Also found in: Dictionary.
Related to SLART: salary
SLARTSecond Language Acquisition, Research and Teaching
Copyright 1988-2018, All rights reserved.
References in periodicals archive ?
We use several baselines to compare the effectiveness of SLARTS. The first baseline is the performance when we simply link all comments to their previous comment (we name it Last baseline) [3,4].
According to [Acc.sub.edge], SLARTS reveals higher accuracy except in Alef in which lots of replies are connected to the root.
Usually, SLARTS and Seo et al.'s methods cannot predict the full path in Thestandard and ENENews, because, according to Figure 15, paths are very long and complex in these datasets.
According to Table 6, for [F.sub.edge], SLARTS performs better than Seo et al.'s method.
On the other hands, [P.sub.edge] and [R.sub.edge] of SLARTS are close.
Several features in SLARTS such as Global similarity, frequent words, Frequent patterns, and authors' language focus on detection of comment-comment relations.
SLARTS has better [F.sub.edge] than Seo et al.'s method in all datasets.
The minimum difference between the results of SLARTS and Seo et al.'s methods appears in Alef datasets.
Since SLARTS method has features which specially focused on detecting comment-root relations (e.g., by adding candidate filtering), [Acc.sub.CTD] of SLARTS is better than Seo et al.
We compare P-value to specify the significance of differences between SLARTS and Seo et al.'s methods on declarative metrics.