IJCAI 2003 Workshop on Learning Statistical Models from Relational Data First-Order Probabilistic Models for Information Extraction Advisor: Hsin-His Chen.

Slides:



Advertisements
Similar presentations
Ontology-Based Computing Kenneth Baclawski Northeastern University and Jarg.
Advertisements

Slide 1 of 18 Uncertainty Representation and Reasoning with MEBN/PR-OWL Kathryn Blackmond Laskey Paulo C. G. da Costa The Volgenau School of Information.
Pat Langley Computational Learning Laboratory Center for the Study of Language and Information Stanford University, Stanford, California USA
Modelling with expert systems. Expert systems Modelling with expert systems Coaching modelling with expert systems Advantages and limitations of modelling.
Autonomic Scaling of Cloud Computing Resources
Big Ideas in Cmput366. Search Blind Search State space representation Iterative deepening Heuristic Search A*, f(n)=g(n)+h(n), admissible heuristics Local.
Rulebase Expert System and Uncertainty. Rule-based ES Rules as a knowledge representation technique Type of rules :- relation, recommendation, directive,
Proceedings of the Conference on Intelligent Text Processing and Computational Linguistics (CICLing-2007) Learning for Semantic Parsing Advisor: Hsin-His.
1/1/ A Knowledge-based Approach to Citation Extraction Min-Yuh Day 1,2, Tzong-Han Tsai 1,3, Cheng-Lung Sung 1, Cheng-Wei Lee 1, Shih-Hung Wu 4, Chorng-Shyong.
Markov Logic Networks Instructor: Pedro Domingos.
Hidden Markov Models Reading: Russell and Norvig, Chapter 15, Sections
Relational Learning with Gaussian Processes By Wei Chu, Vikas Sindhwani, Zoubin Ghahramani, S.Sathiya Keerthi (Columbia, Chicago, Cambridge, Yahoo!) Presented.
Natural Language and Speech Processing Creation of computational models of the understanding and the generation of natural language. Different fields coming.
Automatic Classification of Accounting Literature Nineteenth Annual Strategic and Emerging Technologies Workshop Vasundhara Chakraborty, Victoria Chiu,
A Probabilistic Framework for Information Integration and Retrieval on the Semantic Web by Livia Predoiu, Heiner Stuckenschmidt Institute of Computer Science,
PR-OWL: A Framework for Probabilistic Ontologies by Paulo C. G. COSTA, Kathryn B. LASKEY George Mason University presented by Thomas Packer 1PR-OWL.
LEARNING FROM OBSERVATIONS Yılmaz KILIÇASLAN. Definition Learning takes place as the agent observes its interactions with the world and its own decision-making.
CSE 574 – Artificial Intelligence II Statistical Relational Learning Instructor: Pedro Domingos.
Statistical Relational Learning for Link Prediction Alexandrin Popescul and Lyle H. Unger Presented by Ron Bjarnason 11 November 2003.
Big Ideas in Cmput366. Search Blind Search Iterative deepening Heuristic Search A* Local and Stochastic Search Randomized algorithm Constraint satisfaction.
CSE 574: Artificial Intelligence II Statistical Relational Learning Instructor: Pedro Domingos.
EXPERT SYSTEMS Part I.
Scalable Text Mining with Sparse Generative Models
Building Knowledge-Driven DSS and Mining Data
Statistical Relational Learning Pedro Domingos Dept. Computer Science & Eng. University of Washington.
CS Machine Learning. What is Machine Learning? Adapt to / learn from data  To optimize a performance function Can be used to:  Extract knowledge.
Advisor: Hsin-Hsi Chen Reporter: Chi-Hsin Yu Date:
How to make a presentation (Oral and Poster) Dr. Bernard Chen Ph.D. University of Central Arkansas July 5 th Applied Research in Healthy Information.
Bayesian parameter estimation in cosmology with Population Monte Carlo By Darell Moodley (UKZN) Supervisor: Prof. K Moodley (UKZN) SKA Postgraduate conference,
