# gift-syntax/dsd-tutor-rules

Jump to navigation
Jump to search

- for GIFT Format quizes

Adapted and annotated from figures in del Soldato, T. & du Boulay, B. (1995) "Implementation of Motivational Tactics in Tutoring Systems," *Journal of Artificial Intelligence in Education,* **6**(4): 337-78:

# perfomance modeling - t6p355rP1-2 Rule P1 -- perfomance model successful: if agreements/completed > perf-threshold then perf-value = successful Rule P2 -- perfomance model unsuccessful: if agreements/completed <= perf-threshold then perf-value = unsuccessful # repitition testing - t6p355rP3-4 Rule P3 -- path model lost: if tries > step-repition-limit (from problem type) then path-state = lost Rule P4 -- path model OK: if tries <= step-repition-limit (from problem type) then path-state = OK # confidence modeling - t7p356rC1-2 Rule C1: if answer was low-confidence (either agree or disagree) then decrement conf-value Rule C2: if answer was high-confidence (either agree or disagree) then increment conf-value Rule C3 -- decrease modeled confidence on help request: if answer was help request then decrement conf-value Rule C4: if problem-state = succeeded and without help then increment conf-value by large-conf-inc Rule C5: if problem-state = succeeded and with help then increment conf-value Rule C6: if problem-state = failed and without help then decrement conf-value Rule C7: if problem-state = failed and with help then decrement conf-value by large-conf-dec # effort modeling - t8p357 (non-rule) if path-state = lost then effort-value = "medium" + 1 if problem-state = succeeded then effort-value += 2 if help was asked for then effort-value += 1 # independence modelling - t9p358rI1-I4 Rule I1: if help-state = suggesting then decrement indep-value Rule I2: if help-state = providing and detail = general then decrement indep-value Rule I3: if help-state = providing and detail = specific then decrement indep-value by large-indep-dec Rule I4: if help-state = rejected then increment indep-value by large-indep-inc # domain-based planner rules - t10p359rD1-D7 Rule D1: if problem-state = succeeded then provide right assessment; suggest harder problem Rule D2: if problem-state = failed then provide wrong assessment; suggest same-diff problem Rule D3: if problem-state = given-up then suggest same-diff problem Rule D4: if problem-state = rejected then suggest same-diff problem Rule D5: if help-state = requested then provide help Rule D6: if help-state = rejected then help not-needed Rule D7: if path-state = lost then provide help # motivational planner rules - t11p360rM1-M20 Rule M1: if conf-value < conf-threshold then increase confidence # when and how is this reset? Rule M2: if effort-value < "medium" then increase effort Rule M3: if effort-value > "medium" then maintain effort Rule M4: if help-state = rejected then respect control Rule M5: if problem-state = given-up # "above give-up-lim" then respect control Rule M6: if indep-value < indep-threshold and not increase confidence then increase control Rule M7: if problem-state = succeeded and increase confidence then increase experience success Rule M8: if problem-state = failed and increase confidence then facilitate success Rule M9: if problem-state = given-up and increase effort (M2) and not increase confidence and not respect control then encourage effort Rule M10: if problem-state = given-up and increase confidence and not respect control then facilitate success Rule M11: if problem-state = succeeded and increase effort then stimulate challenge Rule M12: if stimulate challence (M11) and increase confidence (M1) then emphasise promotion Rule M13: if problem-state = failed and increase effort and not increase confidence then stimulate curiosity Rule M14: if perf-value = successful and facilitate success and increase effort then remind success Rule M15: if path-state = lost and increase control then avoid intervention Rule M16: if help-state = requested and increase control then encourage independence Rule M17: if encourage independence then avoid intervention Rule M18: if help-state = rejected and help-skip-next = no (needs to be fitted to "skip next help" or whatever) and respect control then avoid next intervention Rule M19: if help-state != rejected and help-skip-next = yes (ditto) then avoid intervention Rule M20: if help-state != requested and not increase confidence and not stimulate curiosity then share control # negotiated actions - t12p363-rN1-N7 Rule N1 action: suggest harder problem tactics: increase success; not stimulate challenge negotiated: suggest problem, similar difficulty Rule N2 action: suggest harder problem tactic: stimulate challenge; not increase confidence negotiated: suggest much-harder problem Rule N3 action: suggest harder problem tactic: emphasise promotion negotiated: suggest harder problem; provide level-promotion comment Rule N4 action: provide wrong assessment; suggest same-diff problem tactic: emphasize promotion negotiated: suggest harder problem; provide level-promotion comment Rule N5 -- requires suprise result action: provide wrong assessment tactic: stimulate curiosity (M13) negotiated: provide suprise result help Rule N6 action: provide right assessment tactic: maintain effort negotiated: provide right assessment; provide praise performance comment Rule N7 action: provide wrong assessment tactic: maintain effort negotiated: provide wrong assessment; provide praise performance comment # negotiated actions (cont) - t12p364rN8-N15 Rule N8 action: suggest problem; not provide assessment tactic: facilitate success; not respect control negotiated: provide help Rule N9 action: suggest problem; not provide assessment tactic: encourage effort negotiated: provide trying-harder comment; suggest help Rule N10 action: --- tactic: remind successes negotiated: provide previous-successes comment Rule N11 action: provide help tactic: avoid intervention negotiated: skip help Rule N12 action: --- tactic: encourage independence negotiated: provide encourage-indep comment Rule N13 action: provide general help tactic: increase confidence negotiated: provide specific help Rule N14 action: provide help tactic: share control negotiated: suggest help Rule N15 action: help not-needed tactic: avoid next intervention negotiated: skip next-help