NLP for CP
Addressing Constraint Programming with Natural Language Processing
Home
Resources
Publications
Correct
predictions are in
blue
. If we detect only a subset of a labelled sentence, we highlight the caught part as
blue
, the missing part
light blue.
False positives
are in
green
and
false negatives
are in
red
.
Problem Yield_Management — Constraint detection
An
airline
is
selling
tickets
for
flights
to
a
particular
destination
.
The
flight
will
depart
in
three
weeks
'
time
.
It
can
use
up
to
six
planes
each
costing
pounds
50000
to
hire
.
Each
plane
has
the
following
:
37
First
Class
seats
,
38
Business
Class
seats
,
47
Economy
Class
seats
.
Up
to
10
%
of
seats
in
any
one
category
can
be
transferred
to
an
adjacent
category
.
It
wishes
to
decide
a
price
for
each
of
these
seats
.
There
will
be
further
opportunities
to
update
these
prices
after
one
week
and
two
weeks
.
Once
a
customer
has
purchased
a
ticket
,
there
is
no
cancellation
option
.
For
administrative
simplicity
,
three
price
level
options
are
possible
in
each
class
-LRB-
one
of
which
must
be
chosen
-RRB-
.
The
same
option
need
not
be
chosen
for
each
class
.
These
are
given
in
a
table
for
the
current
period
-LRB-
period
1
-RRB-
and
two
future
periods
.
Demand
is
uncertain
but
will
be
affected
by
price
.
Forecasts
have
been
made
of
these
demands
according
to
a
probability
distribution
that
divides
the
demand
levels
into
three
scenarios
for
each
period
.
The
probabilities
of
the
three
scenarios
in
each
period
are
as
follows
:
0.1
-LRB-
scenario
1
-RRB-
,
0.7
-LRB-
scenario
2
-RRB-
,
0.2
-LRB-
scenario
3
-RRB-
.
The
forecast
demands
are
shown
in
a
table
.
Decide
price
levels
for
the
current
period
,
how
many
seats
to
sell
in
each
class
-LRB-
depending
on
demand
-RRB-
,
the
provisional
number
of
planes
to
book
and
provisional
price
levels
and
seats
to
sell
in
future
periods
in
order
to
maximise
expected
yield
.
You
should
schedule
to
be
able
to
meet
commitments
under
all
possible
combinations
of
scenarios
.
Problem Yield_Management — Detection of the decisions and objects to be modeled
An
airline
is
selling
tickets
for
flights
to
a
particular
destination
.
The
flight
will
depart
in
three
weeks
'
time
.
It
can
use
up
to
six
planes
each
costing
pounds
50000
to
hire
.
Each
plane
has
the
following
:
37
First
Class
seats
,
38
Business
Class
seats
,
47
Economy
Class
seats
.
Up
to
10
%
of
seats
in
any
one
category
can
be
transferred
to
an
adjacent
category
.
It
wishes
to
decide
a
price
for
each
of
these
seats
.
There
will
be
further
opportunities
to
update
these
prices
after
one
week
and
two
weeks
.
Once
a
customer
has
purchased
a
ticket
,
there
is
no
cancellation
option
.
For
administrative
simplicity
,
three
price
level
options
are
possible
in
each
class
-LRB-
one
of
which
must
be
chosen
-RRB-
.
The
same
option
need
not
be
chosen
for
each
class
.
These
are
given
in
a
table
for
the
current
period
-LRB-
period
1
-RRB-
and
two
future
periods
.
Demand
is
uncertain
but
will
be
affected
by
price
.
Forecasts
have
been
made
of
these
demands
according
to
a
probability
distribution
that
divides
the
demand
levels
into
three
scenarios
for
each
period
.
The
probabilities
of
the
three
scenarios
in
each
period
are
as
follows
:
0.1
-LRB-
scenario
1
-RRB-
,
0.7
-LRB-
scenario
2
-RRB-
,
0.2
-LRB-
scenario
3
-RRB-
.
The
forecast
demands
are
shown
in
a
table
.
Decide
price
levels
for
the
current
period
,
how
many
seats
to
sell
in
each
class
-LRB-
depending
on
demand
-RRB-
,
the
provisional
number
of
planes
to
book
and
provisional
price
levels
and
seats
to
sell
in
future
periods
in
order
to
maximise
expected
yield
.
You
should
schedule
to
be
able
to
meet
commitments
under
all
possible
combinations
of
scenarios
.
Back to list