1 Shanping Li 12/2003Focuses of Info Integration Group Focuses of Info Integration Group Li Shanping College of Computer Science Zhejiang University (Email:

Presentation on theme: "1 Shanping Li 12/2003Focuses of Info Integration Group Focuses of Info Integration Group Li Shanping College of Computer Science Zhejiang University (Email:"— Presentation transcript:

1 1 Shanping Li 12/2003Focuses of Info Integration Group Focuses of Info Integration Group Li Shanping College of Computer Science Zhejiang University (Email: shan@cs.zju.edu.cn) Dec. 17, 2003

2 2 Shanping Li 12/2003Focuses of Info Integration Group Research Focuses ontology and Semantic Web Overview of ontology ontology based Knowledge System ontology Mapping ontology Definition Methodology XML Documents Annotation Semantic Web Distributed System Grid Computing A Proposal to Pervasive Computing Mobile Computing QoS P2P

3 3 Shanping Li 12/2003Focuses of Info Integration Group The word Ontology was originally derived from philosophy. Aristotle’s Ontology focused on metaphysics Ontology History

4 4 Shanping Li 12/2003Focuses of Info Integration Group Ontology History Since 1980s, IT researchers introduce ONTOLOGY to describe domain conceptualization model 。 Knowledge Engineering (Expression ， Inference) Database Management (Integration) Software Engineering (Reuse) 1. T R Gruber. Towards Principles for the Design of Ontologies Used for Knowledge Sharing. Stanford University, Tech Rep: KSL-93-04, 1993. 2. N Guarino. Formal Ontology and Information Systems. Proc. of the 1st International Conference on Formal Ontology in Information Systems. Trento, Italy: IOS Press, 1998, 3~15.

5 5 Shanping Li 12/2003Focuses of Info Integration Group Ontology History n ontology becomes hot because of Semantic Web 1. T Berners-Lee. Semantic Web Road Map, http://www.w3.org/DesignIssues/Semantic.html, 1998. http://www.w3.org/DesignIssues/Semantic.html 2. T Berners-Lee, J Hendler, O Lassila. The Semantic Web. Scientific American, http://www.sciam.com, May 2001.

6 6 Shanping Li 12/2003Focuses of Info Integration Group Definition to Ontology n An ontology is an explicit specification of a conceptualization.(Gruber,1993) An ontology refers to an engineering artifact, constituted by a specific vocabulary used to describe a certain reality, plus a set of explicit assumptions regarding the intended meaning of the vocabulary words. (Nicola Guarino,1998)

7 7 Shanping Li 12/2003Focuses of Info Integration Group Ontology vs ontology Ontology of philosophy is mostly about cognitive content i. e. Aristotle’s Ontology ontology of AI is only about artifact, emphasizing coherence of content and form i. e. CYC ® Multi-contextual Knowledge Base http://www.cyc.com

8 8 Shanping Li 12/2003Focuses of Info Integration Group ontology vs logic Logic provides the formal structure and rules of inference. ontology defines the kinds of things that exist and share in the application domain.

9 9 Shanping Li 12/2003Focuses of Info Integration Group Ontologies classification 通用本体或常识本体 领域本体 应用本体 语言学本体语言学本体 抽象抽象 细化细化 提供词汇

10 10 Shanping Li 12/2003Focuses of Info Integration Group Why We Need Ontology n Communication n Inter-Operability n Systems Engineering

11 11 Shanping Li 12/2003Focuses of Info Integration Group Communication 本体的核心概念是知识共享。通过减少概念和术语上的歧 义，本体描述为组织间，系统与人之间提供了一个统一框 架或是规范模型，使得来自不同背景，持不同观点和目的 的人员之间，人员与系统间的理解和交流成为可能，并保 持语义上的一致性。如基于本体的语义查询，知识获取， 数据挖掘等。

12 12 Shanping Li 12/2003Focuses of Info Integration Group Inter-Operability 基于本体的信息检索：本体在信息检索引擎中的应用的基 本思路是利用本体对信息的描述来检索知识库，从而提高 检索的效率和精确度。典型的例子为 SHOE 。 基于本体共享的信息访问：不同的应用程序开发人员在共 享本体上达成一致，即定义了一个可以双向转换的数据传 输的格式。典型的例子主要有 EcoCyc [49] 等。 基于本体映射的信息访问：与前一种应用不同的是，不同 的应用程序之间拥有自己独立的本体，而不存在一个共享 的本体，因此它们之间的信息共享或者数据交换只能通过 不同的独立本体之间的映射（ mapping ）。典型的应用主 要有 ONIONS [50] 等。

13 13 Shanping Li 12/2003Focuses of Info Integration Group Systems Engineering n 本体还可以应用到软件系统的设计和实现上。 n 在软件开发的详细说明 (specification) 中，本体论 通过对需要解决的问题和任务的理解描述，可以 帮助我们在需求分析，信息获取中提高明确性， 减小分析代价。 n 同时本体可以作为需求分析基础上软件设计时的 基础，以自动或是半自动的方式检查它们之间的 一致性，从而提高软件系统的可靠性 (reliability) n 本体还可以通过对系统内部各个功能模块和它们 之间的联系的详细描述 ( 分层次的描述）达到软件 的重用性（ reusability ）。

