Question Answering

The shared task is over. Thank you all for participating!

• RESULTS

Participants may submit a paper for publication in the Conference Proceedings "Computational Linguistics and Intellectual Technologies" with a description of the solution and analysis of the results. Deadline for submission of Dialogue Evaluation articles: 8 April 11:59 pm. Please read the publishing rules for the Dialogue Evaluation track.

Links

• Repository on GitHub
• Telegram channel
• CodaLab

Key dates

• 25 December — publication of the train and dev datasets;
• 26 February — publication of the test datasets;
• 13 March — final submission deadline;
• 14-15 March — publication of the results;
• 8 April — paper submission deadline.

Task description

We invite participants to solve the problem of Visual Question Answering and Question Answering.

The corresponding dataset consists of photos of sales receipts with text in English and questions and answers to it. To answer the questions, participants need to apply the simplest aggregation operations to the extracted data (summation, multiplication, averaging, and others).

Example

1. A photo of a sales receipt is given;
2. A question in English is given to the photo, for example, «What is the total cost of tea before taxes?»;
3. In the second track, for all datasets, in addition to the photo, all text information from the photo is available for use when searching for an answer;
4. The taks is to determine the answer to the question based on textual information from the photo.