Relational Probability Models Brian Milch MIT 9.66 November 27, 2007.
1 A Bayesian Method for Guessing the Extreme Values in a Data Set Mingxi Wu, Chris Jermaine University of Florida September 2007.
Artificial Intelligence
Probability and Statistics Required!. 2 Review Outline  Connection to simulation.  Concepts to review.  Assess your understanding.  Addressing knowledge.
Bayesian Hierarchical Clustering Paper by K. Heller and Z. Ghahramani ICML 2005 Presented by HAO-WEI, YEH.
Pattern-directed inference systems
IRCS/CCN Summer Workshop June 2003 Speech Recognition.
14 October, 2010LRI Seminar 2010 (Univ. Paris-Sud)1 Statistical performance analysis by loopy belief propagation in probabilistic image processing Kazuyuki.
Randomized Algorithms for Bayesian Hierarchical Clustering
1 2010/2011 Semester 2 Introduction: Chapter 1 ARTIFICIAL INTELLIGENCE.
Ontology-Based Computing Kenneth Baclawski Northeastern University and Jarg.
Date : 2013/03/18 Author : Jeffrey Pound, Alexander K. Hudek, Ihab F. Ilyas, Grant Weddell Source : CIKM’12 Speaker : Er-Gang Liu Advisor : Prof. Jia-Ling.
BLOG: Probabilistic Models with Unknown Objects Brian Milch, Bhaskara Marthi, Stuart Russell, David Sontag, Daniel L. Ong, Andrey Kolobov University of.
A Word Clustering Approach for Language Model-based Sentence Retrieval in Question Answering Systems Saeedeh Momtazi, Dietrich Klakow University of Saarland,Germany.
Inferring High-Level Behavior from Low-Level Sensors Donald J. Patterson, Lin Liao, Dieter Fox, and Henry Kautz.
Markov Chain Monte Carlo for LDA C. Andrieu, N. D. Freitas, and A. Doucet, An Introduction to MCMC for Machine Learning, R. M. Neal, Probabilistic.
Date: 2013/6/10 Author: Shiwen Cheng, Arash Termehchy, Vagelis Hristidis Source: CIKM’12 Advisor: Jia-ling Koh Speaker: Chen-Yu Huang Predicting the Effectiveness.
IN THE NAME OF GOD. Reference Citing Software.
ECE 8443 – Pattern Recognition ECE 8527 – Introduction to Machine Learning and Pattern Recognition Objectives: Elements of a Discrete Model Evaluation.
Motivation and Overview
A Portrait of the Semantic Web in Action Jeff Heflin and James Hendler IEEE Intelligent Systems December 6, 2010 Hyewon Lim.
1 Scalable Probabilistic Databases with Factor Graphs and MCMC Michael Wick, Andrew McCallum, and Gerome Miklau VLDB 2010.
Refined Online Citation Matching and Adaptive Canonical Metadata Construction CSE 598B Course Project Report Huajing Li.
Pattern Recognition NTUEE 高奕豪 2005/4/14. Outline Introduction Definition, Examples, Related Fields, System, and Design Approaches Bayesian, Hidden Markov.
Daphne Koller Introduction Motivation and Overview Probabilistic Graphical Models.
GoRelations: an Intuitive Query System for DBPedia Lushan Han and Tim Finin 15 November 2011
COMPUTER SYSTEM FUNDAMENTAL Genetic Computer School INTRODUCTION TO ARTIFICIAL INTELLIGENCE LESSON 11.
Probabilistic Reasoning Inference and Relational Bayesian Networks.
Network Management Lecture 13. MACHINE LEARNING TECHNIQUES 2 Dr. Atiq Ahmed Université de Balouchistan.
Warren Shen, Xin Li, AnHai Doan Database & AI Groups University of Illinois, Urbana Constraint-Based Entity Matching.
Learning Bayesian Networks for Complex Relational Data
Artificial Intelligence
School of Computer Science & Engineering
Intelligent Systems (AI-2) Computer Science cpsc422, Lecture 14
Logic for Artificial Intelligence
TA : Mubarakah Otbi, Duaa al Ofi , Huda al Hakami
Intelligent Systems (AI-2) Computer Science cpsc422, Lecture 14
Subject : Artificial Intelligence
Statistical Relational AI
Chapter 14 February 26, 2004.
Intelligent Systems (AI-2) Computer Science cpsc422, Lecture 14
Presentation transcript:

IJCAI 2003 Workshop on Learning Statistical Models from Relational Data First-Order Probabilistic Models for Information Extraction Advisor: Hsin-His Chen Reporter: Chi-Hsin Yu Date: Bhaskara Marthi, Brian Milch, Stuart Russell Computer Science Div. University of California NIPS 15th, 2003 Identity Uncertainty and Citation Matching Hanna Pasula, Bhaskara Marthi, Brian Milch, Stuart Russell, Ilya Shpitser Computer Science Div. University of California

Outlines  Introduction  Related works  Models for the bibliography domain  Experiment on model A  Desiderata for a FOPL  Conclusions 2/18

Introduction – Citation Matching Problem  Citation matching: the problem of deciding which citations correspond to the same publication  Difficulties Different citation styles An imperfect copy of the book ’ s title Different ways to refer an object (identity) Ambiguity  “ Wauchope, K. Eucalyptus: Integrating Natural language Input with a Graphical User Interface ” Author: “ Wauchope, K. Eucalyptus ” or “ Wauchope, K. ” ?  Tasks Parsing Disambiguation Matching 3/ 18

Introduction – Citation Matching Problem: Examples 4/ 18 Journal of Artificial Intelligence Research, or Artificial Intelligence Journal ??

Introduction – First-Order Probabilistic Models 5/ 18 LogicProbabilistic Model Propositio nal Formula A  BP(j  m  a  b  e), Bayesian Network Inference/ Algorithms Resolution, Model Checking, Forward chaining, DPLL, WalkSAT … Bayes ’ rule, Summing-out, smoothing, prediction, approximation (likely-hood, MCMC … ), …Summing-out First-orderFormula x King(x)  Greedy(x)  Evil(x) = Inference/ Algorithms Unification, Resolution, … Learning, Approximation, … System/ Languages Prolog, Rule Engine (JBoss), … FOPL, RPMRPM

Introduction – Probabilistic Model: Inference Back to Introduction – First-Order Probabilistic Models

Introduction – Bayesian Network Back to Introduction – First-Order Probabilistic Models

Introduction – Relational Probabilistic Model Back to Introduction – First-Order Probabilistic Models Compare to: Semantic network Object-Oriented DB

Introduction – Result of Model B 6/ 18

Related Works  IE the Message Understanding Conferences [DARPA,1998]  Bayesian modeling finding stochastically repeated patterns (motifs) in DNA sequences [Xing et al., 2003] Robot localization [Anguelov et al., 2002]  FOPL/RPM (Relational Prob. Model) A. Pfeffer. Probabilistic Reasoning for Complex Systems. PhD thesis, Stanford, / 18

Models for the Bibliography Domain – Model A  [Pasula et al. 2003] 8/ 18

Models for the Bibliography Domain – Model A (Cont.)  Suggest a declarative approach to identity uncertainty using a formal language  Algorithm Steps  Generate objects/instances  Parse and fill attributes  Inference (Approximation, MCMC) Cluster the identity (publication) 9/ 18

Models for the Bibliography Domain – Model A (Cont.)  Attributes using unconditional probability learn several bigram models  letter-based models of first names, surnames, and title words using the following resources  the 2000 Census data on US names  a large A.I. BibTeX bibliography  a hand-parsed collection of 500 citations  Attributes using conditional probability Using noise channels for some attributes  the corruption models of Citation.obsTitle, AuthorAsCited.surname, and AuthorAsCited.fnames  The parameters of the corruption models are learnt online, using stochastic EM Citation.parse  It keeps track of the segmentation of Citation.text  An author segment, a title segment, and three filler segments (one before, one after, and one in between) Citation.text  Be constrained by Citation.parse, Paper.pubType, … These models were learned using our pre-segmented file. 10/ 18

Models for the Bibliography Domain – Model B 11/ 18 Citation Publication TitleAsCited AuthorsAsCited Text Parse Collection Name, Type, Date, Publisher Name, City Authors Name Area+ (Fields) Publication Title Area Type (Book/conf. … ) AuthorList Collection Citation Groups Type (Area, Author) Style PublicationList CitationList

Models for the Bibliography Domain – Model B (Cont.)  Generating objects The set of Author objects, and the set of Collection objects are generated independently. the set of Publication objects is generated conditional on the Authors and Collections. CitationGroup objects are generated conditional on the Authors and Collections. Citation objects are generated from the CitationGroups. 12/ 18

Models for the Bibliography Domain – Model B (Cont.)  Fill attributes Author.Name  is chosen from a mixture of a letter bigram distribution with a distribution that chooses from a set of commonly occurring names Publications.Title  is generated from an n-gram model, conditioned on Publications.area More specific relations and conditions between attributes 13/ 18

Experiment on model A – Experiment Setting  Dataset Citeseer ’ s hand-matched datasets Each of these datasets contains several hundred citations of machine learning papers  Citeseer ’ s phrase matching algorithm a greedy agglomerative clustering method  based on a metric that measures the degrees to which the words and phrases of any two citations overlap half of them in clusters ranging in size from two to twenty-one citations 14/ 18

Experiment on model A – Experiment Result 15/ 18

Desiderata for a FOPL  Contains A probability distribution over possible worlds The expression power to model the relational structure of the world An efficient inference algorithm A learning procedure which allows priors over the parameters  Has the ability to answer queries to make inferences about the existence or nonexistence of objects having particular properties to represent common types of compound objects to represent probabilistic dependencies to incorporate domain knowledge into the inference algorithms 16/ 18

Conclusions  First-order probabilistic models a useful, probably necessary, component of any system that extracts complex relational information from unstructured text data  Some of the directions we plan to pursue in the future defining a representation language that allows such models to be specified declaratively, scaling up the inference procedure to handle large knowledge bases 17/ 18

Thanks!!