Abstract Model Checking for Web Services

Web-services are highly distributed programs, and concurrent software is notoriously error-prone. Model checking is a powerful technique to find bugs in concurrent systems. However, the existing model checkers have no enough ability to support for the programming languages and communication mechanisms used for Web services. We propose to use Kripke structures as means of modeling Web service. This paper presents an automated way to extract formal models from programs implementing Web services using predicate abstraction for abstract model checking. The abstract models are checked by means of a model checker that implements automatic abstraction refinement. These results enable the verification of the applications that implement Web services.

Abstract models for heat engines

We retrospect three abstract models for heat engines which include a classic abstract model in textbook of thermal physics,a primary abstract model for finite-time heat engines,and a refined abstract model for finite-time heat engines.The detailed models of heat engines in literature of finite-time thermodynamics may be mapped into the refined abstract model.The future developments based on the refined abstract model are also surveyed.

Masked Sentence Model Based on BERT for Move Recognition in Medical Scientific Abstracts

Purpose:Mo ve recognition in scientific abstracts is an NLP task of classifying sentences of the abstracts into different types of language units.To improve the performance of move recognition in scientific abstracts,a novel model of move recognition is proposed that outperforms the BERT-based method.Design/methodology/approach:Prevalent models based on BERT for sentence classification often classify sentences without considering the context of the sentences.In this paper,inspired by the BERT masked language model(MLM),we propose a novel model called the masked sentence model that integrates the content and contextual information of the sentences in move recognition.Experiments are conducted on the benchmark dataset PubMed 20K RCT in three steps.Then,we compare our model with HSLN-RNN,BERT-based and SciBERT using the same dataset.Findings:Compared with the BERT-based and SciBERT models,the F1 score of our model outperforms them by 4.96%and 4.34%,respectively,which shows the feasibility and effectiveness of the novel model and the result of our model comes closest to the state-of-theart results of HSLN-RNN at present.Research limitations:The sequential features of move labels are not considered,which might be one of the reasons why HSLN-RNN has better performance.Our model is restricted to dealing with biomedical English literature because we use a dataset from PubMed,which is a typical biomedical database,to fine-tune our model.Practical implications:The proposed model is better and simpler in identifying move structures in scientific abstracts and is worthy of text classification experiments for capturing contextual features of sentences.Originality/value:T he study proposes a masked sentence model based on BERT that considers the contextual features of the sentences in abstracts in a new way.The performance of this classification model is significantly improved by rebuilding the input layer without changing the structure of neural networks.

Influence of Software Modeling and Design on Domain-Specific Abstract Thinking: Student’s Perspective

Software engineering has been taught at many institutions as individual course for many years. Recently, many higher education institutions offer a BSc degree in Software Engineering. Software engineers are required, especially at the small enterprises, to play many roles, and sometimes simultaneously. Beside the technical and managerial skills, software engineers should have additional intellectual skills such as domain-specific abstract thinking. Therefore, software engineering curriculum should help the students to build and improve their skills to meet the labor market needs. This study aims to explore the perceptions of software engineering students on the influence of learning software modeling and design on their domain-specific abstract thinking. Also, we explore the role of the course project in improving their domain-specific abstract thinking. The study results have shown that, most of the surveyed students believe that learning and practicing modeling and design concepts contribute to their ability to think abstractly on specific domain. However, this finding is influenced by the students’ lack of the comprehension of some modeling and design aspects (e.g., generalization). We believe that, such aspects should be introduced to the students at early levels of software engineering curriculum, which certainly will improve their ability to think abstractly on specific domain.

Conversion of Object Oriented System into Software Product Line with Delta Modeling Abstract Behavioral Specification

This article contains a system conversion from object oriented design into Software Product Line (SPL) using delta modeling of Abstract Behavioral Specification (ABS). ABS is a modeling language which targets system with high level of variety and supports SPL development with delta modeling. The case study of this thesis is a digital library system called Library Automation and Digital Archive (LONTAR). Originally, LONTAR only uses SOAP-based web service. With ABS, LONTAR will be converted into SPL and implement another web service called REST. The motivation of this conversion of LONTAR from object oriented into SPL is because it is easier to develop system with ABS than using regular object oriented. Product definition in ABS is relatively easier than creating a new subclass and do customization to make it works well.

