Reasoning-intensive ranking models built on Large Language Models (LLMs) have made notable progress. However, existing approaches often rely on large-scale LLMs and explicit Chain-of-Thought (CoT) reasoning, resulting in high computational cost and latency that limit real-world use. To address this, we propose TFRank, an efficient pointwise reasoning ranker based on small-scale LLMs. To improve ranking performance, TFRank effectively integrates CoT data, fine-grained score supervision, and multi-task training. Furthermore, it achieves an efficient "Think-Free" reasoning capability by employing a "think-mode switch" and pointwise format constraints. Specifically, this allows the model to leverage explicit reasoning during training while delivering precise relevance scores for complex queries at inference without generating any reasoning chains. Experiments show that TFRank achieves performance comparable to models with four times more parameters on the BRIGHT benchmark, and demonstrates strong competitiveness on the BEIR benchmark. Further analysis shows that TFRank achieves an effective balance between performance and efficiency, providing a practical solution for integrating advanced reasoning into real-world systems.

Hang Hu, Ziyan Liu, Rujie Wen, Ruihui Hou, Xueyan Wu, Mu Zhang, Jianxing Yu, Tong Ruan, Jingping Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Medical quality control (MQC) indicators are essential for evaluating the performance of healthcare institutions to ensure high-quality patient care. In this paper, we report the design, implementation, and deployment of the Intelligent EMR-LLM platform for Medical Quality Control (IMQC), a large language model (LLM)-empowered system for automatically computing MQC indicators for enhancing the quality of medical services in Shanghai. It consists of an LLM (i.e., EMR-LLM) for processing electronic medical records (EMRs). With EMR-LLM, IMQC translates existing MQC indicators into a standardized representation language and automatically computes them based on EMRs. Since its deployment in February 2024, IMQC has been adopted by the Shanghai Medical Quality Management Center and associated hospitals. So far, it has processed 1,245 medical quality indicators for secondary- and tertiary-level hospitals, achieving an MQC evaluation accuracy of 93.31%, which is comparable to human experts. It has significantly improved efficiency, increasing from 10 EMRs per hour per human expert to over 1,000 EMRs per hour on average using one single H800 GPU. Over the first round of deployment in Shanghai, it is estimated that IMQC saves around 3.42 million RMB per month in manpower costs compared to traditional reporting methods. The successful deployment of IMQC sets a precedence for other regions to adopt similar AI-driven solutions to enhance medical quality control.

Jingping Liu, Ziyan Liu, Zhedong Cen, Yan Zhou, Yinan Zou, Weiyan Zhang, Haiyun Jiang, Tong Ruan. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Elasticsearch (ES) is a distributed RESTful search engine optimized for large-scale and long-text search scenarios. Recent research on text-to-Query has explored using large language models (LLMs) to convert user query intent to executable code, making it an increasingly popular research topic. To our knowledge, we are the first to introduce the novel semantic parsing task text-to-ES. To bridge the gap between LLM and ES, in detail, we leverage LLMs and employ domain experts to generate ES query bodies, which are Domain-Specific Language (DSL), along with the corresponding post-processing code to support multi-index ES queries. Consequently, we propose the text-to-ES benchmark that consists of two datasets: Large Elasticsearch Dataset (LED), containing 26,207 text-ES pairs derived from a 224.9GB schema-free database, and ElasticSearch (BirdES)with 10,926 pairs sourced from the Bird dataset on a 33.4GB schema-fixed database. Compared with fourteen advanced LLMs and six code-based LLMs, the model we trained outperformed DeepSeek-R1 by 15.64% on the LED dataset, setting a new state-of-the-art, and achieved 78% of DeepSeek-R1’s performance on the BirdES dataset. Additionally, we provide in-depth experimental analyses and suggest future research directions for this task. Our datasets are available at https://huggingface.co/datasets/Barry1915/Text-to-ES.

