Buildings,
Journal Year:
2023,
Volume and Issue:
13(11), P. 2700 - 2700
Published: Oct. 26, 2023
Subway
construction
is
often
in
a
complex
natural
and
human-machine
operating
environment,
that
complicated
setting
leads
to
subway
being
more
prone
safety
accidents,
which
can
cause
substantial
casualties
monetary
losses.
Thus,
it
necessary
investigate
the
risks
of
construction.
The
existing
literature
on
identification
assessment
(SCSR)
susceptible
influence
subjective
factors.
Moreover,
although
studies
have
explored
interrelationships
between
different
risks,
these
usually
analyze
single
lack
study
risk
chain
transfer
relationships,
fail
find
out
key
path
transfer.
Therefore,
this
paper
innovatively
combines
text
mining,
association
rules,
networks
deep
mine
incident
reports
explore
process.
Firstly,
uses
mining
technology
identify
risks.
Then,
rules
are
introduced
causal
relationships
among
Finally,
important
paths
accidents
(SCSA)
obtained
based
network
model.
Research
results
show
(a)
improper
management,
unimplemented
subject
responsibilities,
violation
operation
non-perfect
responsibilities
system
insufficient
education
training
SCSA;
(b)
two
shorter
be
obtained:
training→lower
awareness→violation
rules→safety
accidents;
checks
or
hidden
trouble
investigations→violation
(c)
process
transfer,
controlled
by
controlling
cutting
off
paths.
This
provides
new
ideas
methods
for
SCSR
element
help
managers
propose
accurate
control
measures.
Engineering Construction & Architectural Management,
Journal Year:
2024,
Volume and Issue:
unknown
Published: Feb. 20, 2024
Purpose
The
purpose
is
to
identify
and
evaluate
the
safety
risk
factors
in
coal
mine
construction
process.
Design/methodology/approach
text
mining
technique
was
applied
stage
of
factor
identification.
association
rules
method
used
obtain
associations
with
factors.
Decision-Making
Trial
Evaluation
Laboratory
(DEMATEL)
Interpretative
Structural
Modeling
(ISM)
were
utilized
Findings
results
show
that
18
are
divided
into
6
levels.
There
12
transmission
paths
total.
Meanwhile,
unsafe
behavior
equipment
malfunction
failure
direct
causes
accidents,
inadequate
management
system
basic
determines
status.
Research
limitations/implications
Due
limitation
computational
matrix
workload,
this
article
only
categorizes
numerous
lexical
items
Then,
workshop
relied
on
a
limited
number
experts;
thus,
findings
may
be
potentially
biased.
Next,
accident
report
lacks
universal
standard
for
compilation,
use
further
optimized.
Finally,
since
data
all
from
China,
subsequent
cross-country
studies
should
considered.
Social
implications
can
help
China
project
managers
have
clear
understanding
risks,
efficiently
carry
out
hazard
identification
work
take
timely
measures
cut
off
path
risks
identified
study.
This
helps
reduce
economic
losses
enterprises,
thus
improving
standards
entire
industry
national
policy
formulation.
Originality/value
Coal
projects
characterized
by
complexity
difficulties
construction.
Current
research
assessment
insufficient.
study
combines
objective
systematic
approaches.
contribute
providing
basis
development
measures.
Scientific Reports,
Journal Year:
2024,
Volume and Issue:
14(1)
Published: March 13, 2024
Abstract
A
risk
assessment
method
for
hot
work
based
on
G1-EWM
and
unascertained
measurement
theory
was
proposed
to
prevent
accidents
in
underground
mines.
Firstly,
the
influencing
factors
classification
criteria
operations
mines,
a
single
indicator
matrix
constructed
using
theory;
Secondly,
index
system
mine
established.
The
combination
weight
coefficient
of
each
determined
order
relationship
analysis
(G1)
entropy
(EWM)
coupled
with
evaluation
vector
calculate
multi-index
comprehensive
object;
Finally,
model
validated
examined
engineering
examples,
level
confidence
identification
criteria.
results
showed
that
method,
when
used
evaluate
tunnels
vertical
shafts
metal
produces
levels
are
line
reality
III
(Moderate
Risk)
shaft
IV
(High
tunnels.
consistent
whole
process
on-site
work,
which
verifies
feasibility.
unique
strategy
management
mines
is
provided
by
weighting
measure
models,
has
theoretical
practical
value.
