【作者】 李楠；

【导师】 王忠杰；

【作者基本信息】 哈尔滨工业大学 ， 计算机科学与技术， 2021， 硕士

【摘要】 随着互联网技术的快速发展,大量服务迅速涌现,线上线下的服务互联互通逐渐形成了复杂的服务网络。海量的服务资源在给用户带来便利的同时也导致了用户需求的复杂化与个性化。受服务供求双方信息不对称以及用户认知受限的影响,用户通常无法实时全面的获取服务以及相关领域的信息,很难精确且有针对性的给出需求的细节信息。此外用户在表达需求的过程中还可能会出现不完整或歧义的情况。这些因素导致了对用户的需求理解出现偏差,因此无法为用户提有针对性且优质的服务。因此,如何辅助用户完整、有效地提出需求是大服务研究中的一个重要问题。本文针对该问题进行研究,提出了基于模式的大服务需求工程,并从以下三个方面详细描述。(1)大服务环境下需求建模方法与需求模式挖掘方法:本文首先分析了大服务环境下的用户需求特点,提出了一种意图树的需求模型,该模型既便于用户填充又便于计算机进行自动化分析。在意图树的基础上提出了需求模式,模式代表了需求之间的共性,能提供对于需求的基本理解能力。为了从海量用户需求中发现需求模式,本文提出了一种基于模糊聚类的需求模式挖掘算法。此外由于意图树存在着需求关系的缺失,本文还提出了一种基于混合知识图谱的需求模型,以提升对需求的理解能力。(2)大服务环境下基于需求模式的需求获取方法研究:需求模式能够在不引入其他外部信息的条件下,支持需求的快速识别并提供最低限度的需求理解。需求模式能够在用户提出需求阶段辅助用户提出需求,本文提出了基于混合知识图谱的需求确认方法、基于模式的需求构造与补全方法提升用户需求质量。在大服务需求工程中使用模式也会带来一些问题,即会使大服务解决方案千篇一律。本文还提出了需求模式的自适应自演化策略,使模式能够根据外界服务环境以及用户自身的变化有针对性的做出反馈。(3)大服务下的需求管理工具:为了能够更好的使用本文方法解决大服务需求工程中的问题,本文设计了大服务下的需求管理工具,该工具覆盖了大服务需求工程的处理全过程,包括需求补全、需求模式挖掘等。更多还原

【Abstract】 With the rapid development of Internet technology,many services have emerged rapidly,and the interconnection of online and offline services has gradually formed a complex service network.Massive service resources bring convenience to users and lead to the complexity and personalized of users’ requirements.However,affected by the information asymmetry between the supply and demand of services and the limited cognition of users,users are usually unable to obtain real-time and comprehensive information about services and related fields.So it isn’t easy to provide precise and targeted detailed details on requirements.In addition,users’ requirements may also be incomplete or ambiguous.These factors lead to deviations in understanding users’ requirements,failing to provide targeted and high-quality users.Therefore,assisting users in putting forward their requirements ultimately and effectively is an important problem in big service research.This paper studies this problem proposes a pattern-based requirement engineering method in big service and describes it in detail from the following three aspects.(1)Requirement modeling method and requirement pattern mining method in big service environment: This article first analyzes the characteristics of user requirement in a big service environment and proposes a requirement model based on the intention tree,which is convenient for users to fill and automated analysis.The requirement pattern is presented based on the intention tree,representing the commonness among requirements and providing the essential understanding ability of uses’ requirements.This paper proposes a requirement pattern mining algorithm based on fuzzy clustering to find requirement patterns from massive user requirements.In addition,due to the absence of some relationship between requirements in the intention tree,this paper also proposes a requirements model based on a hybrid knowledge graph to improve the ability to understand requirements.(2)Research on the requirement elicitation method based on requirement pattern in big service environment: requirement pattern can support rapid identification of requirements and provide a minimum understanding of requirements without introducing other external information.Requirement patterns can assist users in the stage of requirement presentation.In this paper,requirement confirmation method based on a mixed knowledge graph,requirement construction,and completion method based on the pattern is proposed to improve user requirement quality.Using patterns in big service requirement engineering makes big service solutions all the same.This paper also proposes an adaptive self-evolution strategy for the requirement patterns,enabling the patterns to respond to changes in the external service environment and users.(3)Requirement management tool in big service: To better use the method proposed in this paper to solve the problems in big service requirement engineering,this paper designs a requirement management tool in big service requirement engineering,which covers the whole process of big service requirement engineering,including requirement completion,requirement pattern mining,etc.更多还原