IvCDS: An End-to-End Driver Simulator for Personal In-Vehicle Conversational Assistant

. 2022 Nov 22;19(23):15493. doi: 10.3390/ijerph192315493

Algorithm 1: The processing step of extracting assistant actions for the weather information retrieval scenario.

$S \leftarrow$ a set of key-value pairs in <slots> ▹ weather_attribute = blizzard, date = Friday, ⋯
$S_{k e y s} \leftarrow$ all keys in S
$S [k e y] \leftarrow$ the value of key in S
$K_{i} \leftarrow$ the i-th item in KB
$K_{k e y s} \leftarrow$ all keys in KB ▹ monday, tuesday, location,⋯
$K_{i} [k e y] \leftarrow$ the value of key in i-th item in KB ▹ "snow, low of 60F, high of 80F", ⋯
$a \leftarrow$ the answer of assistant ▹ On Friday it’s gonna rain in Brentwood, ⋯
$a c t i o n s \leftarrow$ the actions of assistant
if $l o c a t i o n \notin S_{k e y s}$ then
$a c t i o n s \leftarrow$ a question about location
else
get $K_{i}$ where $S [k e y]$ in $K_{i}$
if $d a t e \in S_{k e y s}$ then
$a c t i o n s \leftarrow l o c a t i o n$ = $K_{i} [l o c a t i o n], d a t e$ = $K_{i} [d a t e]$
else
split $K_{i} [k e y]$ into three parts
if $a n y p a r t s \in a$ then
$a c t i o n s \leftarrow l o c a t i o n$ = $K_{i} [l o c a t i o n], k e y$ = $K_{i} [k e y]$
end if
end if
end if