Generating radiology reports from 3D volumetric data remains challenging due to the difficulty of grounding fine-grained pathologies within high-dimensional scans.While retrievalaugmented generation (RAG) offers a potential solution, standard approaches struggle with visual-semantic ambiguity and often introduce irrelevant "normal" context that dilutes pathological signals.To address this limitation, we introduce CPR-RAG, a model-agnostic RAG framework that enhances organ-level grounding by integrating clinical priors into the retrieval process.Specifically, we propose a clinical prior-regularized re-ranking module that leverages corpus-derived co-occurrence statistics to align retrieved candidates with latent disease distributions, ensuring clinical consistency beyond mere visual similarity.Furthermore, we employ clinical relevance context refinement to selectively filter out boilerplate normal descriptions, thereby maximizing the information density of the evidence provided to the generator.Extensive experiments on the RadGenome-ChestCT benchmark demonstrate that CPR-RAG significantly improves clinical efficacy across state-of-theart radiology report generation models.Human evaluation further confirms that our approach achieves superior factual correctness, completeness, and utility compared to the existing models.

Conversational interfaces have emerged as an accessible and user-friendly alternative to traditional touch-based self-service kiosks in food-ordering systems. Despite their promise, building such systems remains challenging due to the need for costly data annotation, store-specific model adaptation, and scalable deployment. In this study, we propose a fully automated, end-to-end framework that transforms structured menu databases into high-quality annotated datasets and efficiently deploys store-specific conversational models using a parameter-efficient fine-tuning method. Our approach fine-tunes only 0.9% of the backbone model parameters per store, enabling cost-effective and plug-and-play deployment across diverse environments. To enhance robustness, we further integrate a recommendation module that suggests alternative items when requested menu options are unavailable. Experimental results on data from 27 stores in South Korea demonstrate that our framework consistently outperforms existing data generation baselines in intent classification and slot filling performance, while maintaining high annotation quality. Simulated real-world voice-ordering scenarios confirm the practicality of our framework for rapid, scalable, and accessible deployment in real-world environments.

Large language models (LLMs) have achieved remarkable performance across diverse natural language processing tasks, but concerns persist regarding their potential political biases.While prior research has extensively explored political biases in LLMs' text generation and perception, limited attention has been devoted to biases associated with media outlet names.In this study, we systematically investigate the presence of media outlet name biases in LLMs and evaluate their impact on downstream tasks, such as political bias prediction and news summarization.Our findings demonstrate that LLMs consistently exhibit biases toward the known political leanings of media outlets, with variations across model families and scales.We propose a novel metric to quantify media outlet name biases in LLMs and leverage this metric to develop an automated prompt optimization framework.Our framework effectively mitigates media outlet name biases, offering a scalable approach to enhancing the fairness of LLMs in news-related applications.GitHub

Recent advancements in large language models have significantly influenced the field of online medical consultations.However, critical challenges remain, such as the generation of hallucinated information and the integration of upto-date medical knowledge.To address these issues, we propose Informatics Llama (ILlama), a novel framework that combines retrievalaugmented generation (RAG) with a structured medical knowledge graph.ILlama incorporates relevant medical knowledge by transforming subgraphs from a structured medical knowledge graph into text for RAG.By generating subgraphs from the medical knowledge graph in advance for RAG, specifically focusing on diseases and symptoms, ILlama enhances the accuracy and relevance of its medical reasoning.This framework enables effective incorporation of causal relationships between symptoms and diseases.Also, it delivers context-aware consultations aligned with user queries.Experimental results on the two medical consultation datasets demonstrate that ILlama outperforms strong baselines, achieving a semantic similarity F1 score of 0.884 when compared to ground-truth consultation answers.Furthermore, qualitative analysis of ILlama's responses reveals significant improvements in hallucination reduction and clinical usefulness.These results suggest that ILlama has strong potential as a reliable tool for real-world medical consultation environments.Our implementation is available at:

As microblogs have become commonplace, recommending relevant hashtags for microblog posts has become increasingly important. However, recommending appropriate hashtags for a post is challenging because it requires a high-level understanding of the context and relationships of the information in the post. In this paper, we propose a novel hashtag recommendation framework that incorporates external knowledge to enrich the context of posts. Using an image of the post, we obtain the hierarchical external knowledge extracted by the Open Directory Project (ODP)-based classifier. Experimental results show that our framework performs better than the baselines on a multimodal hashtag recommendation benchmark dataset. It outperformed the existing state-of-the-art model by providing a 39.86% increase in the average F1-score.

