ibm왓슨과 apple 시리
DESCRIPTION
전문가 토크릴레이 "웹과 플랫폼의 미래를 이야기 하다" 4탄 [IBM 왓슨과 Apple 시리에 대해] : 솔트룩스 이경일 대표TRANSCRIPT
![Page 2: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/2.jpg)
Apple, IBM, Google 비전의
기술적 공통점?
when
BigData met AI
![Page 3: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/3.jpg)
인간 지식 처리를 위한 연구
Artificial Intelligence
Semantic Web
Knowledge Engineering
![Page 4: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/4.jpg)
Knowledge engineering은 어떤 도메인에서 특정 목적을 위해 컴퓨
터가 업무를 처리할 수 있도록 모델을 구성할 때 온톨로지와 로직을
활용하는 과정 - John Sowa
Artificial Intelligence은 컴퓨터를 통해 지능정 행동을 수행하도록
하는 연구로, agent가 어떻게 행동을 할 것인가를 결정하는 과정에 지
식 표현과 지식 이해 과정이 수반됨 – Brachman and Levesque
Semantic Web은 웹 표준 하에서 컴퓨터가 데이터의 의미를 이해하고
처리하는 것이 가능한 데이터의 웹 – Tony
Knowledge representation은 해석될 수 있는 기호(symbolic form)
로 지식을 형식화하는 것을 의미 – Klein and Methlie
인간 지식 처리를 위한 연구
![Page 5: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/5.jpg)
인공 지능 (AI) ?
Systems that think like humans Systems that think rationally
Systems that act like humans Systems that act rationally
AI : The study and design of intelligent agents 인텔리전트 에이전트는 환경을 감지해서, 스스로 행동함으로 기회를 최적화, 자신의 목표 달성할 수 있는 자동 시스템
• Knowledge Representation • Reasoning • Learning • Planning • Natural Language Processing • Social Intelligence • Machine perception and Vision
![Page 6: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/6.jpg)
자연 언어 (Natural Language)
글로 쓰여진 사람의 말 : “지구는 타원 궤도로 태양을 돌고 있다”
시각 언어 (Visual Language)
그림, 구조도, 흐름도, 설계도 등 시각적으로 지식을 표현
주석, 태깅 (Tagging)
개체에 연관된 키워드, 기호, 이미지 등을 부착해 지식을 표현
기호 언어 (Symbolic Language)
수학 등을 포함해 기호로 표현된 지식 : x2/a2 + y2/b2 = 1
의사 결정 나무 (Decision Tree)
복잡한 의사 결정을 위해 구성된 나무 모양의 그래프 구조
규칙 (Rules)
인간 지식을 여러 규칙들의 조건부 결합으로 표현
데이터베이스 (Database System)
개체와 관계로 구성된 테이블 형태의 지식 표현 체계
논리 언어 (Logical Language)
논리 기호, 연산을 통한 지식 표현 : Woman ≡ Person ∩ Female
프레임 언어 (Frame Language)
값 혹은 타 프레임의 포인터를 저장한 슬롯들로 지식 표현
시맨틱 네트워크 (Semantic Network)
개념간의 의미적 관계를 그래프 구조로 구성한 지식 표현
통계적 지식 (Statistical Knowledge)
확률과 통계에 기반한 지식 표현, 기계 학습 기술 접목 가능
사람
기
계
지식 표현 기계와 인간의 협력?
![Page 7: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/7.jpg)
“기업에 종사하는 종업원은 사람들이고, 기업과 종업원은 모두 법적 존재이다.
기업은 직원들을 위해 여행 예약을 할 수 있다. 여행은 한국 내 도시, 혹 미국의
도시를 오고 가는 비행기 혹은 기차를 통해 가능하다. 기업들과 출장지는 도시에
위치하고 있다. 솔트룩스는 홍길동을 위해 서울과 뉴욕 왕복 항공편인 OZ510을
예약하였다.”
자연 언어
규칙 언어
(규칙) 만약 누군가가 날고 있다면, 여행중인 것이다.
(규칙) 만약 누군가의 여행이 한 회사에서 예약되었다면, 그는 그 회사의 종업원이다.
