( : ) [12] [1], [4], [14], [17], [19] Yoshinaga Torisawa [19] Takamura Tsujii [14] ( : ) [11] Kurashima [9] [2], [5], [10], [13], [16] 2 3. 3. 1 (3. 2

DEIM Forum 2016 P2-3,, 113 8654 7 3 1 153 8505 4 6 1 184 8795 4 2 1 101 0003 2 1 2 E-mail: {nari,ynaga,toyoda,kitsure}@tkl.iis.u-tokyo.ac.jp ( : ) ( : ) Twitter 1.,,, [20] () ( ) { } ρ 2 SVM (Ranking Support Vector Machine) [7] SVR (Support Vector Regression) [6] 8 3 Twitter 2. [20] PMI (Point-wise Mutual Information)

( : ) [12] [1], [4], [14], [17], [19] Yoshinaga Torisawa [19] Takamura Tsujii [14] ( : ) [11] Kurashima [9] [2], [5], [10], [13], [16] 2 3. 3. 1 (3. 2 ) 3. 1 4 3 4 2 ( : ) ( : ) ( : ) ( : ) 1 ( : ) 1 3. 2 3. 1 SVM (Ranking Support Vector Machine) [7] SVR (Support Vector Regression) [6] 3. 2. 1 SVM SVM () 2 3. 1 3 () ( ) PMI (1) PMI(x, y) = log p(x, y) p(x) p(y) 2 PMI Turney [15] 2 1 ϕ(x) ( x = (, )) 1 ( : ) (1)

1 SVM ( ) ( ) ( ) ϕ(x) ϕ(x) ϕ(x) / / (2) 2 ϕ(x) (2) = SO () = PMI(, [ or ]) PMI(, [ or ]) = log p(, ) p( ) p(, ) p( ) 4 2 1 { } 1 2 ( 2 ) 1 ( ) SVM 1 3. 2. 2 SVR 3. 2. 1 SVM SVR SVR SVR n n! n min 2 ( ) ( ) SVR () ( 2 ) / N C 2 / N C 2 / N C 2 ( ) / N C 2 ( ) / N C 2 n min! ( ) () 2 1 2 SO () > SO ( ) 1 4 ( ) 5 SVR SVM N 2 4. 4. 1 4. 1. 1 200 2005 2013 2011 2013 Twitter 100 20 Twitter 290 60 Kaji [8] Yoshinaga [18] J.DepP 3 2 PMI 1 0 3 http://www.tkl.iis.u-tokyo.ac.jp/~ynaga/jdepp/

3 (a) ( ),,,,,,,,,, Ruby, Python, Perl, Java, JavaScript, Lisp, Scala, Haskell,,,,,,,,,, ( ),,,,,,, ROOKIES, GOEMON,,,,, ( ),,,,,, NEC,,,,,,,,,,, (b) ( ),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, Ruby, Python, Perl, Java, JavaScript, Lisp, Scala, Haskell,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, ( ),,,,,,, ( ),, TOKIO,, SMAP,, ZARD, EXILE,, ONEPIECE,,,,,,,,,, ROOKIES, GOEMON ( ),,,,,,,,, NEC,,,, (),,,,,,,,,,,,,,,,,,,, Chrome, FireFox, Safari, Opera, Sleipnir,,,,,,,,,,,,, ( ),,,,,,,,,,,,, 4. 1. 2 () 2009 10 1 Brown [3] 8 PMI ( ) 3

1 0.8 0.6 スピアマンの 相 関 係 数 ρ 0.4 0.2 0-0.2-0.4 [ 花, 綺 麗 だ] [ 宝, 上 品 だ] [アルコール, 美 味 い] [スポーツ, 楽 しい] [ 動 物, 賢 い] [ 乗 り 物, 快 適 だ] [ べ 物, 美 味 い] [ 楽 器, 地 よい] [プログラミング 語, 簡 単 だ] [ 動 物, かわいい] [ 野 菜, 美 味 しい] 1 3(a) [ 果 物, い] [ 電 化 製 品, 便 利 だ] [, 好 きだ] [ 類, かわいい] [ 天 候, 嫌 いだ] [ 国, 安 全 だ] [お 寺 ( 京 都 ), 有 名 だ] 20 30 (3 ) (3 ) (1 ) 7 ( 5 2 ) 7 ρ 3(b) 1 1 [] [] [] [] [] [] [] 4. 2 3 [ 歌 ( 本 ), 好 きだ] [アニメ, い] [ 邦 画, い] [ 芸 能 ( 性 ), 可 愛 い] [ 電 気 メーカー, 有 名 だ] [ 野 球 チーム (メジャー), 有 名 だ] [ファストフード, 美 味 い] [ メーカー, 好 調 だ] [コンビニ, 便 利 だ] [ブラウザ, 快 適 だ] [ 都 市 名, 安 全 だ] 1 v b 2 v t 3 (1) (2) ( v b, v t) [カフェ, 好 きだ] [ 地 名 ( 東 京 ), おしゃれだ] [ 出 版 社, 有 名 だ] 6 3 1 leave-one-out ρ SVM SVR LIBLINEAR 4 SVM SVR C 4. 3 4 [20] 4 https://www.csie.ntu.edu.tw/~cjlin/liblinear/

