firefox-main/browser/extensions/newtab/lib/SmartShortcutsRanker
Name Description Size
RankShortcuts.mjs Smart Shortcuts uses experimental prefs on newtabTrainhopConfig. These prefs can be accessed at prefValues.trainhopConfig.smartShortcuts enabled: do smart shortcuts (TopSitesFeed) over_sample_multiplier: number of rows of shortcuts to consider for smart shortcuts a user has n rows, we then query for n*over_sample_multiplier items to rank (TopSitesFeed) force_log: log shortcuts interactions regardless of enabled (SmartShortcutsFeed) features: arry of feature name strings eta: learning rate for feature weights click_bonus: multiplier applied to clicks positive_prior: thompson sampling alpha negative_prior: thompson sampling beta sticky_numimps: number of impressions for sticky clicks. 0 turns off thom_weight: weight of thompson sampling. divided by 100 frec_weight: weight of frecency. divided by 100 hour_weight: weight of hourly seasonality. divided by 100 daily_weight: weight of daily seasonality. divided by 100 bmark_weight: weight of is_bookmark. divided by 100 rece_weight: weight of recency. divided by 100 freq_weight: weight of frequency. divided by 100 refre_weight: weight of re-done frecency. divided by 100 open_weight: weight of is_open. divided by 100 unid_weight: weight of unique days visited. divided by 100 ctr_weight: weight of ctr. divided by 100 bias_weight: weight of bias. divided by 100 32999
RankShortcuts.worker.mjs 827
RankShortcutsWorkerClass.mjs Linear interpolation of values in histogram, wraps from end to beginning @param {number[]} hist Defines histogram of counts @param {number} t Time/Index we are interpolating to @returns {normed: number} Normalized number 18816
ThomSample.mjs This module has utility functions for doing thompson sampling and ranking 4585