Instructor: Smaranda Muresan Columbia University [email protected]

38
Instructor: Smaranda Muresan Columbia University [email protected] Course Introduction

description

Course Introduction. Instructor: Smaranda Muresan Columbia University [email protected]. Natural Language Processing Applications. - PowerPoint PPT Presentation

Transcript of Instructor: Smaranda Muresan Columbia University [email protected]

Page 1: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Instructor: Smaranda MuresanColumbia University

[email protected]

Course Introduction

Page 2: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

10TH DEGREE is a full service advertising agency specializing in direct and interactive marketing. Located in Irvine CA, 10TH DEGREE is looking for an Assistant Account Manager to help manage and coordinate interactive marketing initiatives for a marquee automative account. Experience in online marketing, automative and/or the advertising field is a plus. Assistant Account Manager Responsibilities Ensures smooth implementation of programs and initiatives Helps manage the delivery of projects and key client deliverables … Compensation: $50,000-\$80,000INDUSTRY

POSITIONLOCATIONCOMPANY

AdvertisingAssist. Account Manag.Irvine, CA10th DEGREE

Information Extraction: Identifying the instances of facts names/entities , relations and events from semi-structured or unstructured text; and convert them into structured representations (e.g. databases)

Natural Language Processing Applications

Page 3: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Question Answering IBM’s Watson

• Won Jeopardy on February 16, 2011

Bram Stoker

Page 4: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

• Watson has no discourse understanding

“Watson also tripped up on an “Olympic Oddities” answer, but so imperceptibly that Alex Trebek didn’t notice at first, raising an important point of clarification. After Jennings responded incorrectly that Olympian gymnast George Eyser was “missing a hand”, Watson responded, “What is a leg?”

http://www.wired.com/business/2011/02/watson-wrong-answer-trebek/

Page 5: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

This ClassThe journalist William Finnegan has said about his profession (New Yorker, July 2,2012): ``You fish for facts and instead pull up boatloads of speculation, some of it well informed, much of it trailing tangled agendas. You end up reporting not so much what happened as what people think or imagine or say happened.'’ [Thanks Owen Rambow for this reference] In this class we are interested in understanding communication through the eyes of the authors/speakers.

Page 7: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Syllabus Overview

• http://www1.cs.columbia.edu/~smara/teaching/E6998/S14/

Page 8: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Outline

• Instructor Introduction– Background, Research Interests

• Student Introductions• Class Overview

– Class organization– Website– Office Hours & TA– Topics covered in this class– Grading

Page 9: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Instructor Intro

• Researcher at the Center for Computational Learning Systems

http://www1.cs.columbia.edu/~smara

Broad research interests: computational semantics, language in social media

Page 10: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Some of my current research projectsrelevant to the course

• Detecting Contrary Meaning

Page 11: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Contrary meaning • Explicit: Conflicting statements/beliefs overtly expressed in text

• Implicit: Sarcasm User: I'm so happy I'm going back to the emergency room User: Newspaper faces court over sleazing Facebook ? Facebook

is so defenseless and innocent .

Page 12: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Prelim work on Sarcasm Detection

• Can we automatically distinguish among sarcastic, positive and negative utterances?

• Can we easily build a labeled corpus of

naturally occurring sarcastic, positive and negative utterances?

(Gonzalez, Muresan and Wacholder, 2011; Muresan et al., underreview)

Page 13: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Data collection

Page 14: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

How can we distinguish sarcastic, pos, and negative tweets?

• Lexical Features– Pannebacker et al. (2007) LIWC lexicon (64 word categories grouped

into four general classes: • Linguistic Processes (LP) (e.g., adverbs, pronouns),• Psychological Processes (PP) (e.g., positive and negative emotions)• Personal Concerns (PC) (e.g, work, achievement)• Spoken Categories (SC) (e.g., disfluencies);

– WordNet Affect (WNA) (Strapparava and Valitutti, 2004)– list of interjections (e.g., ah, oh, yeah), and punctuations (e.g., !, ?).

• We merged all of the lists into a single dictionary. • The token overlap between the words in combined dictionary and the

words in the tweets was 85%.

