SLIDE 1
Traditional Pipeline Designs for Task-
- riented Dialogue System
- Intent classifier
– Booking restaurants etc.
- Belief tracker
- Policy maker
- Dialogue generator
Sequicity: Simplifying Task-oriented Dialogue Systems with Single - - PowerPoint PPT Presentation
Sequicity: Simplifying Task-oriented Dialogue Systems with Single - - PowerPoint PPT Presentation
Sequicity: Simplifying Task-oriented Dialogue Systems with Single Sequence- to-Sequence Architectures We Wenqiang Lei , Xisen Jin, Zhaochun Ren, Xiangnan He, Min-Y en Kan, DaweiYin Traditional Pipeline Designs for Task- oriented Dialogue
SLIDE 2
SLIDE 3
- Complex belief trackers
- Fragility
- Templated response
Problems of Traditional Pipeline Designs
3
SLIDE 4
An End-to-end Solution
- Intent classifier
– Booking restaurants etc.
- Belief tracker
- Policy maker
- Response generator
An End-to-end Trainable Dialogue System (NDM) (Wen et al., 2017b) Tsung-Hsien Wen, David Vandyke, Nikola Mrksic, Milica Gasic, Lina M Rojas-Barahona, Pei-Hao Su, Stefan Ultes, and Steve Young. 2017b. A network-based end-to-end trainable task-oriented dialogue system. EACL .
SLIDE 5
- Complex belief trackers
- Pre-trained Belief Tracker
- Fragility
- Templated response
Some Problems Still Remains in NDM
5
SLIDE 6
Complex Belief Tracker In NDM
Food style Price range Open hour … Chinese food Expensive Before 11:00 pm … Japanese food Cheap … … French food … … … … … … … … ... … … Requiring address? Requiring phone number? Requiring name? … Yes Yes Yes … No No Know …
- Informable slots
- Requestable slots
SLIDE 7
Sequicity Solution
- Belief span
– <Inf>Italian;Cheap</Inf> <Req>Address</Req>
SLIDE 8
Sequicity Solution
- Belief span
– <Inf>Italian;Cheap</Inf> <Req>Address</Req>
SLIDE 9
Sequicity Solution
- Belief span
– <Inf>Italian;Cheap</Inf> <Req>Address</Req>
SLIDE 10
Sequicity Solution
- Belief span
– <Inf>Italian;Cheap</Inf> <Req>Address</Req>
SLIDE 11
Sequicity Solution
- Belief span
– <Inf>Italian;Cheap</Inf> <Req>Address; Phone</Req>
SLIDE 12
Sequicity Solution
- Belief span
– <Inf>Italian;Cheap</Inf> <Req>Address; Phone</Req>
- Notation
– Bt: belief span – Ut: user utterance – Rt: machine response
Source sequence Target sequence
SLIDE 13
Sequicity Illustration
SLIDE 14
Sequicity Illustration
Multiple match Single match No match
SLIDE 15
Optimization
- Joint log-likelihood
– Short coming: treating each word equally – E.g., The closest Italian restaurant is at <ad addr_s _slot>
- Reinforcement learning
– Action: decoding a word – State: hidden vectors generated by RNNs – Reward: decoding a correct placeholder +1, decoding each word -0.1
SLIDE 16
Experiments: Datasets
SLIDE 17
Experiment Results
SLIDE 18
Time Expenses on Belief Trackers
SLIDE 19
RL Helps with BLEU and Succ. F1
SLIDE 20
Removing CopyNets
SLIDE 21
Discussions: OOV Experiments
Synthesized OOV data: I would like some Chinese food. à I would like some Chinese_unk food.
SLIDE 22
Discussion: Parameter Scales
SLIDE 23
Discussion: Parameter Scales
SLIDE 24
Conclusion
- Sequicity provides another direction for task-
- riented dialogue systems.
- It is more light-weighted, can handle OOV
requests.
- It learns dialogue action directly from data with
less human interventions
– Requires more training data.
8/26/18 24