Medical visual question answering (MedVQA) requires the model to answer natural language questions based on the given medical images. Recently, large vision-language models (LVLMs) have emerged as the dominant paradigm for this task, enabling open-ended predictions and effectively integrating textual and visual information. However, due to the lack of domain-specific medical knowledge, LVLMs often suffer from hallucination issues, leading to inaccurate answers and flawed chain-of-thought (CoT) reasoning. This limits both the accuracy and interpretability of their responses. Hence, we propose MedKCoT, a framework that integrates a multi-modal medical knowledge graph (MMKG) to mitigate LVLM hallucinations during CoT generation and introduces a novel supervision strategy to train an MMKG retriever without requiring manual annotations. Firstly, we define an LVLM preference score as the supervision signal to train an MMKG retriever. Secondly, the retriever searches entity knowledge related to the medical question-answering pair, which assists the LVLM in generating high-quality CoTs. Finally, the obtained CoTs and original question-answering pairs are used to fine-tune a lightweight VLM as the MedVQA model. We evaluate MedKCoT on two public MedVQA datasets Slake and VQA-RAD. Experimental results show that MedKCoT achieves improvements of 2.93 % and 2.58 % over previous methods. The source code is available at https://github.com/EnjoyFailure/MedKCoT.

Medical Decision Trees (MDTs) are widely used in clinical decision support due to their interpretability. Current methods for constructing clinical decision trees often only capture fragmented segments of the diagnostic process, failing to represent its full scope. In contrast, flowcharts in medical literature typically reflect comprehensive decision-making logic as used by experts in real clinical settings. However, converting flowcharts to decision trees is challenging due to complex structure parsing and irregular rule formats. To address this, we propose Flow2MDT (Flow2Dot & Dot2MDT), an automated framework that constructs high-quality MDTs from medical flowcharts. In the Flow2Dot stage, we first built a dataset of flowchart-Dot scripts and fine-tuned a multimodal model with this dataset to construct the end-to-end flowchart recognition model F2D, which converts the input flowchart into a Dot script output. Then, in the Dot2MDT stage, we optimized the node representation and tree structure to better align with clinical reasoning. To explore the application value of MDTs in the field of medical decisionmaking, we have developed an interactive differential diagnosis method based on diagnostic decision pathways and diagnostic evidence completeness assessment, demonstrating the practical utility of Flow2MDT in clinical decision support. Experimental results on both the public DDxPlus dataset and our self-built MFMDT dataset demonstrate that Flow2MDT outperforms existing methods in constructing MDTs. Ablation studies further validate the contributions of its key components.

Collaboration between multiple Large Language Models (LLMs) has attracted significant attention for its potential to mitigate hallucinations and enhance reasoning capabilities. Previous approaches, such as multi-agent debate and decoding-time integration, either rely on highly capable models with strong self-reflection abilities or are limited to models sharing the same tokenizer. To address these limitations, we introduce PToco (Prefix-based Token-level Collaboration), a novel mechanism that enables effective collaboration among less capable LLMs, independent of tokenizer differences. PToco uses a prefix-grouping method to extract consensus among tokens with varying levels of granularity, ensuring coherent and robust token generation across multiple models. Experimental results on a series of reasoning tasks demonstrate that PToco significantly improves performance over individual models. Furthermore, this approach generalizes well across different quantities and sizes of participating models, providing a more flexible and efficient solution for multi-LLM ensembles.

Biomedical Terminology Normalization aims to identify the standard term in a specified termbase for non-standardized mentions from social media or clinical texts, employing the mainstream “Recall and Re-rank” framework. Instead of the traditional pretraining-finetuning paradigm, we would like to explore the possibility of accomplishing this task through a tuning-free paradigm using powerful Large Language Models (LLMs), hoping to address the costs of re-training due to discrepancies of both standard termbases and annotation protocols. Another major obstacle in this task is that both mentions and terms are short texts. Short texts contain an insufficient amount of information that can introduce ambiguity, especially in a biomedical context. Therefore, besides using the advanced embedding model, we implement a Retrieval-Augmented Generation (RAG) based knowledge card generation module. This module introduces an LLM agent that expands the short texts into accurate, harmonized, and more informative descriptions using a search engine and a domain knowledge base. Furthermore, we present an innovative tuning-free agent collaboration framework for the biomedical terminology normalization task in social media. By leveraging the internal knowledge and the reasoning capabilities of LLM, our framework conducts more sophisticated recall, ranking and re-ranking processes with the collaboration of different LLM agents. Experimental results across multiple datasets indicate that our approach exhibits competitive performance. We release our code and data on the github repository JOHNNY-fans/RankNorm.

