Building a Timeline Action Network for Evacuation in Disaster

Size: px
Start display at page:

Download "Building a Timeline Action Network for Evacuation in Disaster"

Transcription

1 Building a Timeline Action Network for Evacuation in Disaster The-Minh Nguyen, Takahiro Kawamura, Yasuyuki Tahara, and Akihiko Ohsuga Graduate School of Information Systems, University of Electro-Communications, Japan

2

3 Objective Building a service, which recommends action sequence based on social sensors

4 Goal: Timeline Action Network train was stopped 15:13:00 waited a taxi but failed 17:13:00-1 took a bus to go to Shibuya station 18:20:00 walked from Shinjuku station 19:52:09 Success: 9 walked from Shibuya station 20:20:00 received mobile battery at Bic Camera 20:40:00 took refuge 20:22:09 Success: 8

5 Challenge 1: How to extract human activity? In the earthquake, phone lines got stuck, and then twitter was used to exchange information about evacuation, traffic (33 million tweets/day) However, tweets are noisy and syntactically incorrect Challenge 2: How to represent action sequence? To recommend the action sequence, the meaning of the network needs to be understood by both of computer and the human user Challenge 3: How to predict missing activity? Users do NOT post all of their activities in real-time 17:00 20:00 23:00 Roppongi act 1? act 4 Where? Shinjuku

6 Approach 1: Definition of Activity Attributes Attributes to express the human activity time (when) location (where) Relation 1 transition 2 cause object (what) actor (who) Label next act (how) becauseof Tanaka came back home after escaping from the earthquake earthquake escape what who Tanaka next home what come back

7 Approach 1: Difficulty of extraction Frequency or co-occurrence miss most of activities Frequency of activities related to Akihabara electric town Co-occurrence frequency of action and object Consider activity extraction as a sequence labeling problem Output (label) B-What O B-Action O Input ipad を買うよ ipad buy (buy an ipad) B: Begin a phrase, I: Inside a phrase, O: not in a phrase In labeling problem, CRF outperform both MEMMs and HMMs on a number of tasks: text chunking, named entity recognition etc.

8 Approach 1: Process of Activity Extraction 1 NLP method: -deep linguistic parser -syntax patterns -Google Map API parable sentences the other sentences Preprocessing: 3 -remove noise data -convert to simpler sentences Training data 2 Learning model Extract labels of activity attributes 4 Add more training data Feature model of training data Make training data for twitter Extract activity

9 Approach 1: Making training data for twitter easy parsing sentences (initial set of sample data) and twitter and twitter API

10 Approach 2: Design of Timeline Action Network Based on Web Ontology Language (OWL) RDF/N3: compact and readable alternative to RDF s XML Inherit external geo: vcard: tl: < C, I, L are class, instance, label respectively

11 Approach 2: Example of semantic description The train has stopped at Akihabara Station at 16:13:00. :stop a :ActClass ; rdfs:label "stop"@en. :akihabara_station a :TrainStationClass ; rdfs:label Akihabara ; vcard:region ; vcard:locality "Chiyoda-ku"@en ; vcard:street-address " Sotokanda"@en ; geo:lat ; geo:long :act_01 a :ActionClass ; :act :stop ; :what :"train"@en ; :where :akihabara_station ; tl:start T16:13:00^^xsd:dateTime.

12 Approach 1&2: Building of Timeline Action Network Using #jishin (#earthquake) tag to extract activity sentences which relate to earthquake e.g. Earthquake M9.0 was just occurred ( :47) I am taking refuge at Akihabara ( :10) Extract activity attributes Activity ID (Who, Act, What, Where, When) act01 (Null, occur, earthquake M9.0, Null, :47) act02 (I, take refuge, Null, Akihabara, :10) Convert to RDF/N3 :act01 a :ActionClass ; :act :occur ; :what "earthquake M9.0"@en ; tl:start " T14:47:00"^^xsd:dateTime. act02 a :ActionClass ; :act :take_refuge ; :where :Akihabara ; tl:start " T15:10:53"^^xsd:dateTime.

13 Approach 3: Prediction of Missing Activity Can act = {act 1, act 2,.., act t, } is set of candidate actions at t t time P ua act t? 1 based on similar users 2 based on action sequence probability of act t Correspond to user-based collaborative filtering (CF) Correspond to item-based CF We can use CF approach to calculate probability of the active user u a did action act t at time t (Pu a act t )

14 Approach 3: Difficulty of Prediction Traditional CF is trying to recommend interested items on EC site for users Our work is trying to predict missing action in real-world Complexity point of view items on internet actions in real-world Dependence 1 variable 4 variables (act, object, time, location) location NO YES transition weak strong goal NO YES (e.g. want to evaluate) Continuity NO YES It is NOT suitable to use traditional CF for our work

15 Approach 3: Prediction based on user similarity Assumption: (a) similar users also did before/after actions (b) if users had the same goal (e.g. wanted to evacuate in Shinjuku), then they had the same action patterns (c) users did the same action if they were in the same location Similarity between u j & u a : (a) did before/after actions or NOT S(u j,u a ) = βdid({a before,l before },{a after,l after }) + γ Target(a t,l t ) + (1 β γ)samelocation(l) (b) had the same goal or NOT (c) were in the same location or NOT

16 Approach 3: Prediction based on Action Sequence Assumption: Action depends on (a) time (b) location (c) before/after action a before act t a after Probability of act t at time t : (c) frequency of a before act t (c)frequency of act t a after P(act t ) = ρ a + ρ t {F(a F(act before t act,t) + (1 ρ t ) + a ρ F(act t t )F(act t a,l) after )} (a) frequency of act t at time t (b) frequency of act t in location l

17 Approach 3: Prediction based on execution time Assumption: Same actions need same execution time active user u a before action? after action time time_1 similar user u j before action time_b act 1 act 2 after action time time_t t j1 t jb t ab t t j2 tj Probability of act 1 of user u j : time_t time_b T(u j,act1) = 1 time_1 diff if (0 diff else diff < 0 time_1) b diff = time_t time_b

