OnizukaLab / ConferenceProceedings

NLP 班論文読み会用のリポジトリ
8 stars 1 forks source link

Generating More Interesting Responses in Neural Conversation Models with Distributional Constraints #53

Open junya-takayama opened 5 years ago

junya-takayama commented 5 years ago

一言でいうと

発話に対応した内容語を応答文にうまく出せるようなデコード方法を提案

論文リンク

Generating More Interesting Responses in Neural Conversation Models with Distributional Constraints

著者

Ashutosh Baheti, Alan Ritter, Jiwei Li, and Bill Dolan

会議日付

2018/11

概要

最尤推定を用いて text-to-text 生成モデルを構築すると,応答生成などの多対多タスクではありきたりな応答(dull response)を生成しがちになる. 入力文によく対応する内容語を出力させるために,本研究では最尤推定結果に加えて入力文と出力文の ・トピック一貫性 sim(P(T|X),P(T|Y)) ・意味的類似度 sim(Emb(X), Emb(Y)) を考慮したデコード手法を提案

新規性・差分

入力文のトピック推定結果のみを用いた既存モデルに対し,本研究では入力と出力両方のトピック推定結果をデコード時に逐次的に用いており,より内容語が出やすくなっている

評価

使用データ

訓練 :OpenSubtitles テスト:Cornell Movie Dialogue Corpus

評価指標

BLEU, distinct, 人手など

比較手法

MMI: 応答と入力の相互情報量を最大化するモデル TA-seq2seq: 入力文のトピック推定結果を Attention 計算に用いるモデル

結果

2018-12-18 9 42 42 2018-12-18 9 40 56

コメント