Hello!
My name is Avital Zar, and I am a computer science student at Ariel University. As part of my studies, I am required to implement the algorithms from an academic article and contribute the resulting code to a public library.
I am currently working on an article titled Handling Arbitrary Miscalibrations in Ratings (https://arxiv.org/pdf/1806.05085), and I believe the implementation would be a valuable addition to your library. Would you be open to me contributing the resulting code?
The article's core algorithm uses a probabilistic method to estimate the final ranking of a set of items evaluated by different reviewers. It leverages the relative rankings provided by the reviewers (i.e., if a reviewer scores item A higher than item B, A is ranked higher than B), as well as the raw scores.
The method works as follows: It first establishes a baseline ranking based solely on the reviewers' relative preferences. Then, for items that do not conflict with these relative rankings, it adjusts the order by flipping items with a probability x. This probability x is higher when the score difference between the two items being considered for a flip is large. This means an item with a significantly higher score is more likely to be ranked higher.
I would be very happy to contribute my code to your library, as I believe it aligns well with your existing projects.
Thank you for your time and consideration,
Avital Zar:)
Hello!
My name is Avital Zar, and I am a computer science student at Ariel University. As part of my studies, I am required to implement the algorithms from an academic article and contribute the resulting code to a public library.
I am currently working on an article titled Handling Arbitrary Miscalibrations in Ratings (https://arxiv.org/pdf/1806.05085), and I believe the implementation would be a valuable addition to your library. Would you be open to me contributing the resulting code?
The article's core algorithm uses a probabilistic method to estimate the final ranking of a set of items evaluated by different reviewers. It leverages the relative rankings provided by the reviewers (i.e., if a reviewer scores item A higher than item B, A is ranked higher than B), as well as the raw scores.
The method works as follows: It first establishes a baseline ranking based solely on the reviewers' relative preferences. Then, for items that do not conflict with these relative rankings, it adjusts the order by flipping items with a probability x. This probability x is higher when the score difference between the two items being considered for a flip is large. This means an item with a significantly higher score is more likely to be ranked higher.
I would be very happy to contribute my code to your library, as I believe it aligns well with your existing projects.
Thank you for your time and consideration,
Avital Zar:)