diff options
Diffstat (limited to 'gnosis-algorithm.el')
-rw-r--r-- | gnosis-algorithm.el | 45 |
1 files changed, 31 insertions, 14 deletions
diff --git a/gnosis-algorithm.el b/gnosis-algorithm.el index e6a2b1f..48485eb 100644 --- a/gnosis-algorithm.el +++ b/gnosis-algorithm.el @@ -97,23 +97,24 @@ The structure of the given date is (YEAR MONTH DAY)." (+ ef (car gnosis-algorithm-ef))) (t (error "Invalid quality score passed to gnosis-algorithm-e-factor")))) -;; This should be further tested for notes with last-interval of 0 when success 0 -;; For future versions of this algorithm, we should also calculate -;; failures in row to have "leech" like notes as well. -;; TODO: Use initial-interval value instead gnosis-algorithm-interval -(defun gnosis-algorithm-next-interval (last-interval n ef success ff successful-reviews) + +(cl-defun gnosis-algorithm-next-interval (&key last-interval review-num ef success failure-factor successful-reviews successful-reviews-c fails-c fails-t initial-interval) "Calculate next interval. - LAST-INTERVAL : The number of days since the item was last reviewed. -- N : Number of times the item has been reviewed. +-review-num: Number of times the item has been reviewed. - EF : Easiness Factor. - SUCCESS : Success of the recall, ranges from 0 (unsuccessful) to 1 (successful). - FF: Failure factor - SUCCESSFUL-REVIEWS : Number of successful reviews. +- SUCCESSFULL-REVIEWS-C: Successful reviews in a row. +- FAILS-C: Failed reviews in a row. +- FAILS-T: Total failed reviews. +- INITIAL-INTERVAL: Initial intervals for successful reviews. Returns a list of: (INTERVAL N EF) where, - Next review date in (yyyy mm dd) format. -- N : Incremented by 1. +- REVIEW-NUM: Incremented by 1. - EF : Modified based on the recall success for the item." (cl-assert (and (>= success 0) (<= success 1))) @@ -132,23 +133,39 @@ Returns a list of: (INTERVAL N EF) where, ;; First successful review -> first interval ((and (= successful-reviews 0) (= success 1) - (< n 10) + (< review-num 10) (< ef 3.0)) - (car gnosis-algorithm-interval)) + (car initial-interval)) ;; Second successful review -> second interval ((and (= successful-reviews 1) - (< n 10) + (< review-num 10) (= success 1) - (< ef 3.0)) - (cadr gnosis-algorithm-interval)) + (< ef 3.0) + (= fails-c 0) + (cadr initial-interval))) + ;; When successful-reviews-c is above 3, use 150% or 180% + ;; of ef depending on the value of successful-reviews + ((and (= success 1) + (>= successful-reviews-c 3)) + (* (* ef (if (>= successful-reviews 10) 1.8 1.5)) last-interval)) + ((and (= success 0) + (> fails-c 3)) + ;; When fails-c is above 3, use 150% or 180% of + ;; failure-factor depending on the value of total failed + ;; reviews + (* (* failure-factor (if (>= fails-t 10) 1.8 1.5)) last-interval)) ;; For custom review sessions. + ;; When successful-reviews-c is above 0, multiply its value + ;; with ef ((and (= last-interval 0) (= success 1)) - (* ef 1)) + (* ef (if (> successful-reviews-c 0) + successful-reviews-c + 1))) ;; For everything else (t (if (= success 1) (* ef last-interval) - (* ff last-interval)))))) + (* failure-factor last-interval)))))) (list (gnosis-algorithm-date (round interval)) next-ef))) (provide 'gnosis-algorithm) |