18 Approach 3: Handling Minority Action (1/2) Assumption: Minor actions which were not frequent but successful, should be weighted E.g. it was a good decision when staying at company feedback minor action Action stay at company should be weighted, using NLP to extract feedbacks

19 Approach 3: Handling Minority Action (2/2) Probability of u a who did a successful action act t : A. Proportional to percentage of successful actions of u a B. Proportional to percentage of good feedbacks of act t A. B. action success number of success act before 0 0 act act act act after 0 0 f(u a ) = Success(act 2 ) = DidSuccess ua act t = f(u a )*Success(act t )

20 Approach 3: Prediction of Missing Activity Combination of CF + NLP : NLP P * u a act t = f(u a )*Success(act t ) + CF P ua act t P u a act t ω(u,act j j = 1,L = α P(actt ) (1 α) P(act ) 1 user based i t ) S(u L j,u a ) T(u 2 action based 3 execution time based j,act ) + 0 α, β, γ, ρ a, ρ t 1 L is number of all users similar to u a ω(u j, act t ) is a weighting factor t

21 Evaluation: Data set 132,244 tweets which were posted by users in Tokyo from 2011/03/11 to 2011/03/12 416,463 tweets which related to the massive Tohoku earthquake

22 Evaluation: Activity Extraction method activity actor act object time location NLP 81.17% 86.32% 98.13% 84.14% 87.96% 88.25% Precision CRF 57.89% 72.79% 82.98% 67.01% 76.40% 80.20% CRF w/ twitter data 73.21% 82.25% 97.11% 81.23% 80.04% 82.11% NLP 23.86% 26.38% 28.87% 24.77% 26.20% 26.02% Recall CRF 51.13% 69.13% 90.23% 62.11% 73.51% 77.67% CRF w/ twitter data 66.54% 80.11% 93.18% 76.57% 79.75% 81.02% NLP 36.88% 40.41% 44.61% 38.27% 40.37% 40.19% F-measure CRF 54.30% 70.91% 86.45% 64.47% 74.93% 78.91% CRF w/ twitter data 69.72% 81.17% 95.10% 78.83% 79.89% 81.56% NLP has high precision but low recall F-measure is improved by adding more training data and doing pre-processing NLP: use of NLP tools alone CRF: use of CRF, but not trained by twitter data

23 Evaluation: Prediction of Missing Activity Repeat the followings for action data of 3,900 users 10 times 1. Randomly select 39 users (10%) 2. Randomly delete activity data of the above users 3. Try to predict the deleted activity from the other data Method action location action & location Baseline 31.48% 43.09% 27.56% CF+NLP 69.23% 76.92% 43.59% Each attribute has high accuracy, but Accuracy of their combination is still low, and requires the introduction of combinational prediction metrics Baseline: Look up the most similar user, and predict missing activity only from the user

24 Application: Timeline Action Network (1/2) Android App. which recommend action sequence for evacuation based on the timeline action network. Input(time, location) Output (action patterns)

25 Application: Timeline Action Network (2/2)

26 Conclusion Objective: Building a service, which recommends action sequence based on social sensors Challenge: 1. How to extract human activity? 2. How to represent action sequence? 3. How to predict missing activity? Approach: 1. Human Activity Extraction from twitter 2. Building of Timeline Action Network 3. Predict Missing Activity Future Issue: Not only the earthquake, we would apply this method to other disasters and business use, such as typhoon, train accident, travel recommendation, and behavior-based marketing, etc.

Mobility Analytics through Social and Personal Data. Pierre Senellart

Mobility Analytics through Social and Personal Data. Pierre Senellart Mobility Analytics through Social and Personal Data Pierre Senellart Session: Big Data & Transport Business Convention on Big Data Université Paris-Saclay, 25 novembre 2015 Analyzing Transportation and

More information

Twitter s Effectiveness on Blackout Detection during Hurricane Sandy

Twitter s Effectiveness on Blackout Detection during Hurricane Sandy Twitter s Effectiveness on Blackout Detection during Hurricane Sandy KJ Lee, Ju-young Shin & Reza Zadeh December, 03. Introduction Hurricane Sandy developed from the Caribbean stroke near Atlantic City,

More information

Measurement of human activity using velocity GPS data obtained from mobile phones

Measurement of human activity using velocity GPS data obtained from mobile phones Measurement of human activity using velocity GPS data obtained from mobile phones Yasuko Kawahata 1 Takayuki Mizuno 2 and Akira Ishii 3 1 Graduate School of Information Science and Technology, The University

More information

Detecting Heavy Rain Disaster from Social and Physical Sensors

Detecting Heavy Rain Disaster from Social and Physical Sensors Detecting Heavy Rain Disaster from Social and Physical Sensors Tomoya Iwakura, Seiji Okajima, Nobuyuki Igata Kuhihiro Takeda, Yuzuru Yamakage, Naoshi Morita FUJITSU LABORATORIES LTD. {tomoya.iwakura, okajima.seiji,

More information

Natural Disaster :.JP s Experience and Preparation

Natural Disaster :.JP s Experience and Preparation Natural Disaster :.JP s Experience and Preparation 14 March. 2018 Hiro Hotta 1 Where s Japan Japan Puerto Rico earthquakes large enough to feel : 2,000-20,000 times a year typhoons disastrous

More information

The Noisy Channel Model and Markov Models

The Noisy Channel Model and Markov Models 1/24 The Noisy Channel Model and Markov Models Mark Johnson September 3, 2014 2/24 The big ideas The story so far: machine learning classifiers learn a function that maps a data item X to a label Y handle

More information

JP experience of earthquake, tsunami, and nuclear plant accident

JP experience of earthquake, tsunami, and nuclear plant accident JP experience of earthquake, tsunami, and nuclear plant accident ccnso Members Meeting 21 June 2011 Hiro Hotta What happened (1) Earthquake Friday, March 11, 2011 14:46 JST Earthquake