Future
research
could
focus
refineing
this
exploring
applicability
diverse
mining
environments
integrating
advanced
analytical
techniques
enhance
predictive
accuracy
operational
efficiency.
Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing,
Journal Year:
2023,
Volume and Issue:
unknown
Published: Jan. 1, 2023
We
introduce
ClusterLLM,
a
novel
text
clustering
framework
that
leverages
feedback
from
an
instruction-tuned
large
language
model,
such
as
ChatGPT.
Compared
with
traditional
unsupervised
methods
builds
upon
"small"
embedders,
ClusterLLM
exhibits
two
intriguing
advantages:
(1)
it
enjoys
the
emergent
capability
of
LLM
even
if
its
embeddings
are
inaccessible;
and
(2)
understands
user's
preference
on
through
textual
instruction
and/or
few
annotated
data.
First,
we
prompt
ChatGPT
for
insights
perspective
by
constructing
hard
triplet
questions
,
where
A,
C
similar
data
points
belong
different
clusters
according
small
embedder.
empirically
show
this
strategy
is
both
effective
fine-tuning
embedder
cost-efficient
query
Second,
helps
granularity
carefully
designed
pairwise
,
tune
cluster
hierarchies
most
consistent
answers.
Extensive
experiments
14
datasets
consistently
improves
quality,
at
average
cost
~$0.6
per
dataset.
Journal of Construction Engineering and Management,
Journal Year:
2023,
Volume and Issue:
150(3)
Published: Dec. 23, 2023
Highway
work
zones
are
essential
for
the
preservation
and
improvement
of
national
road
system.
Nevertheless,
these
areas
reported
to
be
among
most
hazardous
workplaces.
Thus,
it
is
crucial
develop
appropriate
measures
effectively
mitigate
safety
risks,
which
require
a
good
understanding
critical
causes
accidents.
While
there
many
previous
studies
on
construction
accidents,
none
them
was
specifically
focused
highway
zones.
This
type
workplace
has
its
own
characteristics
(e.g.,
near-passing
traffic),
can
lead
unique
set
study
used
text
mining
extract
root
from
large
narrative
data
accidents
at
obtained
Occupational
Safety
Health
Administration
(OSHA).
The
applied
latent
Dirichlet
allocation
(LDA)
modeling
corpus
12
causes,
were
subsequently
classified
into
five
groups:
management,
human,
unsafe
behavior,
environmental,
material
factors.
In
addition,
social
network
analysis
(SNA)
conducted
gain
further
insights
interrelations
between
determine
their
criticality
degree.
As
result,
four
highly
ranked
identified:
supervision
dereliction
duty,
weak
awareness,
poor
environment,
risk-taking
behavior.
findings
this
offer
new
factors
that
agencies
contractors
should
focus
when
developing
accident
prevention
strategies
Engineering Construction & Architectural Management,
Journal Year:
2023,
Volume and Issue:
32(2), P. 1097 - 1121
Published: Sept. 9, 2023
Purpose
Learning
from
safety
accidents
and
sharing
knowledge
has
become
an
important
part
of
accident
prevention
improving
construction
management.
Considering
the
difficulty
reusing
unstructured
data
in
industry,
it
is
difficult
to
be
used
directly
for
analysis.
The
purpose
this
paper
explore
representation
model
graph
through
deep
learning
methods,
extract
entities
BERT-BiLSTM-CRF
propose
a
management
data–knowledge–services.
Design/methodology/approach
ontology
constructed
by
integrating
entity
relation
logic
evolution.
Then,
database
incidents
architecture,
engineering
(AEC)
industry
established
based
on
collected
incident
reports
related
dispute
cases.
method
studied,
precision
algorithm
information
extraction
verified
comparative
experiments.
Finally,
report
as
example
construct
AEC
domain
(AEC-KG),
which
provides
visual
query
service
verifies
operability
Findings
experimental
results
show
that
combined
84.52%,
recall
92.35%,
F1
value
88.26%
named
recognition
database.
realize
visualization.
Originality/value
proposed
framework
new
approach
improve
practitioners
also
enriches
application
scenarios
graph.
On
one
hand,
innovatively
proposes
integrates
relationship
matter
evolution
logic.
other
legal
adjudication
dimension
added
field
basis
postincident
disposal
measures
accidents,
reference
managers'
decision-making
all
aspects.
