輔助任務(wù)增強(qiáng)的知識(shí)追蹤方法

打開文本圖片集

關(guān)鍵詞：智慧教育；知識(shí)追蹤；輔助任務(wù)；注意力機(jī)制

中圖分類號(hào)：TP391 文獻(xiàn)標(biāo)志碼：A 文章編號(hào)：1001-3695（2025）07-023-2096-07

doi： 10.19734/j . issn.1001-3695.2024.10.0448

Abstract：Knowledge tracingisaneffctiveway toachievesmarteducation，asitasistsstudents intimelyadjustingtheir learning strategiesbyasesing theirlearning status.Theexisting knowledgetracing methods stillhaveshortcomings indealing withthe nternalrelationshipbetweenexercisesandknowledgeconceptsandobtaining students’personalized knowledge level. Toaddressthisissue，thispaperproposed anauxiliarytask-enhanced knowledge tracing（ATEKT）method.Firstly，this paper usedthegroupcalculationmethodandfrequencycalculationmethodtocalculatethedificultyof theexercise，thediscriminationoftheexercise，andthenumberofanswers totherelevantknowledgepoint exercises，andotainedtheirembeddedrepresentation.Secondly，itpresentedtwoauxiliarytasks，exercisetaging predictiontasksandpriorknowledge predictiontasksto dynamicallyaess therelationshipsbetweenexercisesandknowledge points，whileestimatingstudents'knowledgelevelsfrom historicalresponserecords.Finall，theauxiliarytaskslearnedthecharacteristicsof theoptimizationprocesstoenhance knowledge tracing model predicts.Comparedwithsevenknowledge tracing modelsonthreepublicdatasetssuchas ASSIST2009，ASSIST2017 and EdNet，the AUC and ACC of the proposed method are increased by 2.4% to 15.5% and （204 2.0% to 9.9% respectively.The experimental results show that the proposed method can efectively extract information related to knowledgetracing tasks，and itsperformanceismoresuperior.Meanwhile，ablationexperimentsalsodemonstrate the effectivenessofdifferentauxiliarytasks.

Keywords：intelligent education；knowledge tracing；auxiliary task；attention mechanisn

0 引言

智慧教育是指利用現(xiàn)代信息技術(shù)、教育方法和心理學(xué)原理等多種手段，實(shí)現(xiàn)教育過(guò)程個(gè)性化、智能化和高效化的一種新型教育模式1。（剩余15715字）