Covers: implementation of Question Answering System
Estimated time needed to finish: 15 minutes
Questions this item addresses:
  • What is QA over text?
  • How to use BERT for QA systems?
How to use this item?

It explains what it means to have QA functionlity over text. It also provides implementation of QA system using Google BERT

Author(s) / creator(s) / reference(s)
Naga Kiran
0 comment

FeTaQA: A new dataset for training QA systems for tabular data

Total time needed: ~2 hours
Enables readers to understand the advancements in the field of QA systems for tabular data. Also, compares the different datasets available for the same.
Potential Use Cases
Question Answering system for tables
Who is This For ?
INTERMEDIATEPeople interested in knowing how to train models for performing QA over tabules.
Click on each of the following annotated items to see details.
Resource Asset5/7
ARTICLE 1. Intro to Automated Question Answering
  • Introduction to QA systems?
6 minutes
ARTICLE 2. QA over text
  • What is QA over text?
  • How to use BERT for QA systems?
15 minutes
ARTICLE 3. QA over semi structured data
  • How does QA over table generally work?
  • How does TaPaS work?
20 minutes
PAPER 4. A survey on Semantic Parsing techniques
  • What is Semantic parsing task?
  • How is semantic task applicable to QA over tabular data?
10 minutes
ARTICLE 5. Issues with Semantic Parsing using logical formalism
  • What is label bias problem?
15 minutes
PAPER 6. Learning Semantic Parsers from Denotations with Latent Structured Alignments and Abstract Programs
  • What are weak supervision signals that are used in semantic parsers?
  • What are the problems with denotations ?
10 minutes
PAPER 7. FeTaQA : A Free form Table Question Answering dataset
  • What is the difference between Short form QA and Free form QA systems?
  • What are the datasets available for training models for QA systems ?
  • What are the data collection and annotation strategies used ?
  • What are the different modelling strategies used to test the performance on FeTaQA dataset?
36 minutes

Concepts Covered

0 comment