system setting: System metrics

System metrics is a metric monitoring page which is related to ConneXing recommender services.

It can be accessed by opening the side bar >> System settings >> System metrics:

Each section in System metrics, presents available metrics for one of the ConneXing recommendation services. Note that not all metrics might be available for all recommendation services.

Here is a brief summary of each metric:

• Coverage: Coverage is the percentage of recommended user-item pairs over the total number of potential pairs. For example: If we have m user and n items in the system, all possible pair of recommendations would be n * m. Let’s say we have recommended x-pairs in total, so the model coverage is: x / m * n
• Diversity: Diversity measures how dissimilar recommended items are for a user.
  
• Novelty: Novelty is a measure of how popular the items that the system recommends are. Novelty may be similar to diversity. One can achieve high novelty by randomly recommending items since most of the items are not in the popular item list.
  
• Hit rate: In recommender settings, the hit ratio is the fraction of users for which the correct answer is included in the recommendation list. As one can see, the larger the list is, the higher hit ratio becomes, because there is a higher chance that the correct answer is included in the recommendation list. For example: If we generate top-10 recommendations for a user and 5 of those recommendations are items the user currently have, our recommendation hit rate for this user is 50%. By finding this number for all users, the average of those numbers would be our model hit rate.