Learning to Build Accurate Service Representations and Visualization

Learning to Build Accurate Service Representations and Visualization
复制标题

学习构建准确的服务表示和可视化

DOI:
10.1109/tsc.2020.3001307
复制
发表时间:
2022-05
影响因子:
8.1
通讯作者:
Bing Bai
Bing Bai
中科院分区:
计算机科学2区
文献类型:
--
作者:
Junqi Zhang;YuShun Fan;Jia Zhang;Bing Bai

文献摘要

参考文献

相似文献

随着Web服务的蓬勃发展,越来越需要可视化服务生态系统来帮助人们浏览服务并了解其功能和在系统中的位置。构建适当的可视化的一个基本步骤是确保准确表示所包含的服务。然而,这并不是一项简单的任务,因为由于两个重要原因,服务配置文件可能还不够。首先,虽然服务本身用于各种场景,但它们的配置文件可能并不总是准确反映所有场景。其次,服务配置文件通常包含大量无法区分服务的通用背景术语。为了解决这两个问题,我们应用机器学习技术来增量学习整体服务表示。开发了一个定制的主题模型,名为服务表示-潜在狄利克雷分配(SR-LDA)。核心思想是从所涉及的服务组合的配置文件(即混搭配置文件)中了解有关服务的更全面和最新的信息,同时引入全局过滤器来识别和过滤掉背景术语。对真实世界数据集的定量和定性实验表明,与基线相比,所提出的 SR-LDA 构建了更高质量的服务表示。我们进一步生成知识图,以基于学习到的服务表示来可视化服务生态系统。这样的知识图谱直接导致对Web服务的四种典型功能模式的检测,并服务于mashup创建的目的。
With the boom of Web services, there is a growing need for visualizing service ecosystems to help people browse services and understand their functionalities and positions in the systems. One foundational step of building a proper visualization is to ensure accurate representations for the comprising services. However, it is not a trivial task as service profiles may not be sufficient for two significant reasons. First, while the services themselves being used in various scenarios, their profiles may not always precisely reflect all of them. Second, service profiles usually comprise quite a few universal background terms that cannot distinguish services. To address these two issues, we apply machine learning techniques to incrementally learn service representations in a whole. A tailored topic model is developed, named Service Representation-Latent Dirichlet Allocation (SR-LDA). The core idea is to learn more comprehensive and up-to-date information about services from the profiles of the involved service compositions (i.e., mashup profiles), while introducing a global filter to identify and filter out background terms. Both quantitative and qualitative experiments on a real-world dataset demonstrate that the proposed SR-LDA builds higher-quality service representations comparing with baselines. We further generate a knowledge map to visualize a service ecosystem based on the learned service representations. Such a knowledge map directly leads to the detection of four typical functionality patterns of Web services and serves the purpose of mashup creation.
DOI: 10.1201/b10345-5
发表时间: 2010-11
期刊: Encyclopedia of Autism Spectrum Disorders
影响因子: --
作者:
Kim-Anh Lê Cao;Z. Welham
通讯作者: Kim-Anh Lê Cao;Z. Welham
基于BTM主题模型和K-means聚类算法的微博主题检测方法
DOI: 10.3103/s0146411616040040
发表时间: 2016-09
影响因子: 0.9
作者:
Li Weijiang;Feng Yanming;Li Dongjun;Yu Zhengtao
通讯作者: Yu Zhengtao
DOI: 10.1109/tse.2011.22
发表时间: 2012-05
影响因子: 7.4
作者:
V. Andrikopoulos;S. Benbernou;M. Papazoglou
通讯作者: V. Andrikopoulos;S. Benbernou;M. Papazoglou
DOI: 10.1016/j.neucom.2008.12.002
发表时间: 2009-06
期刊: Neurocomputing
影响因子: 6
作者:
Chenping Hou;J. Wang;Yi Wu;Dong-yun Yi
通讯作者: Chenping Hou;J. Wang;Yi Wu;Dong-yun Yi
DOI: 10.5555/1756006.1953039
发表时间: 2010-03
期刊: J. Mach. Learn. Res.
影响因子: --
作者:
Pascal Vincent;H. Larochelle;Isabelle Lajoie;Yoshua Bengio;Pierre-Antoine Manzagol
通讯作者: Pascal Vincent;H. Larochelle;Isabelle Lajoie;Yoshua Bengio;Pierre-Antoine Manzagol