File Information
File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/metho/82/p82-1016_metho.xml
Size: 12,559 bytes
Last Modified: 2025-10-06 14:11:30
<?xml version="1.0" standalone="yes"?> <Paper uid="P82-1016"> <Title>An Improved Heuristic for Ellipsis Processing*</Title> <Section position="3" start_page="0" end_page="87" type="metho"> <SectionTitle> 3. The Heuristic </SectionTitle> <Paragraph position="0"> There are three aspects to our solutien: a mechanism for repetition and replacement ellipsis, an extension for inputs of different types, such as fragmentary answers to questions, and an extension for expansion ellipsis.</Paragraph> <Section position="1" start_page="0" end_page="85" type="sub_section"> <SectionTitle> 3.1 Repetition and Replacement </SectionTitle> <Paragraph position="0"> As noted above, repetition and replacement ellipsis can be viewed as substitution in the previous form. We have implemented this notion in an augmented transition network (ATN) grammar interpreter with the assumption that the &quot;previous form&quot; is the complete ATN path that parsed the previous input and that the lexical items consumed along that path are associated with the arcs that consumed them. In ellipsis mode, the ATN interpreter executes the path using the elliptical input in the following way: I. Words from the elliptical input, i.e., the curren~ input, may be consumed along the path at any point.</Paragraph> <Paragraph position="1"> 2. Any arc requiring a word not found in the current input may be traversed using the lexical item associated with the arc from the previous input.</Paragraph> <Paragraph position="2"> 3. However, once the path consumes the first word from the elliptical input, all words from the elliptical input must be consumed before an arc can use a word from the previous input.</Paragraph> <Paragraph position="3"> 4. Traversing a PUSH arc may be accom ~ plished either by following the sub-path of the previous input or by finding any constituent ef the required type in the current input.</Paragraph> <Paragraph position="4"> The entire ATN can be used in these cases.</Paragraph> <Paragraph position="5"> Suppose that the path for &quot;Were you angry?&quot; is given by Table I. Square brackets are used to indicate subpaths resulting from PUSHes. &quot;...&quot; indicates tests and actions which are irrelevant te the current discussion.</Paragraph> <Paragraph position="6"> An ATN Path for &quot;Were you Angry?&quot; An elliptical input of &quot;Was he?&quot; following &quot;Were you angry?&quot; could be understeed by traversing all of the arcs as in Table I. Following point I above, &quot;was&quot; and &quot;he&quot; would be substituted for &quot;were&quot; and &quot;you&quot;. Following point 3, in traversing the arc (CAT ADJ ... (TO Sz)) the lexical item &quot;angry&quot; from the previous input would be used. Item 4 is illustrated by an elliptical input of &quot;Was the old man?&quot;; this is understood by traversing the arcs at the S level of Table I, but using the appropriate path in the NP network to parse the old man</Paragraph> </Section> <Section position="2" start_page="85" end_page="86" type="sub_section"> <SectionTitle> 3.2 Transformations of the Previous Form </SectionTitle> <Paragraph position="0"> While the approach illustrated in Section 3.1 is useful in a data base query environment where ~\]liptical input typically is a modlfication of the previous query, it does not account for elliptical statements following questions, elliptical questions following statements, etc. Our approach to the problem is to write a set ef transformations which map the parse path of a question (e.g., Table I) into an expected parse path for a declarative response, and the parse ~path for a declarative into a path for an expected question, etc.</Paragraph> <Paragraph position="1"> The left-hand side of a transformation is a pattern which is matched against the ATN path of the previous utterance.</Paragraph> <Paragraph position="2"> Pattern elements include literals referring te arcs, variables which match a single arc or embedded path, variables which match zero or mere arcs, and sets ef alternatives. It is straightforward to construct a discrimination net corresponding to all left-hand sides for efficiently finding what patterns match the ATN path of the previous sentence. The right-hand side ef a transformation is a pattern which constructs an expected path. The form of the pattern en the right-hand side is a list of references to states, arcs, and lexical entries. Such references can be made through items matched on the left-hand side or by explicit construction ef literal path elements.</Paragraph> <Paragraph position="3"> Our technique is to restrict the mapping such that any expected parse path is generated by applying only one transformation and applying it only once. A special feature of our transformational system is the automatic allowance for dialogue diexis. An expected parse path for the answer to &quot;Were you angry?&quot; is given in Declarative for the expected answer for &quot;Were you angry?&quot;.</Paragraph> <Paragraph position="4"> Using this path, the ellipsis interpreter de'scribed in Section 3.1 would understand the ellipses in &quot;a)&quot; and &quot;b)&quot; below, in the same way as &quot;a')&quot; and &quot;b'i&quot; a) I was.</Paragraph> <Paragraph position="5"> a') I was angry.</Paragraph> <Paragraph position="6"> b) ~y spouse was.</Paragraph> <Paragraph position="7"> b') My spouse was angry.</Paragraph> </Section> <Section position="3" start_page="86" end_page="86" type="sub_section"> <SectionTitle> 3.3 Expansions </SectionTitle> <Paragraph position="0"> A large class of expansions are simple adjuncts, such as examples c, d, e, and g in section I. We have handled this by building our ellipsis interpreter to allow departing from the base path at designated states to consume an adjunct from the input string. We mark states in the grammar where adjuncts can occur. For each such state, we list a set of linear (though possibly cyclic) paths, called &quot;expansion paths&quot;. Our interpreter as implemented allows departures from the base path at any state so marked in the grammar; it follows expansion paths by consuming words from the input string, and must return to a state on the base form.