Research Article

A Bichannel Transformer with Context Encoding for Document-Driven Conversation Generation in Social Media

Table 1

Quantitative evaluation results for baselines and the proposed models in terms of BLEU scores and METEOR score.

ModelsBLEU-1BLEU-2BLEU-3BLEU-4METEOR

SEQ [24]§6.121.520.590.304.18
SEQS [24]§6.571.650.670.354.30
D3G [26]§6.321.710.710.414.17
Transformer [25]8.552.491.120.604.53
Incremental transformer [44]0.95
Incremental transformer (our impl)8.192.881.660.855.21

BCTCE9.983.562.051.425.22

Results marked with § are trained and evaluated with the source code from [26], results marked with are trained and evaluated with our implemented code, results marked with are from [44], and results marked with are trained and evaluated with the code published by Li et al. [44].