perm filename PAPER.DOC[4,KMC] blob
sn#061934 filedate 1973-08-31 generic text, type T, neo UTF8
00100 ␈↓ ↓-IDIOLECTIC LANGUAGE-ANALYSIS FOR
00200 ␈↓␈↓ mUNDERSTANDING DOCTOR-PATIENT DIALOGUES*
00300
00400 ␈↓ ↓>Horace Enea and Kenneth Mark Colby
00500
00600 ␈↓ ↓↑Department of Computer Science
00700 ␈↓ αAStanford University
00800 ␈↓ αDStanford, California
00900
01000
01100
01200
01300
01400
01500
01600
01700
01800
01900
02000
02100
02200
02300
02400
02500
02600
02700
02800
02900 ␈↓---------------------------------------------------------
03000 ␈↓*␈α This␈α research␈α is␈α supported␈α by␈α Grant␈α PHS␈α MH␈α 06645-12␈α from
03100 the␈α⊂National␈α⊃Institute␈α⊂of␈α⊂Mental␈α⊃Health,␈α⊂by␈α⊂(in␈α⊃part)␈α⊂Research
03200 Scientist␈α≤Award␈α≤(No.␈α≤ 1-K05-K14,433)␈α≤from␈α≤the␈α≤National
03300 Institute␈α⊃of␈α⊂Mental␈α⊃Health␈α⊂to␈α⊃the␈α⊂second␈α⊃author␈α⊂and␈α⊃(in␈α⊂part)
03400 by␈α⊃the␈α⊃Advanced␈α⊃Research␈α∩Projects␈α⊃Agency␈α⊃of␈α⊃the␈α∩Office␈α⊃of
03500 the␈α Secretary␈α of␈α Defense(SD-183).
03600 ␈↓---------------------------------------------------------
03700
03800 ␈↓␈↓ α\␈↓
03900 ABSTRACT␈↓
04000
04100 ␈↓ A␈α∂programming␈α∂language␈α⊂is␈α∂described␈α∂which␈α∂is␈α⊂designed␈α∂to
04200 simplify␈α↔the␈α⊗construction␈α↔of␈α⊗computer␈α↔programs␈α↔to␈α⊗analyze
04300 English.␈α⊃This␈α⊃system␈α⊂attempts␈α⊃to␈α⊃merge␈α⊂the␈α⊃best␈α⊃features␈α⊂of
04400 pattern␈α≤matchers␈α≤and␈α≥the␈α≤phrase␈α≤structure␈α≥approach␈α≤to
04500 languagα⊂analysisα⊃ Severaα⊂practicaα⊃problemα⊂whicα⊃occuα⊂in
04600 dealing␈α with␈α such␈α a␈α system␈α are␈α described.
04700
04800 ␈↓ α)␈↓
04900 INTRODUCTION␈↓
05000
05100 ␈↓ Why␈α∃is␈α∃it␈α∃so␈α∃difficult␈α∃for␈α∃machines␈α∃to␈α∃understand
05200 natural␈α→language?␈α_Perhaps␈α→it␈α→is␈α_because␈α→machines␈α→do␈α_not
05300 simulate␈α∃sufficiently␈α∀what␈α∃humans␈α∀do␈α∃when␈α∃humans␈α∀process
05400 language.␈α∞ Several␈αyears␈α∞of␈αexperience␈α∞with␈α∞computer␈αscience
05500 and␈α~linguistic␈α≠approaches␈α~have␈α~taught␈α≠us␈α~the␈α≠scope␈α~and
05600 limitations␈α of␈α syntactic␈α and␈α semantic␈αparsers.(Schank,Tesler␈α and
05700 Weber,␈↓↓8␈↓␈α Simmons,␈↓↓9␈↓␈α Winograd,␈↓↓13␈↓␈α Woods␈↓↓14␈↓).␈α!While␈α extant
05800 linguistic␈α∪parsers␈α∩perform␈α∪satisfactorily␈α∩with␈α∪carefully␈α∩edited
05900 text␈α∂sentences␈α∞or␈α∂with␈α∂small␈α∞dictionaries␈α∂,␈α∞they␈α∂are␈α∂unable␈α∞to
06000 deal␈α∂with␈α∂everyday␈α∞language␈α∂behavior␈α∂characteristic␈α∂of␈α∞human
06100 conversation.␈α≡ In␈α≡a␈α∨rationalistic␈α≡quest␈α≡for␈α∨certainty␈α≡and
00100 attracted␈α⊂by␈α⊂an␈α⊂analogy␈α⊂from␈α∂the␈α⊂proof␈α⊂theory␈α⊂of␈α⊂logicians␈α∂in
00200 ␈↓which␈α∂provability␈α∞implied␈α∂computability,␈α∂computational␈α∞linguists
00300 hoped␈α∪to␈α∩develop␈α∪formalisms␈α∩for␈α∪natural␈α∪language␈α∩grammars.
00400 But␈α⊂the␈α⊃hope␈α⊂has␈α⊃not␈α⊂been␈α⊂realized␈α⊃and␈α⊂perhaps␈α⊃in␈α⊂principle
00500 cannot␈α∀be.␈α∀ (It␈α∀is␈α∀difficult␈α∀to␈α∀formalize␈α∀something␈α∀which␈α∪can
00600 hardly␈α be␈α formulated).