14 14 Shanping Li 12/2003Focuses of Info Integration Group ontology descriptive languages Common Logic (Knowledge Interchange Format and Conceptual Graphs) Semantic web ontology language CycL UML Express 。。。。

15 15 Shanping Li 12/2003Focuses of Info Integration Group Related Work to ontology ontology engineering Description, Translation/Mapping, Merging Applications Semantic Web

16 16 Shanping Li 12/2003Focuses of Info Integration Group Our current interests ontology modeling modeling principles ontology acquisition from legacy systems S-B-F product knowledge modeling ontology driven semantic annotation ontology applications in product data model product configuration grid computing

17 17 Shanping Li 12/2003Focuses of Info Integration Group ontology driven semantic annotation

18 18 Shanping Li 12/2003Focuses of Info Integration Group Research Interests in Mobile Computing We are now investigating the performance issues of time-critical data dissemination in a wireless computing environment that supports data broadcast while being backed by a wired content distribution network. We also study time-constrained scheduling algorithms at the wireless gateway for efficient data dissemination over the broadcast-based wireless link. The delay-aware data replication strategies inside the wired network to facilitate data staging for time-constrained data broadcast are our concerns.

19 19 Shanping Li 12/2003Focuses of Info Integration Group Research Interests in P2P n part of PSF project “Research on Mobile Computing Oriented Network Storage System and Algorithms” n Survey most important characteristics in P2P systems, e.g. decentralization, scalability, performance, self-organization, anonymity, cost of ownership, and security n About locating & routing mechanisms 1. centralized directory model 2. flooded requests model 3. document routing model

20 20 Shanping Li 12/2003Focuses of Info Integration Group Research Interests in P2P n proposed a modified locating & routing mechanism based on the mechanism used in Freenet “A Hint-based Locating & Routing Mechanism in Peer-to-Peer File Sharing Systems” n work on mobile P2P system. Precisely L&R algorithm for mobile system “lookup protocol, M-CAN, modification of CAN ”

21 21 Shanping Li 12/2003Focuses of Info Integration Group Unsuitability of classic lookup protocols n limited bandwidth n unreliable connection n and constrains of mobile devices

22 22 Shanping Li 12/2003Focuses of Info Integration Group Projects NSF: Research on Product Information Infrastructure Based On Ontology and Semantic Web PSF: Research on Mobile Computing Oriented Network Storage System and Algorithms PSF: Research on Information Infrastructure Based on Semantic Web

23 23 Shanping Li 12/2003Focuses of Info Integration Group Publications A Hint-based Locating & Routing Mechanism in Peer-to-Peer File Sharing System. Proc. of GCC2003. Secure Grid-based Mobile Agent Platform by Instance-Oriented Delegation. Proc. of GCC2003. An Instance-Oriented Security Mechanism in Grid-based Mobile Agent System. Proc. of CLUSTER2003. M-CAN: a lookup protocol for Mobile Peer-to- Peer environment.

24 24 Shanping Li 12/2003Focuses of Info Integration Group Publications 数据流的活动队列管理算法。电子学报。 用户公平的活动队列管理 。电子学报。 李善平，刘乃若，郭鸣。产品数据标准与 PDM 。清华大 学出版社。 本体论研究综述。计算机研究与发展。 基于本体的产品知识表达。 CAD&CG 学报。 产品数据模型的本体知识表达。 CAD&CG 学报。 n 基于本体论和语义 WEB 的产品信息模型的研究。浙江大 学学报（工学版）。

25 25 Shanping Li 12/2003Focuses of Info Integration Group Publications n Ontology Based Product Data Integration. Proc. of AINA2003. n Semantic Web based distributed Configuration. Proc. of GCC2002. n Some challenges for the Semantic Web. Proc. of GCC2002. n Towards Semantic Web-based Information Integration. Proc. of GCC2002.

26 26 Shanping Li 12/2003Focuses of Info Integration Group Development Focuses Linux and Embedded Systems Vertical Market Oriented 税源分布 GIS 税务局 OA 欠税提示系统 电子报税系统 电子开票系统

27 27 Shanping Li 12/2003Focuses of Info Integration Group Linux and Embedded Systems Set Top Box for VOD over TCP/IP n VPN Router, Broadband Gateway n Brand New Operating System n VOD Server VOD Server n Video Server for DVB Video Server for DVB n HPC Management HPC Management

28 28 Shanping Li 12/2003Focuses of Info Integration Group Set Top Box for VOD over TCP/IP n Project Purpose Implement an all-purpose VOD client High stability High usability Satisfies the commercial use.

