Aid is Out There: Looking for Help from Tweets during a Large Scale Disaster
Varga, István and Sano, Motoki and Torisawa, Kentaro and Hashimoto, Chikara and Ohtake, Kiyonori and Kawai, Takao and Oh, Jong-Hoon and De Saeger, Stijn

Article Structure

Abstract

The 2011 Great East Japan Earthquake caused a wide range of problems, and as countermeasures, many aid activities were carried out.

Introduction

The 2011 Great East Japan Earthquake in March 11, 2011 killed 15,883 people and destroyed over 260,000 households (National Police Agency of Japan, 2013).

Approach

Problem Report and Aid Message Recognizers

We recognize problem reports and aid messages in given tweets using a supervised classifier, SVMs with linear kernel, which worked best in our preliminary experiments.

Problem-Aid Match Recognizer

After problem report and aid message recognition, the positive outputs of the respective classifiers are used as input in this step.

Experiments

We evaluated our problem report recognizer and problem-aid match recognizer.

Related Work

Twitter has been observed as a platform for situational awareness during various crisis situations (Starbird et al., 2010; Vieweg et al., 2010), as sensors for an earthquake reporting system (Sakaki et al., 2010; Okazaki and Matsuo, 2010) or to detect epidemics (Aramaki et al., 2011).

Conclusions

In this paper, we proposed a method to discover matches between problem reports and aid messages from tweets in large-scale disasters.

Topics

dependency relation

Appears in 9 sentences as: dependency relation (7) dependency relations (4)
In Aid is Out There: Looking for Help from Tweets during a Large Scale Disaster
  1. An underlying assumption of our method is that we can find a noun-predicate dependency relation that works as an indicator of problems and aids in problem reports and aid messages, which we refer to as problem nucleus and aid nucleus.1 An example of problem nucleus is “infant formula is sold out” in P1, and that of aid nucleus is “(can) buy infant formula” in A1.
    Page 2, “Introduction”
  2. tweet dependency relation
    Page 2, “Approach”
  3. Then, each tweet is paired with each dependency relation in the tweet, which is a candidate of problem/aid nuclei and given to the problem report and aid message recognizers.
    Page 2, “Approach”
  4. We observed that problem reports in general included either of (A) a dependency relation between a noun referring to some trouble and an excitatory template or (B) a dependency relation between a noun not referring to any trouble and an inhibitory template.
    Page 3, “Approach”
  5. We assume that if we can find such dependency relations in tweets, the tweets are likely to be problem reports.
    Page 3, “Approach”
  6. Contrary, a tweet is more likely to be an aid message when it includes either (C) a dependency relation between a noun referring to some trouble and an inhibitory template or (D) a dependency relation between a noun not referring to any trou-
    Page 3, “Approach”
  7. (We can find the dependency relations sharing “Sendai” but they do not express anything about the contents of problem and aid.)
    Page 4, “Approach”
  8. 5The original similarity was defined over noun pairs and it was estimated from dependency relations .
    Page 6, “Experiments”
  9. Obtaining similarity between template pairs, not noun pairs, is straightforward given the same dependency relations .
    Page 6, “Experiments”

See all papers in Proc. ACL 2013 that mention dependency relation.

See all papers in Proc. ACL that mention dependency relation.

Back to top.

feature set

Appears in 9 sentences as: feature set (9)
In Aid is Out There: Looking for Help from Tweets during a Large Scale Disaster
  1. The feature set given to the SVMs are summarized in the top part of Table 2.
    Page 4, “Problem Report and Aid Message Recognizers”
  2. Note that we used a common feature set for both the problem report recognizer and aid message recognizer and that it is categorized into several types: features concerning trouble expressions (TR), excitation polarity (EX), their combination (TREXl) and word sentiment polarity (WSP), features expressing morphological and syntactic structures of nuclei and their context surrounding problem/aid nuclei (MSA), features concerning semantic word classes (SWC) appearing in nuclei and their context, request phrases, such as “Please help us”, appearing in tweets (REQ), and geographical locations in tweets recognized by our location recognizer (GL).
    Page 4, “Problem Report and Aid Message Recognizers”
  3. We also attempted to represent nucleus template IDs, noun IDs and their combinations directly in our feature set to capture typical templates fre-
    Page 4, “Problem Report and Aid Message Recognizers”
  4. Here also we attempted to capture typical or frequent matches of nuclei using template and noun IDs and their combinations, but we did not observe any improvement so we omit them from the feature set .
    Page 5, “Problem-Aid Match Recognizer”
  5. The bottom part of Table 2 summarizes the additional feature set , some of which are described below in more detail.
    Page 5, “Problem-Aid Match Recognizer”
  6. In both experiments we observed that the performance drops when excitation polarities and trouble expressions are removed from the feature set .
    Page 6, “Experiments”
  7. PROPOSED-*: The proposed method without the feature set denoted by “*”.
    Page 6, “Experiments”
  8. PROPOSED-*z The proposed method without the feature set denoted by “*”.
    Page 8, “Experiments”
  9. did not use excitation polarities and trouble expressions in its feature set .
    Page 8, “Experiments”

See all papers in Proc. ACL 2013 that mention feature set.

