Iryna Gurevych, Marie-Francine Moens, Jonathan Tonglet
We lifted 17 functions out of this paper's own repositories and ran 15 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| UKPLab/5pils | canonical | 15 of 17 |
| Function | Status | Where it lives |
|---|---|---|
| concatenate_entry | Ran | UKPLab/5pils/utils.py code served (permissive licence) · get_code("6ad9e047fb83b6d3") |
| convert_to_date | Ran | UKPLab/5pils/evaluation/evaluation_metrics.py code served (permissive licence) · get_code("bbfbef96b946a60d") |
| cosine_similarity | Ran | UKPLab/5pils/baseline/generation_utils.py code served (permissive licence) · get_code("fc2f37b8acf5766b") |
| date_distance | Ran | UKPLab/5pils/evaluation/evaluation_metrics.py code served (permissive licence) · get_code("dd767a28b135bd80") |
| entry_exists | Ran | UKPLab/5pils/utils.py code served (permissive licence) · get_code("04823be297d284be") |
| get_geonames_hierarchy | Ran | UKPLab/5pils/evaluation/geonames_collection.py code served (permissive licence) · get_code("2a8e9af6866ca933") |
| get_organization | Ran | UKPLab/5pils/dataset_collection/preprocessing_utils.py code served (permissive licence) · get_code("11b07f7f8addccd3") |
| gpt4_prompting | Ran | UKPLab/5pils/baseline/llm_prompting.py code served (permissive licence) · get_code("408a24dade610ca2") |
| gpt4_vision_prompting | Ran | UKPLab/5pils/baseline/llm_prompting.py code served (permissive licence) · get_code("bab48b3fe0c88b5f") |
| is_english_article | Ran | UKPLab/5pils/dataset_collection/scrape_utils.py code served (permissive licence) · get_code("c8af83dd089eb7c5") |
| load_json | Ran | UKPLab/5pils/utils.py code served (permissive licence) · get_code("eb82c897d04b22b2") |
| load_urls | Ran | UKPLab/5pils/dataset_collection/scrape_utils.py code served (permissive licence) · get_code("eaa4649b3cda1658") |
| search_location | Ran | UKPLab/5pils/evaluation/geonames_collection.py code served (permissive licence) · get_code("b73e80aac59eccfa") |
| sort_with_clip_score | Ran | UKPLab/5pils/baseline/generation_utils.py code served (permissive licence) · get_code("12ea3e061c3f8091") |
| sort_with_image_similarity | Ran | UKPLab/5pils/baseline/generation_utils.py code served (permissive licence) · get_code("d78f28d04bd9687a") |
| assemble_prompt_gpt4 | Not yet run | UKPLab/5pils/baseline/llm_prompting.py code served (permissive licence) · get_code("ba1226b6627f813f") |
| scrape_image | Not yet run | UKPLab/5pils/dataset_collection/scrape_utils.py code served (permissive licence) · get_code("16506989e60a009d") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
To assist human fact-checkers, researchers have developed automated approaches for visual misinformation detection. These methods assign veracity scores by identifying inconsistencies between the image and its caption, or by detecting forgeries in the image. However, they neglect a crucial point of the human factchecking process: identifying the original metacontext of the image. By explaining what is actually true about the image, fact-checkers can better detect misinformation, focus their efforts on check-worthy visual content, engage in counter-messaging before misinformation spreads widely, and make their explanation more convincing. Here, we fill this gap by introducing the task of automated image contextualization. We create 5Pils, a dataset of 1,676 fact-checked images with questionanswer pairs about their original meta-context. Annotations are based on the 5 Pillars factchecking framework. We implement a first baseline that grounds the image in its original meta-context using the content of the image and textual evidence retrieved from the open web. Our experiments show promising results while highlighting several open challenges in retrieval and reasoning. We make our code and data publicly available. 1
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2408.09939")
get_code_for_paper("2408.09939")
have("2408.09939")
Connect an agent — have() is free.