Papers by Peinan Zhang
FAST: Fast Annotation tool for SmarT devices (2021.emnlp-demo)
Copied to clipboard
| Challenge: | In real-world applications, annotators with the same attributes are required to annotate whether the outputs of natural language generation systems are fluent or not. |
| Approach: | They propose an annotation tool for application tasks that focuses on the user experience of mobile devices and can be customized to fit various tasks. |
| Outcome: | The proposed tool can annotate faster than existing methods while maintaining the quality of annotation. |
Cross-lingual Transfer or Machine Translation? On Data Augmentation for Monolingual Semantic Textual Similarity (2024.lrec-main)
Copied to clipboard
| Challenge: | Using labeled NLI datasets for learning sentence embeddings leads to improved performance for natural language understanding tasks. |
| Approach: | They compare two data augmentation techniques for learning better sentence embeddings . they use a cross-lingual transfer technique that exploits English resources as training data to yield non-English sentence embeds as zero-shot inference . |
| Outcome: | The proposed techniques yield better performance on Japanese and Korean sentences. |
Aspect-based Analysis of Advertising Appeals for Search Engine Advertising (2022.naacl-industry)
Copied to clipboard
| Challenge: | ad creators must consider various aspects of advertising appeals such as price, product features, and quality in their ac work. |
| Approach: | They propose to use a dataset of ad texts to explore the effective aspects of advertising appeals (A3) for different industries to assist a search engine ap creators. |
| Outcome: | The proposed model can detect aspects of ad texts and help them estimate their performance. |
Beyond Decoder-only: Large Language Models Can be Good Encoders for Machine Translation (2025.findings-acl)
Copied to clipboard
Yingfeng Luo, Tong Zheng, Yongyu Mu, Bei Li, Qinghong Zhang, Yongqi Gao, Ziqiang Xu, Peinan Feng, Xiaoqian Liu, Tong Xiao, JingBo Zhu
| Challenge: | Recent advances in machine translation have focused on a single pre-trained decoder . encoder-decoder architectures have received relatively little attention in NMT . |
| Approach: | They propose a method that leverages LLMs as MT encoders and pairs them with lightweight decoders to develop universal translation models. |
| Outcome: | The proposed method matches or surpasses baselines in terms of translation quality but achieves 75% reduction in memory footprint of the KV cache. |
CAMERA³: An Evaluation Dataset for Controllable Ad Text Generation in Japanese (2024.lrec-main)
Copied to clipboard
| Challenge: | Despite numerous efforts in ad text generation, the aspect of diversifying a text has received limited attention, particularly in non-English languages like Japanese. |
| Approach: | They present a dataset for ad text generation in Japanese using annotators to examine the capabilities of recent NLG models. |
| Outcome: | The proposed dataset includes 3,980 ad texts written by experts taking into account various aspects of ade appeals. |
AdTEC: A Unified Benchmark for Evaluating Text Quality in Search Engine Advertising (2025.naacl-long)
Copied to clipboard
| Challenge: | Existing pre-trained language models outperform them in certain domains, indicating that there is significant potential for further improvement in this area. |
| Approach: | They propose to use pre-trained language models to evaluate ad texts from multiple perspectives within real-world advertising operations to define five tasks and construct a Japanese dataset. |
| Outcome: | The proposed benchmark outperforms existing pre-trained language models in several tasks, but humans outperformed them in certain domains. |
AdParaphrase v2.0: Generating Attractive Ad Texts Using a Preference-Annotated Paraphrase Dataset (2025.findings-acl)
Copied to clipboard
| Challenge: | Identifying factors that make ad text attractive is essential for advertising success . identifying the linguistic factors presents a significant challenge because of the intricate interplay between the semantic content and its linguistic expression. |
| Approach: | They propose to use a dataset for ad text paraphrasing that contains human preference data to enable analysis of linguistic factors. |
| Outcome: | The proposed dataset is 20 times larger than v1.0 and contains 16,460 pairs of ad text paraphrase pairs . it shows that human preference and ade- t attractiveness are related . |
Revealing the Parallel Multilingual Learning within Large Language Models (2024.emnlp-main)
Copied to clipboard
Yongyu Mu, Peinan Feng, Zhiquan Cao, Yuzhang Wu, Bei Li, Chenglong Wang, Tong Xiao, Kai Song, Tongran Liu, Chunliang Zhang, JingBo Zhu
| Challenge: | Large language models (LLMs) can handle multilingual and cross-lingual text within a single input; however, previous studies focusing on using English as the pivot language to enhance language understanding and reasoning focus on using multiple languages. |
| Approach: | They propose to use parallel multilingual input to enhance the model's comprehension of the input and to examine how multilingual processing affects prediction. |