Page 15: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

How can we distinguish sarcastic, pos, and negative tweets?

• Pragmatic features– Emoticons (, )– ToUser (@john)

Page 16: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Classification experiments• Several settings

– S-N-P (900 example each; balanced datasets)– S-NS (NS contain 450 negative and 450 positive)– S-N (900 example each)– S-P (900 example each)

• 2 classifiers– support vector machines (SVM), – and logistic regression (LogR).

• Features used: – 1) unigrams; – 2) presence of the dictionary-based lexical factors and pragmatic

factors (LIWC+_P); – 3) frequency of the dictionary-based lexical factors and pragmatic

factors (LIWC+_F).– 4) combination of unigrams and presence features

Page 17: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Results

Page 18: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

How hard is the task? Can humans do it?

Page 19: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Human performance on the task

• Two studies– 1) we asked 3 judges to classify 10% of our S-P-N

datasets (90 randomly selected tweets per category). we also trained our SVM and LogR classifiers using the remaining 90% of the data.

– 2) we asked another 3 judges to classify 10% of the S-NS dataset (90 per category. The NS category contained 45 positive and 45 negative tweets). We also trained SVM and LogR on the remaining 90% of data

Page 20: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Humans on S-N-P

• overall agreement of 50% was achieved among the three judges, with a Fleiss’ Kappa value of 0.4788 (p<.05). The average accuracy was 62.59%

• When we considered only the 135 of 270 tweets on which all three judges agreed, the accuracy on the set they agreed on was 86.67%. (this can be an upper bound)

Page 21: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Humans on S-NS

• Results showed an agreement of 71.67% among the three judges with a Fleiss’ Kappa value of 0.5861 (p<.05). The average accuracy rate was 66.85% .

• When we considered only cases where all three judges agreed (129 out of 180), the accuracy on the set they agreed on was 82.95%. (this can be un upper bound)

Page 22: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Humans vs Automatic ClassificationS-N-P

S-NS

Page 23: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Discussion

• Hard task both for Automatic Measures and Humans

• Some judges reported specific difficulties:– Lack of context (e.g., world knowledge; context of

conversation)– Brevity of messages

Other issues/observations – We will have a whole class on Sarcasm Detection

Page 24: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Detection Conflicting Information• Explicit Contrary Meaning

User1: A shooting has just occurred at the Occupy Oakland encampment.

User2: Shootings happen in Oakland all the time and it had nothing to do with the Occupy movement.

User1: This shooting does have something to do with the Occupy movement because many of the witness's are the Occupiers and it happened only a few yards away from the encampment.

User3: On Twitter, Occupy Oakland has said the shooting was "related to the occupation. Please keep this man in your thoughts."

Page 25: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Impact• Conflicting statements/beliefs can signal:

– anomalies in events (e.g., different theories about the cause of an event),

– anomalies in beliefs (change in beliefs), – deception/lying

– misinformation– misconception

Page 26: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Recognizing Textual Entailment (RTE)• Given two text fragments – the Text(T) and the Hypothesis(H)–

predict whether a human reader would say:– That the H is true, given T– That the H contradicts T– That it can’t be determined whether or not H is true given T

T: John Smith, who was 65, resigned yesterday.

H: 65-year-old Mr. Smith left office.

T: UberSoft CEO Bill Jobs

H: Frank N. Furter is CEO of Ubersoft

Page 27: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

ApproachFramed as a 2-way Textual Entailment problem (contradict., non-

contradictory). Assume utterances are about the same topic/event

1. Linguisticanalysis

2. Graphalignment

3.Contradiction features &classification

tunedthreshold

contradicts

doesn’tcontradict

score = = –2.001.84

T: A case of indigenously acquired rabies infection has been confirmed.H: No case of rabies was confirmed.

case

No

rabies

det

prep_of

0.10

0.00

–0.75

rabiesPOSNERIDF

NNS--0.027

… … …

Feature fi

wi

Polarity difference - -2.00

case

No rabiesdet prep_of

case

A rabies

det

amod

infection

case

A rabiesdet amod

infection

prep_of

Event coreference

Page 28: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Student Introduction

• Your education: PhD/Master/Undergrad and year• Did you take NLP course? • Did you take ML course?• Are you doing or have done research in NLP? If

yes, briefly say in what area• Any other info you want to share with the class?

