{"id":43003,"date":"2025-07-13T09:32:40","date_gmt":"2025-07-13T13:32:40","guid":{"rendered":"https:\/\/www.bloomberg.com\/company\/stories\/bloomberg-ai-engineers-publish-3-information-retrieval-research-papers-sigir-2025\/"},"modified":"2025-07-13T22:28:26","modified_gmt":"2025-07-14T02:28:26","slug":"bloomberg-ai-engineers-publish-3-information-retrieval-research-papers-sigir-2025","status":"publish","type":"post","link":"https:\/\/www.bloomberg.com\/company\/stories\/bloomberg-ai-engineers-publish-3-information-retrieval-research-papers-sigir-2025\/","title":{"rendered":"Bloomberg’s AI Engineers Publish 3 Information Retrieval Research Papers at SIGIR 2025"},"content":{"rendered":"
\n \n\t\n\t\n\t
\n\t\t\n\t\t\t
\n\t
\n \n

During the <\/span>48th International ACM SIGIR Conference<\/span><\/a> (SIGIR 2025) in Padua, Italy this week (July 13-17, 2025), researchers from Bloomberg\u2019s AI Engineering Group are showcasing their expertise in information retrieval (IR) by publishing three papers at the conference.<\/span><\/p>\n

IR is the process of finding and accessing relevant information from a collection of data, such as documents, images or databases, in response to a user’s query. IR is critical for combating information overload and facilitating access to knowledge in a world full of data. AI is revolutionizing the field, and research presented at SIGIR will be applied toward making information retrieval systems more robust, effective, and efficient.<\/span><\/p>\n\n<\/div>\n

\n \"\"\"\"\n \n<\/figure>\n
\n \n

In addition, Bloomberg AI Research Scientist Shuo Zhang is one of the organizers of the <\/span>2nd Workshop on Financial Information Retrieval in the Era of Generative AI<\/span><\/a> (FinIR) on July 17, 2025. The workshop serves as a forum for researchers and practitioners to explore potential approaches and research directions to address the challenges of using generative AI in finance through the use of advanced IR techniques. Its goal is to deepen understanding, accelerate progress, and support the advancement of IR technology to enhance generative models to address financial challenges.<\/span><\/p>\n

We asked some of the authors to summarize their research and explain why the results were notable:<\/span><\/p>\n

\n

Monday, July 14, 2025<\/u><\/strong><\/h3>\n

Reproducibility: Domain-Specific, Multimodal, and Multilingual Retrieval<\/em>
\n10:30-12:30 CEST<\/em><\/p>\n

Benchmark Granularity and Model Robustness for Image-Text Retrieval: A Reproducibility Study<\/strong><\/span><\/a>
\nMariya Hendriksen (University of Amsterdam), Shuo Zhang (Bloomberg), Ridho Reinanda (Bloomberg), Mohamed Yahya (Bloomberg), Edgar Meij (Bloomberg), Maarten de Rijke (University of Amsterdam)<\/span><\/p>\n\n<\/div>\n

\n \"Click\"Click\n \n<\/figure>\n
\n \n

Please summarize your research. Why are your results notable?<\/strong><\/p>\n

Shuo Zhang:<\/strong> In our study, we examined how well today\u2019s AI models can match images with text descriptions \u2014 a technology behind many modern search and recommendation systems. While these systems are widely used, most are only tested with short, simple captions that don\u2019t reflect how people actually search in real life. Our work looked at two often-overlooked factors: the level of detail in descriptions (caption \u201cgranularity\u201d) and how well these systems handle errors or unexpected input (\u201crobustness\u201d). See Figure 1 as an illustration.<\/p>\n

Granularity refers to how much detail is present in a caption. A simple, coarse caption might be \u201ca red rose is sitting next to a couple of mugs,\u201d while a fine-grained version could be \u201ca single red rose in a glass vase sits on a dining table, next to two ceramic mugs with floral designs.\u201d<\/p>\n

Robustness (illustrated below), on the other hand shows how even a small typo (\u201ccouple\u201d vs. \u201ccoupel,\u201d or \u201cmotorcycles\u201d vs. \u201comtorcycles\u201d) can increase or decrease the quality of retrieved images. This highlights the importance of evaluating models under realistic, imperfect conditions, not just with idealized queries.<\/p>\n\n<\/div>\n

\n \"Examples\"Examples\n
Examples of perturbation effects on R@1 (Recall@1) for image retrieval<\/span><\/figcaption>\n<\/figure>\n
\n \n

We evaluated four of the most advanced vision-language models, <\/span>CLIP<\/b><\/a>, <\/b>ALIGN<\/b><\/a>, <\/b>AltCLIP<\/b><\/a>, <\/b>and <\/span>GroupViT<\/b><\/a>, on both standard datasets and new versions we created, which include richer, more descriptive captions. To test robustness, we introduced a wide range of common mistakes, \u2014 like word order changes, typos, and synonyms \u2014 into the search queries, and measured how each model performed.<\/span><\/p>\n

Key Findings<\/span><\/p>\n