Few-shot table-to-text generation seeks to generate natural language descriptions for the given table in low-resource scenarios. Previous works mostly utilized Pre-trained Language Models (PLMs) even Large Language Models (LLMs) to generate fluent descriptions of the tables. However, they are prone to hallucinations that do not conform to the table. In this work, we propose CoT-Planner, a simple but efficient Chain-of-Thoughts-based approach that can be used to reduce the generation of hallucinations in the few-shot table-to-text generation. We first use a LLM (such as ChatGPT) to automatically generate ten intermediate content plans in the form of a Chain-of-Thoughts (CoT) for each table and corresponding description pair. Then, we refined the most accurate content plan for each sample and used the table and text pairs with the added content plan (CoT-Plan) as demonstrations for In-Context Learning (ICL). Both automatic and human evaluations on the numericNLG dataset show our method can effectively alleviate hallucinations, thereby improving factual consistency in few-shot table-to-text generation. The code and data can be accessed from https://github.com/FXLP/CoT-Planner.

Multimodal entity linking plays a crucial role in a wide range of applications. Recent advances in large language model-based methods have become the dominant paradigm for this task, effectively leveraging both textual and visual modalities to enhance performance. Despite their success, these methods still face two challenges, including unnecessary incorporation of image data in certain scenarios and the reliance only on a one-time extraction of visual features, which can undermine their effectiveness and accuracy. To address these challenges, we propose a novel LLM-based framework for the multimodal entity linking task, called Intra- and Inter-modal Collaborative Reflections. This framework prioritizes leveraging text information to address the task. When text alone is insufficient to link the correct entity through intra- and inter-modality evaluations, it employs a multi-round iterative strategy that integrates key visual clues from various aspects of the image to support reasoning and enhance matching accuracy. Extensive experiments on three widely used public datasets demonstrate that our framework consistently outperforms current state-of-the-art methods in the task, achieving improvements of 3.2%, 5.1%, and 1.6%, respectively. Our code is available at https://github.com/ziyan-xiaoyu/I2CR/.

Decision trees, as a structured representation of medical knowledge, are critical resources for building clinical decision support systems. Their structured decision pathways can be used for retrieval to enhance clinical decision making. Currently, mainstream methods mainly utilize large language models and in-context learning for decision tree extraction. However, these methods often face challenges in understanding the structure of decision trees and accurately extracting the complete content of tree nodes, leading to noise in the extracted trees and ultimately impacting their effectiveness in clinical decision support system. To this end, in this paper, we propose a novel decision tree extraction framework, including two stages. In the first stage, we propose to use the If-Else pseudocode to represent the decision tree structure and design specific constraints on format and content to guide the LLM in generating outputs. In the second stage, we introduce a novel node-filling strategy called PlanSelect to match the extracted triplets with sub-sentences in the generated pseudocode, including four reasoning steps: observation, plan, action, and answer. To evaluate the effectiveness of our proposed method, we construct an English decision tree extraction dataset (EMDT) and conduct extensive experiments on the built and public datasets. Experiments on the Text2DT and EMDT datasets demonstrate that our method outperforms the current state-of-the-art approaches, achieving improvements of 1.37% and 1.54% on the $ER$ metric (which is lower is better), respectively. Furthermore, we use the medical decision trees extracted using our framework to improve the model's performance on clinical decision making tasks, i.e., CMB-Clin and MedQA.

Cross-lingual terminology alignment is an important task in the field of medical terminology. Through cross-lingual alignment, medical terms from different languages can be accurately mapped to their corresponding concepts, thus establishing a unified and multilingual medical terminology fusion system. However, the scarcity of annotated parallel corpora for medical terminology in Chinese and English languages poses a challenge during the training process of Chinese-English terminology alignment models, resulting in decreased alignment accuracy. To address this issue, this paper proposes a hybrid approach that combines Large Language Models(LLMs) and pretrained Language Models(PLMs), leveraging the rich multilingual knowledge embedded in LLMs to obtain more comprehensive Chinese-English terminology information for assisting cross-lingual terminology alignment. However, LLMs require significant computational resources and memory, making them less suitable for large-scale alignment tasks. To tackle this problem, a confidence sampling strategy is introduced, delegating challenging samples to the LLMs for re-ranking, thereby reducing resource costs. Additionally, a prompt strategy tailored for terminology alignment tasks is proposed to enhance the accuracy of predictions made by the LLMs.We evaluate our method on mapping files of two medical open terminologies, and the experimental results demonstrate that our method outperforms baseline methods by 2% in terms of Hits@1 and Hits@10 metrics.Our code and data are available at https://github.com/Bruce-Y12/two-stage-ranking.