竞技运动专家的认知优势及其形成机制——基于自动性特点和抽象化表征

目前对于竞技运动专家如何从长期训练中获得和迁移、提炼和更新专业知识与技能的机制尚未有清晰的阐释。基于专家-新手范式的研究表明,从表征学习的角度看竞技运动专家的认知优势主要体现在注意优势和记忆优势两方面,而竞技运动专家知识与技能的自动性特点和抽象化表征是其认知优势效应形成的两个关键原因。从生成模型的角度探讨竞技运动专家知识系统的抽象化表征机制,为能更好地理解竞技运动专家的认知优势效应,助其突破认知局限性提供了新的理论依据。

数学基本思想在集合教学中的运用

为帮助学生全面深入地理解集合概念,对数学基本思想在集合教学中的运用进行了研究。数学基本思想一般包括抽象、推理与模型,教师从集合概念的抽象性、集合关系的抽象性出发,引导学生理解数学抽象思想;从集合推理运算的基础、集合恒等式的推理与证明出发,引导学生理解数学推理思想;从计数与概率论、数理逻辑中的集合模型出发,引导学生理解数学模型思想。

Review of Software Model-Checking Techniques for Dealing with Error Detection in Program Codes

Debugging software code has been a challenge for software developers since the early days of computer programming. A simple need, because the world is run by software. So perhaps the biggest engineering challenge is finding ways to make software more reliable. This review provides an overview of techniques developed over time in the field of software model checking to solve the problem of detecting errors in program code. In addition, the challenges posed by this technology are discussed and ways to mitigate them in future research and applications are proposed. A comprehensive examination of the various model verification methods used to detect program code errors is intended to lay the foundation for future research in this area.

Lacmia:抗混淆的多民族语言生成式摘要模型

该文提出了一种针对中国多民族低资源语言生成式摘要模型Lacmia(Language-Anti-confusioned Chinese Minority Abstractive Summarization Model)。为了克服以往模型只能处理单一语言的限制,Lacmia采用了一种统一的生成式架构来执行不同民族语言的摘要生成任务。此外,为了解决以往模型在多民族低资源语言处理上的性能不足问题,该模型在框架中加入了语言信息嵌入模块。该文通过在损失函数中引入目标语言偏好性正则化项,有效减轻了多语言摘要中出现的语言混淆现象,从而提升摘要生成准确性和流畅度。广泛的实验表明,Lacmia在包括藏语和维吾尔语在内的多民族低资源语言摘要任务中,取得了卓越成绩。除了在ROUGE评价标准上实现了显著性能提升外,Lacmia在该文新提出的CINOScore和NLCR两项指标上均达到了最佳效果,验证了模型的有效性和先进性。

外部注意力增强语义交互的阅读理解模型

针对传统抽取式阅读理解模型未充分考虑问答样本之间潜在相关性的问题,通过RoBERTa对问题与段落进行编码,利用外部注意力Exatt增强语义交互层特征获取能力,提出外部注意力增强语义交互的阅读理解模型,捕获问题与段落中蕴涵的语义特征和不同问答样本之间的潜在相关性。实验结果表明,在CMRC2018和构建的电力安规问答数据集上,在评价指标EM和F1两方面,该方法较基线模型分别最高提高了0.737%和2.556%。

基于交互引导的问答对联合生成模型

大规模问答对的自动生成在知识问答库构建和机器阅读理解等许多应用具有关键价值.尽管其重要性已得到广泛认可,现有问答对生成方法仍面临着严峻挑战.首先,在传统的问答对生成模型中,抽取式的答案获取方法难以适用于复杂的自然交互场景.相比较而言,生成式模型通过对文本的语义理解,能够自动生成表述更加自然的答案.其次,对于问答对生成任务来说,为了防止生成的答案和问题出现语义上的不匹配,需要更全面地捕捉并增强答案生成和问题生成两个子任务之间的交互.最后,由于答案抽取和问题生成存在任务难度的差异,这两个任务在联合训练的过程中会出现任务之间的优化不平衡问题.为此,本文提出了一个基于交互引导的问答对联合生成模型(Interaction-Guided Joint Abstractive QAPs Generation Model,IGJA-QAP).具体而言,本文设计了一个带有答案引导的多头门机制的联合生成模型,同时对两个子任务进行统一建模并有效地捕获和增强它们之间的信息交互,从而可以生成语义上匹配的问答对.本文在三个大规模数据集SQuAD、NewQA和CoQA上进行了综合全面的实验分析.本文提出的模型在答案生成任务上METEOR值平均分别超出其他最佳方法3.0%、5.9%和4.3%,问题生成任务上METEOR值平均分别超出其他最佳方法1.5%、0.5%和2.1%.实验结果表明,本文提出的模型达到了目前最高的性能.

