wiki:GSoC2017/LetorClickstream/ProjectTimeline

Coding Week 1: May 30–June 5

  • Start working on the implementation of Log template.

Coding Week 2: June 6–June 12

  • Complete the implementation of Log template.
  • Record the click data in the required format from Log template.

Coding Week 3: June 13–June 19

  • Document the click data format.
  • Generate Query file from the click data.

Coding Week 4: June 20–June 26 (Phase 1 Evaluation: June 26–30)

  • Submit PRs for Phase 1 Evaluation.
  • Discuss the prototype for DBN model implementation.

Coding Week 5: June 27–July 3

  • Start the implementation of DBN model.

Coding Week 6: July 4–July 10

  • Continue working on the implementation of DBN model.

Coding Week 7: July 11–July 17

  • Complete the implementation of DBN model.
  • Start testing the DBN model implementation and write automated tests.

Coding Week 8: July 18–July 24 (Phase 2 Evaluation: July 24–28)

  • Complete the testing of DBN model.
  • Document various components of the model and how to use the model.
  • Obtain the relevance judgements dataset for training the DBN model.

Coding Week 9: July 25–July 31

  • Train the implemented DBN model on relevance judgements dataset.
  • Obtain the predicted relevance judgments on the Omega click data.

Coding Week 10: August 1–August 7

  • Implement the mechanism to automatically generate Qrel file.

Coding Week 11: August 8–August 14

  • Train the letor module on training file obtained from Omega click data.
  • Use the letor module for displaying relevant search results on top of SERP.

Coding Week 12: August 15–August 21 (Final Evaluation: August 21–29)

  • Keeping this week as buffer to accommodate delays in the planned schedule due to unprecedented events.

Final Evaluations: August 22–August 29

I will further clean my code and improve documentation. After consulting with my mentors, I’ll submit my code samples. This will mark the successful completion of my project.

Last modified 7 years ago Last modified on 09/06/17 10:39:20
Note: See TracWiki for help on using the wiki.