Data-to-text Generation (D2T) aims to generate textual natural language statements that can fluently and precisely describe the structured data such as graphs, tables, and meaning representations (MRs) in the form of key-value pairs. It is a typical and crucial task in natural language generation (NLG). Early D2T systems generated texts with the cost of human engineering in designing domain specific rules and templates, and achieved acceptable performance in coherence, fluency, and fidelity. In recent years, the data-driven D2T systems based on deep learning have reached state-of-the-art (SOTA) performance in more challenging datasets. In this paper, we provide a comprehensive review on existing neural data-to-text generation approaches. We first introduce available D2T resources, including systematically categorized D2T datasets and mainstream evaluation metrics. Next, we survey existing works based on the taxonomy along two axes: neural end-to-end D2T and neural modular D2T. We also discuss the potential applications and the adverse impacts. Finally, we present readers with the challenges faced by neural D2T and outline some potential future directions in this area.

Relation extraction is an important task in the field of natural language processing. Previous works mainly focus on adopting pipeline methods or joint methods to model relation extraction in general scenarios. However, these existing methods face challenges when adapting to complex relation extraction scenarios, such as handling overlapped triplets, multiple triplets, and cross-sentence triplets. In this paper, we revisit the advantages and disadvantages of the aforementioned methods in complex relation extraction. Based on the in-depth analysis, we propose a novel two-stage bidirectional extract-then-evaluate framework namedBeeRe. In the extraction stage, we first obtain the subject set, relation set, and object set. Then, we design subject- and object-oriented triplet extractors to iteratively recurrent obtain candidate triplets, ensuring high recall. In the evaluation stage, we adopt a relation-oriented triplet filter to determine subject-object pairs based on relations in triplets obtained in the first stage, ensuring high precision. We conduct extensive experiments on three public datasets to show thatBeeReachieves state-of-the-art performance in both complex and general relation extraction scenarios. Even when compared to large language models like closed-source/open-source LLMs,BeeRestill has significant performance gains.

Continual pretraining is a popular way of building a domain-specific pretrained language model from a general-domain language model. In spite of its high efficiency, continual pretraining suffers from catastrophic forgetting, which may harm the model’s performance in downstream tasks. To alleviate the issue, in this paper, we propose a continual pretraining method for the BERT-based model, named Attention-FFN Adapter. Its main idea is to introduce a small number of attention heads and hidden units inside each self-attention layer and feed-forward network. Furthermore, we train a domain-specific language model named AF Adapter based RoBERTa for the Chinese biomedical domain. In experiments, models are applied to downstream tasks for evaluation. The results demonstrate that with only about 17% of model parameters trained, AF Adapter achieves 0.6%, 2% gain in performance on average, compared to strong baselines. Further experimental results show that our method alleviates the catastrophic forgetting problem by 11% compared to the fine-tuning method. Code is available at https://github.com/yanyongyu/AF-Adapter.

Machine reading comprehension (MRC) is an essential task for many question-answering applications. However, existing MRC datasets mainly focus on data with single answer and overlook multiple answers, which are common in the real world. In this paper, we aim to construct an MRC dataset with both data of single answer and multiple answers. To achieve this purpose, we design a novel pipeline method: data collection, data cleaning, question generation and test set annotation. Based on these procedures, we construct a high-quality multi-answer MRC dataset (MA-MRC) with 129K question-answer-context samples. We implement a sequence of baselines and carry out extensive experiments on MA-MRC. According to the experimental results, MA-MRC is a challenging dataset, which can facilitate the future research on the multi-answer MRC task.

