DEIM Forum 2019 D3-5 Web 565 0871 1 5 Yahoo! JAPAN 102 8282 1 3 E-mail: {nakamura.tatsuya,hara}@ist.osaka-u.ac.jp, sufujita@yahoo-corp.jp Q&A Web Web Q&A Web Web 1 Web Web Web [2], [3], [10] Web Web [8], [13] [5] Web X Y Web Q&A Web Web 1
2 3 Wikipedia 1 Wikipedia Wikipedia Wikipedia 2 Web [3] Web Web Web Topic Detection and Tracking (TDT) Wang [9] Hong [5] [13] Web Wikipedia Web 1 https://www.wikipedia.org/ Web Web Bandari [2] Twitter 2 Twitter Yoshida [12] Web Wikipedia Wikipedia Web 3 3. 1 Web Web Twitter 140 3 Twitter Streaming API 4 Yahoo! 5 Q&A Web Yahoo! Yahoo! 6 Web 2 https://twitter.com/ 3 2017 11 7 1 280 4 https://developer.twitter.com/en/docs/tweets/sample-realtime/ overview 5 https://chiebukuro.yahoo.co.jp/ 6 https://search.yahoo.co.jp/
Yahoo! 200,000 6 Twitter Yahoo! Yahoo! Web 2016 10 1 2017 9 31 3. 2 Web Web Wikipedia Web 3. 2. 1 Yahoo! Yahoo! ( ID,, ) #texts/day 150,000 100,000 50,000 1 0 1 Oct-16 Nov-16 Dec-16 Jan-17 Feb-17 Mar-17 Apr-17 May-17 Jun-17 Jul-17 Aug-17 Sep-17 Web Web Twitter Yahoo! Yahoo! 25,700 9,047 98,497 10,832 5,145 42,267 30 1 Twitter Yahoo! e t n(e, t) R µ(e, R) σ(e, R) zscore Yahoo! MeCab 7 zscore(e, t, R) = n(e, t) µ(e, R) σ(e, R) (1) 3. 2. 2 Web TAGME [4] TAGME Wikipedia 2018 8 1 Wikipedia TAGME TAGME [4] TAGME iphone iphone Wikipedia [[ ( )]] [[IPhone XS]] 1 Web Twitter 8 Web Twitter 7 3. 2. 3 Web Web 7 MeCab ipadic 2.7.0 8 (1) 2016 12 24 25 (2) 2017 1 18 2 7 (3) 2017 6 10 7 11 zscore(e, t, R) θ e t R 7 zscore(e, t) θ = 5 1 Web Yahoo! Twitter Yahoo! Twitter 2 3. 2. 4 3. 2. 2 Web Yao [11]
Yao 10 [11] [1] A B W A W B A B R arg min W AR W B F. (2) R s.t. R R=I (2) [7]. Twitter Yahoo! Yahoo! 4 3 Web 4. 1 4. 2 4. 1 [13] 4. 1. 1 2 2 1,584 Twitter Yahoo! 2 E X E all Web 1,587 E t,c Twitter Yahoo! 301 E t,s Twitter Yahoo! 7,150 E c,s Yahoo! Yahoo! 553 E t Twitter 16,665 E c Yahoo! 6,609 E s Yahoo! 89,210 7,150 Web Web 2017 8 2017 9 13 iphone 8 iphone X Web 4. 2 4. 1. 2 Twitter Yahoo! 2 Web E t E c E s Web A t e B t ( = t) t t t t t > 0 e A t t B t t < 0 e A t t B 2 Web 2 Twitter Twitter Yahoo! Twitter
調査対象のトレンド エンティティに占める割合 4.50% 4.00% 3.50% 3.00% 2.50% 2.00% 1.50% 1.00% 0.50% 0.00% -14-12 -10-8 -6-4 -2 0 2 4 6 8 10 12 14 発 の差 (a) Twitter 3.00% 2.50% 2.00% 1.50% 1.00% 0.50% 0.00% -14-12 -10-8 -6-4 -2 0 2 4 6 8 10 12 14 発 の差 (b) Yahoo! 0.60% 0.50% 0.40% 0.30% 0.20% 0.10% 0.00% -14-12 -10-8 -6-4 -2 0 2 4 6 8 10 12 14 発 の差 (c) Yahoo! 2 3 Twitter Yahoo! Yahoo! 50% 1.89 (2.99) 2.67 (4.09) 2.83 (6.91) 30% 2.86 (5.11) 4.75 (9.07) 4.78 (10.12) 20% 4.21 (8.13) 6.77 (11.77) 7.17 (17.52) Yahoo! Yahoo! Web Twitter Yahoo! 7 1 Twitter Yahoo! Yahoo! Twitter Yahoo! Yahoo! Twitter Yahoo! Yahoo! Yahoo! Web Yahoo! Yahoo! 4. 2. 2 4. 1. 3 Twitter 2 Web E all {50, 30, 20}% 3 Web 3 Web 3 t Yahoo! Yahoo! 1% Web 3 Yahoo! Yahoo! Twitter Twitter Yahoo! Yahoo! Twitter Yahoo! [13] Yahoo! Yahoo! Twitter 4. 2. 1 Twitter 4. 2 3. 2. 4 1 Wikipedia 4. 2. 1 4. 1. 3 2 E all
