Systems & Applications: Introduction Ling 573 NLP Systems and Applications March 29, 2011.

Slides:



Advertisements
Similar presentations
A Stepwise Modeling Approach for Individual Media Semantics Annett Mitschick, Klaus Meißner TU Dresden, Department of Computer Science, Multimedia Technology.
Advertisements

Overview of Collaborative Information Retrieval (CIR) at FIRE 2012 Debasis Ganguly, Johannes Leveling, Gareth Jones School of Computing, CNGL, Dublin City.
1 Question Answering in Biomedicine Student: Andreea Tutos Id: Supervisor: Diego Molla.
Question-Answering: Shallow & Deep Techniques for NLP Ling571 Deep Processing Techniques for NLP March 9, 2011 Examples from Dan Jurafsky)
Search Engines and Information Retrieval
Shallow Processing: Summary Shallow Processing Techniques for NLP Ling570 December 7, 2011.
Question-Answering: Overview Ling573 Systems & Applications March 31, 2011.
Information Retrieval in Practice
Reference Collections: Task Characteristics. TREC Collection Text REtrieval Conference (TREC) –sponsored by NIST and DARPA (1992-?) Comparing approaches.
Web Information Retrieval and Extraction Chia-Hui Chang, Associate Professor National Central University, Taiwan Sep. 16, 2005.
Jumping Off Points Ideas of possible tasks Examples of possible tasks Categories of possible tasks.
Web Logs and Question Answering Richard Sutcliffe 1, Udo Kruschwitz 2, Thomas Mandl University of Limerick, Ireland 2 - University of Essex, UK 3.
Information Retrieval and Extraction 資訊檢索與擷取 Chia-Hui Chang National Central University
Web Search – Summer Term 2006 II. Information Retrieval (Basics Cont.) (c) Wolfgang Hürst, Albert-Ludwigs-University.
 Official Site: facility.org/research/evaluation/clef-ip-10http:// facility.org/research/evaluation/clef-ip-10.
DIVINES – Speech Rec. and Intrinsic Variation W.S.May 20, 2006 Richard Rose DIVINES SRIV Workshop The Influence of Word Detection Variability on IR Performance.
Search is not only about the Web An Overview on Printed Documents Search and Patent Search Walid Magdy Centre for Next Generation Localisation School of.
Information Retrieval in Practice
AQUAINT Kickoff Meeting – December 2001 Integrating Robust Semantics, Event Detection, Information Fusion, and Summarization for Multimedia Question Answering.
CONTI’2008, 5-6 June 2008, TIMISOARA 1 Towards a digital content management system Gheorghe Sebestyen-Pal, Tünde Bálint, Bogdan Moscaliuc, Agnes Sebestyen-Pal.
CS598CXZ Course Summary ChengXiang Zhai Department of Computer Science University of Illinois, Urbana-Champaign.
Challenges in Information Retrieval and Language Modeling Michael Shepherd Dalhousie University Halifax, NS Canada.
Introduction to Natural Language Processing Heshaam Faili University of Tehran.
The Evolution of Shared-Task Evaluation Douglas W. Oard College of Information Studies and UMIACS University of Maryland, College Park, USA December 4,
Enterprise & Intranet Search How Enterprise is different from Web search What to think about when evaluating Enterprise Search How Intranet use is different.
Russian Information Retrieval Evaluation Seminar (ROMIP) Igor Nekrestyanov, Pavel Braslavski CLEF 2010.
Search Engines and Information Retrieval Chapter 1.
CLEF Ǻrhus Robust – Word Sense Disambiguation exercise UBC: Eneko Agirre, Oier Lopez de Lacalle, Arantxa Otegi, German Rigau UVA & Irion: Piek Vossen.
“How much context do you need?” An experiment about context size in Interactive Cross-language Question Answering B. Navarro, L. Moreno-Monteagudo, E.
CS523 INFORMATION RETRIEVAL COURSE INTRODUCTION YÜCEL SAYGIN SABANCI UNIVERSITY.
Report on UNSD activities since the last meeting of the Expert Group on International Economic and Social Classifications Meeting of the Expert Group on.
AnswerBus Question Answering System Zhiping Zheng School of Information, University of Michigan HLT 2002.
Information Retrieval and Web Search Lecture 1. Course overview Instructor: Rada Mihalcea Class web page:
Chapter 2 Architecture of a Search Engine. Search Engine Architecture n A software architecture consists of software components, the interfaces provided.
