Presentation is loading. Please wait.

Presentation is loading. Please wait.

Jaeseok Myung, Jongheum Yeon

Similar presentations


Presentation on theme: "Jaeseok Myung, Jongheum Yeon"— Presentation transcript:

1 2007-09-22 Jaeseok Myung, Jongheum Yeon
InfoBroker An Infomediary System Enabling Intelligent Services Jaeseok Myung, Jongheum Yeon

2 Intelligent Service Synonyms Core Concepts
Context-awareness, Personalization, Ubiquitous Core Concepts Predict, Offer Using information that can be understood by a machine Where the intelligence from? The granularity of information that can be understood by a machine decides the degree of intelligence of the machine Center for E-Business Technology

3 The granularity of Information
Search Search & Trace Graph Search Graph Match Metadata Expansion Personal Profile + Context Semantic Extension Relationship Expansion Enriched Personal Profile & Context Personal Profile Keyword Center for E-Business Technology

4 Service Trends Documents vs. Documents about products
The key information is ‘d100 is a product’ The granularity of information is finer and finer than before Center for E-Business Technology

5 InfoBroker Intelligent Service Provider InfoBroker d100 isa dsfadfadafa d100 isa Camera Product Has structured information that can be used for intelligent services Design Intelligent Services Through IB’s Information Center for E-Business Technology

6 InfoBrokers Intelligent Service Provider d100 camera d100 price $850
Design Intelligent Services Through IB’s Information expensive Center for E-Business Technology

7 Big Picture IB에 정보 요청 End-User 서비스 Storage 역할 Entity Name(ID)
Intelligent Service Provider Knowledge Base Knowledge Extraction Engine IB에 정보 요청 Entity 식별 relationship 검색 End-User 서비스 Semantic Search Automatic Tagging Storage 역할 ISP의 요청에 응답 KEE의 입력 처리 Entity Name(ID) Relationship Entity A -> B 플러그인 구조 Preprecessing Entity User-defined Relationship Document 관리 Policy 관리 KB와 통신 및 로깅 Center for E-Business Technology

8 Intelligent Service Provider
Utilize the knowledge (Jaeseok -> Person) People usually don’t have to find it directly ISP <-> KB Send Receive Doc. (URL, Stream) Entity List Entity_A Relationship List Entity_A, Relationship Entity_B Relationship Entity_A, Entity_B All Entities All Relationships IB에 정보 요청 Entity 식별 relationship 검색 End-User 서비스 Semantic Search Automatic Tagging Center for E-Business Technology

9 Relationship (Ea -> Eb)
Information Base Directed Graph Interface Long-term Research (Complex Model) Attributes(datatype, weight, log) Modify, Remove Interface Knowledge Base Entity Relationship (Ea -> Eb) Name(ID) : Literal Entity_A : Entity Entity_B : Entity Storage 역할 ISP의 요청에 응답 KEE의 입력 처리 Entity Name(ID) Relationship Entity A -> B KB <- KEE Func. Param. Create Entity Name Create Relationship Name, Entity_A, Entity_B Center for E-Business Technology

10 Knowledge Extraction Engine
Documents Set 1 Documents Set N Knowledge Extraction Engine Policy Manager Plug-in Manager Document Manager Preprocessing Algorithms Entity Extraction Relationship Knowledge Base Manager Documents Entities Relations Preprocessing Alg. 1 Preprocessing Alg. N Entity Ext. Alg. 1 Entity Ext. Alg. N Relation Ext. Alg. 1 Relation Ext. Alg. N