More information

Introduction to Logistic Regression

Introduction to Logistic Regression Introduction to Logistic Regression Guy Lebanon Binary Classification Binary classification is the most basic task in machine learning, and yet the most frequent. Binary classifiers often serve as the

More information

Your Virtual Workforce. On Demand. Worldwide. COMPANY PRESENTATION. clickworker GmbH 2017

Your Virtual Workforce. On Demand. Worldwide. COMPANY PRESENTATION. clickworker GmbH 2017 Your Virtual Workforce. On Demand. Worldwide. COMPANY PRESENTATION 2017 CLICKWORKER AT A GLANCE Segment: Paid Crowdsourcing / Microtasking Services: Text Creation (incl. SEO Texts), AI-Training Data, Internet

More information

Hurricane Dennis menaces U.S. coast

Hurricane Dennis menaces U.S. coast www.breaking News English.com Ready-to-use ESL / EFL Lessons Hurricane Dennis menaces U.S. coast URL: http://www.breakingnewsenglish.com/0507/050711-dennis-e.html Today s contents The Article 2 Warm-ups

More information

TALP at GeoQuery 2007: Linguistic and Geographical Analysis for Query Parsing

TALP at GeoQuery 2007: Linguistic and Geographical Analysis for Query Parsing TALP at GeoQuery 2007: Linguistic and Geographical Analysis for Query Parsing Daniel Ferrés and Horacio Rodríguez TALP Research Center Software Department Universitat Politècnica de Catalunya {dferres,horacio}@lsi.upc.edu

More information

Information Extraction from Text

Information Extraction from Text Information Extraction from Text Jing Jiang Chapter 2 from Mining Text Data (2012) Presented by Andrew Landgraf, September 13, 2013 1 What is Information Extraction? Goal is to discover structured information

More information

Discovering Geographical Topics in Twitter

Discovering Geographical Topics in Twitter Discovering Geographical Topics in Twitter Liangjie Hong, Lehigh University Amr Ahmed, Yahoo! Research Alexander J. Smola, Yahoo! Research Siva Gurumurthy, Twitter Kostas Tsioutsiouliklis, Twitter Overview

More information

Ontology-Based News Recommendation

Ontology-Based News Recommendation Ontology-Based News Recommendation Wouter IJntema Frank Goossen Flavius Frasincar Frederik Hogenboom Erasmus University Rotterdam, the Netherlands frasincar@ese.eur.nl Outline Introduction Hermes: News

More information

Statistical Methods for NLP

Statistical Methods for NLP Statistical Methods for NLP Sequence Models Joakim Nivre Uppsala University Department of Linguistics and Philology joakim.nivre@lingfil.uu.se Statistical Methods for NLP 1(21) Introduction Structured

More information

Regularity and Conformity: Location Prediction Using Heterogeneous Mobility Data

Regularity and Conformity: Location Prediction Using Heterogeneous Mobility Data Regularity and Conformity: Location Prediction Using Heterogeneous Mobility Data Yingzi Wang 12, Nicholas Jing Yuan 2, Defu Lian 3, Linli Xu 1 Xing Xie 2, Enhong Chen 1, Yong Rui 2 1 University of Science

More information

Mining coreference relations between formulas and text using Wikipedia

Mining coreference relations between formulas and text using Wikipedia Mining coreference relations between formulas and text using Wikipedia Minh Nghiem Quoc 1, Keisuke Yokoi 2, Yuichiroh Matsubayashi 3 Akiko Aizawa 1 2 3 1 Department of Informatics, The Graduate University

More information

Lecture 5 Neural models for NLP

Lecture 5 Neural models for NLP CS546: Machine Learning in NLP (Spring 2018) http://courses.engr.illinois.edu/cs546/ Lecture 5 Neural models for NLP Julia Hockenmaier juliahmr@illinois.edu 3324 Siebel Center Office hours: Tue/Thu 2pm-3pm

More information

DeepLog: Anomaly Detection and Diagnosis from System Logs through Deep Learning. Min Du, Feifei Li, Guineng Zheng, Vivek Srikumar University of Utah

DeepLog: Anomaly Detection and Diagnosis from System Logs through Deep Learning. Min Du, Feifei Li, Guineng Zheng, Vivek Srikumar University of Utah DeepLog: Anomaly Detection and Diagnosis from System Logs through Deep Learning Min Du, Feifei Li, Guineng Zheng, Vivek Srikumar University of Utah Background 2 Background System Event Log 3 Background

More information

Driving Semantic Parsing from the World s Response

Driving Semantic Parsing from the World s Response Driving Semantic Parsing from the World s Response James Clarke, Dan Goldwasser, Ming-Wei Chang, Dan Roth Cognitive Computation Group University of Illinois at Urbana-Champaign CoNLL 2010 Clarke, Goldwasser,

More information

Statistical NLP for the Web

Statistical NLP for the Web Statistical NLP for the Web Neural Networks, Deep Belief Networks Sameer Maskey Week 8, October 24, 2012 *some slides from Andrew Rosenberg Announcements Please ask HW2 related questions in courseworks

More information

A Framework of Detecting Burst Events from Micro-blogging Streams

A Framework of Detecting Burst Events from Micro-blogging Streams , pp.379-384 http://dx.doi.org/10.14257/astl.2013.29.78 A Framework of Detecting Burst Events from Micro-blogging Streams Kaifang Yang, Yongbo Yu, Lizhou Zheng, Peiquan Jin School of Computer Science and

More information

CLRG Biocreative V

CLRG Biocreative V CLRG ChemTMiner @ Biocreative V Sobha Lalitha Devi., Sindhuja Gopalan., Vijay Sundar Ram R., Malarkodi C.S., Lakshmi S., Pattabhi RK Rao Computational Linguistics Research Group, AU-KBC Research Centre