See all papers in Proc. ACL that mention feature set.

Back to top.

sentiment analysis

Appears in 5 sentences as: sentiment analysis (5)
In Aid is Out There: Looking for Help from Tweets during a Large Scale Disaster
  1. An evident alternative to this approach is to use sentiment analysis (Mandel et al., 2012; Tsagkali-dou et al., 2011) assuming that problem reports should include something ‘bad’ while aid messages describe something ‘good’.
    Page 2, “Introduction”
  2. Word Sentiment Polarity (WSP) As we suggested before, full-fledged sentiment analysis to recognize the expressions, including clauses and phrases, that refer to something good or bad was not effective in our task.
    Page 5, “Problem Report and Aid Message Recognizers”
  3. To identify the sentiment polarity of words, we employed the word sentiment polarity dictionary used with a sentiment analysis tool for Japanese, the Opinion Extraction Tool software2, which is an implementation of Nakagawa et al.
    Page 5, “Problem Report and Aid Message Recognizers”
  4. Note that we used the Opinion Extraction Tool in the experiments to check the effectiveness of the full-fledged sentiment analysis in this task.
    Page 5, “Problem Report and Aid Message Recognizers”
  5. This suggests that full-fledged sentiment analysis is not effective at least in this setting.
    Page 7, “Experiments”

See all papers in Proc. ACL 2013 that mention sentiment analysis.

See all papers in Proc. ACL that mention sentiment analysis.

Back to top.

n-grams

Appears in 4 sentences as: n-grams (6)
In Aid is Out There: Looking for Help from Tweets during a Large Scale Disaster
  1. MSAl Morpheme n-grams, syntactic dependency n-grams in the tweet and morpheme n-grams before and after the nucleus template.
    Page 4, “Problem Report and Aid Message Recognizers”
  2. MSA2 Character n-grams of the nucleus template to capture conjugation and modality variations.
    Page 4, “Problem Report and Aid Message Recognizers”
  3. MSA3 Morpheme and part-of-speech n-grams within the bunsetsu containing the nucleus template to capture conjugation and modality variations.
    Page 4, “Problem Report and Aid Message Recognizers”
  4. SWCl 'Ihe semantic class n-grams in the tweet.
    Page 4, “Problem Report and Aid Message Recognizers”

See all papers in Proc. ACL 2013 that mention n-grams.

See all papers in Proc. ACL that mention n-grams.

Back to top.

F-score

Appears in 3 sentences as: F-score (3)
In Aid is Out There: Looking for Help from Tweets during a Large Scale Disaster
  1. The proposed method achieved about 44% recall and nearly 80% precision, outperforming all other systems in terms of precision, F-score and average precision8.
    Page 6, “Experiments”
  2. Table 4: Recall (R), precision (P), F-score (F) and average precision (aP) of the problem report recognizers.
    Page 7, “Experiments”
  3. Table 6: Recall (R), precision (P), F-score (F) and average precision (aP) of the problem-aid match recognizers.
    Page 8, “Experiments”

See all papers in Proc. ACL 2013 that mention F-score.

See all papers in Proc. ACL that mention F-score.

Back to top.

RULE-BASED

Appears in 3 sentences as: RULE-BASED (2) rule-based (1)
In Aid is Out There: Looking for Help from Tweets during a Large Scale Disaster
  1. RULE-BASED : The method that regards only nuclei satisfying the constraint in Table l as problem nuclei.
    Page 6, “Experiments”
  2. The rule-based method achieved relatively high precision despite of the low recall, demonstrating the importance of problem and aid nuclei formulations described in Section 1.
    Page 7, “Experiments”
  3. RULE-BASED : The method that judges only problem-aid nuclei combinations with opposite excitation polarities as proper matches.
    Page 8, “Experiments”

See all papers in Proc. ACL 2013 that mention RULE-BASED.

See all papers in Proc. ACL that mention RULE-BASED.

Back to top.

statistically significant

Appears in 3 sentences as: statistically significant (3)
In Aid is Out There: Looking for Help from Tweets during a Large Scale Disaster
  1. The improvement in precision when using TR&EX is statistically significant (p < 0.05).9 Note that F-measure dropped
    Page 6, “Experiments”
  2. The improvement in precision when using TR&EX is statistically significant (p < 0.01).
    Page 7, “Experiments”
  3. statistically significant in both settings (p < 0.01).
    Page 9, “Experiments”

See all papers in Proc. ACL 2013 that mention statistically significant.

See all papers in Proc. ACL that mention statistically significant.

Back to top.