(규칙 추가) 만약 동일 국가의 근거리 여행이라면, 종업원은 기차를 이용해야 한다.
(추론) 비행 예약이 되어 있는 홍길동은 솔트룩스의 종업원이다
(추론) OZ510은 미국과 한국을 오가는 비행편이다.
지식의 표현
![Page 8: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/8.jpg)
법적 존재
사람 기업
종업원
홍길동
솔트룩스
비행기 기차
도시
위치
한국 도시 미국 도시
뉴욕
서울
OZ510
여행
kin
dO
f
kin
dO
f in
stnace
Of
insta
nce
Of
insta
nce
Of
insta
nce
Of
endsIn
startFrom
books
participatesIn
inst
ance
Of
법적 존재
사람 기업
종업원
홍길동
솔트룩스
subcl
ssO
f
insta
nce
Of
inst
ance
Of
사람 기업
종업원
#3502
#4831
subcl
ssO
f in
stance
Of
inst
ance
Of
법적 존재
이름
고유번호
성별
나이
업종
주소지
직급
홍길동
37 과장
P12345
남자
솔트룩스
서울 삼성동
C98765
소프트웨어
사람 기업
종업원
#3502
#4831
subcl
ssO
f
insta
nce
Of
inst
ance
Of
법적 존재
이름 (필수)
고유번호 (필수)
성별 ⊆ {남,녀}
나이 > 25
업종
주소지 ⊂ 서울
직급 ≠ 임원
홍길동
37 과장
P12345
남자
솔트룩스
서울 삼성동
C98765
소프트웨어
DISJOINT
(a) 시맨틱 네트워크 (b) (a) + 프레임(프로퍼티) (c) (b) + 논리 제약
온톨로지(Ontology)
지식의 표현
![Page 9: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/9.jpg)
Why is Siri more attractive?
Because Siri acts like real human agent including continuous conversation and recommending alternatives.
Functions Other Agent
Apple Siri
Continuous Conversation
Weak Strong
Recommending Alternatives
Weak Strong
Semantic Match Weak Strong
Semantic Disambiguation
Weak Strong
Semantics make it possible in Siri!
![Page 10: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/10.jpg)
Siri vs. S-Voice
![Page 11: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/11.jpg)
• Deductive reasoning Premise 1: All humans are mortal. Premise 2: Socrates is a human. Conclusion: Socrates is mortal.
• Inductive reasoning Premise: The sun has risen in the east every morning up until now. Conclusion: The sun will also rise in the east tomorrow.
• Abductive reasoning
• Analogical reasoning
Ontology and Rules
Machine Learning
추론? : 기존 사실들로부터 새로운 사실을 도출하거나
결론에 도달하는 과정
추론 Reasoning
![Page 12: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/12.jpg)
논리적 추론 발전 방향
![Page 13: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/13.jpg)
학습 Learning
영화, 인류멸망보고서 중
학습(Learning)
• 주어진 여건에 대한 행동이 되풀이
되는 경험으로 인해 생기는 그 여
건에 대한 행동 변화
• 지식의 습득과 기존 지식으로부터
추론된 결과의 재학습 능력 필요
• 궁극적으로 컴퓨터가 새로운 것을
배우고 환경에 적응하는 것
![Page 14: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/14.jpg)
Black Box (learning machine)
Training data Model
• Support vector machines • Inductive logic programming • Decision tree learning • Association rule learning • Artificial neural networks • Genetic programming
Test-data
Prediction
Model
빅 데이터 기계 학습
• Clustering • Bayesian networks • Reinforcement learning • Representation learning • Sparse Dictionary Learning
![Page 15: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/15.jpg)
• 계획(Plan) 목표까지 경로에 있는 아크 연산자들을 하나의 순서로 만든 것
• 계획 수립(Planning) 다양한 순서를 찾아내고, 최적 순서를 확보하는 것
• 투영(Projecting) 어떤 행동 순서의 결과로 나타나는 상태의 순서를 예측
• 계획 시스템 제약조건하에서 목표를 달성 위해 행동을 설계하는 시스템
- 만일 새로운 정보가 생기면 계획되었던 일련의 과업들을 변경시킬 수 있는 유연성을 가져야 함
- 현재까지의 추론 과정을 되돌아 가고, 더 좋은 해결안을 위해 현 추론 결과를 취소할 수 있음
계획 Plan/Planning
(Nils J.Nilsson 1998)
![Page 16: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/16.jpg)
Open window (A) and fly kite (B). String (C) lifts small door (D) allowing moths (E) to escape and eat red flannel shirt (F). As weight of shirt becomes less, shoe (G) steps on switch (H) which heats electric iron (I) and burns hole in pants (J). Smoke (K) enters hole in tree (L), smoking out opossum (M) which jumps into basket (N), pulling rope (O) and lifting cage (P), allowing woodpecker (Q) to chew wood from pencil (R), exposing lead. Emergency knife (S) is always handy in case opossum or the woodpecker gets sick and can't work.
