<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-1642">
  <Title>Fully Automatic Lexicon Expansion for Domain-oriented Sentiment Analysis</Title>
  <Section position="5" start_page="356" end_page="357" type="metho">
    <SectionTitle>
3 Methodology of Clause-level SA
</SectionTitle>
    <Paragraph position="0"> As Figure 1 illustrates, the flow of our sentiment analysis system involves three steps.</Paragraph>
    <Paragraph position="1"> The first step is sentence delimitation: the input document is divided into sentences. The second step is proposition detection: propositions which can form polar clauses are identifiedineachsentence. Thethirdstepis polarity assignment: the polarity of each proposition is examined by considering the polar atoms.</Paragraph>
    <Paragraph position="2"> This section describes the last two processes, which are based on a deep sentiment analysis method analogous to machine translation (Kanayama et al., 2004) (hereafter &amp;quot;the MT method&amp;quot;).</Paragraph>
    <Section position="1" start_page="356" end_page="357" type="sub_section">
      <SectionTitle>
3.1 Proposition Detection
</SectionTitle>
      <Paragraph position="0"> Our basic tactic for clause-level SA is the high-precision detection of polar clauses based on deep syntactic analysis. 'Clause-level' means that only predicative verbs and adjectives such  a subject &amp;quot;nobody&amp;quot; or a modifier &amp;quot;seldom&amp;quot; is rare in Japanese.</Paragraph>
      <Paragraph position="1"> as in (7) are detected, and adnominal (attributive) usages of verbs and adjectives as in (8) are ignored, because utsukushii ('beautiful') in  (8) does not convey a positive polarity.</Paragraph>
      <Paragraph position="2"> (7) E-ga utsukushii.</Paragraph>
      <Paragraph position="3"> 'The picture is beautiful.' (8) Utsukushii hito-ni aitai.</Paragraph>
      <Paragraph position="4"> 'I want to meet a beautiful person.'  Here we use the notion of a proposition as a clause without modality, led by a predicative verb or a predicative adjective. The propositions detected from a sentence are subject to the assignment of polarities.</Paragraph>
      <Paragraph position="5"> Basically, we detect a proposition only at the head of a syntactic tree3. However, this limitation reduces the recall of sentiment analysis to a very low level. In the example (7) above, utsukushii is the head of the tree, while those initial clauses in (9) to (11) below are not. In order to achieve higher recall while maintaininghighprecision, weapplytwotypes of syntactic patterns, modality patterns and conjunctive patterns4, to the tree structures from the full-parsing.</Paragraph>
      <Paragraph position="6">  (9) Sore-ha utsukushii-to omou.</Paragraph>
      <Paragraph position="7"> 'I think it is beautiful.' (10) Sore-ha utsukushiku-nai.</Paragraph>
      <Paragraph position="8"> 'It is not beautiful.' (11) Sore-ga utsukushii-to yoi.</Paragraph>
      <Paragraph position="9"> 'I hope it is beautiful.'  Modality patterns match some auxiliary verbs or corresponding sentence-final expressions, to allow for specific kinds of modality and negation. One of the typical patterns is [ v to omou] ('I think v ')5, which allows utsukushii in (9) to be a proposition. Also negation is handled with a modality pattern, such as [ v nai] ('not v '). In this case a neg feature is attached to the proposition to identify utsukushii in (10) as a negated proposition. On the other hand, no proposition is identified in (11) due to the deliberate absence of a pattern [ v to yoi] ('I hope v '). We used a total of 103 domain-independent modality patterns, most of which are derived from the  coordinative (roughly 'and') -te, -shi, -ueni, -dakedenaku, -nominarazu causal (roughly 'because') -tame, -kara, -node adversative (roughly 'but') -ga, -kedo, -keredo, - monono, -nodaga  junctive patterns.</Paragraph>
      <Paragraph position="10"> MT method, and some patterns are manually added for this work to achieve higher recall. Another type of pattern is conjunctive patterns, which allow multiple propositions in a sentence. We used a total of 22 conjunctive patterns also derived from the MT method, as exemplified in Table 1. In such cases of coordinative clauses and causal clauses, both clauses can be polar clauses. On the other hand, no proposition is identified in a conditional clause due to the absence of corresponding conjunctive patterns.</Paragraph>