Question Answering.  Goal  Automatically answer questions submitted by humans in a natural language form  Approaches  Rely on techniques from diverse.
Internet Information Retrieval Sun Wu. Course Goal To learn the basic concepts and techniques of internet search engines –How to use and evaluate search.
TRECVID Evaluations Mei-Chen Yeh 05/25/2010. Introduction Text REtrieval Conference (TREC) – Organized by National Institute of Standards (NIST) – Support.
1 Information Retrieval Acknowledgements: Dr Mounia Lalmas (QMW) Dr Joemon Jose (Glasgow)
Proposal for Term Project J. H. Wang Mar. 2, 2015.
Interactive Probabilistic Search for GikiCLEF Ray R Larson School of Information University of California, Berkeley Ray R Larson School of Information.
XP New Perspectives on The Internet, Sixth Edition— Comprehensive Tutorial 3 1 Searching the Web Using Search Engines and Directories Effectively Tutorial.
Syllabus CS479(7118) / 679(7112): Introduction to Data Mining Spring-2008 course web site:
Text Based Information Retrieval Text Based Information Retrieval H02C8A H02C8B Marie-Francine Moens Karl Gyllstrom Katholieke Universiteit Leuven.
Course grading Project: 75% Broken into several incremental deliverables Paper appraisal/evaluation/project tool evaluation in earlier May: 25%
CS315-Web Search & Data Mining. A Semester in 50 minutes or less The Web History Key technologies and developments Its future Information Retrieval (IR)
WIRED Week 3 Syllabus Update (next week) Readings Overview - Quick Review of Last Week’s IR Models (if time) - Evaluating IR Systems - Understanding Queries.
1 01/10/09 1 INFILE CEA LIST ELDA Univ. Lille 3 - Geriico Overview of the INFILE track at CLEF 2009 multilingual INformation FILtering Evaluation.
The Structure of Information Retrieval Systems LBSC 708A/CMSC 838L Douglas W. Oard and Philip Resnik Session 1: September 4, 2001.
How Do We Find Information?. Key Questions  What are we looking for?  How do we find it?  Why is it difficult? “A prudent question is one-half of wisdom”
Of 33 lecture 1: introduction. of 33 the semantic web vision today’s web (1) web content – for human consumption (no structural information) people search.
2015/12/121 Extracting Key Terms From Noisy and Multi-theme Documents Maria Grineva, Maxim Grinev and Dmitry Lizorkin Proceeding of the 18th International.
Information Retrieval and Web Search Course overview Instructor: Rada Mihalcea.
Information Retrieval
Comparing Document Segmentation for Passage Retrieval in Question Answering Jorg Tiedemann University of Groningen presented by: Moy’awiah Al-Shannaq
The Cross Language Image Retrieval Track: ImageCLEF Breakout session discussion.
1 13/05/07 1/20 LIST – DTSI – Interfaces, Cognitics and Virtual Reality Unit The INFILE project: a crosslingual filtering systems evaluation campaign Romaric.
AQUAINT AQUAINT Evaluation Overview Ellen M. Voorhees.
1 Question Answering and Logistics. 2 Class Logistics  Comments on proposals will be returned next week and may be available as early as Monday  Look.
Analysis of Experiments on Hybridization of different approaches in mono and cross-language information retrieval DAEDALUS – Data, Decisions and Language,
Information Retrieval Lecture 3 Introduction to Information Retrieval (Manning et al. 2007) Chapter 8 For the MSc Computer Science Programme Dell Zhang.
AQUAINT Mid-Year PI Meeting – June 2002 Integrating Robust Semantics, Event Detection, Information Fusion, and Summarization for Multimedia Question Answering.
Chapter 8: Web Analytics, Web Mining, and Social Analytics
Introduction to Information Retrieval. What is IR? Sit down before fact as a little child, be prepared to give up every conceived notion, follow humbly.
Information Retrieval in Practice
Information Storage and Retrieval Fall Lecture 1: Introduction and History.
Systems & Applications: Introduction
Deliverable #2 Goals: Become familiar with shared task summarization data Implement initial base system with all components Focus on content selection.
Kenneth Baclawski et. al. PSB /11/7 Sa-Im Shin
CSE 635 Multimedia Information Retrieval
CS224N Section 3: Corpora, etc.
Presentation transcript:

