Feedback wanted: Intuitive scoring for just-in-time sets

オープン
#1,791 コメント 4 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 4 日以内に返信

まだ誰も着手していません。

評価

難易度
4/5
見積もり時間
3〜5日
初心者へのやさしさ
15/100
issue の種類
機能追加
明瞭さ
おおむね明確
活発さ
停滞
技術スタック
perl
領域
backend

調査の方向性

The proposal concerns the jitar_problem_adjusted_status routine in lib/WeBWorK/Utils.pm, where child-problem weights are summed into $childScore. Read that routine and the just-in-time set grading logic first, then decide how a new scoring mode would sit beside the current one. Done would mean a maintainer-agreed design, since the thread has no decision and the issue has been open about four years.

索引モデルが issue の本文から書いたものです。

説明

Problem

The current grading algorithm for just-in-time problem sets is rather unintuitive in my opinion, since for child problems it abuses the meaning of the value field to store some factor which - when considering all child problems which count towards the parent - results in a weight. Also, the points for a child problem shown to the student do not correspond to the actually achievable points if I'm not mistaken.

Solution

It would be better in my opinion if the value field retained its meaning (maximum points to reach for the child problem). The relevant jitar_problem_adjusted_status routine in /lib/WeBWorK/Utils.pm could simply proceed as follows when computing the status of a parent problem:

  1. Gather the weighted sum of all subproblems (as is currently done and saved in $childScore)
  2. If this weighted sum is greater than the value of the parent, return one as the status (i.e. when the values of all child problems counting towards the parent problem add up to something at least as large as the value of the parent problem, the score is capped)
  3. Otherwise divide the weighted sum by the parent problem's value and return the maximum of the result and the parent problem's current status.

Instead of the capping in step two, one could simply consider the case where child problems can yield more points than the parent when combined as illegal and throw an error when the user tries to import or save such a set, but this would unfortunately break backwards compatibility (although I'm sure that everybody will switch to the new algorithm...).

Advantages

In addition to being much more intuitive for the end user, this would also enable to apply penalties when a student reaches the child problems after not being able to solve the parent. With the current algorithm, I see no alternative to the insertion of a dummy child problem (always counted as false).

Instead of the capping in step two, it would be better to consider the case where child problems can yield more points than the parent when combined as illegal and throw an error when the user tries to import or save such a set. However, this would break backwards compatibility.

Implementation

I intend to add another routine for the adjusted status to Utils.pm and then offer either a checkbox labelled “use alternative jit scoring” or a dropdown “scoring algorithm for jit sets” with the current algorithm as default to keep backwards compatibility.

Before implementing this, I'd be thankful for some feedback, in particular whether this has a chance to be included in the main tree as I don't want to maintain a local fork.

主要言語
Perl
スター
164
フォーク
167
平均マージ
6日 7時間
マージ済み PR(30日)
27

環境構築

  • Dockerfile または Docker Compose ファイルあり
  • プルリクエストのテンプレートなし
  • コントリビューションガイドなし

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

openwebwork/webwork2 のほかの issue

openwebwork/webwork2 の issue をすべて見る

似ている issue

Perl の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。