Ticket #1869 (assigned defect)

Opened 3 years ago

Last modified 3 years ago

Calculating scores for learning resources

Reported by: hans Assigned to: ayly (accepted)
Priority: critical Milestone:
Component: generic Version:
Keywords: Cc:
Time spent: Time remaining:
Time planned:

Description

I was looking at the scores of some learning resources that we made together with Mart. I will go through the whole collection. I hope that it will help us to improve the way how scores are calculated. This way we can get more meaningful tag clouds and search results.

The collection that I am discussing is here: http://lemill.net/community/people/Hans/collections/too-allikatega-viitamine-ja-viidete-haldamine-e-oppes

First content http://lemill.net/content/webpages/viitamine-ja-viidete-haldamine-e-oppes-pohimotted-moisted-reeglid/ has a score 1

I understand that anything that is created has score 1 in the beginning. I see the following ways how this resource could have more points than one:

  • it is published
  • it has been edited by 2 people
  • it has been edited 5 times by different people (first by me, then by Mart, then by me, then Mart, finally again by me)
  • it is part of 1 collection (it's my collection, if somebody else adds this to a collection, it should give even more points)

Simple way would be calculate scores like this: 1 (created) + 1 (published) + 2 (authors) + 5 (edits) + 1 (collection) = 10

It means that even any draft will also get immediately 3: 1 (created) + 1 (author) + 1 (edit) = 3

It looks better if we don't take into account first author and first edit: 1 (created) + 1 (published) + 1 (additional author) + 4 (additional edits) + 1 (collection) = 8

Second content http://lemill.net/content/webpages/folksonoomia-ja-uhisjarjehoidjad/ has again score 1

If we calculate the score the same way as we did with the first resource we would get: 1 (created) + 1 (published) + 0 (additional author) + 1 (additional edit) + 1 (collection) = 4

This resource has also 3 media pieces that make it look much better. We could get better score by adding points for media pieces: 1 (created) + 1 (published) + 0 (additional author) + 1 (additional edit) + 1 (collection) + 3 (media pieces) = 7

Third content http://lemill.net/content/webpages/viitamissusteemid-ja-viitekirjete-vormistamine/ has also score 1

Based on previous calculations it would have: 1 (created) + 1 (published) + 0 (additional author) + 3 (additional edits) + 1 (collection) + 0 (media pieces) = 6

Fourth content http://lemill.net/content/webpages/naited-korrektselt-vormistatud-apa-viitekirjetest/ has score 1

Based on previous calculations it would have: 1 (created) + 1 (published) + 0 (additional author) + 1 (additional edits) + 1 (collection) + 0 (media pieces) = 4

Fifth content http://lemill.net/content/webpages/viidete-haldamise-vahendid/ has score 1

It does not contain media pieces but it has embedded movies that make it also more attractive.

Let's calculate the score this way: 1 (created) + 1 (published) + 0 (additional author) + 4 (additional edits) + 1 (collection) + 0 (media pieces) + 2 (embedded resources) = 9

Sixth content http://lemill.net/content/webpages/oppematerjalid-ja-autorioigus-2/ has also score 1

The difference between this resource and previous resources is that it is assigned for a group. This is social behaviour that should give points.

The score could be calculated this way: 1 (created) + 1 (published) + 0 (additional author) + 0 (additional edits) + 1 (collection) + 0 (media pieces) + 0 (embedded resources) + 1 (assigned group) = 4

Seventh content http://lemill.net/content/webpages/avatud-sisulitsentsid/ has also score 1

It could be recalculated in the following way: 1 (created) + 1 (published) + 0 (additional author) + 1 (additional edit) + 1 (collection) + 1 (media piece) + 0 (embedded resources) + 1 (assigned group) = 6

Eighth content http://lemill.net/content/webpages/teadusartiklite-andmebaaside-kasutamine/ has a score 1

It could be recalculated in the following way: 1 (created) + 1 (published) + 0 (additional author) + 1 (additional edit) + 1 (collection) + 2 (media pieces) + 0 (embedded resources) + 0 (assigned groups) = 6

Ninth content http://lemill.net/content/webpages/opiobjektide-repositooriumid-ja-referatooriumid/ has score 1

It could be recalculated in the following way: 1 (created) + 1 (published) + 0 (additional author) + 2 (additional edits) + 1 (collection) + 0 (media pieces) + 0 (embedded resources) + 0 (assigned groups) = 5

Conclusions

With this kind of scoring we will get much better distribution of scores: 8-7-6-4-9-4-6-6-5 is better than 1-1-1-1-1-1-1-1-1.

If I remember correctly, then it gives 10 points when somebody adds your resource to a collection? The difference between 6 and 16 looks much more fair than the difference between 1 and 11. So I would leave this value to 10 points.

Another social activity is editing somebody's learning resource. If I look at my calculations, then maybe it makes sense to give more points for additional authors. This way we encourage collaborative editing and resources that have more authors stand out better. I suggest that we should try with giving 5 points for additional authors. Sometimes additional authors don't show the goodness of content, but simply show that there was something wrong and it had to be fixed.

At the end let's see how our calculations would work for one of the most popular resources in this http://lemill.net/content/browse_resources tag cloud (everything looks so f...ing equal there!!!).

The most popular resource seems to be http://lemill.net/content/webpages/moistekaardi-loomine-tarkvaraga-cmaptools that has a score 23.

I'm not sure how this 23 comes - this resource is part of 7 collections, 5 of these are by other users.

Based on our calculation formula it would get the following score: 1 (created) + 1 (published) + 1x5 (additional author) + 1 (additional edit) + 2 (collections) + 0 (media pieces) + 0 (embedded resources) + 0 (assigned groups) + 5 x 10 (collections by other people) = 60

What do you think about this?

Change History

01/05/09 14:14:07 changed by pjotr

This looks quite interesting and should be a lot better than the thing we have now.

So for comparison I will write here how we calculate scores atm:

For all the cases >>> Minimum score is always 1.

GroupBlog >>> 1 point for each member; max 1 point for materials (is devided by number of months old plus 1); max one point for every post in the blog (same ideas as with previous)

MemberFolder? >>> 1 point for each piece; 10 points for each learning resource, method or tool; 10 point for every teaching and learning story; 1 point for each post; 1 point for each contact.

Resource >>> 1 point for each unique member to have current resource in collection; 10 point for every story.

Piece >>> 1 point for each resource using piece; 1 point in case of non-default title; 1 point if piece has tags; 1 point if piece has description.

One additional thing that might also need some score to be set is Collection. ATM it will always get it's score as 1.

01/08/09 17:38:13 changed by ayly

  • owner changed from anonymous to ayly.
  • status changed from new to assigned.

02/17/09 14:34:29 changed by ayly

(In [2777]) References #1874, collection description has to be at least 200 character long for collection being displayed on front page and References #1869, recalculating scores for resources.