Rube Goldberg의 연필 깎는 기계
계획 수립 Rube Goldberg Machine?
![Page 17: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/17.jpg)
Apple의 Siri 들여다 보기
![Page 18: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/18.jpg)
View Points for Siri-like Service
Unstructured Big Data Structured Big Data
Human Interaction
Natural Language Understanding / Generation
Search & Reasoning (incl. computation)
Knowledge Base
Knowledge Acquisition and Modeling
Lin
ked
Ser
vice
s
![Page 19: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/19.jpg)
Dynamic Context
Inferred Context
Conte
xt
Model
Conte
xt
Rule
s
CONTEXT
Device User
CONTEXT OWNER
Smart Service
Service Personalization
SMART MOBILE SERVICE
Service Adaptation
Service Discovery
QoC
CONTEXT MANAGER SENSOR / NETWORK
Filter
Collector
Context Driven Mobile Service
![Page 20: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/20.jpg)
A virtual personal assistant is a SW system that
• Helps the user find or do something (focus on tasks, rather
than information)
• Understands the user’s intent (interpreting language) and
context (location, schedule, history)
• Works on the user’s behalf, orchestrating multiple services
and information sources to help complete the task
In other words, an assistant helps me do things by understanding
me and working for me. (Tom Gruber, 2010)
Virtual Personal Assistance?
![Page 21: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/21.jpg)
Intelligent Agent is an autonomous entity which observes through sensors and acts upon an environment using actuators.
IA directs its activity towards achieving goals.
Intelligent agents may also learn or use knowledge to achieve their goals.
- Russell & Norvig
Intelligent Agent?
![Page 22: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/22.jpg)
Intelligent Agent?
Simple reflex agent
General learning agent
![Page 23: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/23.jpg)
Intelligent Agent?
Model based reflex agent
Model and goal based agent
![Page 24: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/24.jpg)
Siri?
Siri is an intelligent software assistant and
knowledge navigator functioning as a
personal assistant application for iOS.
Siri uses a natural language UI to
• answer questions
• make recommendations
• perform actions with web services.
Siri adapts to the user's individual
preferences over time and personalizes
results
![Page 25: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/25.jpg)
Why Siri is different from others before…
Task focus. Siri is very focused on a bounded set of specific human tasks, like finding something to do, going out with friends, and getting around town.
Structured data focus. The kinds of tasks that Siri is particularly good at involve semi-structured data, usually on tasks involving multiple criteria and drawing from multiple sources.
Architecture focus. Siri is built from deep experience in integrating multiple advanced technologies into a platform designed expressly for virtual assistants. The CALO project taught Siri a lot about what works and doesn’t when applying AI to build a virtual assistant.
![Page 26: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/26.jpg)
What exactly can you ask Siri to do?
• Ask for a reminder.
• Ask to send a text.
• Ask about the weather.
• Ask to set a meeting.
• Ask to send an email.
• Ask for a number.
• Ask for information from Yelp, Wolfram|Alpha, or Wikipedia
1. Does Things for you focus on task completion
2. Gets What you Say intent understanding via conversation
3. Gets to Know You learns and applies personal information
• Ask to set an alarm.
