Reward Models To Improve Question Specificity In Deep Learning Question Generation - ON-123
Preferred Disciplines and Level: Computer Science, Computational Linguistics, Computer Engineering. Post-Doc or Existing Ph.D. Student.
Company: Crater Labs Inc.
Project Length: 8-12 months (2 units)
Desired start date: As soon as possible
Location: Toronto, Ontario
No. of Positions: 1
Preferences: We would prefer to work with a university in the Greater Toronto Area or Southern Ontario. Language: English, Bilingual
About the Company:
We are a Toronto-based studio specializing in the use of computer vision, predictive analytics and natural language processing to build intelligence into business applications.
Recent research has indicated that a recurrent neural network can be trained to generate questions from documents. We seek to examine how common-sense reasoning methods can be employed in conjunction with these question generation models to create abstract and well posed questions.
- Implement a common-sense reasoning method within the RNN model as proposed in Machine Comprehension by Text-to-Text Neural Question Generation, to determine the degree to which such an approach can generate abstract and well-posed question.
- Apply this model to semi-structured business documents to determine if such a model can generate abstract and well-posed documents from provided source materials.
- Implement and train RNN based on reference contained in Machine Comprehension by Text-to-Text Neural Question Generation
- Examine efficacy of model on provided sample business documents
- Examine approaches to implement commonsense reasoning that may impact the ability of the RNN to generate abstract/interesting questions
Expertise and Skills Needed:
- Familiarity with basic natural language processing concepts
- Familiarity with deep learning libraries such as TensorFlow or Caffe
- Knowledge of Python, matplotlib, numpy, pandas and Jupyter Notebooks
For more info or to apply to this applied research position, please