310 likes | 414 Views
This study explores various strategies for selecting news based on specific criteria, such as individual ratings and group lists, to determine the most suitable news items. The experiment involved 36 subjects who rated news items based on importance, relevance, and recency. Results from different strategies like Average, Least Misery, and Average Without Misery were compared to evaluate their effectiveness in news selection. Insights were gained on dealing with inequality in news importance and adapting strategies to address these issues.
E N D
Selecting news to suit a group of criteria: an exploration Judith Masthoff University of Brighton
Average Strategy Use average of the individual ratings Group list: (E, F) H (D, J) A I B G C
Least Misery Strategy Use minimum of the individual ratings Group list: F E (H, J, D) G B I C A
Average Without Misery Strategy Use average of the individual ratings, excluding items below threshold (say 4) Group list: (E, F) H (D, J) B G
Experiment: How people do it • 36 subjects • Given ratings of news items on three criteria • Which to watch if only time for 1 item, 2 items, …, 7 items
Criteria used • Importance of the news in general • Relevance of the location to you • Recency of the news
Criteria used • Importance of the news in general • Relevance of the location to you • Recency of the news
Results - Part 1 Red = Average Strategy
Results - Part 1 Red = Average Strategy Shaded =Least Misery Strategy
Results - Part 2 Red = Average Strategy
Results - Part 2 Red = Average Strategy Red dotted = Average Without Misery Strategy
Introduction of A and I A (10, 1, 10) I (10, 3, 7) Grey = This experiment Black = Old experiment
Average Strategy Use average of the individual ratings Group list: (E, F) H (D, J) A I B G C
Average over Important Criteria Use average of the individual ratings Group list: A E (F, I) (H, J) D G B C
Least Misery Strategy Use minimum of the individual ratings Group list: F E (H, J, D) G B I C A
Least Misery For Important Criteria Use minimum of the individual ratings Group list: A F E I (H, J, D) G B C
Average Strategy Use average of the individual ratings Group list: (E, F) H (D, J) A I B G C
Average Strategy with Weights Use average of the individual ratings Group list: E A F I H J D G B C (3-1-3) E F A (H,I) J D G B C (2-1-2)
Average Without Misery Strategy Use average of the individual ratings, excluding items below threshold (say 6) Group list: (E, F) H (D, J)
Average Without Misery for Some Use average of the individual ratings, excluding items below threshold (say 6) Group list: (E, F) H (D, J) A I (threshold 6) (E, F) A I (threshold 7)
Results - Part 1 Red = Average Strategy Shaded =Least Misery Strategy
Results - Part 1 Red = Average Strategy Shaded =Least Misery Strategy Blue dotted = Average Without Miseryfor Some
Results - Part 2 Red = Average Strategy Dotted = Average Without Misery
Results - Part 2 Red = Average Strategy Blue = Average With Weights Dotted = Average Without Misery (blue: for Some)
Results - Part 3 Yellow shading = Least Misery for Important Criteria
Conclusions • Similarity between multiple criteria and multiple people • Main issue is inequality in importance • Adapted strategies help to explain a bit • But more work needed