;">
<img src="https://img.shields.io/badge/Project%20Page-Visit-blue" alt="Project Page">
</a>
<a href="https://github.com/georgianagmanolache/biointeract" target="_blank" style="display:inline-block;">
<img src="https://img.shields.io/badge/GitHub-Visit-lightgrey" alt="GitHub">
</a>
</div>

## Description

[BioInteract](https://georgianagmanolache.github.io/biointeract/) comprising richly annotated images depicting interactions between organsims, or biotic interactions, provides a natural testbed for tasks involving images and unconstrained, free-form natural language, as interacting organisms are discerned from images alone and their relationship can be expressed through multiple linguistic forms.

## BioInteract Dataset
BioInteract, the largest publicly available multimodal dataset of biotic interaction, specifically curated for vision and machine learning application in the context of AI-driven ecological research. BioInteract includes 256K images annotated with 15.4K unique biotic interactions knowledge graphs which represent the semantic relationship between entities as triplets—source taxon, interaction type, target taxon—across five kingdoms (*Animalia*, *Plantae*, *Fungi*, *Chromista*, and *incertae sedis*) and nine ecologically standardized interaction types.
A key contribution of BioInteract is that it can generate semantically controlled linguistic variations directly from the underlying knowledge graph.
By leveraging structured triplets, we can systematically construct both meaning-preserving and contradictory query variants, enabling explicit control over semantic similarity.
This allows us to disentangle correctness from consistency and to rigorously evaluate model robustness under targeted linguistic transformations.

### Directory
```plaintext
main/
├── BioInteract/
│ └── train-00000-of-00001.parquet
├── BioInteract-benchmark/
│ ├── BioInteractCommon.csv
│ └── BioInteract100.csv
├──README.md
└──.gitattributes
```