More information

Applied Natural Language Processing

Applied Natural Language Processing Applied Natural Language Processing Info 256 Lecture 7: Testing (Feb 12, 2019) David Bamman, UC Berkeley Significance in NLP You develop a new method for text classification; is it better than what comes

More information

CS460/626 : Natural Language Processing/Speech, NLP and the Web

CS460/626 : Natural Language Processing/Speech, NLP and the Web CS460/626 : Natural Language Processing/Speech, NLP and the Web Lecture 23: Binding Theory Pushpak Bhattacharyya CSE Dept., IIT Bombay 8 th Oct, 2012 Parsing Problem Semantics Part of Speech Tagging NLP

More information

Analysis Based on SVM for Untrusted Mobile Crowd Sensing

Analysis Based on SVM for Untrusted Mobile Crowd Sensing Analysis Based on SVM for Untrusted Mobile Crowd Sensing * Ms. Yuga. R. Belkhode, Dr. S. W. Mohod *Student, Professor Computer Science and Engineering, Bapurao Deshmukh College of Engineering, India. *Email

More information

Geographical Bias on Social Media and Geo-Local Contents System with Mobile Devices

Geographical Bias on Social Media and Geo-Local Contents System with Mobile Devices 212 45th Hawaii International Conference on System Sciences Geographical Bias on Social Media and Geo-Local Contents System with Mobile Devices Kazunari Ishida Hiroshima Institute of Technology k.ishida.p7@it-hiroshima.ac.jp

More information

Department of Computer Science and Engineering Indian Institute of Technology, Kanpur. Spatial Role Labeling

Department of Computer Science and Engineering Indian Institute of Technology, Kanpur. Spatial Role Labeling Department of Computer Science and Engineering Indian Institute of Technology, Kanpur CS 365 Artificial Intelligence Project Report Spatial Role Labeling Submitted by Satvik Gupta (12633) and Garvit Pahal

More information

Applied Natural Language Processing

Applied Natural Language Processing Applied Natural Language Processing Info 256 Lecture 3: Finding Distinctive Terms (Jan 29, 2019) David Bamman, UC Berkeley https://www.nytimes.com/interactive/ 2017/11/07/upshot/modern-love-what-wewrite-when-we-write-about-love.html

More information

Maschinelle Sprachverarbeitung

Maschinelle Sprachverarbeitung Maschinelle Sprachverarbeitung Parsing with Probabilistic Context-Free Grammar Ulf Leser Content of this Lecture Phrase-Structure Parse Trees Probabilistic Context-Free Grammars Parsing with PCFG Other

More information

Maschinelle Sprachverarbeitung

Maschinelle Sprachverarbeitung Maschinelle Sprachverarbeitung Parsing with Probabilistic Context-Free Grammar Ulf Leser Content of this Lecture Phrase-Structure Parse Trees Probabilistic Context-Free Grammars Parsing with PCFG Other

More information

Yahoo! Labs Nov. 1 st, Liangjie Hong, Ph.D. Candidate Dept. of Computer Science and Engineering Lehigh University

Yahoo! Labs Nov. 1 st, Liangjie Hong, Ph.D. Candidate Dept. of Computer Science and Engineering Lehigh University Yahoo! Labs Nov. 1 st, 2012 Liangjie Hong, Ph.D. Candidate Dept. of Computer Science and Engineering Lehigh University Motivation Modeling Social Streams Future work Motivation Modeling Social Streams

More information

Geo-information and Disaster Risk Reduction in the Hindu Kush-Himalayan region

Geo-information and Disaster Risk Reduction in the Hindu Kush-Himalayan region Geo-information and Disaster Risk Reduction in the Hindu Kush-Himalayan region Basanta Shrestha, Division Head MENRIS, International Centre for Integrated Mountain Development Kathmandu, Nepal The International

More information

Spatial Role Labeling CS365 Course Project

Spatial Role Labeling CS365 Course Project Spatial Role Labeling CS365 Course Project Amit Kumar, akkumar@iitk.ac.in Chandra Sekhar, gchandra@iitk.ac.in Supervisor : Dr.Amitabha Mukerjee ABSTRACT In natural language processing one of the important

More information

SBI PO 2015 PRELIM (TIER I) EXAM MODEL PAPER SOLUTIONS

SBI PO 2015 PRELIM (TIER I) EXAM MODEL PAPER SOLUTIONS SBI PO 2015 PRELIM (TIER I) EXAM MODEL PAPER SOLUTIONS Part 1 SOLUTIONS ENGLISH LANGUAGE Answers (1-10) 1. (5) 2. (1) 3. (2) 4. (3) 5. (4) 6. (4) nurturing 7. (1) drive 8. (4) hurdle 9. (2) manageable

More information

Session-Based Queueing Systems

Session-Based Queueing Systems Session-Based Queueing Systems Modelling, Simulation, and Approximation Jeroen Horters Supervisor VU: Sandjai Bhulai Executive Summary Companies often offer services that require multiple steps on the

More information

D2D SALES WITH SURVEY123, OP DASHBOARD, AND MICROSOFT SSAS

D2D SALES WITH SURVEY123, OP DASHBOARD, AND MICROSOFT SSAS D2D SALES WITH SURVEY123, OP DASHBOARD, AND MICROSOFT SSAS EDWARD GAUSE, GISP DIRECTOR OF INFORMATION SERVICES (ENGINEERING APPS) HTC (HORRY TELEPHONE COOP.) EDWARD GAUSE, GISP DIRECTOR OF INFORMATION

More information

* Abstract. Keywords: Smart Card Data, Public Transportation, Land Use, Non-negative Matrix Factorization.