Systems & Applications: Introduction Ling 573 NLP Systems and Applications March 29, 2011

Roadmap Motivation 573 Structure Question-Answering Shared Tasks

Motivation Information retrieval is very powerful Search engines index and search enormous doc sets Retrieve billions of documents in tenths of seconds

Motivation Information retrieval is very powerful Search engines index and search enormous doc sets Retrieve billions of documents in tenths of seconds But still limited!

Motivation Information retrieval is very powerful Search engines index and search enormous doc sets Retrieve billions of documents in tenths of seconds But still limited! Technically – keyword search (mostly)

Motivation Information retrieval is very powerful Search engines index and search enormous doc sets Retrieve billions of documents in tenths of seconds But still limited! Technically – keyword search (mostly) Conceptually User seeks information Sometimes a web site or document

Motivation Information retrieval is very powerful Search engines index and search enormous doc sets Retrieve billions of documents in tenths of seconds But still limited! Technically – keyword search (mostly) Conceptually User seeks information Sometimes a web site or document Very often, the answer to a question

Why Question-Answering? People ask questions on the web

Why Question-Answering? People ask questions on the web Web logs: Which English translation of the bible is used in official Catholic liturgies? Who invented surf music? What are the seven wonders of the world?

Why Question-Answering? People ask questions on the web Web logs: Which English translation of the bible is used in official Catholic liturgies? Who invented surf music? What are the seven wonders of the world? 12-15% of queries

Why Question Answering? Answer sites proliferate:

Why Question Answering? Answer sites proliferate: Top hit for ‘questions’ :

Why Question Answering? Answer sites proliferate: Top hit for ‘questions’ : Ask.com

Why Question Answering? Answer sites proliferate: Top hit for ‘questions’ : Ask.com Also: Yahoo! Answers, wiki answers, Facebook,… Collect and distribute human answers

Why Question Answering? Answer sites proliferate: Top hit for ‘questions’ : Ask.com Also: Yahoo! Answers, wiki answers, Facebook,… Collect and distribute human answers Do I Need a Visa to Go to Japan?

Why Question Answering? Answer sites proliferate: Top hit for ‘questions’ : Ask.com Also: Yahoo! Answers, wiki answers, Facebook,… Collect and distribute human answers Do I Need a Visa to Go to Japan? eHow.com Rules regarding travel between the United States and Japan are governed by both countries. Entry requirements for Japan are contingent on the purpose and length of a traveler's visit.travel Passport Requirements Japan requires all U.S. citizens provide a valid passport and a return on "onward" ticket for entry into the country. Additionally, the United States requires a passport for all citizens wishing to enter or re-enter the country.ticket

Search Engines & QA Who was the prime minister of Australia during the Great Depression?