• Ask for directions.
• Ask about stocks.
• Ask to set the timer.
• Ask Siri about Siri.
![Page 27: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/27.jpg)
History of Siri
Siri is using the results of over 40 years of research funded by DARPA via
SRI International’s Artificial Intelligence Center through CALO
project (2003~2008).
Siri technology has come a long way with dialog and natural language
understanding, machine learning, evidential and probabilistic
reasoning, ontology and knowledge representation, planning,
reasoning and service delegation.
Siri was founded in 2007 (spin-off from SRI international) by Dag Kittlaus
(CEO), Adam Cheyer (VP Engineering), and Tom Gruber (CTO/VP
Design).
$150 million – DARPA funds (4.5 years)
$8.5 million - series A (2009)
$15.5 million - series B
$200 million - purchased by apple (2010)
![Page 28: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/28.jpg)
dialog and natural language understanding
machine learning
evidential and probabilistic reasoning
ontology and knowledge representation
planning, reasoning
service delegation
Technology of Siri
Conversation Interface
Personal Context
Awareness
Service Delegation
![Page 29: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/29.jpg)
Overview of Siri Technology
![Page 30: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/30.jpg)
The interface is a Conversation
![Page 31: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/31.jpg)
Task-oriented NL Understanding
![Page 32: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/32.jpg)
Ontology Unifies all Models
![Page 33: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/33.jpg)
Semantic Autocomplete
![Page 34: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/34.jpg)
Dialog modules organize by generic task and domain
![Page 35: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/35.jpg)
What happened in Apple Siri?
Active Ontology is a brain to understand user’s intention and make conversation under the semantics
• Heterogeneous data integration
• Managing short and long term personal memory
• Improving speech recognition quality
• Semantic disambiguation
• Dialog generation and management
![Page 36: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/36.jpg)
IBM의 Watson Computer
들여다 보기
![Page 37: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/37.jpg)
The Jeopardy! Challenge
A compelling and notable way to drive and measure the technology of automatic Question Answering along 5 Key Dimensions
Broad/Open Domain
Complex Language
High Precision
Accurate
Confidence
High
Speed
$800 In cell division, mitosis splits the nucleus & cytokinesis splits this liquid cushio
ning the nucleus
$200 If you're standing, it's the di
rection you should look to c
heck out the wainscoting.
$1000 Of the 4 countries in the world that the U.S. does not have diplomatic relations with, the one that’s farthest no
rth
![Page 38: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/38.jpg)
Q&A The Domain
![Page 39: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/39.jpg)
The Big Idea Evidence-Based Reasoning over Natural Language Content
Deep Analysis of clues/questions AND content
Search for many possible answers based on different interpretations of question
Find, analyze and score EVIDENCE from many different sources (not just one document) for each answer using many advanced NLP and reasoning algorithms
Combine evidence and compute a confidence value for each possibility using statistical machine learning
Rank answers based on confidence
If top answer is above a threshold – buzz in else keep quiet
![Page 40: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/40.jpg)
IBM 왓슨 Deep QA 시스템
![Page 41: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/41.jpg)
Hardware Infrastructure
![Page 42: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/42.jpg)
Through training Watson Evaluates and Selects documents worth analyzing for a given task.
Too much irrelevant
content requires unnecessary compute power
For Jeopardy! Watson has analyzed
and stored the equivalent of about 1
million books (e.g., encyclopedias,
dictionaries, news articles, reference
texts, plays, etc)
![Page 43: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/43.jpg)
Auto. Learning & Semantic Frame
![Page 44: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/44.jpg)
UIMA Framework & UIMA-AS
![Page 45: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/45.jpg)
The Difference Between Search & DeepQA
Decision Maker
Search Engine
Finds Documents containing Keywords
Delivers Documents based on Popularity
Has Question
Distills to 2-3 Keywords
Reads Documents, Finds Answers
Finds & Analyzes Evidence
Expert
Understands Question
Produces Possible Answers & Evidence
Delivers Response, Evidence & Confidence
Analyzes Evidence, Computes Confidence
Asks NL Question
Considers Answer & Evidence
Decision Maker
![Page 46: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/46.jpg)
Keyword Search vs. Deep Reasoning for finding Evidences
![Page 47: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/47.jpg)
Natural Language Processing in Watson
![Page 48: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/48.jpg)
Deep QA Process One Jeopardy! question can take 2 hours on a single 2.6Ghz Core
2880-Core IBM Power750’s using UIMA-AS, Watson is answering in 2-6 sec.
