English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Currently, open-domain generative dialog systems have attracted considerableattention in academia and industry. Despite the success of single-turn dialoggeneration, multi-turn dialog generation is still a big challenge. So far,there are two kinds of models for open-domain multi-turn dialog generation:hierarchical and non-hierarchical models. Recently, some works have shown thatthe hierarchical models are better than non-hierarchical models under theirexperimental settings; meanwhile, some works also demonstrate the oppositeconclusion. Due to the lack of adequate comparisons, it's not clear which kindof models are better in open-domain multi-turn dialog generation. Thus, in thispaper, we will measure systematically nearly all representative hierarchicaland non-hierarchical models over the same experimental settings to check whichkind is better. Through extensive experiments, we have the following threeimportant conclusions: (1) Nearly all hierarchical models are worse thannon-hierarchical models in open-domain multi-turn dialog generation, except forthe HRAN model. Through further analysis, the excellent performance of HRANmainly depends on its word-level attention mechanism; (2) The performance ofother hierarchical models will also obtain a great improvement if integratingthe word-level attention mechanism into these models. The modified hierarchicalmodels even significantly outperform the non-hierarchical models; (3) Thereason why the word-level attention mechanism is so powerful for hierarchicalmodels is because it can leverage context information more effectively,especially the fine-grained information. Besides, we have implemented all ofthe models and already released the codes.

Which Kind Is Better in Open-domain Multi-turn Dialog,Hierarchical or Non-hierarchical Models? An Empirical Study