<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-1305">
  <Title>A Model for Processing Temporal References in Chinese</Title>
  <Section position="4" start_page="2" end_page="4" type="metho">
    <SectionTitle>
METBY;; OVERLAPS;; OVERLAPPED,
DURING;; CONTAINS;; STAREDBY,
STARTS;; FINISHES;; FINISHEDBY,
SAME ASg
</SectionTitle>
    <Paragraph position="0"> For an absolute relation of a single event, T is an indispensable parameter, which includes event  OCCURis a predicate for the happening of a single event. Under the situations where there are no ambiguity, E</Paragraph>
    <Paragraph position="2"> is simplied as TR(T).</Paragraph>
    <Paragraph position="3">  There maybe exist more than one reference time in a statement.</Paragraph>
    <Paragraph position="5"> Some Chinese words can function as the temporal indicators. These include time word (TW), time position word (F), temporal adverb (ADV ), auxiliary word (AUX), preposition word (P), auxiliary verb (VA), trend verb (VC) and some special verbs (VV). They are all regarded as the elements of the temporal indicator TI: TI= fTW;; F;; ADV;; AUX;; VA;;VC;;VV;;Pg Each type of the indicators, e.g. TW, contains a set of words, such as TW = twlist =</Paragraph>
    <Paragraph position="7"> g, with eachword having an temporal attribute, indicated by ATT.</Paragraph>
    <Paragraph position="8"> The core of the model is thus a rule set R which maps the combinational eects of all the indicators, TI,inasentence to its corresponding</Paragraph>
    <Paragraph position="10"> Regarding to the temporal relations, the language has three basic forms in representation: Single event statement: in which only one single event is stated.</Paragraph>
    <Paragraph position="11"> Multiple events statement: in which two or more events are stated.</Paragraph>
    <Paragraph position="12"> declaration statement: in which the event(s) are declared by a person or an organization.</Paragraph>
  </Section>
  <Section position="5" start_page="4" end_page="4" type="metho">
    <SectionTitle>
3 Rule Construction
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="4" end_page="4" type="sub_section">
      <SectionTitle>
3.1 General Rules for Temporal
References (GR)
</SectionTitle>
      <Paragraph position="0"> Some general temporal characteristics, approximations and assumptions are examined to understand and uncover the hidden reference times, or to polish the identied solutions in order to make them more nature. For example, PAST(cw I, reporting date) is probably better than ON( , a few days ago). Or, when no explicit reference time is given, the default value of T, i.e. the \reporting date&amp;quot; (thereafter referred to as RD), would be assumed. It must be noting that the rules given below are not for the immediate use of extracting TR. But they are necessary to design  the TICS program.</Paragraph>
      <Paragraph position="1"> 1. TR(T) (single event) supports the following rules: (1) Approximation:</Paragraph>
      <Paragraph position="3"> (2) Negation:</Paragraph>
      <Paragraph position="5"> the following rules:  (3) Symmetry:</Paragraph>
      <Paragraph position="7"/>
    </Section>
  </Section>
  <Section position="6" start_page="4" end_page="7" type="metho">
    <SectionTitle>
3. TR
</SectionTitle>
    <Paragraph position="0"/>
    <Paragraph position="2"/>
    <Section position="1" start_page="4" end_page="4" type="sub_section">
      <SectionTitle>
3.2 Impact Coecients of Temporal
</SectionTitle>
      <Paragraph position="0"> Indicators (R0) The combined eect of all the temporal indicators in a sentence determines its temporal relation. However, in dierent situations, a certain indicator may have dierent eects. Compared (a) ;ThAEec (This morning, he read the newspaper) and (b) %*OeUyV (I read two books yesterday), the two sentences are alike as they both embody an indicator e, which implies PAST in principle. The sole dierence is that a denite time present in (b). (a) means the reading is nished at the speech time and the per-son must have known some news or information from his reading. Thus TR= PAST(t s ) for (a).</Paragraph>
      <Paragraph position="1"> However, (b) means the event took place yesterday but not before yesterday. Consequently, for (b), TR= ON(%, yesterday) is appropriate. In the database, the impact coecients are dened for the temporal indicators when T does or does not present in the sentence.</Paragraph>
      <Paragraph position="2"> Remark: It is likely for a sentence to contain two or more indicators. For example, adverbs .2 and aspectual auxiliary word e together express a past tense and they both share the same reference  time t r . The same kind instances include R...</Paragraph>
      <Paragraph position="3"> (will) and t...o (being) etc. Another example,  such as )? .2 (before the National Day, one has already), however includes two reference times. Here, )? (National Day) is t</Paragraph>
      <Paragraph position="5"> The current proposed algorithm is unable to mine the implicit reference time (i.e. t  Thus, for this example, a relation of PAST()?, National Day) is enough to provide sucient information to the users. To cater for these cases, we dene a general rule: if all the indicators in a sentence indicate the same relation, then it is identied as TR(hereafter this rule together with impact coecients is referred as R0).</Paragraph>
    </Section>
    <Section position="2" start_page="4" end_page="7" type="sub_section">
      <SectionTitle>
