Chenghao Yang, Yuning Zhang, Zhoufutu Wen, Tao Gong, Jiaheng Liu, Qi Chu, Nenghai Yu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Sunbowen Lee, Junting Zhou, Chang Ao, Kaige Li, Xeron Du, Sirui He, Haihong Wu, Tianci Liu, Jiaheng Liu, Hamid Alinejad-Rokny, Min Yang, Yitao Liang, Zhoufutu Wen, Shiwen Ni. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Understanding multimodal video ads is crucial for improving query-ad matching and relevance ranking on short video platforms, enhancing advertising effectiveness and user experience. However, the effective utilization of multimodal information with high commercial value still largely constrained by reliance on highly compressed video embeddings-has long been inadequate. To address this, we propose SUMMA (the abbreviation of SUmmarizing MultiModalAds), a multimodal model that automatically processes video ads into summaries highlighting the content of highest commercial value, thus improving their comprehension and ranking in Douyin search-advertising systems. SUMMA is developed via a two-stage training strategy-multimodal supervised fine-tuning followed by reinforcement learning with a mixed reward mechanism-on domain-specific data containing video frames and ASR/OCR transcripts, generating commercially valuable and explainable summaries. We integrate SUMMA-generated summaries into our production pipeline, directly enhancing the candidate retrieval and relevance ranking stages in real search-advertising systems. Both offline and online experiments show substantial improvements over baselines, with online results indicating a statistically significant 1.5% increase in advertising revenue. Our work establishes a novel paradigm for condensing multimodal information into representative texts, effectively aligning visual ad content with user query intent in retrieval and recommendation scenarios.

The increasing application of multi-modal large language models (MLLMs) across various sectors have spotlighted the essence of their output reliability and accuracy, particularly their ability to produce content grounded in factual information (e.g. common and domain-specific knowledge). In this work, we introduce SimpleVQA, the first comprehensive multi-modal benchmark to evaluate the factuality ability of MLLMs to answer natural language short questions. SimpleVQA is characterized by six key features: it covers multiple tasks and multiple scenarios, ensures high quality and challenging queries, maintains static and timeless reference answers, and is straightforward to evaluate. Our approach involves categorizing visual question-answering items into 9 different tasks around objective events or common knowledge and situating these within 9 topics. Rigorous quality control processes are implemented to guarantee high-quality, concise, and clear answers, facilitating evaluation with minimal variance via an LLM-as-a-judge scoring system. Using SimpleVQA, we perform a comprehensive assessment of leading 18 MLLMs and 8 text-only LLMs, delving into their image comprehension and text generation abilities by identifying and analyzing error cases.

In this paper, we introduce Knowledge-Orthogonal Reasoning (KOR), a concept aimed at minimizing reliance on domain-specific knowledge, enabling more accurate evaluation of models' reasoning abilities in out-of-distribution settings. Based on this concept, we propose the Knowledge-Orthogonal Reasoning Benchmark (KOR-Bench), encompassing five task categories: Operation, Logic, Cipher, Puzzle, and Counterfactual. KOR-Bench emphasizes models' effectiveness in applying new rule descriptions to solve novel rule-driven questions. O1-Preview and O1-Mini achieve accuracies of 72.88\% and 70.16\%, surpassing Claude-3.5-Sonnet and GPT-4o (58.96\% and 58.00\%), highlighting the effectiveness of KOR-Bench. We perform detailed analyses, identifying bottlenecks in the Cipher task with Stepwise Prompting, where two rounds of Self-Correction yield optimal results. We evaluate performance across three integrated tasks, explore the impact of Tricks on the Puzzle task, and visualize rule-focused attention. Additionally, we conduct an ablation study on dataset size, benchmark correlations, and zero-shot and three-shot "only questions" experiments. KOR-Bench aims to enhance reasoning evaluation and support further research in this area.

Recent advancements in large language models (LLMs) underscore the need for more comprehensive evaluation methods to accurately assess their reasoning capabilities. Existing benchmarks are often domain-specific and thus cannot fully capture an LLM’s general reasoning potential. To address this limitation, we introduce the **Knowledge Orthogonal Reasoning Gymnasium (KORGym)**, a dynamic evaluation platform inspired by KOR-Bench and Gymnasium. KORGym offers over fifty games in either textual or visual formats and supports interactive, multi-turn assessments with reinforcement learning scenarios. Using KORGym, we conduct extensive experiments on 19 LLMs and 8 VLMs, revealing consistent reasoning patterns within model families and demonstrating the superior performance of closed-source models. Further analysis examines the effects of modality, reasoning strategies, reinforcement learning techniques, and response length on model performance. We expect KORGym to become a valuable resource for advancing LLM reasoning research and developing evaluation methodologies suited to complex, interactive environments.

In the multimedia era, image becomes an effective medium in search advertising. Dynamic Image Advertising (DIA), a system that matches queries with appropriate ad images and generates multimodal ads, is introduced to improve user experience and ad revenue. The core of DIA is a query-image matching module performing ad image retrieval and relevance modeling. Current query-image matching suffers from data scarcity and inconsistency, and insufficient cross-modal fusion. Also, the retrieval and relevance models are separately trained, affecting overall performance. In this paper, we propose a vision-language framework for query-image matching. It consists of two parts. First, we design a base model combining different encoders and tasks, and train it on large-scale image-text pairs to learn general multimodal representation. Then, we fine-tune the base model on advertising business data, unifying relevance modeling and retrieval through multi-objective learning. Our framework has been implemented in Baidu search advertising system "Phoneix Nest". Online evaluation shows that it improves cost per mille (CPM) and click-through rate (CTR) by 1.04% and 1.865% on the system main traffic.