flow: a first-language-oriented writing assistant system

44
Mei-Hua Chen*, Shih-Ting Huang+, Hung-Ting Hsieh*, Ting-Hui Kao+, Jason S. Chang+ * Institute of Information Systems and Applications + Department of Computer Science National Tsing Hua University HsinChu, Taiwan, R.O.C. 30013 ACL 2012 1

Upload: lizina

Post on 09-Jan-2016

33 views

Category:

Documents


1 download

DESCRIPTION

FLOW: A First-Language-Oriented Writing Assistant System. Mei-Hua Chen*, Shih-Ting Huang+, Hung-Ting Hsieh*, Ting-Hui Kao+, Jason S. Chang+ * Institute of Information Systems and Applications + Department of Computer Science - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: FLOW: A First-Language-Oriented Writing Assistant System

Mei-Hua Chen*, Shih-Ting Huang+, Hung-Ting Hsieh*, Ting-Hui Kao+, Jason S. Chang+

* Institute of Information Systems and Applications + Department of Computer Science National Tsing Hua University HsinChu, Taiwan, R.O.C. 30013

ACL 2012

1

Page 2: FLOW: A First-Language-Oriented Writing Assistant System

Feature• First-Language-Oriented• Translations• Paraphrases• N-grams (N=5)

2

Page 3: FLOW: A First-Language-Oriented Writing Assistant System

Introduction composing stage We propose a method to ” 解決問題“ .

solve the problem tackle the problem revising stage We propose a method to solve the

problem

3

盡力

try our bestdo our best

Page 4: FLOW: A First-Language-Oriented Writing Assistant System

Translation-based N-gram Prediction• {e1, e2, …em, f1, f2 …fn}1.predict the possible translations (Och and Ney, 2003)

bilingual phrase alignments

2. disambiguous (correct the alignment error)

ex. ...on ways to identify tackle 洗錢 money laundering money His forum entitled money laundry

4

Page 5: FLOW: A First-Language-Oriented Writing Assistant System

Paraphrase Suggestion• {e1, e2,…ek}• pivot-based method proposed by Bannard

and Callison-Burch (2005).

5

Page 6: FLOW: A First-Language-Oriented Writing Assistant System

Experiment• Training data: Hong Kong Parallel Text (2,220,570 Chinese-English sentence pairs)

• 10 Chinese sentences• two students to translate the Chinese

sentences to English sentences using FLOW

6

Page 7: FLOW: A First-Language-Oriented Writing Assistant System

Result

7

•Paraphrase performance well•N-gram tends to produce shorter phrases

Page 8: FLOW: A First-Language-Oriented Writing Assistant System

Keke Cai, Jiajun Bu, Chun Chen, Kangmiao LiuCollege of Computer Science, Zhejiang University

Hangzhou, 310027, China

ACL 2007

8

Page 9: FLOW: A First-Language-Oriented Writing Assistant System

Sentence Retrieval • Limited information

• Application:• document summarization• question answering• novelty detection

9

Page 10: FLOW: A First-Language-Oriented Writing Assistant System

Term Dependence• Query:{Everest, highest , mountain}• Q ={TS1, TS2, …, TSn}• Term combinations:{Everest highest,

highest mountain, Everest mountain}• further evaluated in each retrieved

sentence• Ex. Everest is the highest mountain

10

Page 11: FLOW: A First-Language-Oriented Writing Assistant System

MINIPAR• a dependency parser• Ex. Everest is the highest mountain• :{Everest highest, highest mountain,

Everest mountain}

11

Distance=(3+1+2)/3

Page 12: FLOW: A First-Language-Oriented Writing Assistant System

Association Strength

12

: Size of

D( ) :

Page 13: FLOW: A First-Language-Oriented Writing Assistant System

Discussion• Query:{ Everest, highest , mountain}• TS1:{ Everest, highest , mountain} TS2:{ highest , mountain}

AS(TS1, S1)= 0.5^(1/3)*0.5^2=0.1984AS(TS2, S2)= 0.5^(1/2)*0.5^1=0.35355

• Dependency distance tend to small set pairs

13

Page 14: FLOW: A First-Language-Oriented Writing Assistant System

Experiments• Testing data: TREC novelty track 2003 and

2004• Average precision of each different retrieval

models

14

Page 15: FLOW: A First-Language-Oriented Writing Assistant System

15

Page 16: FLOW: A First-Language-Oriented Writing Assistant System

Colin Bannard , Chris Callison-BurchSchool of Informatics

University of Edinburgh2 Buccleuch Place

Edinburgh, EH8 9LW

ACL 2005

16

Page 17: FLOW: A First-Language-Oriented Writing Assistant System

Parallel Corpora• Monolingual

• Bilingual (German-English)

17

Page 18: FLOW: A First-Language-Oriented Writing Assistant System

Bilingual Parallel Corpora• much more commonly available resource• one language can be identified using a

phrase in another language as a pivot.