29 29 Shanping Li 12/2003Focuses of Info Integration Group Set Top Box for VOD over TCP/IP

30 30 Shanping Li 12/2003Focuses of Info Integration Group

31 31 Shanping Li 12/2003Focuses of Info Integration Group STB Features ---- Play stream data RTSP@ISMA Control panel seek, pause, stop, quit, etc. User friendly http, HTML, etc. Transport Stream more than 1.5Mb

32 32 Shanping Li 12/2003Focuses of Info Integration Group STB Features ---- Browser Standards compatible e.g. HTML, CSS, etc. Programmable e.g. JavaScript Safe e.g. SSL Configurable ( see following )

33 33 Shanping Li 12/2003Focuses of Info Integration Group STB Features ---- Internet Access Static IP PPPoE DHCP, dynamic IP address

34 34 Shanping Li 12/2003Focuses of Info Integration Group STB Features ---- Configuration Network Config Access Config Web site Config Online help Config ……

35 35 Shanping Li 12/2003Focuses of Info Integration Group STB implementation ---- Hardware CPU CODEC Memory External storage Interface Ethernet 、 S-Video 、 USB (x2) 、 RCA for Composite video 、 RCA for analog stereo audio output (x2) Remote control

36 36 Shanping Li 12/2003 STB implementation ---- Software Linux + Libraries and drivers + Qt/Embedded + Konqueror/Embedded + Http server + CODEC based Player

37 37 Shanping Li 12/2003Focuses of Info Integration Group VOD Server n Suitable for hotels, entertainment sites, etc. n Function Components: video on Demand video searcher interface to authentication and billing system video administration system with B/S Architecture system supervisor system with B/S Architecture

38 38 Shanping Li 12/2003Focuses of Info Integration Group VOD Server n Stable capability: Able to support at least 200 nodes and 60 concurrencies n Scalability: freely add streaming servers to the system n Failure diagnosis system, and n Failure backup system

39 39 Shanping Li 12/2003Focuses of Info Integration Group System Work Flow

40 40 Shanping Li 12/2003Focuses of Info Integration Group System Modules Web Server: IPG (Interactive Program Guide) for video browsing and demanding; System configuration interface; System supervisor interface and business management interface. Video Administration: explore, insert, delete and modify video information. n Control Server: runs as the heart of VOD system. It is responsible for supervising and maintaining the state of system, load balance, calling billing module.

41 41 Shanping Li 12/2003Focuses of Info Integration Group System Modules n Streaming Server: It is responsible for notifying Control of his register and logout; Reporting its state to Control Server. Notifying Control Server when connection between clients and it establishes or closes; Serving clients streaming media; Calling authentication module to perform authentication.

42 42 Shanping Li 12/2003Focuses of Info Integration Group Video Server for DVB n The Video Server is a Cluster Server and the draft system architecture is from HKU. n The system consists of a set of nodes connected by two switches. One-node servers as backup node, the others are working nodes. Every two working nodes are connected to share a video card. n There are two switches in the system. One of the switches is also regarded as backup switch.

43 43 Shanping Li 12/2003Focuses of Info Integration Group Video Server Architecture

44 44 Shanping Li 12/2003Focuses of Info Integration Group Software System Target n The system software focused on load balance, high I/O throughput and high availability. n The video data is divided into blocks and the blocks are placed on disks across nodes. n We also use software RAID 5 scheme to provide enough redundant information for data recovery when node failure or disk failure occurs. n As the video data isn’t stored in one node, the system has a buffer layer and uses two-level prefetch policy (local prefetch and global prefetch) to make sure that the application can get video data quickly and smoothly.

45 45 Shanping Li 12/2003Focuses of Info Integration Group Software System Target n Another important module of the system is failured, which is used as failure detection. Failured can detect network failure or node failure through heartbeat information. When an error is detected, the failured informs the other modules and corresponding modules will do some switch work and make backup devices active. For customers’ convenience, we’ve also developed a set of libraries.

46 46 Shanping Li 12/2003Focuses of Info Integration Group Project Status n Now most of the work has been done. n The server (5 nodes, 4 nodes are work nodes and the rest one is backup node) can provide 216 Mbits/s MPEG 2 data stream stably (each working node can provide 54Mbits/s data stream, 54*4=216). n The system could also scale to 12 nodes and provide much high I/O throughput. n Besides, the high available module also works well. Various tests have been done, including the simulation of NIC failure, switch failure and node failure. When never a single point failure occurs, the redundant device could take over the job and the service would continue.

47 47 Shanping Li 12/2003Focuses of Info Integration Group HPC management Using OSCAR(Open Source Cluster Application Resources), LAM, MPICH, PVM (Parallel Virtual Machine) n Demo system: 32 nodes with 2 Xeon CPUs Built-in n Future work: GUI facilitated OSCAR-like toolkit

48 48 Shanping Li 12/2003Focuses of Info Integration Group Publications Linux 操作系统及实验教程。机械工业出版社。 1999 Linux 内核 2.4 版源代码分析大全。机械工业出版 社。 2002 Linux 与嵌入式系统。清华大学出版社。 2003 边干边学 —— LINUX 内核指导。浙江大学出版社 。 2002 4.4BSD 操作系统设计与实现（翻译）。中国电力 出版社。 2003 n Linux: The Textbook （翻译）。清华大学出版社

49 49 Shanping Li 12/2003Focuses of Info Integration Group Publications 嵌入式 Linux 操作系统研究。浙江大学学报（工学 版）。

50 50 Shanping Li 12/2003Focuses of Info Integration Group 谢谢大家！ THANK YOU