swap-uniba/EVWSD-ITA-eval

This is the evaluation set for the EVWSD-ITA EVALITA 2026 task.

Each instance contains a "query" (manually written following the procedure outlined on the main site) and a list of candidate images (image paths to the images directory). We also upload the images as zip file.

Some notes:

• Queries and images have been manually cheked to increase robustness of the data
• For synsets where the sum of the co-Hyponyms and the synsets with the same lemma was less than 9, we have sampled other candidates from other instances within the test set (so that there were a total of 10 possible images for each instance). Sampling was done by leveraging cosine similarity of vision embeddings, so to choose images that were still challenging to disambiguate w.r.t. the target image
• The test set consists of 222 instances in total
• We will be releasing baseline results using a Multilingual CLIP model

UPDATE

The dataset has been updated following a warning regarding the quality of queries with compund words as lemmas (shuffled in a non-meaningful way). Please refer to the latest version of the dataset.

Submission Guidelines

To submit your results, please send a file to the EVWSD-ITA task organizers via mail containing the output provided by your system for each instance, with the same ordering as of the original test set (the instance with index 0 in the test set corresponds to the output with index 0 in your system). Any format, such as .csv or .json, is acceptable.

Note that only three submissions per participant are allowed. In your mail also briefly describe the system rationale and provide the team name.

If your system provides only the prediction for the correct image: For each instance in the test set, provide the integer index from the candidate list OR the string corresponding to the correct image name.

If your system provides a ranking for candidate images: For each instance in the test set, please provide the ordered list of integer indexes from the candidate list OR the list of strings of the image names sorted in descending order (the item with index 0 in the list corresponds to the item with the highest score).

Example:

{ "query": "patologia calcolo organo", "candidates": [ "1133.jpg", "850.jpg", "743.jpg", "1266.jpg", "1367.jpg", "948.jpg", "549.jpg", "695.jpg", "1504.jpg", "1148.jpg" ] }

System providing prediction, expected output: "743.jpg" OR 2

System providing ranking, expected output: ["948.jpg", "1504.jpg", ..., "1148.jpg"] OR [5, 8, ..., 9]

To submit your results send an email to the following addresses:

• elio.musacchio@uniba.it
• lucia.siciliani@uniba.it
• pierpaolo.basile@uniba.it

Baseline Results

The baseline is a Multilingual CLIP model. We used the pre-trained model without further fine-tuning.

HIT@1: 0.4505
MRR: 0.6244