MOTIVATION: In the medical field, multiple terminology bases coexist across different institutions and contexts, often resulting in the presence of redundant terms. The identification of overlapping terms among these bases holds significant potential for harmonizing multiple standards and establishing unified framework, which enhances user access to comprehensive and well-structured medical information. However, the majority of terminology bases exhibit differences not only in semantic aspects but also in the hierarchy of their classification systems. The conventional approaches that rely on neighborhood-based methods such as GCN may introduce errors due to the presence of different superordinate and subordinate terms. Therefore, it is imperative to explore novel methods to tackle this structural challenge. RESULTS: To address this heterogeneity issue, this paper proposes a multi-view alignment approach that incorporates the hierarchical structure of terminologies. We utilize BERT-based model to capture the recursive relationships among different levels of hierarchy and consider the interaction information of name, neighbors, and hierarchy between different terminologies. We test our method on mapping files of three medical open terminologies, and the experimental results demonstrate that our method outperforms baseline methods in terms of Hits@1 and Hits@10 metrics by 2%. AVAILABILITY AND IMPLEMENTATION: The source code will be available at https://github.com/Ulricab/Bert-Path upon publication.

MOTIVATION: Biomedical relation extraction is a vital task for electronic health record mining and biomedical knowledge base construction. Previous work often adopts pipeline methods or joint methods to extract subject, relation, and object while ignoring the interaction of subject-object entity pair and relation within the triplet structure. However, we observe that entity pair and relation within a triplet are highly related, which motivates us to build a framework to extract triplets that can capture the rich interactions among the elements in a triplet. RESULTS: We propose a novel co-adaptive biomedical relation extraction framework based on a duality-aware mechanism. This framework is designed as a bidirectional extraction structure that fully takes interdependence into account in the duality-aware extraction process of subject-object entity pair and relation. Based on the framework, we design a co-adaptive training strategy and a co-adaptive tuning algorithm as collaborative optimization methods between modules to promote better mining framework performance gain. The experiments on two public datasets show that our method achieves the best F1 among all state-of-the-art baselines and provides strong performance gain on complex scenarios of various overlapping patterns, multiple triplets, and cross-sentence triplets. AVAILABILITY AND IMPLEMENTATION: Code is available at https://github.com/11101028/CADA-BioRE.

Most real-world applications in EHRs involve temporal data with skewed distributions. The imbalanced classification problem becomes more difficult in sporadic temporal data that variables exist on correlation and have some missing values. A common solution to classification tasks with imbalanced data is the oversampling methods, which generate new samples to re-balancing the classes. However, traditional oversampling methods usually change the distribution, thereby leading to bias. This paper proposed a self-adaptive integrated oversampling method for imbalanced sporadic temporal data in EHRs. The masking vectors and density vectors have been introduced to measure missing value distribution of samples, and the minority samples are divided into high density samples and sparse density samples. We extend the resampling strategies combining a subsample alignment method and structure preserving oversampling method. The weight of sample difference is used to improve classification performance. Furthermore, the filter mechanism is proposed to remove the noise samples with good efficiency. The experimental results show that the proposed method increases performance compared to traditional resampling methods in terms of AUC, F1, and G-mean evaluation metrics.

MOTIVATION: Medical terminology normalization aims to map the clinical mention to terminologies coming from a knowledge base, which plays an important role in analyzing electronic health record and many downstream tasks. In this article, we focus on Chinese procedure terminology normalization. The expressions of terminology are various and one medical mention may be linked to multiple terminologies. Existing studies based on learning to rank does not fully consider the quality of negative samples during model training and the importance of keywords in this domain-specific task. RESULTS: We propose a combined recall and rank framework to solve these problems. A pair-wise Bert model with deep metric learning is used to recall candidates. Previous methods either train Bert in a point-wise way or based on a multi-class classification problem, which may lead serious efficiency problems or not be effective enough. During model training, we design a novel online negative sampling algorithm to activate the pair-wise method. To deal with multi-implication scenarios, we train the task of implication number prediction together with the recall task in a multi-task learning setting, since these two tasks are highly complementary. In rank step, we propose a keywords attentive mechanism to focus on domain-specific information such as procedure sites and procedure types. Finally, a fusion block merges the results of the recall and the rank model. Detailed experimental analysis shows our proposed framework has a remarkable improvement on both performance and efficiency. AVAILABILITY AND IMPLEMENTATION: The source code will be available at https://github.com/sxthunder/CMTN upon publication.

Clinical text structuring is a critical and fundamental task for clinical research. Traditional methods such as task-specific end-to-end models and pipeline models usually suffer from the lack of dataset and error propagation. In this paper, we present a question answering based clinical text structuring (QA-CTS) task to unify different specific CTS tasks and make dataset shareable. A novel model that aims to introduce domain-specific features (e.g., clinical named entity information) into pre-trained language model is also proposed for QA-CTS task. Experimental results on Chinese pathology reports collected from Ruijing Hospital demonstrate our presented QA-CTS task is very effective to improve the performance on specific tasks. Our proposed model also competes favorably with strong baseline models in specific tasks.