    </Section>
    <Section position="2" start_page="357" end_page="357" type="sub_section">
      <SectionTitle>
3.2 Polarity Assignment Using Polar
Atoms
</SectionTitle>
      <Paragraph position="0"> To assign a polarity to each proposition, polar atoms in the lexicon are compared to the proposition. A polar atom consists of polarity, verb or adjective, and optionally, its arguments. Example (12) is a simple polar atom, where no argument is specified. This atom matches any proposition whose head is utsukushii. Example (13) is a complex polar atom, which assigns a negative polarity to any proposition whose head is the verb kaku and where the accusative case is miryoku.</Paragraph>
      <Paragraph position="1">  (12) [+] utsukushii 'to be beautiful' (13) [[?]] kaku - miryoku-wo 'to lack - attraction-ACC'  A polarity is assigned if there exists a polar atom for which verb/adjective and the arguments coincide with the proposition, and otherwise no polarity is assigned. The opposite polarity of the polar atom is assigned to a proposition which has the neg feature.</Paragraph>
      <Paragraph position="2"> We used a total of 3,275 polar atoms, most of which are derived from an English sentiment lexicon (Yi et al., 2003).</Paragraph>
      <Paragraph position="3"> According to the evaluation of the MT method (Kanayama et al., 2004), high-precision sentiment analysis had been achieved using the polar atoms and patterns, where the  system never took positive sentiment for negative and vice versa, and judged positive or negative to neutral expressions in only about 10% cases. However, the recall is too low, and most of the lexicon is for domain-independent expressions, and thus we need more lexical entries to grasp the positive and negative aspects in a specific domain.</Paragraph>
    </Section>
  </Section>
  <Section position="6" start_page="357" end_page="359" type="metho">
    <SectionTitle>
4 Context Coherency
</SectionTitle>
    <Paragraph position="0"> This section introduces the intra- and intersententialcontextsinwhichweassume context coherency for polarity, and describes some preliminary analysis of the assumption.</Paragraph>
    <Section position="1" start_page="357" end_page="358" type="sub_section">
      <SectionTitle>
4.1 Intra-sentential and
Inter-sentential Context
</SectionTitle>
      <Paragraph position="0"> The identification of propositions described in Section 3.1 clarifies our viewpoint of the contexts. Here we consider two types of contexts: intra-sentential context and inter-sentential context. Figure 2 illustrates the context coherency in a sample discourse (14), where the polarities are perfectly coherent.</Paragraph>
      <Paragraph position="1"> (14) Kono kamera-ha subarashii-to omou.</Paragraph>
      <Paragraph position="2"> 'I think this camera is splendid.' Karui-shi, zuumu-mo tsuite-iru.</Paragraph>
      <Paragraph position="3"> 'It's light and has a zoom lens.' Ekishou-ga chiisai-kedo, manzoku-da.</Paragraph>
      <Paragraph position="4"> 'Though the LCD is small, I'm satisfied.' Tada, nedan-ga chotto takai.</Paragraph>
      <Paragraph position="5"> 'But, the price is a little high.' The intra-sentential context is the link between propositions in a sentence, which are detected as coordinative or causal clauses. If there is an adversative conjunction such as -kedo ('but') in the third sentence in (14), a flag is attached to the relation, as denoted with '[?]' in Figure 2. Though there are differences in syntactic phenomena, this is sim- null shikashi ('however'), demo ('but'), sorenanoni ('even though'), tadashi ('on condition that'), dakedo ('but'), gyakuni ('on the contrary'), tohaie ('although'), keredomo ('however'), ippou ('on the other hand')  used in this paper. The &amp;quot;Post.&amp;quot; and &amp;quot;Sent.&amp;quot; columns denote the numbers of postings and sentences, respectively. &amp;quot;Len.&amp;quot; is the average length of sentences (in Japanese characters). ilar to the semantic orientation proposed by Hatzivassiloglou and McKeown (1997).</Paragraph>
