检索结果-内蒙古大学图书馆

A Combined Usage of NLP Libraries Towards Analyzing Software Documents

international JOURNAL OF SOFTWARE engineering AND knowledge engineering 2023年第9期33卷 1387-1404页

作者： Kong, Xianglong Zhuo, Hangyi Gu, Zhechun Cheng, Xinyun Zhang, Fan Purple Mt Labs Nanjing Peoples R China Southeast Univ Sch Comp Sci & Engn Nanjing Peoples R China Wuxi Taihu Univ Scotland Acad Wuxi Peoples R China Natl Digital Switching Syst & Engn Technol Res Ctr Zhengzhou Peoples R China

Software documents are commonly processed by natural language processing (NLP) libraries to extract information. The libraries provide similar functional APIs to achieve NLP tasks, numerous toolkits result in a problem of selection. In this work, we propose a method to combine the strengths of different NLP libraries to avoid the subjective selection of a specific NLP library. The combined usage is conducted through two steps, i.e. document-level selection of primary NLP library and sentence-level overwriting. The primary NLP library is determined according to the overlap degree of the results. The highest overlap degree indicated the most effective NLP library on a specific NLP task. Through sentence-level overwriting, the possible fine-gained improvements from other libraries are extracted to overwrite the outputs of primary library. We evaluate the combined method with six widely used NLP libraries and 200 documents from three different sources. The results show that the combined method can generally outperform all the studied NLP libraries in terms of accuracy. The finding means that our combined method can be used instead of individual NLP library for more effective results.

关键词： natural language processing NLP library selection software document

来源：评论

学校读者我要写书评

暂无评论

CyGPT: knowledge Graph-Based Enhancement Techniques for Large language Models in Cybersecurity 9

CyGPT: Knowledge Graph-Based Enhancement Techniques for Larg...

引用

9th IEEE international conference on Data Science in Cyberspace, DSC 2024

作者： Ou, Lu Ni, Xiaoya Wu, Wei Tian, Zhihong Guangzhou University Cyberspace Institute of Advanced Technology Guangzhou510006 China

ISBN: (纸本)9798350391367

Large language Models (LLMs) excel in numerous natural language processing (NLP) tasks but encounter significant challenges in practical applications, including hallucinations, outdated information, and a lack of domain-specific external knowledge. This study proposes a collaborative, training-free reasoning approach, leveraging close cooperation between knowledge Graphs (KG) and LLMs for cybersecurity applications. Our approach employs the 'Joint Reasoning Chain,' which dynamically integrates information from network security-specific knowledge graphs, serving as an external knowledge base to enhance the domain-specific external knowledge of LLMs. This cooperative method not only improves reliable knowledge-based reasoning but also enhances the traceability of decision-making processes. Furthermore, we introduce a novel GPT-based technique to evaluate answer quality and have performed systematic experiments on a purpose-built test set. The results confirm that our method significantly boosts GPT's performance in network security knowledge, demonstrating the potential of knowledge graphs to augment LLMs' reasoning abilities and their applicability in specialized fields. © 2024 IEEE.

关键词： knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Do Large language Models Mirror Cognitive language processing? 31

Do Large Language Models Mirror Cognitive Language Processin...

引用

31st international conference on Computational Linguistics, COLING 2025

作者： Ren, Yuqi Jin, Renren Zhang, Tongxuan Xiong, Deyi College of Intelligence and Computing Tianjin University China College of Computer and Information Engineering Tianjin Normal University China

ISBN: (纸本)9798891761964

Large language Models (LLMs) have demonstrated remarkable abilities in text comprehension and logical reasoning, indicating that the text representations learned by LLMs can facilitate their language processing capabilities. In neuroscience, brain cognitive processing signals are typically utilized to study human language processing. Therefore, it is natural to ask how well the text embeddings from LLMs align with the brain cognitive processing signals, and how training strategies affect the LLM-brain alignment? In this paper, we employ Representational Similarity Analysis (RSA) to measure the alignment between 23 mainstream LLMs and fMRI signals of the brain to evaluate how effectively LLMs simulate cognitive language processing. We empirically investigate the impact of various factors (e.g., pre-training data size, model scaling, alignment training, and prompts) on such LLM-brain alignment. Experimental results indicate that pre-training data size and model scaling are positively correlated with LLM-brain similarity, and alignment training can significantly improve LLM-brain similarity. Explicit prompts contribute to the consistency of LLMs with brain cognitive language processing, while nonsensical noisy prompts may attenuate such alignment. Additionally, the performance of a wide range of LLM evaluations (e.g., MMLU, Chatbot Arena) is highly correlated with the LLM-brain similarity. © 2025 Association for Computational Linguistics.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prompt Fix: Vulnerability Automatic Repair Technology Based on Prompt engineering

Prompt Fix: Vulnerability Automatic Repair Technology Based ...