Search Engines & QA Who was the prime minister of Australia during the Great Depression? Rank 1 snippet: The conservative Prime Minister of Australia, Stanley Bruce

Search Engines & QA Who was the prime minister of Australia during the Great Depression? Rank 1 snippet: The conservative Prime Minister of Australia, Stanley Bruce Wrong! Voted out just before the Depression

Perspectives on QA TREC QA track ( ) Initially pure factoid questions, with fixed length answers Based on large collection of fixed documents (news) Increasing complexity: definitions, biographical info, etc Single response

Perspectives on QA TREC QA track (~ ) Initially pure factoid questions, with fixed length answers Based on large collection of fixed documents (news) Increasing complexity: definitions, biographical info, etc Single response Reading comprehension (Hirschman et al, ) Think SAT/GRE Short text or article (usually middle school level) Answer questions based on text Also, ‘machine reading’

Perspectives on QA TREC QA track (~ ) Initially pure factoid questions, with fixed length answers Based on large collection of fixed documents (news) Increasing complexity: definitions, biographical info, etc Single response Reading comprehension (Hirschman et al, ) Think SAT/GRE Short text or article (usually middle school level) Answer questions based on text Also, ‘machine reading’ And, of course, Jeopardy! and Watson

Natural Language Processing and QA Rich testbed for NLP techniques:

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval Named Entity Recognition

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval Named Entity Recognition Tagging

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval Named Entity Recognition Tagging Information extraction

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval Named Entity Recognition Tagging Information extraction Word sense disambiguation

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval Named Entity Recognition Tagging Information extraction Word sense disambiguation Parsing

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval Named Entity Recognition Tagging Information extraction Word sense disambiguation Parsing Semantics, etc..

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval Named Entity Recognition Tagging Information extraction Word sense disambiguation Parsing Semantics, etc.. Co-reference

Natural Language Processing and QA Rich testbed for NLP techniques: Information retrieval Named Entity Recognition Tagging Information extraction Word sense disambiguation Parsing Semantics, etc.. Co-reference Deep/shallow techniques; machine learning

573 Structure Implementation:

573 Structure Implementation: Create a factoid QA system

573 Structure Implementation: Create a factoid QA system Extend existing software components Develop, evaluate on standard data set

573 Structure Implementation: Create a factoid QA system Extend existing software components Develop, evaluate on standard data set Presentation:

573 Structure Implementation: Create a factoid QA system Extend existing software components Develop, evaluate on standard data set Presentation: Write a technical report Present plan, system, results in class

573 Structure Implementation: Create a factoid QA system Extend existing software components Develop, evaluate on standard data set Presentation: Write a technical report Present plan, system, results in class Give/receive feedback

Implementation: Deliverables Complex system: Break into (relatively) manageable components Incremental progress, deadlines

Implementation: Deliverables Complex system: Break into (relatively) manageable components Incremental progress, deadlines Key components: D1: Setup

Implementation: Deliverables Complex system: Break into (relatively) manageable components Incremental progress, deadlines Key components: D1: Setup D2: Query processing, classification

Implementation: Deliverables Complex system: Break into (relatively) manageable components Incremental progress, deadlines Key components: D1: Setup D2: Query processing, classification D3: Document, passage retrieval

Implementation: Deliverables Complex system: Break into (relatively) manageable components Incremental progress, deadlines Key components: D1: Setup D2: Query processing, classification D3: Document, passage retrieval D4: Answer processing, final results

Implementation: Deliverables Complex system: Break into (relatively) manageable components Incremental progress, deadlines Key components: D1: Setup D2: Query processing, classification D3: Document, passage retrieval D4: Answer processing, final results Deadlines: Little slack in schedule; please keep to time Timing: ~12 hours week; sometimes higher

Presentation Technical report: Follow organization for scientific paper Formatting and Content

Presentation Technical report: Follow organization for scientific paper Formatting and Content Presentations: minute oral presentation for deliverables

Presentation Technical report: Follow organization for scientific paper Formatting and Content Presentations: minute oral presentation for deliverables Explain goals, methodology, success, issues

Presentation Technical report: Follow organization for scientific paper Formatting and Content Presentations: minute oral presentation for deliverables Explain goals, methodology, success, issues Critique each others’ work

Presentation Technical report: Follow organization for scientific paper Formatting and Content Presentations: minute oral presentation for deliverables Explain goals, methodology, success, issues Critique each others’ work Attend ALL presentations

Working in Teams Why teams?