Models
Answer & Co
nfidence
Question
Evidence
Sources
Models
Models
Models
Models
Models Primary
Search
Candidate
Answer
Generation
Hypothesis Generation
Hypothesis and Evidence Scoring
Final Confidence Merging&Ranking
Answer
Sources
Question & Topic Analy
sis
Evidence
Retrieval
Evidence
Scoring
Learned Models
help combine and
weigh the Evidence
Hypothesis Generation
Hypothesis and
Evidence Scoring
Question Decomposition
Merging &
Ranking
Synthesis
Multiple
Interpretations 100’s
sources
1000’s of
Pieces of Evidence
100’s Possible
Answers
100,000’s Scores from
many Deep Analysis
Algorithms
Balance
& Combine
![Page 49: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/49.jpg)
Performances
![Page 50: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/50.jpg)
Organizations
![Page 51: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/51.jpg)
Future of Watson?
![Page 52: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/52.jpg)
• 5 years R&D from 2009
• Computes answers to natural language questions
• Integrates disconnected trusted data sources
• Sophisticated automated algorithm and visualization selection
• General and domain-specific linguistic and presentation development
Wolfram|Alpha Computation Knowledge Engine
![Page 53: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/53.jpg)
• Any systematic data can be curated
• Human-driven curation includes tools, processes, and methodologies
• Thousands of domains curated falling into about 50-100 domain models
• Ontology is at a meta level
• Hierarchical knowledge included with entity classes, attributes
• Relates things at computation time
Capability & Data Curation
• 10+ trillion of pieces of data
• 50,000+ types of algorithms and models
• linguistic capabilities for 1000+ domains
• Built with Mathematica
![Page 54: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/54.jpg)
Infrastructures
• Mathematica 7 : 2500 built-in functions
• Super Computer Clusters
- DCS(Dell Data Center Solutions)
and R Systems Cluster
- World 44th powerful super computer
- Clustered 5 super computer
- Windows HPC server 2008, Windows Computer Cluster Server
- Platform LSF, Altair PBS, Sun Grid
![Page 55: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/55.jpg)
Examples
![Page 56: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/56.jpg)
ziny.us
똑똑한 소셜 매거진 “지니어스”
빅 데이터와 인공지능 기반의 스마트 미디어
![Page 57: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/57.jpg)
iPhone : Reinvention of Phone
ziny.us : Reinvention of Social Media
IBM Watson
관심기반 퍼블리싱 인공지능 지니어스
![Page 58: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/58.jpg)
The Three Happiness
모으는 즐거움 보는 즐거움 나누는 즐거움
![Page 59: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/59.jpg)
Feeding,
Crawling,
Wrapping,
Open API
Bookmarklet,
File upload,
Camera
Search & Discover Publish & Share Filter & Organize
Hybrid Classification,
Automatic Clustering
Clip/Re-Clip,
Love/Comment
Machine Learning,
Recommendation
Auto-Publishing,
Personalization
HTML5,
App, PDF
Smart Curation?