Entity and relation extraction is the necessary step in structuring medical text. However, the feature extraction ability of the bidirectional long short term memory network in the existing model does not achieve the best effect. At the same time, the language model has achieved excellent results in more and more natural language processing tasks. In this paper, we present a focused attention model for the joint entity and relation extraction task. Our model integrates well-known BERT language model into joint learning through dynamic range attention mechanism, thus improving the feature representation ability of shared parameter layer. Experimental results on coronary angiography texts collected from Shuguang Hospital show that the F1-scores of named entity recognition and relation classification tasks reach 96.89% and 88.51%, which outperform state-of-the-art methods by 1.65% and 1.22%, respectively.

Image sentiment analysis has been studied for many years, and most of algorithms take the image sentiment as independent and discrete labels to predict by machine learning. Actually, as a product of multiple hormone combinations, emotions are generated by mutual suppression signal in brain. Inspired by neural microcircuit in amygdala, we propose a novel Multi-Subnet Neural Network (MSNN) that simulates the human brain mechanism for image sentiment classification. Different from traditional neural network, MSNN extends a new domain channel to imitate the way that images stimulate the brain through different neural circuits and produce sentimental semantic information by multi-subnet and signal reforming network. Experiments show that MSNN is well adapted to multi-class image sentiment classification task, and outperforms other multi-class sentiment classification models.

Clinical named entity recognition aims to identify and classify clinical terms such as diseases, symptoms, treatments, exams, and body parts in electronic health records, which is a fundamental and crucial task for clinical and translational research. In recent years, deep neural networks have achieved significant success in named entity recognition and many other natural language processing tasks. Most of these algorithms are trained end to end, and can automatically learn features from large scale labeled datasets. However, these data-driven methods typically lack the capability of processing rare or unseen entities. Previous statistical methods and feature engineering practice have demonstrated that human knowledge can provide valuable information for handling rare and unseen cases. In this paper, we propose a new model which combines data-driven deep learning approaches and knowledge-driven dictionary approaches. Specifically, we incorporate dictionaries into deep neural networks. In addition, two different architectures that extend the bi-directional long short-term memory neural network and five different feature representation schemes are also proposed to handle the task. Computational results on the CCKS-2017 Task 2 benchmark dataset show that the proposed method achieves the highly competitive performance compared with the state-of-the-art deep learning methods.

In this article, the authors design two metric sets considering Richness and Correctness based on a quasi-formal conceptual representation. They also design a novel metric set on overlapped instances of different KBs to make the metric results comparable. Finally, they use random sampling techniques to reduce human efforts for assessing the correctness. The authors evaluate three large Chinese KBs including DBpedia Chinese, Zhishi.me and SSCO comparatively, and further compare them with English KBs in terms of data set qualities. They also compare different versions of DBpedia and YAGO. The findings in these KBs not only give a detailed report of the current situation of extracted KBs, but also show the effectiveness of their methods in assessing the quality of Web-Scale KBs comparatively.

While a large number of well-known knowledge bases (KBs) in life science have been published as Linked Open Data, there are few KBs in Chinese. However, KBs of life science in Chinese are necessary when we want to automatically process and analyze electronic medical records (EMRs) in Chinese. Of all, the symptom KB in Chinese is the most seriously in need, since symptoms are the starting point of clinical diagnosis. Furthermore, expressions used in describing symptoms in clinical practice are diverse, which makes it hard to collect such a KB. In this paper, we publish a public KB of symptoms in Chinese. The KB is constructed by fusing data automatically extracted from eight mainstream healthcare websites, three Chinese encyclopedia sites, and symptoms extracted from a large number of EMRs as supplements. As a result, the KB has more than 26,000 distinct symptoms in Chinese including 3,968 symptoms in traditional Chinese medicine (TCM) and 1,029 synonym pairs for symptoms. The KB also includes concepts such as diseases and medicines as well as relations between symptoms and the above related entities. We also link our KB to the Unified Medical Language System (UMLS) and analyze the differences between symptoms in the two KBs. We released the KB as Linked Open Data and a demo at https://datahub.io/dataset/symptoms-in-chinese.