      <Paragraph position="6"> The inter-sentential context is the link between propositions in the main clauses of pairs of adjacent sentences in a discourse. The polarities are assumed to be the same in the inter-sentential context, unless there is an adversative expression as those listed in Table 2. If no proposition is detected as in a nominal sentence, the context is split. That is, there is no link between the proposition of the previous sentence and that of the next sentence.</Paragraph>
    </Section>
    <Section position="2" start_page="358" end_page="359" type="sub_section">
      <SectionTitle>
4.2 Preliminary Study on Context
Coherency
</SectionTitle>
      <Paragraph position="0"> We claim these two types of context can be used for unsupervised learning as clues to assign a tentative polarity to unknown expressions. To validate our assumption, we conducted preliminary observations using various corpora.</Paragraph>
      <Paragraph position="1">  Throughout this paper we used Japanese corpora from discussion boards in four different domains, whose features are shown in Table 3. All of the corpora have clues to the boundaries of postings, so they were suitable to identify the discourses.</Paragraph>
      <Paragraph position="2">  How strong is the coherency in the context proposed in Section 4.1? Using the polar clauses detected by the SA system with the initial lexicon, we observed the coherent precision of domain d with lexicon L, defined as:</Paragraph>
      <Paragraph position="4"> where #(Coherent) and #(Conflict) are occurrence counts of the same and opposite polarities observed between two polar clauses as observed in the discourse. As the two polar clauses, we consider the following types: Window. A polar clause and the nearest polar clause which is found in the preceding n sentences in the discourse.</Paragraph>
      <Paragraph position="5"> Context. Two polar clauses in the intra-sentential and/or inter-sentential context described in Section 4.1. This is the view-point of context in our method.</Paragraph>
      <Paragraph position="6"> Table 4 shows the frequencies of coherent pairs, conflicting pairs, and the coherent precision for half of the digital camera domain corpus. &amp;quot;Baseline&amp;quot; is the percentage of positive clauses among the polar clauses6.</Paragraph>
      <Paragraph position="7"> For the &amp;quot;Window&amp;quot; method, we tested for n=0, 1, 2, and [?]. &amp;quot;0&amp;quot; means two propositions within a sentence. Apparently, the larger the window size, the smaller the cp value. When the window size is &amp;quot;[?]&amp;quot;, implying anywhere within a discourse, the ratio is larger than the baseline by only 2.7%, and thus these types of coherency are not reliable even though the number of clues is relatively large.</Paragraph>
      <Paragraph position="8"> &amp;quot;Context&amp;quot; shows the coherency of the two types of context that we considered. The cp values are much higher than those in the &amp;quot;Window&amp;quot; methods, because the relationships between adjacent pairs of clauses are handled more appropriately by considering syntactic trees, adversative conjunctions, etc. The cp values for inter-sentential and intra-sentential contexts are almost the same, and thus both contexts can be used to obtain 2.5 times more clues for the intra-sentential context. In the rest of this paper we will use both contexts.</Paragraph>
      <Paragraph position="9"> We also observed the coherent precision for each domain corpus. The results in the center column of Table 5 indicate the number is slightly different among corpora, but all of them are far from perfect coherency.</Paragraph>
      <Paragraph position="10"> 6Ifthereisapolarclausewhosepolarityisunknown, the polarity is correctly predicted with at least 57.0% precision by assuming &amp;quot;positive&amp;quot;.</Paragraph>
      <Paragraph position="11">  Besides the conflicting cases, there are many more cases where a polar clause does not appear in the polar context. We also observed the coherent density of the domain d with the lexicon L defined as:</Paragraph>
      <Paragraph position="13"> This indicates the ratio of polar clauses that appear in the coherent context, among all of the polar clauses detected by the system.</Paragraph>
      <Paragraph position="14"> The right column of Table 5 shows the coherent density in each domain. The movie domain has notably higher coherent density than the others. This indicates the sentiment expressions are more frequently used in the movie domain.</Paragraph>
