Large Language Models (LLMs) hold significant potential in the medical domain but are hindered by the “hallucination” phenomenon, where generated content may be plausible but factually incorrect. Retrieval-Augmented Generation has emerged as an effective paradigm to mitigate this by integrating external knowledge, and gradually migrates to storing and retrieving relevant information within graph structure. Currently, most Graph RAG methods in medical domain face critical limitations such as semantic poverty of the graph and the fragility of the retrieval process. We propose IRG-RAG, an Iterative Reflective Graph Retrieval-Augmented Generation framework tailored for Chinese medical question-answering, focuses on addressing these two key challenges. Our framework introduces a knowledge-enhanced graph construction method, enriching edges with contextual descriptions to endow the graph with richer semantic information. Additionally, we propose an iterative reflective retrieval strategy, which dynamically travels on the graph to retrieve relevant information, backtracks from ineffective paths to reduce noise, and incorporates negative feedback to guide subsequent explorations, mimicking humanlike problem-solving. Experiments on CMExam and CMB-Exam datasets demonstrate that our method significantly enhances the reliability of the model's answer, offering a robust solution to hallucination in domain-specific applications. Codes and data will be available at https://github.com/Damon1Zhang/IRG-RAG.

Denoising is the essential step for distant supervision based named entity recognition. Previous denoising methods are mostly based on instance-level confidence statistics, which ignore the variety of the underlying noise distribution on different datasets and entity types. This makes them difficult to be adapted to high noise rate settings. In this paper, we propose Hypergeometric Learning (HGL), a denoising algorithm for distantly supervised NER that takes both noise distribution and instance-level confidence into consideration. Specifically, during neural network training, we naturally model the noise samples in each batch following a hypergeometric distribution parameterized by the noise-rate. Then each instance in the batch is regarded as either correct or noisy one according to its label confidence derived from previous training step, as well as the noise distribution in this sampled batch. Experiments show that HGL can effectively denoise the weakly-labeled data retrieved from distant supervision, and therefore results in significant improvements on the trained models.

Map Generalization is one of the most fundamental technologies for modern digital maps. It can effectively reduce the storage space and fit to different applications according to their scale requirement. This paper presents an efficient solution for this problem that won the ACM SIGSPATTAL CUP 2014. Given the original geometries which are represented by sampling points sequence, this method divides the boundaries into many small segments based on their topological characteristics and constriants. It attempts to minimize the number of sampling points by simplifying the given map and constraining points. In addition, the method also employs many optimization techniques to reduce the total latency, like memory pool, parallel computing and string parsing. Experimental results on real datasets demonstrate the effectiveness and efficiency of the proposed method.

A handwritten Tibetan database, MRG-OHTC, is presented to facilitate the research of online handwritten Tibetan character recognition. The database contains 910 Tibetan character classes written by 130 persons from Tibetan ethnic minority. These characters are selected from basic set and extension set A of Tibetan coded character set. The current version of this database is collected using electronic pen on digital tablet. We investigate some characteristic of writing style from different writers. We evaluate MRG-OHTC database using existing algorithms as a baseline. Experimental results reveal a big challenge to higher recognition performance. To our knowledge, MRG-OHTC is the first publicly available database for handwritten Tibetan research. It provides a basic database to compare empirically different algorithms for handwritten Tibetan character recognition.