Effective question answering over heterogeneous documents requires not only powerful reasoning models but also careful decisions about how evidence is structured and retrieved. In many real-world settings, information is inherently multimodal: a researcher assessing a scientific finding must reason jointly over paragraphs, figures, and result tables; a financial analyst combines textual summaries with tabular data; and a clinician interprets diagnostic imaging alongside patient records. Yet the question of how to decompose such heterogeneous documents into meaningful, retrievable units of evidence remains largely unexplored.

Recent advances in interactive text-to-image retrieval (I-TIR) use diffusion models to bridge the modality gap between the textual information need and the images to be searched. However, existing frameworks fuse multi-modal user feedback by simple embedding addition. In this work, we show that this basic fusion strategy indiscriminately incorporates generative noise produced by the diffusion model, leading to performance degradation for up to 55.62% of samples. We further propose ADaFuSE (Adaptive Diffusion-Text Fusion with Semantic-aware Experts), a lightweight fusion model designed to align and calibrate multi-modal views for diffusion-augmented I-TIR, which can be plugged into existing frameworks without modifying the backbone encoder. Specifically, we introduce a dual-branch fusion mechanism that employs an adaptive gating branch to dynamically balance modality reliability, alongside a semantic-aware mixture-of-experts branch to capture fine-grained cross-modal nuances. Via thorough evaluation over four standard I-TIR benchmarks, ADaFuSE achieves state-of-the-art performance, surpassing the DAR model by up to 3.49% in Hits@10 with only a 5.29% parameter increase, while exhibiting stronger robustness to noisy and longer interactive queries. These results show that generative augmentation coupled with principled fusion provides a simple, generalizable alternative to fine-tuning for tasks like product search.

Interactive Text-to-image retrieval (I-TIR) is an important enabler for a wide range of state-of-the-art services in domains such as e-commerce and education.However, current methods rely on finetuned Multimodal Large Language Models (MLLMs), which are costly to train and update, and exhibit poor generalizability.This latter issue is of particular concern, as: 1) finetuning narrows the pretrained distribution of MLLMs, thereby reducing generalizability; and 2) I-TIR introduces increasing query diversity and complexity.As a result, I-TIR solutions are highly likely to encounter queries and images not well represented in any training dataset.To address this, we propose leveraging Diffusion Models (DMs) for text-to-image mapping, to avoid finetuning MLLMs while preserving robust performance on complex queries.Specifically, we introduce Diffusion Augmented Retrieval (DAR), a framework that generates multiple intermediate representations via LLM-based dialogue refinements and DMs, producing a richer depiction of the user's information needs.This augmented representation facilitates more accurate identification of semantically and visually related images.Extensive experiments on four benchmarks show that for simple queries, DAR achieves results on par with finetuned I-TIR models, yet without incurring their tuning overhead.Moreover, as queries become more complex through additional conversational turns, DAR surpasses finetuned I-TIR models by up to 7.61% in Hits@10 after ten turns, illustrating its improved generalization for more intricate queries.