11 Knowledge Extraction Engine
Document 3 Document 2 Document 1 2006 독일 월드컵때 신은 박지성 축구화가 뭔가요?? 프랑스전인가.. 박지성 신발이 벗겨져서 다시신고있는걸 카메라로 찍었었는데 축구화 뒤쪽에 대한민국 국기가 있는 축구화요 답변 이거 말하시는거죠? 2006 독일월드컵 대한민국:프랑스 이 축구화 이름은 에어레전드<티엠포>라는 나이키 제품입니다. 이름하고 태극기는 선수들만 할 수 있다고 들었습니다. 예를 들어서 C.호날두 축구화 옆에 보면 CR.7이라 써있을거예요. 각자 자기 축구화에 자기꺼라 이름 새기는 겁니다. Knowledge Extraction Engine Policy Manager Plug-in Manager Document Manager Entity Extraction Algorithms Relationship Knowledge Base Manager Preprocessing Preprocessed Result Question 2006/독일/월드컵/박지성/축구화/프랑스전/박지성/신발/카메라/찍었었는데/축구화/뒤쪽/대한민국/국기/ 있는/축구화 Answer 2006/독일/월드컵/대한민국:프랑스/ /축구화/이름/에어레전드<티엠포>/나이키/제품/입니다/이름/하고/태극기/선수/할 수 있다/들었습니다/예/들어/C.호날두/축구화 /옆/보면/CR.7/써있다/각자/자기/축구화/자기꺼라/이름/새기다 Result 박지성 에어레전드<티엠포> 나이키 축구화 제품

12 Knowledge Extraction Engine
Document 3 Document 2 Document 1 박지성, '불사조 축구화' 신고 축구 레전드로 한국에서는 박지성(27, 맨체스터 유나이티드)이 나이키가 아시아 선수 중 최초라 자신만을 위해 특별히 제작한 ‘불사조 축구화, 박지성 티엠포 레전드 FG’를 신고 달리게 됐다. 나이키의 마케팅적 판단이라고 할 수도 있지만 그 사실조차도 세계무대에서 박지성의 위상을 알 수 있는 부분이다. 지난 8월 30일 제니트와의 UEFA 슈퍼컵에 교체 투입된 박지성이 30여 분간 활약하며 첫 선을 보인 이 신발은 검정 바탕에 붉은 색 나이키 로고와 골드 컬러로 박지성의 열정을 강조했다. 나이키는 박지성의 이름을 딴 축구화뿐만 아니라 축구복, 축구공 그리고 가방까지 제작해 ‘Be The Legend’라는 캠페인을 진행할 예정이다. Knowledge Extraction Engine Policy Manager Plug-in Manager Document Manager Entity Extraction Algorithms Relationship Knowledge Base Manager Preprocessing Preprocessed Result Title 박지성/불사조/축구화/신다/축구/레전드 Contents 한국/박지성/27/맨체스터 유나이티드/‘불사조 축구화, 박지성 티엠포 레전드 FG’/ 신다/달리다/되었다/나이키/마케팅/판단/사실/세계무대/박지성/위상/부분/지난/8월 30일/제니트/UEFA/슈퍼컵/교체/투입/박지성/30여분간/활약/첫 선/보이다/신발/검정/바탕/붉은색/나이키로고/골드컬러/박지성/열정/강조/하다 Result 박지성 검정 축구화 바탕 제품 티엠포 레전드 FG 나이키

13 GATE a General Architecture for Text Engineering Architecture GDM : the GATE Document Manager Language Resource(LR) : refers to data-only resources such as lexicons, corpora, thesauruses or ontologies CREOLE : a Collection of REusable Objects for Language Engineering Processing Resource(PR) : refers to resources whose character is principally programmatic or algorithmic, such as lemmatisers, generators, translators, parsers or speech recognisers GGI : the GATE Graphical Interface Visual Resource(VR) : Each resource (PR and LR) can have its own associated visual resource Define resources and pipelined processing order to make text analysis applications

14 CREAM Creating relational metadata with a component-based, ontology-driven annotation framework Architecture

15 Attensity Server Automatically extracts valuable data from free-form text Designed to run on multiple machines over large volumes of data

16 To Do Discussion InfoBroker? Vs. Extraction Algorithms?
Center for E-Business Technology


Download ppt "Jaeseok Myung, Jongheum Yeon"

Similar presentations


Ads by Google