3.3 Rules for Resolving Conicts (R1)
</SectionTitle>
      <Paragraph position="0"> In many situations, the indicators in a sentence mayintroduce more than one relation. For example, adverbs .2 (have already) and o (being) indicate PAST and ONGING, respectively. But they could be collocated to represent some event which began in the past and continued to the reference time. For example AE .2oc (He has been reading newspaper). Such a problem is regarded as conict. In the following, ve cases are illustrated with examples. To resolve this conict, asetof rules are being dened in Table</Paragraph>
      <Paragraph position="2"> {IAA(They believed that the stock market will still be the major motivation for the HK</Paragraph>
      <Paragraph position="4"/>
      <Paragraph position="6"> uC(It has laid agood foundation for the market</Paragraph>
      <Paragraph position="8"> See the last rule in Table 1  (See the last rule in Table 3. To t for this case,</Paragraph>
      <Paragraph position="10"> See the eighth rule in Table 3. For those rules in Table 3, the parameters t</Paragraph>
      <Paragraph position="12"> insurance business, especially general insurance, has been aected by the Asian nancial crisis.)</Paragraph>
      <Paragraph position="14"/>
    </Section>
    <Section position="3" start_page="7" end_page="7" type="sub_section">
      <SectionTitle>
3.4 Rules for Discovering the Relevance
</SectionTitle>
      <Paragraph position="0"> of TwoEvents (R2 &amp; R3) To express two relevantevents is straightforward. In general, one of them is treated as the reference  , i.e. the event concerned, is expression by the main clause. The position words (F), suchas(before) and(after), and some special nouns, suchas (when) and I- (during) between the twoevent expressions play an important role in determining their relevance in time. Also, it is noticed that the im- null . Especially for the latter, the inuence of TR(E  ) is indispensable.</Paragraph>
      <Paragraph position="1"> The rules for this are being dened in the rule set R2. In addition, some special templates are also necessary for relating twoevent, which are being dened in the rule set R3, when F is absent.  month of nancial pages of LUc (Ta Kung Bao). In total, 7924 temporal relations are discovered from the data. The distribution of temporal relations in test data is shown in Table 4. Considering the ultimate objective of this researchisto nd out the temporal relations embedded in sentences, the focus of the evaluation is therefore to gure out the number of the temporal relations of single event (i.e. TR(E)) and of multiple events (i.e. TR(E</Paragraph>
      <Paragraph position="3"> )), which are correctly marked by the program. Table 5 shows the results. Table 6 gives the order of TRclassied by the program.</Paragraph>
      <Paragraph position="4"> After analyzing the outputs, it was discovered that most errors were due to: (1) t as a noun modier;; Since the proposed method does not integrate the mechanism of parsing, the association between a modier and its corresponding modifyee is not clear. In view of the task engaged, a time expression (indicated by t) could either modify a verb as an adverb, or modify a noun as a restricted modier. Only the adverb t, determines the temporal reference of the event described by the verb. Thus, the mistakeisunavoidable when a noun modier t appears in the text.</Paragraph>
    </Section>
  </Section>
  <Section position="7" start_page="7" end_page="7" type="metho">
    <SectionTitle>
(2) Ambiguous rules
</SectionTitle>
    <Paragraph position="0"> All the rules are dened on the basis of indicators' attributes. The majority attributes is taken to be the nal inferences. However, some special words may lead to exceptional results.</Paragraph>
    <Paragraph position="1"> These special words are possible sources of errors.</Paragraph>
    <Paragraph position="2"> Following is the example of a typical ambiguous rule.</Paragraph>
    <Paragraph position="3">  continue concentrating on the development of computer monitors, and the related in order to widen the product.</Paragraph>
    <Paragraph position="4">  reason: The word AZ has the essence of CON-TINUED, but it is independent to any reference time.</Paragraph>
  </Section>
  <Section position="8" start_page="7" end_page="7" type="metho">
    <SectionTitle>
(3) Noisy Annotation
</SectionTitle>
    <Paragraph position="0"> Some errors are resulted from noisy annotations. For example, noun or verb? #&amp;quot; (vnA)(auxe) )I2{ (v) (?) (did push the growth of national economy) vs or v?  (ty+-) IqM,ccC(vaI) (vsUY) (?) (v3/4$) C(the price of the new island houses will be annonced this week) vv or v? (t+-) UAIO\?OF(vvq) (?) (v, u) C(the two departments will vote for the suggestion of the mergence on next Monday) (4) Coreference  The program searches for a solution within a sentence bounded by a full stop. As the connections between two sentences are ignored, it is incapable to resolvethecoreference problem. As such, the following two sentences are all identied as TR= PAST, which is acceptable for the rst and correct for the second. Nevertheless, since 3 links the currenteventtothe event described in the last sentence (indicated by ?), a solution of SAME AS(?, zI) would be more accurate. Similarly, BEFORE($?,n) is more proper in the second sentence with $ refering to the event stated before. The problem of coreference will be studied in our future work.</Paragraph>
  </Section>
class="xml-element"></Paper>