Masked language modeling (MLM) has been widely used for pre-training effective bidirectional representations, but incurs substantial training costs.In this paper, we propose a novel concept-based curriculum masking (CCM) method to efficiently pre-train a language model.CCM has two key differences from existing curriculum learning approaches to effectively reflect the nature of MLM.First, we introduce a carefully-designed linguistic difficulty criterion that evaluates the MLM difficulty of each token.Second, we construct a curriculum that gradually masks words related to the previously masked words by retrieving a knowledge graph.Experimental results show that CCM significantly improves pre-training efficiency.Specifically, the model trained with CCM shows comparative performance with the original BERT on the General Language Understanding Evaluation benchmark at half of the training cost.Code is available at https://github.com/KoreaMGLEE/Concept- based-curriculum-masking.

Deep neural networks typically have extensive parameters and computational operations. Pruning and quantization techniques have been widely used to reduce the complexity of deep models. Both techniques can be jointly used for realizing significantly higher compression ratios. However, separate optimization processes and difficulties in choosing the hyperparameters limit the application of both the techniques simultaneously. In this study, we propose a novel compression framework, termed as quantized sparse training, that prunes and quantizes networks jointly in a unified training process. We integrate pruning and quantization into a gradient-based optimization process based on the straight-through estimator. Quantized sparse training enables us to simultaneously train, prune, and quantize a network from scratch. The empirical results validate the superiority of the proposed methodology over the recent state-of-the-art baselines with respect to both the model size and accuracy. Specifically, quantized sparse training achieves a 135 KB model size in the case of VGG16, without any accuracy degradation, which is 40% of the model size feasible based on the state-of-the-art pruning and quantization approach.

Distributed representations of words have been an indispensable component for natural language processing (NLP) tasks.However, the large memory footprint of word embeddings makes it challenging to deploy NLP models to memory-constrained devices (e.g., selfdriving cars, mobile devices).In this paper, we propose a novel method to adaptively compress word embeddings.We fundamentally follow a code-book approach that represents words as discrete codes such as (8, 5, 2, 4).However, unlike prior works that assign the same length of codes to all words, we adaptively assign different lengths of codes to each word by learning downstream tasks.The proposed method works in two steps.First, each word directly learns to select its code length in an end-to-end manner by applying the Gumbel-softmax tricks.After selecting the code length, each word learns discrete codes through a neural network with a binary constraint.To showcase the general applicability of the proposed method, we evaluate the performance on four different downstream tasks.Comprehensive evaluation results clearly show that our method is effective and makes the highly compressed word embeddings without hurting the task accuracy.Moreover, we show that our model assigns word to each code-book by considering the significance of tasks.

Pre-trained word embeddings are widely used in various fields. However, the coverage of pre-trained word embeddings only includes words that appeared in corpora where pre-trained embeddings are learned. It means that the words which do not appear in training corpus are ignored in tasks, and it could lead to the limited performance of neural models. In this paper, we propose a simple yet effective method to represent out-of-vocabulary (OOV) words. Unlike prior works that solely utilize subword information or knowledge, our method makes use of both information to represent OOV words. To this end, we propose two stages of representation learning. In the first stage, we learn subword embeddings from the pre-trained word embeddings by using an additive composition function of subwords. In the second stage, we map the learned subwords into semantic networks (e.g., WordNet). We then re-train the subword embeddings by using lexical entries on semantic lexicons that could include newly observed subwords. This two-stage learning makes the coverage of words broaden to a great extent. The experimental results clearly show that our method provides consistent performance improvements over strong baselines that use subwords or lexical resources separately.

