<?xml version="1.0" standalone="yes"?>
<Paper uid="C88-2102">
  <Title>Maintaining Consistency and Plausibility in Integra,ted Natm-al Langu~ge Understanding</Title>
  <Section position="4" start_page="0" end_page="0" type="metho">
    <SectionTitle>
2 Maintaining Cons~si, e:c~cy
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
The CME (Consistency Maintenm~ce Engixtc)
</SectionTitle>
      <Paragraph position="0"> is a component of the httegrated pa:rsmg (mghte  re~;ponsib\]e fl)r maintaining consistency among beliefs. Basic design principles of the CME is b~L, md on de Kleer's ATMS (Assumptionbz~sed '\]!i'uth Maintenance Engine) \[de 86\]. The CME maintains a set of alternative betlet~ eae~ of which consists of a set of as~mp~ion~ m~d their conclusions, as follows: alter'na~,ive I {./ilj~odeg,,A!,rt~} Bll~ .... ,Blm~ alternative re (A,,i~. * . ,Aura,, } Bul~ ...~ Bunt,, en vlr o~tme~t couclu sious An extc:c~fl problem solver is assumed to exist which makeu a~sumpfion, adds conclusion, and dctcd;s contx~a~li(:tiondeg ~\['he mv~n ~ask of CME is to maintain alternative bc~i('2~ by removing all alternatives whose :;ct of a:~'~mmptions has turned out contradictorydeg Lik(, ATMS, the CME takes advantage of the followi~,g monotonic property: if ~ contr~dictlo** is derived from a set of assumptions A, then contradiction is Mso derived from any set of assumptions B such that B D A.</Paragraph>
      <Paragraph position="2"/>
      <Paragraph position="4"> Thus, if contradiction is derived from a set of as.,mm~)tions { t~-~, D ), alternative in terpretatiol~s depending on sets of assumptions such as {B,C,D}, {A,B,D}, \[A,B,C,D}, ... are removed. \[n addition~ t, he GME keeps records of contradictory sets of assumptions to prevent any interpretation depending on them from being considered in future.</Paragraph>
      <Paragraph position="5"> Unlike ATMS whose control regime is breadfirst, our CME uses a tree called the envh'onment tree, or the E-tree for short, to guide the search process. Each node of the E-tree represents an environment, a set of assumptions.</Paragraph>
      <Paragraph position="6"> \]i;alch arc of the E4ree represents that a lower node is derived from the upper node by making one :more assumption. Thus in figure 1, E0 is the root node, and it represents an environmnet without any assumption. Nodes below -5;0 :represent environments with one or more assumption added to its parent node's envi-</Paragraph>
      <Paragraph position="8"> We assume that a set of assumptions made at ~he same parent node axe mutually exclusive.</Paragraph>
      <Paragraph position="9"> Although this is a rather strong assumption, it, makes sense in ~tatural language tmderstanding :~ince many assmuptions being made duri~g the natural language mlderstanding process are mutuMly exclusive. Even if this is not the c~se, any set of assumptions can be transformed into a set of mutually exclusive assumptions by adding appropriate conditions. Although this is a cumbersome solution, it does not often take place in natural language understanding and most importantly it saves tile amottnt of computation. null Note that the CME alone cannot determine which way to go when there is more than one possibility of extending the set of beliefs. This information is provided by the PME, as described in the next section.</Paragraph>
    </Section>
  </Section>
  <Section position="5" start_page="0" end_page="4133" type="metho">
    <SectionTitle>
3 Maintaining Plausibility
</SectionTitle>
    <Paragraph position="0"> The PME (Plausibility Maintenance Engine) inaintains estimations of how plausible each environment is. This information is given as conditional probabilities and it is kept as annotations to each arc of the F,-tree. Thus, in figure 2, which is a slightly more precise version of figure 1, Pl stands for P(EI), pq for P(EjIAi), pi./~ for P(Ek, IAi, Aj), etc.</Paragraph>
    <Paragraph position="1"> It follows from the property of conditional probability that = O, if i ~ j and El and Ej are immediate children  (a) initial E-tree. (b) The F~tree after -,E~ is</Paragraph>
    <Paragraph position="3"> of the same parent. Furthermore, if Ej is a parent node of Ei.</Paragraph>