| Outcome: | The proposed model can handle multilingual and cross-lingual text within a single input, but previous studies focused on using English as the pivot language to enhance language understanding and reasoning. |
An Empirical Study of Generating Texts for Search Engine Advertising (2021.naacl-industry)
Copied to clipboard
| Challenge: | Existing studies on neural language generation have not evaluated the effect of generated ads with actual serving included because it requires a large amount of training data and a particular environment. |
| Approach: | They propose to integrate a reinforcement learning framework into an end-to-end sequence-tosequence (Seq2S) model and demonstrate how to improve the ads’ impact, deploy models to a product, and evaluate the generated ads. |
| Outcome: | The proposed method improves the ads’ impact, deploys the models to a product, and evaluates the generated ads. |
Does Self-Consistency Improve the Recall of Encyclopedic Knowledge? (2026.acl-short)
Copied to clipboard
| Challenge: | a lack of evaluation grounds for self-consistency on symbolic reasoning is unclear . however, it is unclear whether it improves performance on non-math questions involving encyclopedic knowledge. |
| Approach: | They establish a knowledge recall split for the popular MMLU benchmark by applying a data-driven heuristic from prior work. |
| Outcome: | The proposed knowledge recall split achieves an 89% accuracy on the MMLU benchmark. |
Generating Diverse and High-Quality Texts by Minimum Bayes Risk Decoding (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing decoding algorithms to generate diverse outputs are based on beam search or random sampling, thus their output quality is capped by these underlying decoding methods. |
| Approach: | They propose to add a diversity penalty to MBR decoding and a clustering problem to create diversity-promoting decoding algorithms by enforcing diversity objectives. |
| Outcome: | The proposed method achieves a better trade-off than the diverse beam search and sampling algorithms overall. |
Distilling Many-Shot In-Context Learning into a Cheat Sheet (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) enable effective in-context learning with many-shot examples, but at the cost of high computational demand due to longer input tokens. |
| Approach: | proposed cheat-sheet ICL distills information from many-shot ICL into a concise textual summary . experiment shows cheat- sheet ICL achieves comparable or better performance than many- shot ICL . |
| Outcome: | Experiments on reasoning tasks show that cheat-sheet ICL achieves comparable or better performance than many-shot ICL with far fewer tokens. |
Progressive Visual Refinement for Multi-modal Summarization (2026.eacl-short)
Copied to clipboard
| Challenge: | Multi-modal summarization (MMS) is a critical research area driven by the proliferation of multimedia content. |
| Approach: | They propose a patch-refined visual information network to exploit multimodal information . they propose combining visual information with textual information to generate concise summaries . |
| Outcome: | Extensive experiments on two public MMS datasets show the superiority of the proposed model. |
Striking Gold in Advertising: Standardization and Exploration of Ad Text Generation (2024.acl-long)
Copied to clipboard
| Challenge: | Existing benchmarks and problem sets for automatic ad text generation are lacking . however, the growing volume of search queries has fueled research on the automatic generation of ads. |
| Approach: | They propose to standardize the task of automatic ad text generation (ATG) using a benchmark dataset, CAMERA, to enable the utilization of multi-modal information and facilitate industry-wise evaluations. |
| Outcome: | The proposed dataset standardizes the task of automatic ad text generation (ATG) it shows that existing metrics align with human evaluations and that the proposed methods can be used to improve the quality of the results. |
AdParaphrase: Paraphrase Dataset for Analyzing Linguistic Features toward Generating Attractive Ad Texts (2025.findings-naacl)
Copied to clipboard
| Challenge: | Effective linguistic choices that attract potential customers play crucial roles in advertising success. |
| Approach: | They propose to use a paraphrase dataset to explore linguistic features of ad texts that influence human preferences to maximize the potential success of advertisements. |
| Outcome: | The proposed model improves the attractiveness of ad texts by focusing on human preferences. |
BannerBench: Benchmarking Vision Language Models for Multi-Ad Selection with Human Preferences (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Web banner advertisements are often selected manually because of human preferences . a new benchmark evaluates the degree of alignment with human preferences in two tasks . |
| Approach: | a benchmark was developed to evaluate the human preference-driven banner selection process using vision-language models. |
| Outcome: | The proposed benchmark assesses the degree of alignment with human preferences in two tasks using vision-language models. |