【摘要】：推薦系統(tǒng)作為解決信息過(guò)載問(wèn)題的重要工具,逐漸滲透并改變著人們的生活方式。個(gè)性化習(xí)題推薦作為推薦系統(tǒng)在教育領(lǐng)域的分支,近年來(lái),吸引了不少研究學(xué)者展開(kāi)了廣泛的研究。本文針對(duì)個(gè)性化習(xí)題推薦算法準(zhǔn)確率不高的問(wèn)題,提出一種基于知識(shí)點(diǎn)的個(gè)性化習(xí)題推薦方案,主要工作和創(chuàng)新點(diǎn)如下:(1)針對(duì)認(rèn)知診斷模型不能對(duì)知識(shí)點(diǎn)掌握概率化的問(wèn)題,提出學(xué)生知識(shí)點(diǎn)掌握概率模型,進(jìn)而提出TopN個(gè)性化習(xí)題推薦算法。實(shí)驗(yàn)表明,該算法在準(zhǔn)確率方面具有較好的效果,在FrcSub數(shù)據(jù)集比認(rèn)知診斷模型提高了 10.1%,在Math1數(shù)據(jù)集比認(rèn)知診斷模型提高了 2.84%,在Math2數(shù)據(jù)集不僅比認(rèn)知診斷模型提高了 1.39%,而且運(yùn)行效率是認(rèn)知診斷模型的12.1倍。(2)提出一種表征知識(shí)點(diǎn)層次關(guān)系的權(quán)重圖,構(gòu)建知識(shí)點(diǎn)之間的層次關(guān)系和學(xué)生知識(shí)點(diǎn)失分率矩陣,解決了學(xué)生知識(shí)點(diǎn)之間相互孤立造成的數(shù)據(jù)稀疏性引起的推薦準(zhǔn)確率不高的問(wèn)題。實(shí)驗(yàn)表明,該算法在自建數(shù)據(jù)集上具有較好的效果。較基于知識(shí)點(diǎn)無(wú)層次圖的個(gè)性化習(xí)題推薦算法在準(zhǔn)確率提高了 12.35%,召回率提高了 5.49%,F1 提高了 11.91%。(3)為了解決現(xiàn)有的評(píng)價(jià)指標(biāo)不能準(zhǔn)確反映學(xué)生對(duì)知識(shí)點(diǎn)掌握情況的問(wèn)題,提出一種基于知識(shí)點(diǎn)的評(píng)價(jià)指標(biāo),以便更準(zhǔn)確地為學(xué)生推薦掌握薄弱的知識(shí)點(diǎn)習(xí)題。為了驗(yàn)證提出的習(xí)題推薦方法的合理性,本文設(shè)計(jì)并實(shí)現(xiàn)了一個(gè)基于知識(shí)點(diǎn)個(gè)性化習(xí)題推薦系統(tǒng)原型。試用結(jié)果表明,該系統(tǒng)可以作為一種重要的輔助教學(xué)手段,不僅彌補(bǔ)了不同程度的學(xué)生的知識(shí)漏洞,而且提高了學(xué)生自主學(xué)習(xí)效率。
[Abstract]:As an important tool to solve the problem of information overload, recommendation system is gradually infiltrating and changing people's way of life. As a branch of recommendation system in the field of education, personalized exercise recommendation has attracted a lot of researchers to carry out extensive research in recent years. Aiming at the problem that the accuracy of personalized exercise recommendation algorithm is not high, this paper proposes a personalized exercise recommendation scheme based on knowledge point. The main work and innovations are as follows: (1) aiming at the problem that the cognitive diagnosis model can not grasp the probability of knowledge point, the paper puts forward the probability model of students' knowledge point mastery, and then puts forward the TopN personalized exercise recommendation algorithm. The experimental results show that the algorithm has a good effect on accuracy, which is 10.1% higher in FrcSub data set than in cognitive diagnosis model, 2.84% higher in Math1 data set than cognitive diagnosis model. The Math2 dataset is not only 1.39 times more efficient than the cognitive diagnostic model, but also is 12.1 times more efficient than the cognitive diagnostic model. (2) A weight map representing the hierarchical relationship of knowledge points is proposed. Constructing the hierarchical relationship between knowledge points and the rate matrix of students' knowledge points can solve the problem of low recommendation accuracy caused by the data sparsity caused by the isolation of students' knowledge points. Experiments show that the algorithm has good effect on self-built data set. Compared with the personalized exercise recommendation algorithm based on the knowledge point without hierarchy graph, the accuracy of the algorithm is increased 12.35%, and the recall rate increases 5.49%. F1 raised 11.911.In order to solve the problem that the existing evaluation indexes could not accurately reflect the students' mastery of knowledge points, a kind of evaluation index based on knowledge point was put forward. In order to more accurately recommend students to grasp the weak points of knowledge exercises. In order to verify the rationality of the proposed exercise recommendation method, this paper designs and implements a prototype of personalized exercise recommendation system based on knowledge point. The experimental results show that the system can be used as an important auxiliary teaching method, which not only makes up for the gaps of students' knowledge, but also improves the efficiency of students' autonomous learning.
【學(xué)位授予單位】：西北大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.3

【相似文獻(xiàn)】

相關(guān)碩士學(xué)位論文 前1條

1 王子靜;中日初中數(shù)學(xué)教科書(shū)的比較研究[D];上海師范大學(xué);2017年

，

本文編號(hào)：2298536