[B! NAACL][2015] sassanoのブックマーク

Semi-supervised Sentiment-aware LDA - skozawa's blog

NAACL 2015の論文のLCCT: A Semi-supervised Model for Sentiment Classificationを読んだ。メモと気になった部分を軽く実装して試してみた。感情分析で、コーパスベースのアプローチと辞書ベースのアプローチでco-trainingする話。コーパスベースのアプローチだと、精度は高いが再現率が低い。逆に辞書ベースのアプローチだと再現率は高いが精度が低いのでこれを解決したいというもの。ドメインに依存しない汎用的な素性をコーパスベースのアプローチで取得し、ドメインに依存するような素性は辞書ベースのアプローチでカバーする。ドメイン依存の単語はSemi-supervised Sentiment-aware LDAというアプローチを使って獲得する。ポジティブ、ネガティブ、ニュートラルの3分類でそれぞれに対していくつかシードを与えてることでド

sassano 2015/12/17

NAACL
2015

リンク

NAACL HLT 2015: Best papers at NAACL HLT 2015

Best papers at NAACL HLT 2015 Congratulations to the best paper award winners. This year NAACL HLT 2015 will have three best papers. One best long paper and two best student papers. The best paper awards were chosen by the best paper award committee which included Claire Cardie, Daniel Gildea, Daniel Marcu, Fernando Pereira and the program co-chairs: Joyce Chai and Anoop Sarkar who decided on the

sassano 2015/06/07

リンク

http://aclweb.org/anthology/N/N15/

sassano 2015/06/02

NAACL
2015

リンク

NAACL HLT 2015: NAACL-HLT 2015 Main Program Schedule

455: Unsupervised Induction of Semantic Roles within a Reconstruction-Error Minimization Framework, by Ivan Titov and Ehsan Khoddam We introduce a new approach to unsupervised estimation of feature-rich semantic role labeling models. Our model consists of two components: (1) an encoding component: a semantic role labeling model which predicts roles given a rich set of syntactic and lexical feature

sassano 2015/04/27

リンク

NAACL HLT 2015 on Twitter: "@NAACL2015 got 299 short paper and 385 long paper submissions! http://t.co/emiFRvd4Zu"

sassano 2015/03/11

NAACL
2015

リンク

NAACL HLT 2015: Accepted papers

List of accepted papers 9 TACL Posters 10 TACL Talks 55 Long Posters 62 Long Talks 45 Short Posters 24 Short Talks At the suggestion of the TACL co-editors-in-chief, the decision between talk and poster for TACL papers was made using random selection, with the exception of one paper that specifically requested a poster presentation. The decision between talk and poster for the NAACL 2015 long and

sassano 2015/02/28

リンク

Vector Space Modeling for NLP (NAACL 2015 workshop)

Workshop on Vector Space Modeling for NLP NAACL 2015, Denver, Colorado (June 5, 2015) Modern NLP started with methods based on pure symbolic analysis of language. Statistical methods were introduced to NLP in its current form in the 1980s/1990s, allowing "soft" reasoning about language, and made NLP more data-driven. Over the last decade another step has been taken in this direction -- it was prop