基于误差分治的神经网络验证

随着神经网络技术的快速发展,其在自动驾驶、智能制造、医疗诊断等安全攸关领域得到了广泛应用,神经网络的可信保障变得至关重要.然而,由于神经网络具有脆弱性,轻微的扰动经常会导致错误的结果,因此采用形式化验证的手段来保障神经网络安全可信是非常重要的.目前神经网络的验证方法主要关注分析的精度,而易忽略运行效率.在验证一些复杂网络的安全性质时,较大规模的状态空间可能会导致验证方法不可行或者无法求解等问题.为了减少神经网络的状态空间,提高验证效率,提出一种基于过近似误差分治的神经网络形式化验证方法.该方法利用可达性分析技术计算非线性节点的上下界,并采用一种改进的符号线性松弛方法减少了非线性节点边界计算过程中的过近似误差.通过计算节点过近似误差的直接和间接影响,将节点的约束进行细化,从而将原始验证问题划分为一组子问题,其混合整数规划(MILP)公式具有较少的约束数量.所提方法已实现为工具NNVerifier,并通过实验在经典的3个数据集上训练的4个基于ReLU的全连接基准网络进行性质验证和评估.实验结果表明,NNVerifier的验证效率比现有的完备验证技术提高了37.18%.

基于过滤注意力机制的生成式自动社交媒体文本摘要模型

[目的/意义]为了帮助用户在浩如烟海的社交媒体文本中快速获取所需信息,本研究创新地设计了一种基于过滤注意力机制的自动摘要生成模型Filter Unit Model(FUM)。[方法/过程]首先,采用微调BERT对于输入的社交媒体文本进行向量嵌入;其次,设计了过滤注意力机制来滤除嵌入社交媒体文本中的无用信息,基于字词层面和语句层面两种过滤注意力机制,旨在从不同角度对于嵌入向量进行过滤;最后,采用Transformer的解码器部分进行解码工作,并根据过滤注意力机制的不同设计相应的解码策略。[结果/结论]本研究在微博数据集上与摘要生成领域中经典、优秀的基线模型进行对比实验。实验结果表明,本研究所设计的FUM模型有着比其他基线方法更出色的表现。同时,基于语句层面的过滤注意力机制比基于字词层面的有更好的过滤效果。

基于Transformer模型的文本自动摘要生成

论文探讨文本摘要的自动生成技术,其任务是产生能够表达文本主要含义的简明摘要。传统的Seq2Seq结构模型对长期特征和全局特征的捕获和存储能力有限,导致所生成的摘要中缺乏重要信息。因此,论文基于Transformer模型提出了一种新的生成式文本摘要模型RC-Transformer-PGN(RCTP)。该模型首先使用了一个附加的基于双向GRU的编码器来扩展Transformer模型,以捕获顺序上下文表示并提高局部信息的捕捉能力,其次引入指针生成网络以及覆盖机制缓解未登录词和重复词问题。在CNN/Daily Mail数据集上的实验结果表明论文模型与基线模型相比更具竞争力。

面向不可移动文物自然灾害风险图生成的领域特定语言

针对当前不可移动文物自然灾害风险图生成领域需求快速增长、经常变更,而现有程序及工具无法满足实际使用需要的问题,提出了一种构建语义模型的方法,在语义模型的基础之上,为不可移动文物领域专家设计了一种贴近自然语言的领域特定语言(DSL)。首先,通过对不可移动文物本体各项指标数据、风险图生成方法和流程进行深入研究,提取出业务模型。之后,通过分析业务模型,抽象出风险值计算规则的元计算单元,构建语义模型。在此基础之上,设计出一种可表述语义模型中所有语义的领域特定语言。该语言脚本可由领域专家自行编写,能快速高效地生成风险图,且易于扩展,满足需求经常变更的要求。与目前国内外主流的通过地理信息系统(GIS)方法生成风险图相比,利用DSL生成风险图可缩短66.7%以上的工作时长。

