LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May 2006 1 Cross-Media Indexing in the Reveal-This System Murat Yakici,

Slides:



Advertisements
Similar presentations
Large Scale Knowledge Management across Media Prof. Fabio Ciravegna, Department of Computer Science University of Sheffield
Advertisements

GMD German National Research Center for Information Technology Darmstadt University of Technology Perspectives and Priorities for Digital Libraries Research.
A Human-Centered Computing Framework to Enable Personalized News Video Recommendation (Oh Jun-hyuk)
Pseudo-Relevance Feedback For Multimedia Retrieval By Rong Yan, Alexander G. and Rong Jin Mwangi S. Kariuki
Distributed search for complex heterogeneous media Werner Bailer, José-Manuel López-Cobo, Guillermo Álvaro, Georg Thallinger Search Computing Workshop.
Interoperability Scenarios All Working Groups Meeting May, Rome, Italy.
A Stepwise Modeling Approach for Individual Media Semantics Annett Mitschick, Klaus Meißner TU Dresden, Department of Computer Science, Multimedia Technology.
DL:Lesson 11 Multimedia Search Luca Dini
1 Texmex – November 15 th, 2005 Strategy for the future Global goal “Understand” (= structure…) TV and other MM documents Prepare these documents for applications.
Discussion on Video Analysis and Extraction, MPEG-4 and MPEG-7 Encoding and Decoding in Java, Java 3D, or OpenGL Presented by: Emmanuel Velasco City College.
Mining the web to improve semantic-based multimedia search and digital libraries
Information Retrieval in Practice
Chapter 11 Beyond Bag of Words. Question Answering n Providing answers instead of ranked lists of documents n Older QA systems generated answers n Current.
1 CS 430: Information Discovery Lecture 22 Non-Textual Materials 2.
Video retrieval using inference network A.Graves, M. Lalmas In Sig IR 02.
Multimedia Search and Retrieval: New Concepts, System Implementation, and Application Qian Huang, Atul Puri, Zhu Liu IEEE TRANSACTION ON CIRCUITS AND SYSTEMS.
Image Search Presented by: Samantha Mahindrakar Diti Gandhi.
Supervised by Prof. LYU, Rung Tsong Michael Department of Computer Science & Engineering The Chinese University of Hong Kong Prepared by: Chan Pik Wah,
Presentation Outline  Project Aims  Introduction of Digital Video Library  Introduction of Our Work  Considerations and Approach  Design and Implementation.
1 MPEG-21 : Goals and Achievements Ian Burnett, Rik Van de Walle, Keith Hill, Jan Bormans and Fernando Pereira IEEE Multimedia, October-November 2003.
Architecture & Data Management of XML-Based Digital Video Library System Jacky C.K. Ma Michael R. Lyu.
Real-time and Retrospective Analysis of Video Streams and Still Image Collections using MPEG-7 Ganesh Gopalan, College of Oceanic and Atmospheric Sciences,
1 Information Retrieval and Extraction 資訊檢索與擷取 Chia-Hui Chang, Assistant Professor Dept. of Computer Science & Information Engineering National Central.
Philips Research France Delivery Context in MPEG-21 Sylvain Devillers Philips Research France Anthony Vetro Mitsubishi Electric Research Laboratories.
Outline of Presentation Introduction of digital video libraries Introduction of the CMU Informedia Project Informedia: user perspective Informedia:
Overview of Search Engines
Enabling Access to Sound Archives through Integration, Enrichment and Retrieval WP3 – Retrieval systems.
Knowledge Science & Engineering Institute, Beijing Normal University, Analyzing Transcripts of Online Asynchronous.
Information Retrieval in Practice
Data Exchange Tools (DExT) DExT PROJECTAN OPEN EXCHANGE FORMAT FOR DATA enables long-term preservation and re-use of metadata,