*  Abstract. Keywords: Smart Card Data, Public Transportation, Land Use, Non-negative Matrix Factorization. Analysis of Activity Trends Based on Smart Card Data of Public Transportation T. N. Maeda* 1, J. Mori 1, F. Toriumi 1, H. Ohashi 1 1 The University of Tokyo, 7-3-1 Hongo Bunkyo-ku, Tokyo, Japan *Email:

More information

A DOP Model for LFG. Rens Bod and Ronald Kaplan. Kathrin Spreyer Data-Oriented Parsing, 14 June 2005

A DOP Model for LFG. Rens Bod and Ronald Kaplan. Kathrin Spreyer Data-Oriented Parsing, 14 June 2005 A DOP Model for LFG Rens Bod and Ronald Kaplan Kathrin Spreyer Data-Oriented Parsing, 14 June 2005 Lexical-Functional Grammar (LFG) Levels of linguistic knowledge represented formally differently (non-monostratal):

More information

Advanced Natural Language Processing Syntactic Parsing

Advanced Natural Language Processing Syntactic Parsing Advanced Natural Language Processing Syntactic Parsing Alicia Ageno ageno@cs.upc.edu Universitat Politècnica de Catalunya NLP statistical parsing 1 Parsing Review Statistical Parsing SCFG Inside Algorithm

More information

CS626: NLP, Speech and the Web. Pushpak Bhattacharyya CSE Dept., IIT Bombay Lecture 14: Parsing Algorithms 30 th August, 2012

CS626: NLP, Speech and the Web. Pushpak Bhattacharyya CSE Dept., IIT Bombay Lecture 14: Parsing Algorithms 30 th August, 2012 CS626: NLP, Speech and the Web Pushpak Bhattacharyya CSE Dept., IIT Bombay Lecture 14: Parsing Algorithms 30 th August, 2012 Parsing Problem Semantics Part of Speech Tagging NLP Trinity Morph Analysis

More information

CS 6375 Machine Learning

CS 6375 Machine Learning CS 6375 Machine Learning Decision Trees Instructor: Yang Liu 1 Supervised Classifier X 1 X 2. X M Ref class label 2 1 Three variables: Attribute 1: Hair = {blond, dark} Attribute 2: Height = {tall, short}

More information

Spatial Data Science. Soumya K Ghosh

Spatial Data Science. Soumya K Ghosh Workshop on Data Science and Machine Learning (DSML 17) ISI Kolkata, March 28-31, 2017 Spatial Data Science Soumya K Ghosh Professor Department of Computer Science and Engineering Indian Institute of Technology,

More information

Collaborative NLP-aided ontology modelling

Collaborative NLP-aided ontology modelling Collaborative NLP-aided ontology modelling Chiara Ghidini ghidini@fbk.eu Marco Rospocher rospocher@fbk.eu International Winter School on Language and Data/Knowledge Technologies TrentoRISE Trento, 24 th

More information

Collaborative Recommendation with Multiclass Preference Context

Collaborative Recommendation with Multiclass Preference Context Collaborative Recommendation with Multiclass Preference Context Weike Pan and Zhong Ming {panweike,mingz}@szu.edu.cn College of Computer Science and Software Engineering Shenzhen University Pan and Ming

More information

Empirical Methods in Natural Language Processing Lecture 11 Part-of-speech tagging and HMMs

Empirical Methods in Natural Language Processing Lecture 11 Part-of-speech tagging and HMMs Empirical Methods in Natural Language Processing Lecture 11 Part-of-speech tagging and HMMs (based on slides by Sharon Goldwater and Philipp Koehn) 21 February 2018 Nathan Schneider ENLP Lecture 11 21

More information

WMO. Key Elements of PWS and Effective EWS. Haleh Haleh Kootval Chief, PWS Programme

WMO. Key Elements of PWS and Effective EWS. Haleh Haleh Kootval Chief, PWS Programme WMO Key Elements of PWS and Effective EWS Haleh Haleh Kootval Chief, PWS Programme Workshop Objectives This workshop is all about Service Delivery and becoming excellent at it through: Sharing experiences

More information

Collaborative Hotel Recommendation based on Topic and Sentiment of Review Comments

Collaborative Hotel Recommendation based on Topic and Sentiment of Review Comments DEIM Forum 2017 P6-2 Collaborative Hotel Recommendation based on Topic and Sentiment of Abstract Review Comments Zhan ZHANG and Yasuhiko MORIMOTO Graduate School of Engineering, Hiroshima University Hiroshima

More information

Natural Language Processing

Natural Language Processing Natural Language Processing Word vectors Many slides borrowed from Richard Socher and Chris Manning Lecture plan Word representations Word vectors (embeddings) skip-gram algorithm Relation to matrix factorization

More information

Data mining, 4 cu Lecture 7:

Data mining, 4 cu Lecture 7: 582364 Data mining, 4 cu Lecture 7: Sequential Patterns Spring 2010 Lecturer: Juho Rousu Teaching assistant: Taru Itäpelto Sequential Patterns In many data mining tasks the order and timing of events contains

More information

Semantic GIS GeoSPARQL

Semantic GIS GeoSPARQL Semantic GIS GeoSPARQL Michal Med michal.med@fel.cvut.cz December 20, 2018 Michal Med (michal.med@fel.cvut.cz) Semantic GIS GeoSPARQL December 20, 2018 1 / 47 Outline 1 WTF is GIS?? 2 Usage of GIS 3 Linked

More information

Semantics, ontologies and escience for the Geosciences

Semantics, ontologies and escience for the Geosciences Semantics, ontologies and escience for the Geosciences Femke Reitsma¹, John Laxton², Stuart Ballard³, Werner Kuhn, Alia Abdelmoty ¹femke.reitsma@ed.ac.uk: School of Geosciences, Edinburgh University ²

More information

NLP Programming Tutorial 11 - The Structured Perceptron

NLP Programming Tutorial 11 - The Structured Perceptron NLP Programming Tutorial 11 - The Structured Perceptron Graham Neubig Nara Institute of Science and Technology (NAIST) 1 Prediction Problems Given x, A book review Oh, man I love this book! This book is

More information

Intelligent Systems (AI-2)

Intelligent Systems (AI-2) Intelligent Systems (AI-2) Computer Science cpsc422, Lecture 19 Oct, 23, 2015 Slide Sources Raymond J. Mooney University of Texas at Austin D. Koller, Stanford CS - Probabilistic Graphical Models D. Page,

More information

What s so Hard about Natural Language Understanding?

