Sign In

UniER: A Unified Benchmark for Item-level and Path-level Exercise Recommendation

์ž‘์„ฑ์ž
  • Haebom
์นดํ…Œ๊ณ ๋ฆฌ
Empty

์ €์ž

Xinghe Cheng, Guiyong Zhuang, Yusheng Xie, Jiapu Wang, Yixin Liu, Quanlong Guan, Liangda Fang, Shirui Pan

๐Ÿ’ก ๊ฐœ์š”

๋ณธ ๋…ผ๋ฌธ์€ ๊ฐœ๋ณ„ ํ•™์Šต์ž์˜ ๋™์  ํ•™์Šต ์š”๊ตฌ๋ฅผ ์ถฉ์กฑ์‹œํ‚ค๊ธฐ ์œ„ํ•œ ๊ฐœ์ธ ๋งž์ถคํ˜• ์—ฐ์Šต ๋ฌธ์ œ ์ถ”์ฒœ ์—ฐ๊ตฌ์—์„œ, ๊ฐœ๋ณ„ ๋ฌธ์ œ ์ถ”์ฒœ(ILER)๊ณผ ํ•™์Šต ๊ฒฝ๋กœ ์ถ”์ฒœ(PLER)์ด๋ผ๋Š” ๋‘ ๊ฐ€์ง€ ์ฃผ์š” ์—ฐ๊ตฌ ํ๋ฆ„์ด ์„œ๋กœ ๋‹จ์ ˆ๋˜์–ด ์žˆ์–ด ํ†ตํ•ฉ์ ์ธ ๋ฒค์น˜๋งˆํ‚น๊ณผ ๊ณต์ •ํ•œ ๋น„๊ต๊ฐ€ ์–ด๋ ต๋‹ค๋Š” ๋ฌธ์ œ๋ฅผ ์ œ๊ธฐํ•ฉ๋‹ˆ๋‹ค. ์ด๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•ด, ๋…ผ๋ฌธ์€ ILER๊ณผ PLER์„ ํ†ตํ•ฉํ•˜๋Š” ํฌ๊ด„์ ์ธ ํ‰๊ฐ€ ํ”„๋ ˆ์ž„์›Œํฌ์ธ UniER์„ ์ œ์•ˆํ•˜๋ฉฐ, ๋‘ ํŒจ๋Ÿฌ๋‹ค์ž„ ๊ฐ„์˜ ์•Œ๊ณ ๋ฆฌ์ฆ˜ ์„ฑ๋Šฅ์„ ์ธก์ •ํ•˜๊ธฐ ์œ„ํ•œ ํ†ตํ•ฉ ์ง€ํ‘œ์ธ WCG(Weighted Cognitive Gain)๋ฅผ ๋„์ž…ํ•ฉ๋‹ˆ๋‹ค.

๐Ÿ”‘ ์‹œ์‚ฌ์  ๋ฐ ํ•œ๊ณ„

โ€ข
ILER๊ณผ PLER์„ ํ†ตํ•ฉํ•˜๋Š” UniER ๋ฒค์น˜๋งˆํฌ๋ฅผ ํ†ตํ•ด ๋‘ ์—ฐ๊ตฌ ๋ถ„์•ผ ๊ฐ„์˜ ๊ณต์ •ํ•œ ๋น„๊ต ๋ฐ ํ†ตํ•ฉ์ ์ธ ํ‰๊ฐ€๊ฐ€ ๊ฐ€๋Šฅํ•ด์กŒ์Šต๋‹ˆ๋‹ค.
โ€ข
์ œ์•ˆ๋œ WCG ์ง€ํ‘œ๋Š” ์„œ๋กœ ๋‹ค๋ฅธ ์ ‘๊ทผ ๋ฐฉ์‹์˜ ์„ฑ๋Šฅ์„ ๊ฐ๊ด€์ ์œผ๋กœ ์ธก์ •ํ•˜๋Š” ๋ฐ ๊ธฐ์—ฌํ•ฉ๋‹ˆ๋‹ค.
โ€ข
๊ธฐ์กด ์—ฐ๊ตฌ ๊ฒฐ๊ณผ์— ๋Œ€ํ•œ ๋ถ„์„์„ ํ†ตํ•ด, PLER์ด ILER๋ณด๋‹ค ์ „๋ฐ˜์ ์œผ๋กœ ์šฐ์ˆ˜ํ•˜๋ฉฐ, ํŠนํžˆ ๊ทน์‹ฌํ•œ ๋ฐ์ดํ„ฐ ํฌ์†Œ์„ฑ ๋ฐ ๋…ธ์ด์ฆˆ ํ™˜๊ฒฝ์—์„œ ILER์˜ ๋‹จํŽธ์ ์ธ ์ถ”์ฒœ ๋ฐฉ์‹์€ ๊ต์œก์  ์‹คํŒจ๋กœ ์ด์–ด์งˆ ์ˆ˜ ์žˆ์Œ์„ ๋ฐํ˜”์Šต๋‹ˆ๋‹ค.
โ€ข
์˜คํ”ˆ ์†Œ์Šค ์ฝ”๋“œ๋ฅผ ์ œ๊ณตํ•˜์—ฌ ์žฌํ˜„ ๊ฐ€๋Šฅํ•œ ์—ฐ๊ตฌ๋ฅผ ์ด‰์ง„ํ•˜๊ณ  ํ–ฅํ›„ ์—ฐ๊ตฌ ๋ฐฉํ–ฅ์„ ์ œ์‹œํ–ˆ์Šต๋‹ˆ๋‹ค.
โ€ข
(ํ•œ๊ณ„์  ๋˜๋Š” ํ–ฅํ›„ ๊ณผ์ œ) ์‹ค์ œ ๊ต์œก ํ˜„์žฅ์˜ ๋ณต์žก์„ฑ์„ ์™„์ „ํžˆ ๋ฐ˜์˜ํ•˜๋Š” ๋ฐ๋Š” ํ•œ๊ณ„๊ฐ€ ์žˆ์„ ์ˆ˜ ์žˆ์œผ๋ฉฐ, ๋‹ค์–‘ํ•œ ๊ต์œก์  ๋งฅ๋ฝ๊ณผ ํ•™์Šต์ž ํŠน์„ฑ์„ ๊ณ ๋ คํ•œ ์ถ”๊ฐ€์ ์ธ ๋ฒค์น˜๋งˆํ‚น ๋ฐ ์ง€ํ‘œ ๊ฐœ๋ฐœ์ด ํ•„์š”ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
๐Ÿ‘
Made with Slashpage