    <Paragraph position="4"> Initial value of pi's are to be given from the external problem solver. The PME's role is to maintain estimation of prausibility by taking into account given observations. Currently we only take -~E, the event of environment E running into contradiction, as an observation. We use a Bayes' law to modify P(A) into P(AI-E).</Paragraph>
    <Paragraph position="5"> Thus,</Paragraph>
    <Paragraph position="7"> if El and Ej are brothers, (1) is further simplified to:</Paragraph>
    <Paragraph position="9"> For example, suppose it has turned out that environment E4 is in contradiction and hence -E4 is observed (figure 3(a)). The annotations to the E-tree are updated as in figure 3(b).</Paragraph>
    <Paragraph position="10"> Notice that the update of conditional probability can be done based on local information.  lag Engine as a Subsys tern The integrated parsing engine consists of the CME and the PME. The architecture of a natural language understanding system with the integrated parsing engine as a subsystem is shown in figure 4.</Paragraph>
    <Paragraph position="11"> The knowledge base contains various types of information for language comprehension, including lexicon, morphology, syntax, semantics, discourse, pragmatics, commonsenses, and so on. The whole system is controled by the problem solving engine (PSE). The PSE can access to the knowledge base and use the integrated parsing engine as an aid to seek for most plausible interpretation. Input texts are analyzed in a sentence-by-sentence manner. The discourse structure is maintained as a previous topic in the working memorydeg When it scans a new sentence, the PSE tirs~ initialize the F~tree with only the root nodedeg Then the PSE repeats the following cycle: (step 1) choose a leaf node with the highest probability as a working envirb  believed p~'opositions until either (a) the goal is achieved, (b) contra, diction is derived~ or (c) no more conchlsion is derived ~mless making more assumption.</Paragraph>
    <Paragraph position="12"> In case (a), the process hMts.</Paragraph>
    <Paragraph position="13"> In case (b), the process is passed to the PME~ which modifies current estimation of plausibility so that this f,~:t is reflected, then nat alternative of meximum plausibility is dmsen ~(l is suggested to the CME.</Paragraph>
    <Paragraph position="14"> In case (c)~ the process also is passed to the PME, which assigns plausibib ity to new nodes, and working cnvironment is chosen agMn.</Paragraph>
    <Paragraph position="15"> The integrated parsing engine has been written in Lisp. It is running with a small exmerlmental grammar for Japanese. The next section shows how it works.</Paragraph>
  </Section>
  <Section position="6" start_page="4133" end_page="4133" type="metho">
    <SectionTitle>
5 .A.n Example
</SectionTitle>
    <Paragraph position="0"> Suppose a dialog envh'omne~tt in which a pro..</Paragraph>
    <Paragraph position="1"> fesso~&amp;quot; speaks to a clerk to borrow a key of some rooms (figure 5) and utters the following  ambiguous if there is more than one key in a given situation. Suppose three keys are there: key1 for a hbrary room, key2 for a xerox room, and key3 for a meeting room.</Paragraph>
    <Paragraph position="2"> Although sentence (3) is ambiguous in norreal contexts, it becomes much lcss so if it follows sentences like: (4) HO N WO KO PI I SHI TA I NO DE SU GA &amp;quot;I'd like to xerox some books.&amp;quot; Even if no previous sentence is spoken, sen-. fence (3) is acceptable in a situation where the speaker and the hearer rmltually believe that the xerox room is accessed so often that &amp;quot;the key&amp;quot; is usually uscd to refer to key&amp; the one for the xerox room.</Paragraph>
    <Paragraph position="3"> Note that the omission of the patient case does not matter in usual situations, since there is a strong defa~flt that the filler of this case is the speaker.</Paragraph>
    <Paragraph position="4"> Now let us show how sentence (3) is analyzed in a context where sentence (4) was previously uttered. The task of analyzing input starts from recognizing words. Lots of ambiguities arise in this phase. For sentence (3), 'KA' might be a single word 'KA' (postposition marking interrogative) or a part of a longer word 'KAGI' (key). Since longer match is considered to be more plausible in generM case in Japanese analysis, we assign larger number of probability to the latter possibility. Following this anMysis, the PSE makes the assumptions to the integrated parsing engine: @toord-1 (t~ke the sequence ~KA t as a word): ~-~ probability 1/3.</Paragraph>