What s so Hard about Natural Language Understanding? What s so Hard about Natural Language Understanding? Alan Ritter Computer Science and Engineering The Ohio State University Collaborators: Jiwei Li, Dan Jurafsky (Stanford) Bill Dolan, Michel Galley, Jianfeng

More information

Introduction to Spatial Big Data Analytics. Zhe Jiang Office: SEC 3435

Introduction to Spatial Big Data Analytics. Zhe Jiang Office: SEC 3435 Introduction to Spatial Big Data Analytics Zhe Jiang zjiang@cs.ua.edu Office: SEC 3435 1 What is Big Data? Examples Internet data (images from the web) Earth observation data (nasa.gov) wikimedia.org www.me.mtu.edu

More information

Part of Speech Tagging: Viterbi, Forward, Backward, Forward- Backward, Baum-Welch. COMP-599 Oct 1, 2015

Part of Speech Tagging: Viterbi, Forward, Backward, Forward- Backward, Baum-Welch. COMP-599 Oct 1, 2015 Part of Speech Tagging: Viterbi, Forward, Backward, Forward- Backward, Baum-Welch COMP-599 Oct 1, 2015 Announcements Research skills workshop today 3pm-4:30pm Schulich Library room 313 Start thinking about

More information

Degree (k)

Degree (k) 0 1 Pr(X k) 0 0 1 Degree (k) Figure A1: Log-log plot of the complementary cumulative distribution function (CCDF) of the degree distribution for a sample month (January 0) network is shown (blue), along

More information

搜索中的带区域化和个性化的自动补全和自动建议技术 叶旭刚

搜索中的带区域化和个性化的自动补全和自动建议技术 叶旭刚 搜索中的带区域化和个性化的自动补全和自动建议技术 叶旭刚 Probabilistic Autocomplete Suggestion Ranking with Localization and Personalization Xugang Ye General Search vs. Vertical Search General Search General data General users Many

More information

Intelligent Systems (AI-2)

Intelligent Systems (AI-2) Intelligent Systems (AI-2) Computer Science cpsc422, Lecture 19 Oct, 24, 2016 Slide Sources Raymond J. Mooney University of Texas at Austin D. Koller, Stanford CS - Probabilistic Graphical Models D. Page,

More information

Categorization ANLP Lecture 10 Text Categorization with Naive Bayes

Categorization ANLP Lecture 10 Text Categorization with Naive Bayes 1 Categorization ANLP Lecture 10 Text Categorization with Naive Bayes Sharon Goldwater 6 October 2014 Important task for both humans and machines object identification face recognition spoken word recognition

More information

ANLP Lecture 10 Text Categorization with Naive Bayes

ANLP Lecture 10 Text Categorization with Naive Bayes ANLP Lecture 10 Text Categorization with Naive Bayes Sharon Goldwater 6 October 2014 Categorization Important task for both humans and machines 1 object identification face recognition spoken word recognition

More information

Exploiting Geographic Dependencies for Real Estate Appraisal

Exploiting Geographic Dependencies for Real Estate Appraisal Exploiting Geographic Dependencies for Real Estate Appraisal Yanjie Fu Joint work with Hui Xiong, Yu Zheng, Yong Ge, Zhihua Zhou, Zijun Yao Rutgers, the State University of New Jersey Microsoft Research

More information

Text Analysis. Week 5

Text Analysis. Week 5 Week 5 Text Analysis Reference and Slide Source: ChengXiang Zhai and Sean Massung. 2016. Text Data Management and Analysis: a Practical Introduction to Information Retrieval and Text Mining. Association

More information

Trade Show Planning Timeline by Cory Ann Ellis

Trade Show Planning Timeline by Cory Ann Ellis Trade Planning Timeline Action +2 9-2 6-9 3-6 -3 At the < After Action +2 9-2 6-9 3-6 -3 At the < After Select your show(s) Clarify Value of Research: cost, history, # of attendees, ROI, past vendors,

More information

Jointly Extracting Event Triggers and Arguments by Dependency-Bridge RNN and Tensor-Based Argument Interaction

Jointly Extracting Event Triggers and Arguments by Dependency-Bridge RNN and Tensor-Based Argument Interaction Jointly Extracting Event Triggers and Arguments by Dependency-Bridge RNN and Tensor-Based Argument Interaction Feng Qian,LeiSha, Baobao Chang, Zhifang Sui Institute of Computational Linguistics, Peking

More information

Propositional Logic: Methods of Proof (Part II)

Propositional Logic: Methods of Proof (Part II) Propositional Logic: Methods of Proof (Part II) You will be expected to know Basic definitions Inference, derive, sound, complete Conjunctive Normal Form (CNF) Convert a Boolean formula to CNF Do a short

More information

Boolean and Vector Space Retrieval Models CS 290N Some of slides from R. Mooney (UTexas), J. Ghosh (UT ECE), D. Lee (USTHK).

