Extractive summarization aims to produce a concise version of a document by extracting information-rich sentences from the original texts. The graph-based model is an effective and efficient approach to rank sentences since it is simple and easy to use. However, its performance depends heavily on good text representation. In this paper, an integrated graph model (iGraph) for extractive text summarization is proposed. An enhanced embedding model is used to detect the inherent semantic properties at the word level, bigram level and trigram level. Words with part-of-speech (POS) tags, bigrams and trigrams were extracted to train the embedding models. Based on the enhanced embedding vectors, the similarity values between the sentences were calculated in three perspectives. The sentences in the document were treated as vertexes and the similarity between them as edges. As a result, three different types of semantic graphs were obtained for every document, with the same nodes and different edges. These three graphs were integrated into one enriched semantic graph in a naive Bayesian fashion. After that, TextRank, which is a graph-based ranking algorithm, was applied to rank the sentences, before the top scored sentences were selected for the summary according to the compression rate. Evaluated on the DUC 2002 and DUC 2004 datasets, our proposed method shows competitive performance compared to the state-of-the-art methods.
3 March 2019 in Concurrency and Computation: Practice and Experience
The recognition of Application Programming Interface (API) mentions in the software-related texts is a prerequisite task for extracting API-related knowledge. Previous studies have demonstrated the superiority of deep learning-based methods in accomplishing this task. However, such techniques still meet their bottlenecks due to their inability to effectively handle the following three challenges: (1) differentiating APIs from common words; (2) identifying APIs in morphological variants of the standard APIs; and (3) the lack of high-quality labeled data for training. To overcome these challenges, this paper proposes a context-aware API recognition method named CAREER. This approach utilizes two key components, namely Bidirectional Encoder Representations from Transformers (BERT) and Bi-directional Long Short-Term Memory (BiLSTM), to extract context information at both the word-level and sequence-level. This strategic combination empowers the method to dynamically capture both syntactic and semantic information, effectively addressing the first challenge. To tackle the second challenge, CAREER introduces a character-level BiLSTM component, enriched with an attention mechanism. This enables the model to grasp character-level global context information, thereby enhancing the recognition of morphological attributes within API mentions. Furthermore, to address the third challenge, the paper introduces three data augmentation techniques aimed at generating new data samples. Accompanying these techniques is a novel sample selection algorithm designed to screen out high-quality instances. This dual-pronged approach effectively mitigates the requirement for data labeling. Experiments demonstrate that CAREER significantly improves F1-score by 11.0% compared with state-of-the-art methods. We also construct specific datasets to assess CAREER's capacity to tackle the aforementioned challenges. Results confirm that (1) CAREER significantly outperforms baseline methods in addressing the first and second challenges, and (2) with the aid of data augmentation techniques and sample selection algorithms, high-quality samples can be generated to improve the performance, and alleviate the third challenge.
In view of the fact that Dempster-Shafer (D-S) evidence theory is unable to fuse data of multiple different kinds of sensors, an improved D-S evidence theory method based on the fusion of support and confidence entropy is proposed. Firstly, the identification framework of evidence theory is improved; secondly, Spearman correlation coefficient is introduced to represent the correlation between evidences; thirdly, a new confidence entropy is defined to describe the inconsistent uncertainty and non-specific uncertainty between evidences; then, the evidence set is modified by the combination of correlation and confidence entropy; finally, Dempster combination rule is used for information fusion. The simulation results confirm that the improved method of D-S evidence theory is feasible and more effective than the traditional algorithm.
26 September 2025 in IEEE Transactions on Software Engineering
Code generation has advanced with large language models (LLMs), but LLMs still struggle with complex tasks, especially in competitive programming. These tasks require understanding complex problems, generating correct code that passes numerous test cases, and meeting tight time and memory limits. We observed that there are some critical hints provided by competition platforms, which often point to the most critical information needed to solve the problem, thus guiding participants to accurate solutions. Inspired by these observations, we propose TEACH1, an approach that tutors LLMs in solving competitive programming by combining critical hints with a structured Chain of Thought (CoT). The key insight of TEACH is to employ a domain-specialized hint generator that is fine-tuned on curated data from competitive programming platforms, enabling it to produce concise and targeted algorithmic hints. By integrating these hints into the reasoning process of LLMs, TEACH helps LLMs bridge the gap between complex tasks and solutions. Furthermore, TEACH simulates human problem-solving through a structured CoT that covers problem understanding, analysis, algorithm selection, and coding. We extensively evaluate TEACH on both proprietary (GPT-3.5, GPT-4o, Claude-3.5-Sonnet, Gemini-2.5-Flash) and open-source (DeepSeek-V3) LLMs. TEACH achieves up to 6.56 absolute (17.4% relative) gain in pass@1 on LeetCode, and demonstrates strong generalization to APPS and ASAC, with maximum pass@1 relative improvements of 17.6% and 26.9%, respectively. Furthermore, existing CoT methods with the hints generated from TEACH yield additional gains, demonstrating its compatibility and extensibility across models and prompting strategies.