    <Paragraph position="5"> @word-2 (ta&amp;e the sequence 'KAGI' as a word): probability 2/3.</Paragraph>
    <Paragraph position="6"> Accordingly, 'the CME extends the initial E-tree as in figure 6. Since, the enviromnent E1 has the highest plausibility, the CME chooses it for the next environment and control is returned to the PSE.</Paragraph>
    <Paragraph position="7"> 4~3 5 k , the libra W cy t ............ room ,,,,., .</Paragraph>
    <Section position="1" start_page="4133" end_page="4133" type="sub_section">
      <SectionTitle>
Concepts
</SectionTitle>
      <Paragraph position="0"> Now the PSE tries to derive further conclusion in the chosen environment. After having i'ccognized that the pm't of speech of the word 'KA(~I' i~ noun, the PSE tries to find out the referent of the noun and reahzes that thi'ee ambigtAties arise lit this situation. Again, the PSE calls the CME to make assumptions. At the same time, the PSE is called for to assign estimated conditional probabihties to each assumptiondeg null Currently, the system uses an associative network as shown in figure 7 to determine plausL bility. Nodes of this network represent either a concept or art instzatce, and arcs mean that the two concepts or instants at its both ends have a certain relation. Those items which have dense conuections to previous subjects are considered to be plausible as a referent. In our example, since the node xerox is marked as the previous subject key2 is considered most plausible, while key1 is less plausible and key3 much less. Thus, the following assumptions are made: 1 @re fereni-1 (consider 'KAGI' to refer to keyl): =~ probabiliy 1/3.</Paragraph>
      <Paragraph position="1"> @referent-2 (consider 'KAGI' to refer to key$): --~ probabiliy 1/2.</Paragraph>
      <Paragraph position="2"> @re\[erenl-3 (consider 'KAGI' to refer to key3): =ez probabiliy 1/6.</Paragraph>
      <Paragraph position="3"> In case no previous utterance is given, the PSE will consult information given as a priori measurements.</Paragraph>
      <Paragraph position="4"> The E-4ree now becomes as in figure 8, a~td {@word-2, @referent-2}, which is the most 1 Currently we use a very simple algorithm for assigning those value: when there are three alternatives, the densest connection receives the vMue (1/3), the second (1/2), and the third (1\]6), regardless of how closely they are related to each other. We plan to develop a much more precise method in a near future.</Paragraph>
      <Paragraph position="5">  plausible enviromnent at tiffs point, is chosen as the next environment. The analysis is continued this way until the semantic representation is obtained for the whole sentence. The interpretation obtained tlds case is:</Paragraph>
      <Paragraph position="7"> Figure 9 shows the dependency structtu'e of befiefs related to this analysis.</Paragraph>
      <Paragraph position="8"> Notice that the efficiency of the analysis is significantly improved when strong expectation exists. For example, although character 'sin' h~ sentence (3) has many possible interpretations in Japanese, the system is not annoyed by those ambiguities, since this part of the sentence just goes as expected. The system may come to suspect it only when most of its expectation faik.</Paragraph>
      <Paragraph position="10"> addition~ the integrated paxsign engine provides a concise and high level mechatdsm for abduc~ tire reasoning. We have carefully chosen a set of reasonably high-level functions necessary for abductive reasoning. This serves to much simplifying natur~ langu.age mtdersta~tding system than otherwise.</Paragraph>
      <Paragraph position="11"> li'ig,~re 10: Gtree after assumptions about the proposed interpretation based on {@word-2, @referent-.2} is rejected Now suppose the above interpretation is rejected for some ~'eason, say by expficitly negated by the speaker. Th.e~ the system will eventually produce an alte~atative interpretation taking key1 as a referent, by changing ammtations to the E4ree as lit figm'e 10.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>