Boolean and Vector Space Retrieval Models CS 290N Some of slides from R. Mooney (UTexas), J. Ghosh (UT ECE), D. Lee (USTHK). Boolean and Vector Space Retrieval Models 2013 CS 290N Some of slides from R. Mooney (UTexas), J. Ghosh (UT ECE), D. Lee (USTHK). 1 Table of Content Boolean model Statistical vector space model Retrieval

More information

Spatial Data, Spatial Analysis and Spatial Data Science

Spatial Data, Spatial Analysis and Spatial Data Science Spatial Data, Spatial Analysis and Spatial Data Science Luc Anselin http://spatial.uchicago.edu 1 spatial thinking in the social sciences spatial analysis spatial data science spatial data types and research

More information

ACS Introduction to NLP Lecture 2: Part of Speech (POS) Tagging

ACS Introduction to NLP Lecture 2: Part of Speech (POS) Tagging ACS Introduction to NLP Lecture 2: Part of Speech (POS) Tagging Stephen Clark Natural Language and Information Processing (NLIP) Group sc609@cam.ac.uk The POS Tagging Problem 2 England NNP s POS fencers

More information

Information Retrieval and Organisation

Information Retrieval and Organisation Information Retrieval and Organisation Chapter 13 Text Classification and Naïve Bayes Dell Zhang Birkbeck, University of London Motivation Relevance Feedback revisited The user marks a number of documents

More information

Statistical NLP for the Web Log Linear Models, MEMM, Conditional Random Fields

Statistical NLP for the Web Log Linear Models, MEMM, Conditional Random Fields Statistical NLP for the Web Log Linear Models, MEMM, Conditional Random Fields Sameer Maskey Week 13, Nov 28, 2012 1 Announcements Next lecture is the last lecture Wrap up of the semester 2 Final Project

More information

10/17/04. Today s Main Points

10/17/04. Today s Main Points Part-of-speech Tagging & Hidden Markov Model Intro Lecture #10 Introduction to Natural Language Processing CMPSCI 585, Fall 2004 University of Massachusetts Amherst Andrew McCallum Today s Main Points

More information

Final Exam, Fall 2002

Final Exam, Fall 2002 15-781 Final Exam, Fall 22 1. Write your name and your andrew email address below. Name: Andrew ID: 2. There should be 17 pages in this exam (excluding this cover sheet). 3. If you need more room to work

More information

Analysing Soft Syntax Features and Heuristics for Hierarchical Phrase Based Machine Translation

Analysing Soft Syntax Features and Heuristics for Hierarchical Phrase Based Machine Translation Analysing Soft Syntax Features and Heuristics for Hierarchical Phrase Based Machine Translation David Vilar, Daniel Stein, Hermann Ney IWSLT 2008, Honolulu, Hawaii 20. October 2008 Human Language Technology

More information

Chunking with Support Vector Machines

Chunking with Support Vector Machines NAACL2001 Chunking with Support Vector Machines Graduate School of Information Science, Nara Institute of Science and Technology, JAPAN Taku Kudo, Yuji Matsumoto {taku-ku,matsu}@is.aist-nara.ac.jp Chunking

More information

Where to Find My Next Passenger?

Where to Find My Next Passenger? Where to Find My Next Passenger? Jing Yuan 1 Yu Zheng 2 Liuhang Zhang 1 Guangzhong Sun 1 1 University of Science and Technology of China 2 Microsoft Research Asia September 19, 2011 Jing Yuan et al. (USTC,MSRA)

More information

Exploit your geodata to enable smart cities and countries

Exploit your geodata to enable smart cities and countries Exploit your geodata to enable smart cities and countries Carsten Rönsdorf, Head of Advisory Services Middle East Practice Manager Future Cities, standards and 3D 24 November 2014 Importance of Standards

More information

Data Mining: Concepts and Techniques. (3 rd ed.) Chapter 8. Chapter 8. Classification: Basic Concepts

Data Mining: Concepts and Techniques. (3 rd ed.) Chapter 8. Chapter 8. Classification: Basic Concepts Data Mining: Concepts and Techniques (3 rd ed.) Chapter 8 1 Chapter 8. Classification: Basic Concepts Classification: Basic Concepts Decision Tree Induction Bayes Classification Methods Rule-Based Classification

More information

Better! Faster! Stronger*!

Better! Faster! Stronger*! Jason Eisner Jiarong Jiang He He Better! Faster! Stronger*! Learning to balance accuracy and efficiency when predicting linguistic structures (*theorems) Hal Daumé III UMD CS, UMIACS, Linguistics me@hal3.name

More information

United States Forest Service Makes Use of Avenza PDF Maps App to Reduce Critical Time Delays and Increase Response Time During Natural Disasters

United States Forest Service Makes Use of Avenza PDF Maps App to Reduce Critical Time Delays and Increase Response Time During Natural Disasters CASE STUDY For more information, please contact: Christine Simmons LFPR Public Relations www.lfpr.com (for Avenza) 949.502-7750 ext. 270 Christines@lf-pr.com United States Forest Service Makes Use of Avenza

More information

Probabilistic Neighborhood Selection in Collaborative Filtering Systems

Probabilistic Neighborhood Selection in Collaborative Filtering Systems Probabilistic Neighborhood Selection in Collaborative Filtering Systems Panagiotis Adamopoulos and Alexander Tuzhilin Department of Information, Operations and Management Sciences Leonard N. Stern School

More information

Disaster Preparedness in Shizuoka

Disaster Preparedness in Shizuoka Disaster Preparedness in Shizuoka Shizuoka is a beautiful, excellently located prefecture. It s got sea views and mountains (including Mount Fuji, the tallest mountain in Japan), and is relatively close

More information

Liangjie Hong, Ph.D. Candidate Dept. of Computer Science and Engineering Lehigh University Bethlehem, PA

Liangjie Hong, Ph.D. Candidate Dept. of Computer Science and Engineering Lehigh University Bethlehem, PA Rutgers, The State University of New Jersey Nov. 12, 2012 Liangjie Hong, Ph.D. Candidate Dept. of Computer Science and Engineering Lehigh University Bethlehem, PA Motivation Modeling Social Streams Future

More information

Penn Treebank Parsing. Advanced Topics in Language Processing Stephen Clark

Penn Treebank Parsing. Advanced Topics in Language Processing Stephen Clark Penn Treebank Parsing Advanced Topics in Language Processing Stephen Clark 1 The Penn Treebank 40,000 sentences of WSJ newspaper text annotated with phrasestructure trees The trees contain some predicate-argument