Existing keyphrase extraction approaches often suffer from issues such as the sparsity and brevity of short text (e.g., headlines, queries, and tweets). In this paper, we propose a novel keyphrase extraction method for short text by utilizing recurrent neural networks. The main idea behind our approach is to classify short text into a relevant class or category and extract keyphrases from important words in the class or category. Unlike previous supervised approaches that need the information of annotated keyphrases, our approach requires only a text classification dataset (i.e., DBpedia), which is easier to use and requires less human effort. In our approach, we first feed short text into the attention-based neural network for text classification. We then compute attention weights of each word in input short text. Subsequently, we detect keyphrase candidates by chunking phrases and summing the attention weights of compositional words in the chunked phrase. The experimental results clearly show the efficacy of our approach on real-world datasets, such as headlines, queries, and tweets. The proposed method outperforms the Microsoft Cognitive Services and IBM Watson Natural Language Understanding service for keyphrase extraction in terms of F1-score and acceptable percentage on the NYT and Question datasets. Further, we confirm that the proposed method is comparable to supervised methods for keyphrase extraction from short text in the Tweet dataset.

In this demo, we demonstrate an in-device conversational photo sharing service, termed meChat, which helps users share in-device photos easily in messaging applications by searching conversation-related photos automatically. In particular, meChat understands the semantics of on-going conversation and in-device photos by projecting both of them into a single semantic space. Subsequently, it retrieves in-device photos related to the conversation context. Through this process, meChat makes it easy for the users to share the photos while communicating in messaging applications. In addition, it is worth noting that meChat works in a stand-alone, privacy-protecting manner without sending out any in-device photos and conversations to the external servers. This is different from existing photo services (e.g., Google Photos) which resort on the cloud server.

Neural network models have achieved impressive results in the field of text classification. However, existing approaches often suffer from insufficient training data in a large-scale text classification involving a large number of categories (e.g., several thousands of categories). Several neural network models have utilized multi-task learning to overcome the limited amount of training data. However, these approaches are also limited to small-scale text classification. In this paper, we propose a novel neural network-based multi-task learning framework for large-scale text classification. To this end, we first treat the different scales of text classification (i.e., large and small numbers of categories) as multiple, related tasks. Then, we train the proposed neural network, which learns small- and large-scale text classification tasks simultaneously. In particular, we further enhance this multi-task learning architecture by using a gate mechanism, which controls the flow of features between the small- and large-scale text classification tasks. Experimental results clearly show that our proposed model improves the performance of the large-scale text classification task with the help of the small-scale text classification task. The proposed scheme exhibits significant improvements of as much as 14% and 5% in terms of micro-averaging and macro-averaging F1-score, respectively, over state-of-the-art techniques.

Distributed representations of words play a major role in the field of natural language processing by encoding semantic and syntactic information of words. However, most existing works on learning word representations typically regard words as individual atomic units and thus are blind to subword information in words. This further gives rise to a difficulty in representing out-of-vocabulary (OOV) words. In this paper, we present a character-based word representation approach to deal with this limitation. The proposed model learns to generate word representations from characters. In our model, we employ a convolutional neural network and a highway network over characters to extract salient features effectively. Unlike previous models that learn word representations from a large corpus, we take a set of pre-trained word embeddings and generalize it to word entries, including OOV words. To demonstrate the efficacy of the proposed model, we perform both an intrinsic and an extrinsic task which are word similarity and language modeling, respectively. Experimental results show clearly that the proposed model significantly outperforms strong baseline models that regard words or their subwords as atomic units. For example, we achieve as much as 18.5% improvement on average in perplexity for morphologically rich languages compared to strong baselines in the language modeling task.

In-app advertising has become a significant source of revenue for mobile app. In order to improve the effectiveness of in-app ads, most ad networks focus on targeting a user based on the user's personal information collected from their ad library inside mobile apps and the global knowledge built from big data on ad servers. However, sharing user's sensitive information with the ad servers may raise privacy concerns. As opposed to targeting users, mobile contextual advertising seeks to target the app page a user is viewing. In this demo, we present a novel mobile contextual advertising platform, called MoCA, which is designed to improve the semantic relevance of in-app ads in a stand-alone, privacy-protecting manner on mobile devices. MoCA understands the semantics of app page and ads, and then matches semantically relevant ads to the page inside mobile devices. To the best of our knowledge, this is the first work to implement the mobile contextual advertising platform based on the semantic approach without resort to ad servers.