Working in Teams Why teams? Too much work for a single person Representative of professional environment

Working in Teams Why teams? Too much work for a single person Representative of professional environment Team organization: Form groups of 3 (possibly 2) people

Working in Teams Why teams? Too much work for a single person Representative of professional environment Team organization: Form groups of 3 (possibly 2) people Arrange coordination Distribute work equitably

Working in Teams Why teams? Too much work for a single person Representative of professional environment Team organization: Form groups of 3 (possibly 2) people Arrange coordination Distribute work equitably All team members receive the same grade End-of-course evaluation

First Task Form teams: Ryan with the team

Resources Readings: Current research papers in question-answering

Resources Readings: Current research papers in question-answering Jurafsky & Martin/Manning & Schutze text Background, reference, refresher

Resources Readings: Current research papers in question-answering Jurafsky & Martin/Manning & Schutze text Background, reference, refresher Software:

Resources Readings: Current research papers in question-answering Jurafsky & Martin/Manning & Schutze text Background, reference, refresher Software: Build on existing system components, toolkits NLP, machine learning, etc Corpora, etc

Resources: Patas System should run on patas Existing infrastructure Software systems Corpora Repositories

Shared Task Evaluations Goals: Lofty:

Shared Task Evaluations Goals: Lofty: Focus research community on key challenges ‘Grand challenges’.

Shared Task Evaluations Goals: Lofty: Focus research community on key challenges ‘Grand challenges’ Support the creation of large-scale community resources Corpora: News, Recordings, Video Annotation: Expert questions, labeled answers,..

Shared Task Evaluations Goals: Lofty: Focus research community on key challenges ‘Grand challenges’ Support the creation of large-scale community resources Corpora: News, Recordings, Video Annotation: Expert questions, labeled answers,..

Shared Task Evaluations Goals: Lofty: Focus research community on key challenges ‘Grand challenges’ Support the creation of large-scale community resources Corpora: News, Recordings, Video Annotation: Expert questions, labeled answers,.. Develop methodologies to evaluate state-of-the-art Retrieval, Machine Translation, etc Facilitate technology/knowledge transfer b/t industry/acad.

Shared Task Evaluations Goals: Lofty: Focus research community on key challenges ‘Grand challenges’ Support the creation of large-scale community resources Corpora: News, Recordings, Video Annotation: Expert questions, labeled answers,.. Develop methodologies to evaluate state-of-the-art Retrieval, Machine Translation, etc

Shared Task Evaluation Goals: Pragmatic:

Shared Task Evaluation Goals: Pragmatic: Head-to-head comparison of systems/techniques Same data, same task, same conditions, same timing

Shared Task Evaluation Goals: Pragmatic: Head-to-head comparison of systems/techniques Same data, same task, same conditions, same timing Centralizes funding, effort

Shared Task Evaluation Goals: Pragmatic: Head-to-head comparison of systems/techniques Same data, same task, same conditions, same timing Centralizes funding, effort Requires disclosure of techniques in exchange for data Base:

Shared Task Evaluation Goals: Pragmatic: Head-to-head comparison of systems/techniques Same data, same task, same conditions, same timing Centralizes funding, effort Require disclosure of techniques in exchange for data Base: Bragging rights

Shared Task Evaluation Goals: Pragmatic: Head-to-head comparison of systems/techniques Same data, same task, same conditions, same timing Centralizes funding, effort Require disclosure of techniques in exchange for data Base: Bragging rights Government research funding decisions

Shared Tasks: Perspective Late ‘80s-90s:

Shared Tasks: Perspective Late ‘80s-90s: ATIS: spoken dialog systems MUC: Message Understanding: information extraction

Shared Tasks: Perspective Late ‘80s-90s: ATIS: spoken dialog systems MUC: Message Understanding: information extraction TREC (Text Retrieval Conference) Arguably largest ( often >100 participating teams) Longest running (1992-current) Information retrieval (and related technologies) Actually hasn’t had ‘ad-hoc’ since ~2000, though Organized by NIST