4 (a) [] Blog Tweet Blog + Tweet baseline svm svr baseline svm svr svm svr [] 0.943 0.886 0.943 0.600 0.829 0.600 0.771 0.886 [ ] 0.257 0.600 0.257 0.257 0.543 0.600 0.600 0.600 [ ] -0.024 0.071 0.119-0.024 0.190 0.500 0.048 0.619 [] 0.286 0.536 0.321 0.464 0.536 0.607 0.393 0.321 [] 0.900 0.600 0.700 0.700 0.900 0.900 0.700 0.900 ( ) [ ] 0.500 0.595 0.571 0.976 0.952 0.952 0.762 0.595 [ ] 0.321 0.357 0.250-0.071 0.429 0.429 0.321 0.571 ( ) [ ] -0.179 0.321 0.179 0.143 0.143 0.143 0.071 0.321 [ ] 0.190 0.310 0.262 0.738 0.429 0.786 0.262 0.452 [ ] 0.300 0.300 0.300-0.100-0.100-0.600-0.100-0.100 0.350 0.458 0.390 0.368 0.485 0.492 0.383 0.517 (b) ( human ) [] Blog Tweet Blog + Tweet human baseline svm svr baseline svm svr svm svr [] 0.749 0.286 0.167 0.095 0.190 0.476 0.476 0.238 0.214 [] 0.667 0.238 0.476 0.357 0.214 0.381 0.524 0.619 0.857 [] 0.648 0.167 0.690 0.857 0.524 0.762 0.762 0.667 0.786 [] 0.412 0.238 0.310 0.381 0.333 0.286 0.024 0.524 0.333 [ ] 0.578-0.200 0.143 0.257 0.600 0.086 0.257 0.143 0.143 [] 0.683 0.371 0.257 0.257 0.486 0.486 0.486 0.486 0.257 [] 0.639 0.143 0.393 0.679 0.143 0.500 0.286 0.393 0.214 [] 0.570-0.048 0.095 0.119-0.595-0.333-0.357-0.190-0.214 [] 0.826 0.476 0.619 0.786 0.762 0.881 0.905 0.786 0.810 [] 0.790 0.738 0.571 0.667 0.214 0.500 0.643 0.524 0.738 [] 0.451 0.524 0.429 0.262 0.071-0.286-0.048-0.429-0.024 [ ] 0.729 0.964 0.607 0.643 0.857 0.607 0.643 0.750 0.643 [] 0.772 0.536 0.679 0.679 0.143 0.750 0.857 0.714 0.786 [] 0.662-0.429 0.179 0.107-0.607-0.286-0.607 0.000 0.500 [] 0.800 0.881 0.929 0.905 0.929 0.810 0.929 0.905 0.905 [] 0.651 0.738 0.690 0.833 0.810 0.857 0.595 0.738 0.595 [] 0.804-0.500-0.200 0.000-0.300-0.600-0.300-0.700 0.200 ( ) [] 0.841 0.190 0.643 0.762 0.524 0.429 0.643 0.762 0.690 ( ) [] 0.614 0.762 0.667 0.857 0.857 0.571 0.095 0.548 0.048 [] 0.633-0.167 0.429-0.238 0.738 0.738 0.738 0.524 0.571 [] 0.649-0.071-0.107-0.036 0.107 0.286 0.107 0.000 0.286 ( ) [] 0.699 0.000 0.071 0.000 0.250 0.286 0.214 0.429 0.214 [] 0.644 0.381 0.643 0.333 0.810 0.738 0.643 0.857 0.690 () [] 0.864 0.976 0.905 0.952 0.762 0.762 0.810 0.929 0.952 [] 0.774 0.486-0.086-0.371 0.771 0.486-0.086 0.086-0.029 [] 0.665-0.900-0.700 0.000-0.100 0.400 0.900 0.700 0.000 [] 0.791 0.400 0.600 0.300 0.100 0.400 0.100 0.500 0.500 [] 0.856-0.200-0.100-0.100 0.700 0.700 0.700-0.100 0.500 [] 0.649 0.429 0.762 0.214-0.071 0.190 0.143 0.048 0.310 [] 0.405 0.071 0.571 0.393 0.357 0.179 0.179 0.607 0.286 ( ) [ ] 0.658 0.524 0.595 0.738 0.571 0.619 0.714 0.571 0.333 [] 0.916 0.786 0.964 0.857 0.893 0.929 0.893 0.929 0.893 0.690 0.275 0.403 0.392 0.376 0.425 0.402 0.424 0.437 SVM SVR