引用

international conference on Computing, Networking and Communications (ICNC)

作者： Liu, Peng Wang, He Zheng, Chen Zhang, Yuqing Xidian Univ Guangzhou Inst Technol Guangzhou Peoples R China Chinese Acad Sci Inst Software Beijing Peoples R China Univ Chinese Acad Sci Nanjing Peoples R China Xidian Univ Xian Peoples R China Hainan Univ Haikou Hainan Peoples R China

ISBN: (纸本)9798350371000;9798350370997

With the emergence of large-scale language models (LLM), the powerful capabilities of LLM in natural language processing have attracted attention. Based on programming language LLM (Programming language Model, PLM), we use prompt templates to explore its potential in the field of automatic vulnerability repair, and combine it with a special workflow to improve its efficiency in automatic vulnerability repair tasks. Specifically, we design four prompt templates for handling vulnerable code, and design an iterative reasoning method to improve the efficiency of vulnerability fixing. We selected multiple typical LLMs for evaluation on multiple data sets. The results show that reasonable prompt templates can effectively improve the efficiency of automatic vulnerability repair, which is significantly improved compared with neural machine translation technology. In addition, we also discussed previous bug fixing related work and our work, and pointed out some of our shortcomings and directions for future improvements.

关键词： automatic program repair vulnerability repair large language model program language model

来源：评论

学校读者我要写书评

暂无评论

The Evolution of Large language Model: Models, Applications and Challenges 12

The Evolution of Large Language Model: Models, Applications ...

引用

12th international conference on Current Trends in Advanced Computing, ICCTAC 2024

作者： Sindhu, B. Prathamesh, R.P. Sameera, M.B. KumaraSwamy, S. Global Academy of Technology Department of Computer Science and Engineering Karnataka Bangalore India

ISBN: (数字)9798350395808

ISBN: (纸本)9798350395808

Large language Models (LLMs) have attracted a lot of attention due to their success in natural language processing tasks. This paper provides a thorough overview by examining the architecture, applications, problems, assessment techniques, and future directions of LLM. With the constantly growing body of literature, a succinct yet comprehensive overview of recent developments is essential. Following the development of NLP, it highlights the move from rule-based systems to sophisticated transformer structures like as BERT and GPT. Important LLMs for text creation, translation, and summarization are mentioned, including T5, BART, and BioGPT. LLM performance is evaluated using metrics including as accuracy, perplexity, BLEU score, and ROUGE score. Research is still being done because of issues with bias, overfitting, and real-Time processing. Future directions include managing longer contexts, lowering bias, and increasing efficiency through methods like federated learning. Continuous learning and multimodal LLMs are promising fields, as well as interpretive AI. In conclusion, LLMs have transformed natural language processing (NLP) and brought up both technical and ethical issues about the future of AI. © 2024 IEEE.

关键词： natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

The Development of Individualized Assignment Generator 17

The Development of Individualized Assignment Generator

引用

17th international conference on Developments in eSystems engineering, DeSE 2024

作者： Zaripova, Rinata R. Danilov, Andrew V. Salekhova, Leila L. Fazliakhmetov, Timur R. The Dept. of Bilingual and Digital Education Kazan Federal University Kazan Russia

ISBN: (纸本)9798350368697

The article discusses the development of a system that uses artificial intelligence (AI) to generate individualized mathematics assignments for bilingual students in Tatarstan, Russia. The goal is to enhance learning by tailoring assignments to students’ linguistic preferences, cognitive styles, and knowledge levels. The system employs machine learning techniques and GPT-based models to create personalized tasks that align with curriculum goals while addressing linguistic diversity, particularly for Tatar-Russian bilinguals. The study evaluates several large language models (LLMs), including GPT-4, GPT-3.5 Turbo, YandexGPT, and GigaChat, based on their ability to generate math problems and content in the Tatar language. While GPT-4 and GPT-3.5 Turbo show superior performance in producing accurate and semantically correct problems, their proficiency in Tatar remains inconsistent. The research underscores the need for further development of LLMs to enhance content generation for bilingual educational contexts and highlights the potential of AI in advancing adaptive learning for mathematics education. Future directions include expanding the system’s functionality and testing its effectiveness across diverse educational settings. © 2024 IEEE.

关键词： Adaptive Learning Bilingual Education Large language Models natural language processing Tatar-Russian Bilingualism

来源：评论

学校读者我要写书评

暂无评论

引用

2nd international conference on Emerging Trends in Information Technology and engineering, ic-ETITE 2024

作者： Bali, Aadeesh Bhagwat, Aniket Bhise, Aditya Joshi, Sarang SCTR's Pune Institute of Computer Technology Savitribai Phule Pune University Department of Computer Engineering Pune India

ISBN: (纸本)9798350328202

This paper delves into the critical task of measuring semantic similarity in text documents, a fundamental need in today's data-rich landscape. Efficiently gauging semantic con-nections is vital for applications such as information retrieval, content recommendation, and document clustering. Traditional methods, centered on lexical and syntactic features, often fall short, especially when dealing with synonyms, abbreviations, or diverse language use. Consequently, there's a growing demand for advanced techniques that can better capture the underlying semantic relationships in text documents. The study introduces a comprehensive approach to semantic similarity analysis for text documents. This approach involves the development and evaluation of innovative methodologies that utilize advanced natural language processing and machine learning techniques. It explores the integration of domain-specific knowledge bases, advanced feature representations, and semantic embeddings to enhance the accuracy of similarity detection. Additionally, adaptable synonym repositories are investigated to align with evolving language trends. These efforts collectively aim to advance the field of semantic similarity detection, ultimately improving the precision and efficiency of document analysis across a broad spectrum of text-based applications. © 2024 IEEE.