TREC Tracks Track: Basic task organization

TREC Tracks Track: Basic task organization Previous tracks: Ad-hoc – Basic retrieval from fixed document set

TREC Tracks Track: Basic task organization Previous tracks: Ad-hoc – Basic retrieval from fixed document set Cross-language – Query in one language, docs in other English, French, Spanish, Italian, German, Chinese, Arabic

TREC Tracks Track: Basic task organization Previous tracks: Ad-hoc – Basic retrieval from fixed document set Cross-language – Query in one language, docs in other English, French, Spanish, Italian, German, Chinese, Arabic Genomics

TREC Tracks Track: Basic task organization Previous tracks: Ad-hoc – Basic retrieval from fixed document set Cross-language – Query in one language, docs in other English, French, Spanish, Italian, German, Chinese, Arabic Genomics Spoken Document Retrieval

TREC Tracks Track: Basic task organization Previous tracks: Ad-hoc – Basic retrieval from fixed document set Cross-language – Query in one language, docs in other English, French, Spanish, Italian, German, Chinese, Arabic Genomics Spoken Document Retrieval Video search

TREC Tracks Track: Basic task organization Previous tracks: Ad-hoc – Basic retrieval from fixed document set Cross-language – Query in one language, docs in other English, French, Spanish, Italian, German, Chinese, Arabic Genomics Spoken Document Retrieval Video search Question Answering

Current TREC tracks TREC 2011: Chemical IR Crowdsourcing (Web) Entity Legal Medical records Microblog Session Web

Other Shared Tasks International: CLEF (Europe); NTCIR (Japan); FIRE (India)

Other Shared Tasks International: CLEF (Europe); NTCIR (Japan); FIRE (India) Other NIST: DUC (Document Summarization) Machine Translation Topic Detection & Tracking

Other Shared Tasks International: CLEF (Europe); NTCIR (Japan); FIRE (India) Other NIST: DUC (Document Summarization) Machine Translation Topic Detection & Tracking Various: CoNLL (NE, parsing,..); SENSEVAL: WSD; PASCAL (morphology); BioNLP (biological entities, relations) Mediaeval (multi-media information access)

Other Shared Tasks International: CLEF (Europe); NTCIR (Japan); FIRE (India) Other NIST: DUC (Document Summarization) Machine Translation Topic Detection & Tracking Various: CoNLL (NE, parsing,..); SENSEVAL: WSD; PASCAL (morphology); BioNLP (biological entities, relations)

TREC Question-Answering Several years ( ) Started with pure factoid questions from news sources

TREC Question-Answering Several years ( ) Started with pure factoid questions from news sources Extended to lists, relationship

TREC Question-Answering Several years ( ) Started with pure factoid questions from news sources Extended to lists, relationship Extended to blog data Employed question series

TREC Question-Answering Several years ( ) Started with pure factoid questions from news sources Extended to lists, relationship Extended to blog data Employed question series Final: ‘complex, interactive’ evaluation

TREC Question-Answering Provides: Lists of questions Document collections (licensed via LDC) Ranked document results Evaluation tools: Answer verification patterns Derived resources: E.g. Roth and Li’s question categories, training/test Reams of related publications

Questions Number: 894 Description: How far is it from Denver to Aspen?

Questions Number: 894 Description: How far is it from Denver to Aspen? Number: 895 Description: What county is Modesto, California in?

Documents APW NEWS STORY :05 19 charged with drug trafficking UTICA, N.Y. (AP) - Nineteen people involved in a drug trafficking ring in the Utica area were arrested early Wednesday, police said. Those arrested are linked to 22 others picked up in May and comprise ''a major cocaine, crack cocaine and marijuana distribution organization,'' according to the U.S. Department of Justice.

Answer Keys 1394: French 1395: Nicole Kidman 1396: Vesuvius 1397: 62, : : Brigadoon

Reminder Team up!