Page 29: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Outline

• Instructor Introduction– Background, Research Interests

• Student Introductions• Class Overview

– Class organization– Office Hours & TA– Website/details of topics covered in this class– Grading

Page 30: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Class organization (except first two lectures)

• 50 min discussion of research articles led by students on topic of the week (intro on topic done previous week)– There will be 2 papers per class for discussion– 25 min each (15 min presentation, 10 min discussion)

• 5 minutes break• 30 minutes in depth lecture/open questions for topic

of the week • 25 Intro lecture to topic of following week (to facilitate

paper discussion)

Page 31: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Office Hours

• Instructor Office hours”– Thursday 6:00-7:00 (after class) or by appointment

if needed• TA: Arpit Gupta• TA office ours: TBA

Page 32: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Class Website

• http://www1.cs.columbia.edu/~smara/teaching/E6998/S14/

• Pay attention to top of page for announcements.

Page 33: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Extracting social/interactional meaning• Sentiment (Positive or negative)

– Movie or Products or Politics: is a text positive or negative? “The movie was great”- How can we automatically detect sentiment? (word level and text level)

• Emotion(sad, happy) and Mood (depressed)– Detecting expression of emotion/mood in language– Applications:

• Annoyance in talking to dialog systems• Uncertainty of students in tutoring• Detecting Trauma or Depression

• Hedging & Beliefs– Committed Belief (CB): W/S firmly believes p “John will arrive at 6” - Non-committed Belief (NCB): W/S weakly believes p “John may arrive at 6”- Reported Belief (RB): W/S is reporting someone else’s belief “John said he would arrive at 6”How can we automatically detect/tag beliefs?

Page 34: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Extracting social/interactional meaning

• Sarcasm– Contrary of people’s actual sentiments or beliefs “I love shopping on Black Friday” “A Shooting in Oakland? That NEVER happend”

• Agreement/Disagreement– Agreement vs. disagreement with propositions (and

people)• Perspective

– An aggregate of a person’s beliefs and sentiments w.r.t topic/event/proposition

– How can we detect perspective automatically? • Deception

– Automatic ways to identify deceptive language

Page 35: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Extracting social/interactional meaning

• Power– Different types of power: e.g. hierarchical, influence– Applications:

• Find influential people in online communities or those who want to become influential

• target ads to influential people in community

• Extracting Social Networks from text– Analyze online discussion and identify who are the people, and

how are they related (beyond metadata)– Social network of characters from novels

• Personality and Interpersonal Stance– Romantic interest, flirtation, friendliness

Page 36: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Grading• Critical Discussion of one of the research articles (40% of grade)

– Brief Presentation in class about the paper – Lead a critical discussion on key positive and negative aspects– Full list of papers up by Tuesday, Jan 28 11:59pm. – Students select their top 5 papers before class on Jan 30. – TA/Instructor assigns papers based on preference and in case of conflict first-come-

first served, by Feb 1 5pm. • Project about a topic discussed in class (or related) (60%)

– Computational Implementation– Can be individual or team of 2-3– Project Proposal (5th week of classes; receive feedback by week 6)– Literature review on the chosen topic (9th week of classes; receive feedback by week

10 )– Final paper – conference/workshop format (8 pages) (last week of classes)– Final project presentation (last week of classes)

Page 37: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Next Class

• Computational Models for learning semantic lexicons

• 2 papers for reading/discussion– I will lead the discussion of one of the research

articles to set up a model of what’s expected– Second paper will be free discussion (unless there

is a volunteer to present on of the papers )

Page 38: Instructor:  Smaranda Muresan Columbia University smara@ccls.columbia.edu

Resources• ACL anthology

– All the proceedings of main conferences in NLP as well as major journals.

– http://aclweb.org/anthology/(Recent years authors are encourage to submit datasets and code)

• Linguistic Data Consortium – Annotated corpora – http://catalog.ldc.upenn.edu/(If interested to have access to some corpora for your project ask Instructor, most likely we have it)