SVR 0.517 0.437 32 18 1 [ ] SVM SVR SVR SVM SVR 4. 4 [] [ ] [] SVM SVR 1 [] SVM SVR [] 1 [] [] [ ] ( : ) ( : ) 1 [ ] [ ] 5. ( : ) ( : ) SVM SVR SVM SVR

1 () () ( : ) ( ) ( : ) ( : ) ( : ) JSPS 25280111 [1] Sören Auer and Jens Lehmann. What Have Innsbruck and Leipzig in Common? Extracting Semantics from Wiki Content. In Proceedings of the 4th European Conference on The Semantic Web (ESWC), pp. 503 517, 2007. [2] Ralph A. Bradley and Milton E. Terry. Rank analysis of incomplete block designs: I. the method of paired comparisons. Vol. 39, pp. 324 345, December 1952. [3] Peter F. Brown, Peter V. desouza, Robert L. Mercer, Vincent J. Della Pietra, and Jenifer C. Lai. Class-Based n-gram Models of Natural Language. Computational Linguistics, Vol. 18, No. 4, pp. 467 479, December 1992. [4] Hsin-Hsi Chen, Shih-Chung Tsai, and Jin-He Tsai. Mining tables from large scale HTML texts. In Proceedings of the 18th conference on Computational linguistics (COLING), pp. 166 172, 2000. [5] Xi Chen, Paul N. Bennett, Kevyn Collins-Thompson, and Eric Horvitz. Pairwise ranking aggregation in a crowdsourced setting. In Proceedings of the Sixth ACM International Conference on Web Search and Data Mining (WSDM), pp. 193 202, 2013. [6] Harris Drucker, Christopher J. C. Burges, Linda Kaufman, Alexander J. Smola, and Vladimir N. Vapnik. Support Vector Regression Machines. In Advances in Neural Information Processing Systems 9, NIPS 1996, pp. 155 161. MIT Press, 1997. [7] Thorsten Joachims. Optimizing Search Engines Using Clickthrough Data. In Proceedings of the Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), pp. 133 142, 2002. [8] Nobuhiro Kaji and Masaru Kitsuregawa. Efficient Word Lattice Generation for Joint Word Segmentation and POS Tagging in Japanese. In Proceedings of the Sixth International Joint Conference on Natural Language Processing, pp. 153 161, Nagoya, Japan, October 2013. Asian Federation of Natural Language Processing. [9] Takeshi Kurashima, Katsuji Bessho, Hiroyuki Toda, Toshio Uchiyama, and Ryoji Kataoka. Ranking Entities Using Comparative Relations. In Proceedings of the 19th Conference on Database and Expert Systems Applications (DEXA), pp. 124 133, 2008. [10] Shuzi Niu, Yanyan Lan, Jiafeng Guo, and Xueqi Cheng. Stochastic rank aggregation. In Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence (UAI), pp. 478 487, 2013. [11] Bo Pang and Lillian Lee. Opinion Mining and Sentiment Analysis. Now Publishers Inc., 2008. [12] John Prager. Open-Domain Question Answering. Now Publishers Inc., 2007. [13] Karthik Raman and Thorsten Joachims. Methods for ordinal peer grading. In Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), pp. 1037 1046, 2014. [14] Hiroya Takamura and Jun ichi Tsujii. Estimating Numerical Attributes by Bringing Together Fragmentary Clues. In Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 1305 1310, 2015. [15] Peter Turney. Thumbs Up or Thumbs Down?: Semantic Orientation Applied to Unsupervised Classification of Reviews. In Proceedings of the 40th Annual Meeting on Association for Computational Linguistics, pp. 417 424, 2002. [16] Maksims N. Volkovs and Richard S. Zemel. A flexible generative model for preference aggregation. In Proceedings of the 21st International Conference on World Wide Web (WWW), pp. 479 488, 2012. [17] Fei Wu and Daniel S. Weld. Autonomously semantifying Wikipedia. In Proceedings of the sixteenth ACM conference on Conference on information and knowledge management (CIKM), pp. 41 50, 2007. [18] Naoki Yoshinaga and Masaru Kitsuregawa. Kernel Slicing: Scalable Online Training with Conjunctive Features. In COLING, pp. 1245 1253. Tsinghua University Press, 2010. [19] Naoki Yoshinaga and Kentaro Torisawa. Open-domain attribute-value acquisition from semi-structured texts. In Proceedings of the 6th International Semantic Web Conference (ISWC-07), Workshop on Text to Knowledge: The Lexicon/Ontology Interface (OntoLex-2007), pp. 55 66, 2007. [20],,,.. IPSJ SIG NL, Vol. 2013, No. 8, pp. 1 7, nov 2013.