      <Paragraph position="15"> The next section describes the method of our unsupervised learning using this imperfect context coherency.</Paragraph>
    </Section>
  </Section>
  <Section position="7" start_page="359" end_page="360" type="metho">
    <SectionTitle>
5 Unsupervised Learning for
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="359" end_page="359" type="sub_section">
      <SectionTitle>
Acquisition of Polar Atoms
</SectionTitle>
      <Paragraph position="0"> Figure 3 shows the flow of our unsupervised learning method. First, the runtime SA system identifies the polar clauses, and the candidate polar atoms are collected. Then, each candidate atom is validated using the two metrics in the previous section, cp and cd, which are calculated from all of the polar clauses found in the domain corpus.</Paragraph>
      <Paragraph position="1">  and their frequencies. '*' denotes that it should not be added to the lexicon. f(a), p(a), and n(a) denote the frequency of the atom and in positive and negative contexts, respectively.</Paragraph>
    </Section>
    <Section position="2" start_page="359" end_page="359" type="sub_section">
      <SectionTitle>
5.1 Counts of Candidate Polar Atoms
</SectionTitle>
      <Paragraph position="0"> From each proposition which does not have a polarity, candidate polar atoms in the form of simple atoms (just a verb or adjective) or complex atoms (a verb or adjective and its right-most argument consisting of a pair of a noun and a postpositional) are extracted. For each candidate polar atom a, the total appearances f(a), and the occurrences in positive contexts p(a) and negative contexts n(a) are counted, based on the context of the adjacent clauses (using the method described in Section 4.1).</Paragraph>
      <Paragraph position="1"> If the proposition has the neg feature, the polarity is inverted. Table 6 shows examples of candidate polar atoms with their frequencies.</Paragraph>
    </Section>
    <Section position="3" start_page="359" end_page="360" type="sub_section">
      <SectionTitle>
5.2 Determination for Adding to
Lexicon
</SectionTitle>
      <Paragraph position="0"> Among the located candidate polar atoms, how can we distinguish true polar atoms, which should be added to the lexicon, from fake polar atoms, which should be discarded? As shown in Section 4, both the coherent precision (72-77%) and the coherent density (7-19%) are so small that we cannot rely on each single appearance of the atom in the polar context. One possible approach is to set the threshold values for frequency in a polar context, max(p(a),n(a)) and for the ratio of appearances in polar contexts among the to- null tal appearances, max(p(a),n(a))f(a) . However, the optimum threshold values should depend on the corpus and the initial lexicon.</Paragraph>
      <Paragraph position="1"> In order to set general criteria, here we assume that a true positive polar atom a should have higher p(a)f(a) than its average i.e. coherent density, cd(d,L+a), and also have higher p(a) p(a)+n(a) than its average i.e. coherent precision, cp(d,L+a) and these criteria should be met with 90% confidence, where L+a is the initial lexicon with a added. Assuming the binomial distribution, a candidate polar atom is adopted as a positive polar atom7 if both (17) and (18) are satisfied8.</Paragraph>
      <Paragraph position="3"> We can assume cd(d,L+a) similarequal cd(d,L), and cp(d,L+a) similarequal cp(d,L) when L is large. We compute the confidence interval using approximation with the F-distribution (Blyth, 1986).</Paragraph>
      <Paragraph position="4"> These criteria solve the problems in minimum frequency and scope of the polar atoms simultaneously. In the example of Table 6, the simple atom chiisai (ID=1) is discarded because it does not meet (18), while the complex atom chiisai - bodii-ga (ID=3) is adopted as a positive atom. shikkari-suru (ID=2) is adopted as a positive simple atom, even though 10 cases out of 64 were observed in the negative context. On the other hand, todoku - mokuyou-ni (ID=4) is discarded because it does not meet (17), even though n(a)f(a) = 1.0, i.e. always observed in negative contexts.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>