基于Abstract Factory模式的C#实现

随着面向对象设计技术的日益广泛运用,作为可复用技术基础的设计模式也得到了深入研究。本文结合C#语言,就创建模式中的AbstractFactory给出了一种可行的实现方式。

基于关键词异构图的生成式摘要研究

生成式摘要是自然语言处理中的重要任务,它帮助人们从海量文本中提取简洁而重要的信息.目前主流的生成式摘要模型是基于深度学习的序列到序列模型,这类模型生成的摘要质量更高.但由于缺乏对原文中关键词和句子之间的依赖关系的关注,现有模型生成的摘要仍然存在语义不明、重要信息含量低等问题.针对这个问题,提出了一种基于关键词异构图的生成式摘要模型.该模型通过从原始文本中提取关键词,将其与句子共同作为输入构建异构图,进而学习关键词和句子之间的依赖关系.文档编码器和图编码器分别用于学习文本知识和异构图中的依赖关系.此外,在解码器中采用分层图注意力机制来提高模型在生成摘要时对显著信息的关注.在CNN/Daily Mail和XSum数据集上进行了充分的实验,实验结果表明,所提模型在ROUGE评价指标上有了显著的提升.进一步的人类评估结果显示,所提模型所生成的摘要比基线模型包含更多的关键信息,并具有更高的可读性.

长文本摘要生成:基于Pegasus模型的STM32论文摘要生成分割策略研究

研究探讨了使用预训练的Pegasus模型进行长文本摘要时,不同文本分割方法对摘要质量的影响。收集来自知网的200篇关于STM32单片机的学术论文作为实验文本,比较了滑动窗口、句子分割、段落分割及滑动窗口加句子分割四种分割法的长文本摘要生成效果。实验使用ROUGE(Recall-Oriented Understudy for Gisting Evaluation)指标对生成的摘要进行评估,并对实验结果进行了详细分析。在生成摘要的质量方面,段落分割法表现出色,其ROUGE-1、ROUGE-2和ROUGE-L评分分别达到了30.85、7.60和20.15,轻微超过了句子分割法的评分,且显著优于句子分割加滑动窗口法。该研究旨在为研究者和开发者提供关于长文本摘要的实践经验和见解。

Intelligent Transformation: General Intelligence Theory

This paper aims to formalize a general definition of intelligence beyond human intelligence. We accomplish this by re-imagining the concept of equality as a fundamental abstraction for relation. We discover that the concept of equality = limits the sensitivity of our mathematics to abstract relationships. We propose a new relation principle that does not rely on the concept of equality but is consistent with existing mathematical abstractions. In essence, this paper proposes a conceptual framework for general interaction and argues that this framework is also an abstraction that satisfies the definition of Intelligence. Hence, we define intelligence as a formalization of generality, represented by the abstraction ∆∞Ο, where each symbol represents the concepts infinitesimal, infinite, and finite respectively. In essence, this paper proposes a General Language Model (GLM), where the abstraction ∆∞Ο represents the foundational relationship of the model. This relation is colloquially termed “The theory of everything”.

摘要AI文本检测中“猫鼠游戏”的行为界定和能力分析

[研究目的]辨别学术论文中的AI内容已成为学术出版界面临的新问题。研究人类对AI文本的界定依据和分辨能力,以及机器猫(AI文本检测工具)对AI鼠(AI文本)的检测能力,为学术出版中AI文本检测提供理论参考。[研究方法]使用GPT-3.5-Turbo、GPT-4和GPT-4-Turbo三个不同版本的ChatGPT对中文权威学术期刊《计算机学报》的150篇中文论文进行了摘要AI生成和AI改写;对英文权威学术期刊Artificial intelligence的150篇英文论文进行了摘要AI翻译,分析人类和4款主流AI文本检测工具对于AI生成摘要、通过AI改写和AI翻译的伪装摘要的识别能力。[研究结论]根据调查问卷结果统计,人类一般会从直觉、表达习惯、逻辑性、语句过渡和数据合理性5个方面来界定AI文本。该实验研究范围内,人类对AI生成摘要、AI改写摘要和AI翻译摘要的分辨能力有限,识别率在50%~65%之间;而该文选用的4款AI文本检测工具识别率均在80%左右,但对于由高版本的GPT-4-Turbo模型生成的摘要,以及经过ChatGPT改写或翻译的摘要,目前主流AI文本检测工具也都表现不佳。因此在当前的学术出版过程中,出版者还应采用多种方式识别AI文本。