Skip to main content
. 2021 Jun 5;5(4):474–496. doi: 10.1007/s41666-021-00100-z

Table 10.

Direct comparison to the work of Sushil et al. [17] considering methods Bag-of-Words (BoW) [10], doc2vec [30], and CUI-based; using a regular Feed-Forward network and a stacked denoising autoencoder (SDAE) [32] network

Architecture P@1
FFN (CUI + CCS) 0.778
BoW 0.701
SDAE-BoW 0.650
doc2vec 0.681
[doc2vec, SDAE-BoW] 0.679
BoCUI 0.710
SDAE-BoCUI 0.665

Highest marks in bold face