Information Technology Reference
In-Depth Information
in precision and a 6% increase in recall, mostly in the Hotel domain, which is
rich in adjectives (
e.g.
,“
clean room
,” “
soft bed
”).
2.3.5 Finding Opinion Phrases and Their Polarity
This subsection describes how opine extracts potential opinion phrases, dis-
tinguishes between opinions and non-opinions, and finds the
polarity
of each
opinion in the context of its associated feature in a particular review sentence.
opine uses explicit features to identify potential opinion phrases. Our
intuition is that an opinion phrase associated with a product feature will
occur in its vicinity. This idea is similar to that of [9] and [2], but instead
of using a window of size
k
or the output of a noun phrase chunker, opine
takes advantage of the dependencies computed by the MINIPAR parser. Our
intuition is embodied by a set of
extraction rules
, the most important of which
are shown in Table 2.7. If an explicit feature is found in a sentence, opine
applies the extraction rules in order to find the heads of potential opinion
phrases. Each head word, together with its modifiers, is returned as a potential
opinion phrase.
Table 2.7. Domain-Independent Rules for Potential Opinion Phrase
Extraction.
Notation: po=potential opinion, M=modifier, NP=noun phrase,
S=subject, P=predicate, O=object. Extracted phrases are enclosed in parenthe-
ses. Features are indicated by the typewriter font. The equality conditions on the
left-hand side use
po
's head.
Extraction Rules Examples
if
∃
(
M, NP
=
f
)
→ po
=
M
(expensive)
scanner
if
po
=
O
lamp
has (problems)
if
∃
(
S, P, O
=
f
)
→ po
=
P
∃
(
S
=
f, P, O
)
→
I (hate) this
scanner
if
∃
(
S
=
f, P
)
→ po
=
P
program
(crashed)
Table 2.8. Dependency Rule Templates For Finding Words
w
,
w
with
Related Semantic Orientation Labels
Notation: v,w,w'=words; f, f'=feature
names; dep=dependent; m=modifier
Rule Templates
Example Rules
dependent
(
w, w
)
modifier
(
w, w
)
∃v
s.t.
dep
(
w, v
)
,dep
(
v, w
)
∃v
s.t.
m
(
w, v
)
, object
(
v, w
)
∃v
s.t.
dep
(
w, v
)
,dep
(
w
,v
)
∃v
s.t.
m
(
w, v
)
, object
(
w
,v
)
∃f, f
s.t.
dep
(
w, f
)
,dep
(
w
,f
)
,dep
(
f, f
)
∃f, f
s.t.
m
(
w, f
)
,m
(
w
,f
)
,and
(
f, f
)
opine examines the potential opinion phrases in order to identify the ac-
tual opinions. First, the system finds the semantic orientation for the lexical
head of each potential opinion phrase. Every phrase whose head word has a
positive
or
negative
semantic orientation is then retained as an
opinion phrase
.
Search WWH ::
Custom Search