Generation Sedigheh Mahdavi, Aijun An, Heidar Davoudi Davoudi, - - PowerPoint PPT Presentation

Sedigheh Mahdavi, Aijun An, Heidar Davoudi Davoudi, Marjan Delpisheh, Emad Gohari Gohari York University

Question-Worthy Sentence Selection for Question Generation

Automatic Question Generation

❑ Teaching machines to ask questions

2

Model

If there is an engine overheat warning , but no steam is seen or heard , the problem may not be too serious. What if there is an engine

• verheat warning but no

steam is seen or heard ?

Application

 Chat bot interfaces and smart conversational systems  Educational environments  Reading comprehension assessment

3

Application

 iNAGO Inc. Company

 Chat bot interfaces and AI conversational assistance  Car Manual

4

Neural Question Generation

❑ Seq2Seq models

5

Seq2SeqModel

If there is an engine overheat warning , but no steam is seen or heard , the problem may not be too serious. What if there is an engine

• verheat warning but no

steam is seen or heard ?

Paragraph-level Neural Question Generation

6

Challenge of Car Manual

❑ A small set of training data ❑ Lots of irrelevant sentences

7

Why not summarization methods?

8

This section provides a brief overview about some of the important features that may or may not be on your specific vehicle. for more detailed information, refer to each of the features which can be found later in this

• wner’s manual. The remote keyless entry (rke) transmitter is used to

remotely lock and unlock the doors from up to 60m (197ft) away from the vehicle. press to unlock the driver door. press unlock-symbol again within three seconds to unlock all remaining doors. press to lock all

• doors. lock and unlock feedback can be personalized. see vehicle

personalization.

Car manual data example

9

Feature-based Question-worthy Sentence Extraction

10

❑ Ranking features obtained from summarization methods ❑ POS-tag features: (verbs in a sentence, (nouns, adjectives, adverbs,… ) ❑ Tf/idf ❑ The length of a sentence ❑ The stop words ratio

context-based and sentence- based features The corresponding passage of the input sentence A classifier Question-worthy context (C)

Experiments

11

Experiments

12

The proposed Model

13

Feature-based question-worthy sentence selection context-based and sentence- based features The corresponding passage of the input sentence A classifier Question-worthy context (C) Question generation model Attention

Decoder Encoder

LSTM LSTM LSTM LSTM LSTM LSTM

The input sentence (S) Generated question ( Q) context-based and sentence- based features The corresponding passage of the input sentence A classifier Question-worthy context (C)

Experiments

14

Experiments

15

Thank you for your attention