Digital Magazine
Facebook/Twitter
Mail Sharing
Real-time Chatting
Learning
![Page 60: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/60.jpg)
소셜 데이터 수집
• 클라우드에 기반한 대용량 분산/병렬처리, 1일 500만건 수집
• 클라우드 스토리지에 데이터 저장과 실시간 인덱싱 수행
•450 Cores, 1.5TB Ram, 200TB HDD
•원시 소셜 데이터 : 총 5억 건, 2.5TB
•수집 속도 : 500만 건 / 일
•수집 방식 : Hybrid Model (크롤링 + Open API + Agent)
•저장 구조 : 클라우드(NoSQL+DFS), 데이터 3중화
트위터
57% 블로그
24%
뉴스
1%
미투
데이
18%
수집 데이터 구성 1일 수집, 인덱싱 로그
![Page 61: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/61.jpg)
소셜 토픽의 추출
• Social Topic간 Co-occurrence 거리를 Weigh w로 할 때, 중요도 WS(Vi) 정의,
• Google PageRank 개념이 적용된 TextRank를 발전, 소셜 토픽을 추출
• Social co-occurrence 분석 통해 특성 벡터의 품질 향상과 실시간 처리
• Graph system G = (V, E)에 대해 각 vertex Vi의 중요도 S(vi)를 정의,
![Page 62: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/62.jpg)
소셜 데이터의 분류
• SVM 기반 학습 모델과 VSM 기반의 규칙 모델 통합
• 대규모 실시간 소셜 아티클 분류를 위해 병렬, 분산처리
학습기반 분류 (SVM)
규칙기반 분류 (VSM+RULE)
피드백 학습
A 분류체계 B 분류체계 C 분류체계
소셜 데이터
아티클1
아티클7 아티클20 아티클51
…
실시간
병렬,
분산처리
![Page 63: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/63.jpg)
소셜 이슈 학습
• 소셜 아티클의 실시간 군집을 통한 사회적 이슈 도출
• 주제별 사회적 관심 트랜드 분석과 예측, 추론
𝑊𝑔= 𝐷𝐹 +𝑊𝑆+𝑀𝑒𝑎𝑛 𝑇𝐹∗𝑊𝐹𝑢𝑐(𝐷𝐹)
Wfunc : Skewed Distrib.
Social Article Retrieval
Global Features Selection
Hierarchical Word clustering
Article clustering (cosine similarity)
Cluster Labeling
Clusters Ranking/Grouping
![Page 64: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/64.jpg)
사용자 생성 매거진 학습 아티클 자동추천(ziny 추천) 사용자 피드백(Clip, Love)
쓰면 쓸수록 똑똑해지는 소셜 매거진
매거진 별 SP Feature vector 생성
Social Feature- Vector Index
Fast Similarity Calculation on Vector Space Model
약 5억
건
사용자 관심 학습과 추천
![Page 65: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/65.jpg)
Knowledge Network Analysis
![Page 66: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/66.jpg)
![Page 67: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/67.jpg)
e-Discovery Solution
![Page 68: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/68.jpg)
VOC(Voice of Customer) Analysis
![Page 69: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/69.jpg)
Technology Sensing
![Page 70: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/70.jpg)
Personalized Android Mobile App
Real-time Recommendation Service
Originally developed in CogFrame proj.
Improved to work on LarKC Platform
Based on Location-based Social Media
Analysis (incl. Sentiment Analysis)
Applying Hybrid (Stream) Reasoning
BOTTARI Mobile App
![Page 71: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/71.jpg)
BOTTARI 보따리
• 트위터 등 소셜 빅 데이터에 대한 실시간 분석 (트랜드, 평판)
• AR이 적용된 Android App. / 시맨틱웹첼린지 그랑프리
![Page 72: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/72.jpg)
Communicating Knowledge 72
미래,예측하는 것이 아닌
만들어 가는 것...
![Page 73: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/73.jpg)
(matthew Komorwski, 2010)
1/1억
기술 혁신 > 낭비 하도록 만들기
Transistors in a CPU
100만 배
Enterprise Strategy Group, 2010
지난 30년간
1천만 배
![Page 74: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/74.jpg)
저장 가격 1/100,
반도체 집적도 X100
앞으로 10년 후의 왓슨?
<IBM Power 750>
- 10 full racks
- 2880 CPU cores
- 15 TB RAM
- 80 teraflops / sec
- 10 GE ethernet
![Page 75: Ibm왓슨과 apple 시리](https://reader033.vdocuments.net/reader033/viewer/2022051610/54811212b37959a22b8b5c1f/html5/thumbnails/75.jpg)
Peter Drucker
“유일한 성공 방법은, 미래를 예측하는 것이 아니라
이미 시작된 변화를 이해, 그 시간차를 이용하는 것!"