关键词： Information retrieval

来源：评论

学校读者我要写书评

暂无评论

A Survey on Machine and Deep Learning Approaches in Sign language Recognition: Techniques and Future Trends

A Survey on Machine and Deep Learning Approaches in Sign Lan...

引用

2024 international conference on Signal processing, Computation, Electronics, Power and Telecommunication, IConSCEPT 2024

作者： Harshitha, C. Sendil Vadivu, D. Rajagopalan, Narendran National Institute of Technology Puducherry Dept. of Computer Science and Engineering Karaikal India

ISBN: (纸本)9798331540685

Sign language is an essential communication medium for individuals with hearing impairments. It enables them to convey messages, disseminate knowledge, and transfer ideas within the deaf community. However, not everyone understands sign language, making it challenging for individuals with hearing impairments to communicate effectively with other common people. Sign language Recognition (SLR) is vital for bridging the communication gap between persons with hearing loss and non-sign language speakers. This survey paper aims to explore Machine Learning (ML) and Deep Learning (DL) techniques for recognition of sign languages. Investigate further research scope in this field. © 2024 IEEE.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

AQLoRA: An Adaptive Quantization-Based Efficient Fine-Tuning Method for LLMs 13th

AQLoRA: An Adaptive Quantization-Based Efficient Fine-Tuning...

引用

13th international conference on natural language processing and Chinese Computing

作者： Huang, Xingchen Hu, Yujia Wong, Derek F. Wang, Yao Cal, Liqiong Jiang, Yonghong Guizhou Minzu Univ Sch Data Sci & Informat Engn Guiyang Peoples R China Univ Macau NLP2CT Lab Macau Peoples R China Univ Macau Dept Comp & Informat Sci Macau Peoples R China Guizhou SiSo Elect Co LTD Guiyang Peoples R China

ISBN: (纸本)9789819794331;9789819794348

Large language models (LLMs) have shown exceptional performance in the domain of composite artificial intelligence tasks, offering a preliminary insight into the potential of general artificial intelligence. The fine-tuning process for LLMs necessitates significant computational resources, often surpassing those available from standard consumer-grade GPUs. To this end, we introduce the Adaptive Quantization Low-Rank Adaptation fine-tuning (AQLoRA), a method that reduces memory demands during fine-tuning by utilizing quantization coupled with pruning techniques. This dual strategy not only reduces memory usage but also preserves accuracy. AQLoRA refines the original Low-Rank Adaptation fine-tuning (LoRA) method by efficiently quantizing LLMs weights, prioritizing computational resource allocation based on weight importance, and effectively integrating the quantized model with auxiliary weights post fine-tuning. Applying AQLoRA to the ChatGLM2-6B model, we demonstrate its effectiveness in both natural language generation (NLG) and natural language understanding (NLU) across diverse fine-tuning datasets and scenarios. Our findings reveal that AQLoRA achieves balance between performance and memory efficiency, reducing memory consumption by 25% in NLG tasks. For NLU tasks, it enhances performance by 10% and reduces memory consumption by 10% compared to state-of-the-art methods.

关键词： Large language Model Fine-tuning LoRA

来源：评论

学校读者我要写书评

暂无评论

Utilizing Process Models in the Requirements engineering Process Through Model2Text Transformation 32

Utilizing Process Models in the Requirements Engineering Pro...

引用

32nd IEEE international Requirements engineering conference (RE)

作者： Klievtsova, Nataliia Mangler, Juergen Kampik, Timotheus Rinderle, Stefanie Tech Univ Munich Tum Sch Computat Informat & Technol Garching Germany SAP Signavio Berlin Germany

ISBN: (纸本)9798350395129;9798350395112

With the advent of large language models (LLMs), requirements engineers have gained a powerful natural language processing tool to analyze, query, and validate a wide variety of textual artifacts, thus potentially supporting the whole requirements engineering process from requirements elicitation to management. However, the input for the requirements engineering process often encompasses a variety of potential information sources in various formats, especially graphical models such as process models. Hence, this work aims to contribute to the state of the art by assessing the feasibility of utilizing graphical process models and their textual representations in the requirements engineering process. In particular, we focus on the extraction of textual process descriptions from process models as i) input for the requirements engineering process and ii) documentation as the result of process-oriented requirements engineering. To this end, we explore, quantify, and compare traditional deterministic and LLM-based extraction methods where the latter includes GPT3, GPT3.5, GPT4, and LLAMA. The evaluation assesses output quality and information loss based on one data set. The results indicate that LLMs produce human-like process descriptions based on the predefined patterns, but apparently lack true comprehension of the process models.

关键词： AI4RE Process Models Process Descriptions Large language Models

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：