Teaching Metadata and Networked Information Organization & Retrieval The UNT SLIS Experience William E. Moen School of Library and Information Sciences.
Aurora: A Conceptual Model for Web-content Adaptation to Support the Universal Accessibility of Web-based Services Anita W. Huang, Neel Sundaresan Presented.
Katanosh Morovat.   This concept is a formal approach for identifying the rules that encapsulate the structure, constraint, and control of the operation.
Dr. Kurt Fendt, Comparative Media Studies, MIT MetaMedia An Open Platform for Media Annotation and Sharing Workshop "Online Archives:
A Motivating Scenario for Designing an Extensible Audio- Visual Description Language Monday 25 th of October, 2004 Raphaël Troncy, Jean Carrive, Steffen.
An Overview of MPEG-21 Cory McKay. Introduction Built on top of MPEG-4 and MPEG-7 standards Much more than just an audiovisual standard Meant to be a.
MPEG-21 : Overview MUMT 611 Doug Van Nort. Introduction Rather than audiovisual content, purpose is set of standards to deliver multimedia in secure environment.
Information Systems & Semantic Web University of Koblenz ▪ Landau, Germany Semantic Web - Multimedia Annotation – Steffen Staab
The PrestoSpace Project Valentin Tablan. 2 Sheffield NLP Group, January 24 th 2006 Project Mission The 20th Century was the first with an audiovisual.
Chapter 2 Architecture of a Search Engine. Search Engine Architecture n A software architecture consists of software components, the interfaces provided.
CHAPTER TEN AUTHORING.
Understanding The Semantics of Media Chapter 8 Camilo A. Celis.
2007. Software Engineering Laboratory, School of Computer Science S E Web-Harvest Web-Harvest: Open Source Web Data Extraction tool 이재정 Software Engineering.
Prof. Thomas Sikora Technische Universität Berlin Communication Systems Group Thursday, 2 April 2009 Integration Activities in “Tools for Tag Generation“
Structure of IR Systems INST 734 Module 1 Doug Oard.
Data and Applications Security Developments and Directions Dr. Bhavani Thuraisingham The University of Texas at Dallas Lecture #15 Secure Multimedia Data.
DATA RESOURCE MANAGEMENT
Digital Libraries1 David Rashty. Digital Libraries2 “A library is an arsenal of liberty” Anonymous.
MMDB-9 J. Teuhola Standardization: MPEG-7 “Multimedia Content Description Interface” Standard for describing multimedia content (metadata).
Information Retrieval
1 MPEG-7 Overview - part 2. 2 Review Descriptor (D) - 對內容的特徵作定義。 - 通常用以描述 low-level features 。 Description Scheme (DS) - 通常用以描述 high-level features 。
MPEG-4: Multimedia Coding Standard Supporting Mobile Multimedia System Lian Mo, Alan Jiang, Junhua Ding April, 2001.
Digital Video Library Network Supervisor: Prof. Michael Lyu Student: Ma Chak Kei, Jacky.
SAPIR Search in Audio-Visual Content using P2P Information Retrival For more information visit: Support.
MPEG-7 Audio Overview Ichiro Fujinaga MUMT 611 McGill University.
A Reduced Yet Extensible Audio- Visual Description Language: How to Escape From The MPEG-7 Bottleneck Thursday 28 th of October, 2004 Raphaël Troncy, Jean.
Introduction to MPEG  Moving Pictures Experts Group,  Geneva based working group under the ISO/IEC standards.  In charge of developing standards for.
MPEG 7 &MPEG 21.
Multi-Source Information Extraction Valentin Tablan University of Sheffield.
Information Retrieval in Practice
Working meeting of WP4 Task WP4.1
Digital Video Library - Jacky Ma.
Visual Information Retrieval
CS644 Advanced Topics in Networking
An Overview of MPEG-21 Cory McKay.
MPEG-7 Video Retrieval using Bayesian Networks
Peggy van der Kreeft Deutsche Welle
Multimedia Content Description Interface
Searching and browsing through fragments of TED Talks
Ying Dai Faculty of software and information science,
Presentation transcript:

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Cross-Media Indexing in the Reveal-This System Murat Yakici, Fabio Crestani Dept. Computer & Information Sciences University of Strathclyde, Glasgow, UK

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Overview Application Scenario Reveal-This Project Cross-Media Indexing –Process Model –Information Model –Indexing Model Evaluation Future work

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Application Scenario Persistency MPEG-7 Coded … How about the Armani jacket that I saw on Fashion TV

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May The Reveal-This (R-T) Project Aims at: Developing content programming technology able to: 1.capture 2.semantically index 3.categorise 4.cross-link … Multiplatform, multimedia and multilingual digital content…

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May The Reveal-This Project (2) … as well as provide the system user with: 1.semantic search 2.retrieval 3.summarisation 4.translation… functionalities Clearly an ambitious project!

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Digital Content in R-T Digital Content is –Distributed over different platforms –Repurposed and delivered to diverse devices –Can be in a range of media types –Rapidly consumed (on demand provision) Key issues –Managing meta-data –Managing data

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Reveal-This Architecture

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Cross-Media Indexing Component (CMIC) Addresses the 2 nd scientific objective and in part also the 4 th Part of Cross-Media Indexing and Analysis Subsystem Media meta-data integration and indexing Service

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May CMIC Overview Builds relationships among concepts extracted from different processors –such as video, speech and text analysis How do we do it?

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May CMIC Overview (2) Two interpretations of “media” 1.Different sources on same topic 2.Single source but different type on the same topic

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May CMIC Features Transform source XML (feature) streams into an internal cross-media representation (MPEG-7) in order to make comparisons in the same similarity space Capture relations across media Cross-link media (within document as cross- media) Augment digital objects with semantic information (MPEG-7) Store meta-data and relations Online indexing and retrieval Support to various languages English, Greek, French Enable push and pull of events

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May CMIC Process Model

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May CMIC Information Model Structures and patterns for representing digital content at any level –What is the unit of information processing across R-T system? –Are there any common patterns out there? –What are the emerging standards? We adoped MPEG-7 as the information model

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May MPEG-7 Overview Standard is finalised “How to describe content” Provides diverse and large set of description elements Tools for –Multiplexing of descriptions –Synchronization of descriptions with content

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May MPEG-7 How to describe content –A set of description schemes (DS) and descriptors (D) –A language to specify description schemes (Description Definition Language - DDL) –A scheme for binary coding the descriptions Consider DSs as a library of descriptions –Feel free to pick and use appropriate subset of relevant DSs depending on your requirements.

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Subtitles Face 1 Face 2 Speaker 2 Speaker 1 Relevant Segment And Others… Studio Setting Transitions Zoom in Closed captions Noise Music Transcription … Task of Cross-Media Indexing

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Subtitles Face 1 Face 2 Speaker 2 Speaker 1 Relevant Segment Semantic Indexing Person 2 Sea Person 1 Boat Sailing

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Challenge Signal and Semantic Gap problem Task entails dealing with –uncertainty –imprecision –inconsistency from each single media analysis module

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Indexing Models supported by CMIC –Tf = term frequency –Tf*Idf = gives less importance to a term, if it appears in high number of stories –Modal Tf = the term frequency provided from each processing module (such as image analysis, text processing etc.) is incorporated to the previous approaches –Dempster-Shafer Multi-Evidence Approach

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Dempster-Shafer (1) Dempster-Shafer combines two or more bodies of evidence defined with in the same frame of discernment into one body of evidence. Every modality individually gives a support for a single story, a term’s existence in one modality is counted as an evidence to support the topical similarity hypothesis. Each processing module is treated as a probability function also called as Source of Evidence or Base Probability Assignment (BPA).

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Dempster-Shafer (2) Combination

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Initial experience Accuracy of processing units – In general Granularity of Topics and Categories –Topics and Categories are not describing on the same level –Terms in categories/Topics do not appear in text (Inconsistencies between values received) Confidence scores & Ranking Faces are detected but… Indexing in other languages (Greek and French) Indexing Model Indexing time depends on the indexing model

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Evaluation Task oriented user experiment. –The users are given a test video sequence and then asked to detect-recognize faces, transcribe text and describe certain aspects. Their descriptions are regarded as confidence level 1. –The tasks are introduced. The users try to find out information needed to accomplish each individual task and find most relevant segments –Compare this with CMIC’s output.

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Evaluation (2) Data annotation by experts: relevance assessments: A three-hour multimedia test collection Covering politics, travel and news In English, French and Greek languages

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Towards complex models After we finish testing the simpler indexing models we will explore more complex models: –Bayesian Networks –Kernel Canonical Correlation Analysis –Gaussian Mixture Models

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Future work Research –Add and improve indexing models –Benchmark performance with current version on updated data sets –Evaluation with annotated data set Software engineering tasks –Add management and administration services –Use in push and pull services –Adapt to user profile –Integrate with other services (summarisation, etc.)

LREC – Workshop on Crossing media for Improved Information Access, Genova, Italy, 23 May Thank you Questions?