More information

Spam ain t as Diverse as It Seems: Throttling OSN Spam with Templates Underneath

Spam ain t as Diverse as It Seems: Throttling OSN Spam with Templates Underneath Spam ain t as Diverse as It Seems: Throttling OSN Spam with Templates Underneath Hongyu Gao, Yi Yang, Kai Bu, Yan Chen, Doug Downey, Kathy Lee, Alok Choudhary Northwestern University, USA Zhejiang University,

More information

Point-of-Interest Recommendations: Learning Potential Check-ins from Friends

Point-of-Interest Recommendations: Learning Potential Check-ins from Friends Point-of-Interest Recommendations: Learning Potential Check-ins from Friends Huayu Li, Yong Ge +, Richang Hong, Hengshu Zhu University of North Carolina at Charlotte + University of Arizona Hefei University

More information

The Challenge of Geospatial Big Data Analysis

The Challenge of Geospatial Big Data Analysis 288 POSTERS The Challenge of Geospatial Big Data Analysis Authors - Teerayut Horanont, University of Tokyo, Japan - Apichon Witayangkurn, University of Tokyo, Japan - Shibasaki Ryosuke, University of Tokyo,

More information

Lecture 13: Structured Prediction

Lecture 13: Structured Prediction Lecture 13: Structured Prediction Kai-Wei Chang CS @ University of Virginia kw@kwchang.net Couse webpage: http://kwchang.net/teaching/nlp16 CS6501: NLP 1 Quiz 2 v Lectures 9-13 v Lecture 12: before page

More information

Recommendation Systems

Recommendation Systems Recommendation Systems Popularity Recommendation Systems Predicting user responses to options Offering news articles based on users interests Offering suggestions on what the user might like to buy/consume

More information

Dialogue Systems. Statistical NLU component. Representation. A Probabilistic Dialogue System. Task: map a sentence + context to a database query

Dialogue Systems. Statistical NLU component. Representation. A Probabilistic Dialogue System. Task: map a sentence + context to a database query Statistical NLU component Task: map a sentence + context to a database query Dialogue Systems User: Show me flights from NY to Boston, leaving tomorrow System: [returns a list of flights] Origin (City

More information

The GapVis project and automated textual geoparsing

The GapVis project and automated textual geoparsing The GapVis project and automated textual geoparsing Google Ancient Places team, Edinburgh Language Technology Group GeoBib Conference, Giessen, 4th-5th May 2015 1 1 Google Ancient Places and GapVis The

More information

Complete Weather Intelligence for Public Safety from DTN

Complete Weather Intelligence for Public Safety from DTN Complete Weather Intelligence for Public Safety from DTN September 2017 White Paper www.dtn.com / 1.800.610.0777 From flooding to tornados to severe winter storms, the threats to public safety from weather-related

More information

N-grams. Motivation. Simple n-grams. Smoothing. Backoff. N-grams L545. Dept. of Linguistics, Indiana University Spring / 24

N-grams. Motivation. Simple n-grams. Smoothing. Backoff. N-grams L545. Dept. of Linguistics, Indiana University Spring / 24 L545 Dept. of Linguistics, Indiana University Spring 2013 1 / 24 Morphosyntax We just finished talking about morphology (cf. words) And pretty soon we re going to discuss syntax (cf. sentences) In between,

More information

A Convolutional Neural Network-based

A Convolutional Neural Network-based A Convolutional Neural Network-based Model for Knowledge Base Completion Dat Quoc Nguyen Joint work with: Dai Quoc Nguyen, Tu Dinh Nguyen and Dinh Phung April 16, 2018 Introduction Word vectors learned

More information

You are Building Your Organization s Geographic Knowledge

You are Building Your Organization s Geographic Knowledge You are Building Your Organization s Geographic Knowledge And Increasingly Making it Available Sharing Data Publishing Maps and Geo-Apps Developing Collaborative Approaches Citizens Knowledge Workers Analysts

More information

Attention Based Joint Model with Negative Sampling for New Slot Values Recognition. By: Mulan Hou

Attention Based Joint Model with Negative Sampling for New Slot Values Recognition. By: Mulan Hou Attention Based Joint Model with Negative Sampling for New Slot Values Recognition By: Mulan Hou houmulan@bupt.edu.cn CONTE NTS 1 2 3 4 5 6 Introduction Related work Motivation Proposed model Experiments

More information

Ryosuke Kojima & Taisuke Sato Tokyo Institute of Technology

Ryosuke Kojima & Taisuke Sato Tokyo Institute of Technology Ryosuke Kojima & Taisuke ato Tokyo Institute of Technology PRIM [ato et al. 97] (http://sato-www.cs.titech.ac.jp/prism/) Probabilistic Prolog for machine and subsumes BN,HMM,PCFG, Bayesian networks HMMs

More information

ANALYSIS OF EVACUATION BEHAVIORS IN DIFFERENT AREAS BEFORE AND AFTER THE GREAT EAST JAPAN EARTHQUAKE

ANALYSIS OF EVACUATION BEHAVIORS IN DIFFERENT AREAS BEFORE AND AFTER THE GREAT EAST JAPAN EARTHQUAKE Fifth International Conference on Geotechnique, Construction Materials and Environment, Osaka, Japan, Nov. 16-18, 2015, ISBN: 978-4-9905958-4-5 C3051 ANALYSIS OF EVACUATION BEHAVIORS IN DIFFERENT AREAS

More information

Part A. P (w 1 )P (w 2 w 1 )P (w 3 w 1 w 2 ) P (w M w 1 w 2 w M 1 ) P (w 1 )P (w 2 w 1 )P (w 3 w 2 ) P (w M w M 1 )

Part A. P (w 1 )P (w 2 w 1 )P (w 3 w 1 w 2 ) P (w M w 1 w 2 w M 1 ) P (w 1 )P (w 2 w 1 )P (w 3 w 2 ) P (w M w M 1 ) Part A 1. A Markov chain is a discrete-time stochastic process, defined by a set of states, a set of transition probabilities (between states), and a set of initial state probabilities; the process proceeds

More information