</Paragraph> <Paragraph position="1"> Each of the examples in c, d, e, and g of section I can be handled by expansion paths only one arc long. They are given in Table 3.</Paragraph> </Section> <Section position="4" start_page="86" end_page="87" type="sub_section"> <SectionTitle> Example Expansion Paths </SectionTitle> <Paragraph position="0"> Since this is an extension to the ellipsis interpreter, combinations of repetition, replacement, and expansion can all be handled by the one mechanism. For instance, in response to &quot;Were you angry?&quot;, &quot;Yesterday you were (angry)&quot; would be treated using the expansion and replacement mechanisms.</Paragraph> <Paragraph position="1"> ~. Special Cases and Limitations The ideal model of contextual ellipsis would correctly predict what are appropriate elliptical forms in context, what their interpretation is, and what forms are not meaningful in context. We believe this requires structural restrictions, semantic constraints, and a model of the goals of the speaker. Our heuristic does not meet these criteria in a number of cases.</Paragraph> <Paragraph position="2"> Only two classes of structural constraints are captured. One relates the ellipsis to the previous form as a combination of repetition, replacement, and expansion. The o~her constraint is that the input must be consumed as a contiguous string. This constraint is violated, for instance, in &quot;I was (angry) yesterday&quot; as a response to &quot;Were you angry?&quot; Nevertheless, the constraint is computationally useful, since allowing arbitrary gaps in consuming the elliptical input produces a very large space of correct interpretations. A ludicrous example is the following question and elliptical response: Has the boss given our mutual friend a raise? A fat raise.</Paragraph> <Paragraph position="3"> Allowing arbitrary gaps between the sub-strings of the ellipsis allows an interpretation such as &quot;A (boss has given our) fat (friend a) raise.&quot; While it may be possible to view all contextual ellipsis as combinations of the operations repetition, replacement, and expansion applied to something, our model makes the strong assumption that these operations may be viewed as applying to an ATN path rather straightforwardly related to the previous utterance. Not all expansions can be viewed that way, as example f in Section I illustrates. Also, answers of &quot;No&quot; require special processing; that response in answer to &quot;Were you angry&quot; should not be interpreted as &quot;No, I was angry.&quot; One should be able to account for such examples within the heuristic described in this paper, perhaps by allowing the transformation system described in section 3.2 to be completely general rather than strongly restricted to one and only one transformation application. Rowever, we propose handling such cases by special purpose rules we are developing.</Paragraph> <Paragraph position="4"> These rules for the special cases, plus the mechanism described in section 3 together will be formally equivalent in predictive power to a grammar for elliptical forms.</Paragraph> <Paragraph position="5"> Though the heuristic is independent of the individual grammar, designating expansion paths and transformations obviously is not. The grammar may make this an easy oz&quot; difficult task. For instance in the grammar we are using, a subnetwork that collects all tense, aspect, and modality elements would simplify some of the transformations and expansion paths.</Paragraph> <Paragraph position="6"> ~aturally, semantics must play an important part in ellipsis processing.</Paragraph> <Paragraph position="7"> Consider the utterance pair below: Did the bess have a martini at lunch? Some wine.</Paragraph> <Paragraph position="8"> Though syntactically this could be interpreted either as &quot;Some wine (did have a martini at lunch)&quot;, &quot;(The boss did have) some wine (at lunch)&quot;, or &quot;(The boss did have a martini at) some wine&quot;. Semantics should prefer the second reading. We are testing our heuristic using the RUS grammar (Bebrow, 1978) which has frequent calls from the grammar requesting that the semantic component decide whether to build a semantic interpretation for the partial parse found or to veto that partial parse.</Paragraph> <Paragraph position="9"> This should aid performance.</Paragraph> <Paragraph position="10"> ~. Summary and Conclusion There are three aspects te our solution: a mechanism for repetition and replacement ellipsis, an extension for inputs of different types, such as fragmentary answers to questions, and an extension for expansion ellipsis.</Paragraph> <Paragraph position="11"> Our heuristic deals with the three types of expansion ellipsis as follows: Repetition ellipsis is processed by repeating specific parts of a transformed previous path using the same phrases as in the transformed form (&quot;I was angry&quot;). Replacement ellipsis is processed by substituting the elliptical input for contiguous constituents on a transformed previous path. Expansion ellipsis may be processed by taking specially marked paths that detour from a given state in that path. Combinations of the three types of ellipsis are represented by combinations of the three variations in a transformed previous path.</Paragraph> <Paragraph position="12"> There are two contributions of the work. First, our method allows for expansion ellipsis. Second, it accounts for combinations of previous sentence form and ellided form, e.g., statement following question, question following statement, question following question. Furthermore, the method works without any constraints on the ATN grammar. The heuristics carry over to formalisms similar to the ATN, such as context-free grammars and augmented phrase structure grammars.</Paragraph> <Paragraph position="13"> Our study of ellipsis is part of a much broader framework we are developing for processing syntactically and/or semantically ill-formed input; see Weischedel and Sondheimer (1981).</Paragraph> </Section> </Section> class="xml-element"></Paper>