Matches in SemOpenAlex for { <https://semopenalex.org/work/W3201429331> ?p ?o ?g. }
- W3201429331 abstract "Controllable text generation is an appealing but challenging task, which allows users to specify particular attributes of the generated outputs. In this paper, we propose a controllable dialogue generation model to steer response generation under multi-attribute constraints. Specifically, we define and categorize the commonly used control attributes into global and local ones, which possess different granularities of effects on response generation. Then, we significantly extend the conventional seq2seq framework by introducing a novel two-stage decoder, which first uses a multi-grained style specification layer to impose the stylistic constraints and determine word-level control states of responses based on the attributes, and then employs a response generation layer to generate final responses maintaining both semantic relevancy to the contexts and fidelity to the attributes. Furthermore, we train our model with an attribute consistency reward to promote response control with explicit supervision signals. Extensive experiments and in-depth analyses on two datasets indicate that our model can significantly outperform competitive baselines in terms of response quality, content diversity and controllability." @default.
- W3201429331 created "2021-09-27" @default.
- W3201429331 creator A5005043718 @default.
- W3201429331 creator A5025886127 @default.
- W3201429331 creator A5037949684 @default.
- W3201429331 creator A5053105539 @default.
- W3201429331 creator A5066326238 @default.
- W3201429331 creator A5067689496 @default.
- W3201429331 creator A5069371409 @default.
- W3201429331 date "2021-09-14" @default.
- W3201429331 modified "2023-09-25" @default.
- W3201429331 title "Controllable Dialogue Generation with Disentangled Multi-grained Style Specification and Attribute Consistency Reward" @default.
- W3201429331 cites W1522301498 @default.
- W3201429331 cites W1591706642 @default.
- W3201429331 cites W1958706068 @default.
- W3201429331 cites W2101105183 @default.
- W3201429331 cites W2123442489 @default.
- W3201429331 cites W2159640018 @default.
- W3201429331 cites W2250539671 @default.
- W3201429331 cites W2410983263 @default.
- W3201429331 cites W2411447339 @default.
- W3201429331 cites W2574872930 @default.
- W3201429331 cites W2581637843 @default.
- W3201429331 cites W2626778328 @default.
- W3201429331 cites W2784400615 @default.
- W3201429331 cites W2798888952 @default.
- W3201429331 cites W2884970917 @default.
- W3201429331 cites W2890276793 @default.
- W3201429331 cites W2890969459 @default.
- W3201429331 cites W2913755428 @default.
- W3201429331 cites W2916772188 @default.
- W3201429331 cites W2949888546 @default.
- W3201429331 cites W2950635152 @default.
- W3201429331 cites W2952420867 @default.
- W3201429331 cites W2955421862 @default.
- W3201429331 cites W2962717182 @default.
- W3201429331 cites W2962805889 @default.
- W3201429331 cites W2963084599 @default.
- W3201429331 cites W2963167310 @default.
- W3201429331 cites W2963332597 @default.
- W3201429331 cites W2963544536 @default.
- W3201429331 cites W2963939249 @default.
- W3201429331 cites W2963986868 @default.
- W3201429331 cites W2964199361 @default.
- W3201429331 cites W2964308564 @default.
- W3201429331 cites W2965718149 @default.
- W3201429331 cites W2970383515 @default.
- W3201429331 cites W2970419266 @default.
- W3201429331 cites W2970785611 @default.
- W3201429331 cites W2973049837 @default.
- W3201429331 cites W2973940173 @default.
- W3201429331 cites W3021582395 @default.
- W3201429331 cites W3030711095 @default.
- W3201429331 cites W3035451444 @default.
- W3201429331 cites W3074476581 @default.
- W3201429331 cites W3100236260 @default.
- W3201429331 doi "https://doi.org/10.48550/arxiv.2109.06717" @default.
- W3201429331 hasPublicationYear "2021" @default.
- W3201429331 type Work @default.
- W3201429331 sameAs 3201429331 @default.
- W3201429331 citedByCount "0" @default.
- W3201429331 crossrefType "posted-content" @default.
- W3201429331 hasAuthorship W3201429331A5005043718 @default.
- W3201429331 hasAuthorship W3201429331A5025886127 @default.
- W3201429331 hasAuthorship W3201429331A5037949684 @default.
- W3201429331 hasAuthorship W3201429331A5053105539 @default.
- W3201429331 hasAuthorship W3201429331A5066326238 @default.
- W3201429331 hasAuthorship W3201429331A5067689496 @default.
- W3201429331 hasAuthorship W3201429331A5069371409 @default.
- W3201429331 hasBestOaLocation W32014293311 @default.
- W3201429331 hasConcept C111472728 @default.
- W3201429331 hasConcept C127413603 @default.
- W3201429331 hasConcept C138885662 @default.
- W3201429331 hasConcept C154945302 @default.
- W3201429331 hasConcept C178790620 @default.
- W3201429331 hasConcept C185592680 @default.
- W3201429331 hasConcept C201995342 @default.
- W3201429331 hasConcept C2775924081 @default.
- W3201429331 hasConcept C2776436953 @default.
- W3201429331 hasConcept C2776459999 @default.
- W3201429331 hasConcept C2779227376 @default.
- W3201429331 hasConcept C2779530757 @default.
- W3201429331 hasConcept C2780451532 @default.
- W3201429331 hasConcept C28826006 @default.
- W3201429331 hasConcept C33923547 @default.
- W3201429331 hasConcept C41008148 @default.
- W3201429331 hasConcept C48209547 @default.
- W3201429331 hasConcept C76155785 @default.
- W3201429331 hasConcept C94124525 @default.
- W3201429331 hasConceptScore W3201429331C111472728 @default.
- W3201429331 hasConceptScore W3201429331C127413603 @default.
- W3201429331 hasConceptScore W3201429331C138885662 @default.
- W3201429331 hasConceptScore W3201429331C154945302 @default.
- W3201429331 hasConceptScore W3201429331C178790620 @default.
- W3201429331 hasConceptScore W3201429331C185592680 @default.
- W3201429331 hasConceptScore W3201429331C201995342 @default.
- W3201429331 hasConceptScore W3201429331C2775924081 @default.
- W3201429331 hasConceptScore W3201429331C2776436953 @default.
- W3201429331 hasConceptScore W3201429331C2776459999 @default.
- W3201429331 hasConceptScore W3201429331C2779227376 @default.