70% 60% 50% 40% 30% 20% 10% 0% 60% 50% 40% 30% 20% 10% 0% 調査対象のトレンド エンティティに占める割合 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% (a) 50% (b) 30% (c) 20% 3 4 Twitter Yahoo! Yahoo! 0.5 1.69 (0.95) 1.63 (0.89) 1.88 (1.07) 0.3 2.25 (1.38) 2.06 (1.25) 2.33 (1.62) 0.2 2.59 (1.64) 2.35 (1.53) 2.63 (1.93) {0.5, 0.3, 0.2}. 4 Web 4 Web 4 t 50% Twiiter Yahoo! 30,20% Twitter Yahoo! 1% 4. 1. 3 Twitter Yahoo! Yahoo! 4. 2. 2 Twitter Yahoo! Twitter Yahoo! 5 Twitter Twitter E all 0.55 (0.21) 0.63 (0.16) 0.65 (0.18) E t,c 0.53 (0.27) 0.59 (0.20) 0.55 (0.24) E t,s 0.47 (0.28) 0.64 (0.18) 0.53 (0.29) E c,s 0.51 (0.24) 0.55 (0.24) 0.66 (0.19) E t 0.34 (0.33) 0.44 (0.31) 0.35 (0.35) E c 0.32 (0.34) 0.29 (0.35) 0.42 (0.34) E s 0.19 (0.33) 0.29 (0.34) 0.27 (0.35) 2 5 5 5 t Twitter Yahoo! E all E t,c E t,c E c,s Twitter Yahoo! E all E t,s E c E s Yahoo! Yahoo! E all E c,s E t,c E c,s 1% Twitter Yahoo! E t Yahoo! Yahoo! E c 4. 1. 2 Twitter
調査対象のトレンド エンティティに占める割合 0.6 0.5 0.4 0.3 0.2 0.1 0 調査対象のトレンド エンティティに占める割合 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 (a) 0.5 (b) 0.3 (c) 0.2 4 コサイン類似度 コサイン類似度 コサイン類似度 E all E t,c E t,s E c,s E t E c E s E all E t,c E t,s E c,s E t E c E s E all E t,c E t,s E c,s E t E c E s (a) Twitter Yahoo! (b) Twitter Yahoo! (c) Yahoo! Yahoo! 5 Yahoo! E t E c Yahoo! 4. 2. 3 Wikipedia Web Wikipedia Wikipedia 50 Wikipedia Wikipedia [6] 6 Wikipedia 6 Wikipedia 6 6 Wikipedia Twitter Yahoo! Yahoo! E all 0.70 (0.18) 0.73 (0.19) 0.66 (0.17) E t,c 0.75 (0.21) 0.77 (0.21) E t,s 0.55 (0.21) 0.56 (0.22) E c,s 0.66 (0.22) 0.61 (0.21) E t 0.50 (0.24) E c 0.59 (0.24) E s 0.45 (0.24) t 1% Yahoo! E all E t,c E c,s E c Wikipedia Yahoo! Wikipedia E all Wikipedia Wikipedia Wikipedia Wikipedia
分布間類似度 分布間類似度 分布間類似度 E all E t,c E t,s (a) Twitter E t E all E t,c E c,s (b) Yahoo! E c E all E t,s E c,s E s (c) Yahoo! 6 Wikipedia Wikipedia Wikipedia Wikipedia 5 Twitter Yahoo! Yahoo! Web Social Media: Forecasting Popularity, ICWSM (2012). [3] Chen, Y., Amiri, H., Li, Z. and Chua, T.-S.: Emerging Topic Detection for Organizations from Microblogs, SIGIR, pp. 43 52 (2013). [4] Ferragina, P. and Scaiella, U.: TAGME: On-the-fly Annotation of Short Text Fragments (by Wikipedia Entities), CIKM, pp. 1625 1628 (2010). [5] Hong, L., Dom, B., Gurumurthy, S. and Tsioutsiouliklis, K.: A Time- Dependent Topic Model for Multiple Text Streams, KDD, pp. 832 840 (2011). [6] Milne, D. and Witten, I. H.: An Effective, Low-cost Measure of Semantic Relatedness Obtained from Wikipedia Links, WikiAI, pp. 25 30 (2008). [7] Schönemann, P. H.: A Generalized Solution of the Orthogonal Procrustes Problem, Psychometrika, Vol. 31, No. 1, pp. 1 10 (1966). [8] Teevan, J., Ramage, D. and Morris, M. R.: #TwitterSearch: A Comparison of Microblog Search and Web Search, WSDM, pp. 35 44 (2011). [9] Wang, X., Zhai, C., Hu, X. and Sproat, R.: Mining Correlated Bursty Topic Patterns from Coordinated Text Streams, KDD, pp. 784 793 (2007). [10] Xie, W., Zhu, F., Jiang, J., Lim, E.-P. and Wang, K.: TopicSketch: Real-Time Bursty Topic Detection from Twitter, TKDE, Vol. 28, No. 8, pp. 2216 2229 (2016). [11] Yao, Z., Sun, Y., Ding, W., Rao, N. and Xiong, H.: Dynamic Word Embeddings for Evolving Semantic Discovery, WSDM, pp. 673 681 (2018). [12] Yoshida, M., Arase, Y., Tsunoda, T. and Yamamoto, M.: Wikipedia Page View Reflects Web Search Trend, WebSci, pp. 65:1 65:2 (2015). [13] (TOD) Vol. 9, No. 1, pp. 20 30 (2016). A(26240013) [1] Artetxe, M., Labaka, G. and Agirre, E.: Learning Principled Bilingual Mappings of Word Embeddings while Preserving Monolingual Invariance, EMNLP, pp. 2289 2294 (2016). [2] Bandari, R., Asur, S. and Huberman, B. A.: The Pulse of News in