18

German is pivot, use it to find English phrase

Page 19: FLOW: A First-Language-Oriented Writing Assistant System

Paraphrases• Application multidocument summarization machine translation question answering

19

Page 20: FLOW: A First-Language-Oriented Writing Assistant System

Aligning phrase pairs• statistical machine translation• phrase alignment• Och and Ney(2003)

20

Giza++

Page 21: FLOW: A First-Language-Oriented Writing Assistant System

Assigning probabilities

21

: original English phrase

: candidate English phrase

: foreign language phrase

Page 22: FLOW: A First-Language-Oriented Writing Assistant System

Experimental Design1• 46 English phrases (occurred multiple times in the first 50,000 sentences)

• Corpus:

German-English section of the Europarl corpus (1,036,000 German-English sentence pairs)

• Manually aligned• 289 evaluation sets (each contain 2~10)

• Judgment: (meaning and grammar)

two native English speakers• Precision: 0.605

22

Page 23: FLOW: A First-Language-Oriented Writing Assistant System

23

Page 24: FLOW: A First-Language-Oriented Writing Assistant System

24

Page 25: FLOW: A First-Language-Oriented Writing Assistant System

Experimental Design2• evaluated the accuracy of top ranked

paraphrases• conditions 1. manual alignments 2. automatic alignments 3. automatic alignments & multiple corpora in different languages (French-English, Spanish-English, Italian-

English) (4,000,000 sentence pairs)

4. re-ranking 5. limited to the same sense

25

Page 26: FLOW: A First-Language-Oriented Writing Assistant System

26

IgnoreGrammar

trigram language model

Page 27: FLOW: A First-Language-Oriented Writing Assistant System

27

Page 28: FLOW: A First-Language-Oriented Writing Assistant System

Hao Xu† Jingdong Wang‡ Xian-Sheng Hua‡ Shipeng Li‡ †MOE-MS KeyLab of MCC, University of Science and Technology of China, Hefei, 230026, P. R. China ‡Microsoft Research Asia, Beijing 100190, P. R. China

SIGIR 2010

28

Page 29: FLOW: A First-Language-Oriented Writing Assistant System

29

Page 30: FLOW: A First-Language-Oriented Writing Assistant System

30

Image search schemes

Page 31: FLOW: A First-Language-Oriented Writing Assistant System

Flowchart

31

Page 32: FLOW: A First-Language-Oriented Writing Assistant System

Visual Instance Transformation• text-based image search (Top 50)• affinity propagation (AP) clustering

algorithm

• sort the obtained centers in a descending order of their groups sizes

32

Page 33: FLOW: A First-Language-Oriented Writing Assistant System

• snoopy

33

Visual Instance

Side view

Front view

Page 34: FLOW: A First-Language-Oriented Writing Assistant System

Spatial Intention Estimation• position• influence scope• Use 2D Gaussian distribution

34

Page 35: FLOW: A First-Language-Oriented Writing Assistant System

Layout Sensitive Relevance Evaluation• Sum up the relevance score for each

concept• Appearance consistency -the count of common visual words

• Spatial consistency -desired spatial distribution of the concept k

-spatial distribution of visual instance v in the image

35

Page 36: FLOW: A First-Language-Oriented Writing Assistant System

Quantitative Search Performance

36

Page 37: FLOW: A First-Language-Oriented Writing Assistant System

Visual Results (1)

37

Page 38: FLOW: A First-Language-Oriented Writing Assistant System

38

Visual Results (2)

Page 39: FLOW: A First-Language-Oriented Writing Assistant System

User Study• participants : 20 college students

• To the question “have you ever had any image search intention concerning the concept layout?”

• 20% of respondents replied with “yes” and 50% of respondents replied with “no, but probably in

the future”.

39

Page 40: FLOW: A First-Language-Oriented Writing Assistant System

Michael Bendersky , W. Bruce CroftDept. of Computer Science

Univ. of Massachusetts AmherstAmherst, MA

SIGIR 2012

40

Page 41: FLOW: A First-Language-Oriented Writing Assistant System

Feature

41

a more accurate modeling of the dependencies between the query terms Query concepts n-grams, term proximities, noun phrases, named entities

verbose natural language queries (grammatical complexity)

Page 42: FLOW: A First-Language-Oriented Writing Assistant System

42

Example

Provide information on the use of dogs worldwide for law enforcement purposes.

sequential dependence model (dog, “law enforcement”) (information, “law enforcement”)

Page 43: FLOW: A First-Language-Oriented Writing Assistant System

43

Hypergraph structure

Query: “ international art crime “

Page 44: FLOW: A First-Language-Oriented Writing Assistant System

44

Evaluation