00700
00800 Linguistic␈α~parsers␈α~use␈α~morphographemic␈α~analyses,
00900 parts-of-speech␈α'assignments␈α'and␈α'dictionaries␈α'containing
01000 multiple␈α≥word-senses␈α≡each␈α≥possessing␈α≡semantic␈α≥features,
01100 programs␈α⊗or␈α⊗rules␈α⊗for␈α⊗restricting␈α⊗word␈α↔combinations.␈α⊗ Such
01200 parsers␈α∩perform␈α⊃a␈α∩detailed␈α⊃analysis␈α∩of␈α⊃every␈α∩word,␈α⊃valiantly
01300 disambiguating␈α↔at␈α↔each␈α↔step␈α↔in␈α↔an␈α↔attempt␈α↔to␈α↔construct␈α⊗a
01400 meaningful␈α≤interpretation.␈α≤ While␈α≤it␈α≤may␈α≤be␈α≠sophisticated
01500 computationally,␈α∪a␈α∀conventional␈α∪parser␈α∀is␈α∪quite␈α∀at␈α∪a␈α∀loss␈α∪to
01600 deal␈α∩with␈α⊃the␈α∩caprice␈α∩of␈α⊃ordinary␈α∩conversation.␈α∩ In␈α⊃everyday
01700 discourse␈α⊂people␈α⊂speak␈α⊃colloquially␈α⊂and␈α⊂idiomatically␈α⊃using␈α⊂all
01800 sorts␈α_of␈α_pat␈α_phrases,␈α_ slang␈α_and␈α_cliches.␈α_ The␈α_number␈α_of
01900 special-case␈α↔expressions␈α_is␈α↔indefinitely␈α↔large.␈α_ Humans␈α↔are
02000 cryptic␈α∞and␈α∂elliptic.␈α∞ They␈α∞lard␈α∂even␈α∞their␈α∂written␈α∞expressions
02100 with␈α≤meaningless␈α≤fillers␈α≤and␈α≤fragments.They␈α≤convey␈α≠their
02200 intentions␈α∀and␈α∀ideas␈α∀in␈α∀idiosyncratic␈α∀and␈α∃metaphorical␈α∀ways,
02300 blithely␈αviolating␈α∞rules␈αof␈α'correct'␈α∞grammar␈αand␈α∞syntax.␈α Given
02400 these␈αdifficulties,␈α∞how␈αis␈α∞it␈αthat␈α∞people␈αcarry␈α∞on␈αconversations
02500 easily␈α∂most␈α∞of␈α∂the␈α∞time␈α∂while␈α∞machines␈α∂thus␈α∞far␈α∂have␈α∂found␈α∞it
02600 extremely␈α∃difficult␈α∃to␈α∀continue␈α∃to␈α∃make␈α∃appropriate␈α∀replies
02700 indicating␈α some␈α degree␈α of␈α understanding?
02800
02900 It␈α_seems␈α→that␈α_people␈α→`get␈α_the␈α→message'␈α_without
03000 always␈α∩analyzing␈α∩every␈α∩single␈α∩word␈α∩in␈α∩the␈α∩input.␈α∩They␈α∩even
03100 ignore␈α_some␈α↔of␈α_its␈α_terms.␈α↔People␈α_make␈α_individualistic␈α↔and
03200 idiosyncratic␈α⊃selections␈α⊃from␈α⊃highly␈α⊃redundant␈α⊃and␈α⊂repetitious
03300 communications.␈α⊃ These␈α⊃personal␈α⊃selective␈α⊃operations,␈α⊃ based
03400 on␈α∃idiosyncratic␈α∃intentions,␈α∃produce␈α∃a␈α∃transformation␈α∃of␈α∃the
03500 input␈α∂by␈α∂destroying␈α∂and␈α∂even␈α∂distorting␈α∂information.␈α∂ In␈α∞speed
03600 reading,␈α∪for␈α∪example,␈α∪only␈α∪a␈α∪small␈α∪percentage␈α∪of␈α∪contentive
03700 words␈α≠on␈α≠each␈α≠page␈α~need␈α≠be␈α≠looked␈α≠at.␈α≠ These␈α~words
03800 somehow␈α→resonate␈α~with␈α→the␈α→readers␈α~relevant␈α→conceptual-
03900 belief␈α⊃structure␈α⊃whose␈α∩processes␈α⊃enable␈α⊃him␈α∩to␈α⊃`understand'
04000 not␈α⊂simply␈α⊂the␈α⊂language␈α⊂but␈α⊂all␈α⊂sorts␈α⊂of␈α⊂unmentioned␈α∂aspects
04100 about␈α↔the␈α↔situations␈α↔and␈α↔events␈α↔being␈α↔referred␈α↔to␈α_in␈α↔the
04200 language.␈α∂ Normal␈α∞written␈α∂English␈α∞text␈α∂is␈α∞estimated␈α∂to␈α∂be␈α∞5/6
04300 redundant␈α9 (Rubenstein␈α8 and␈α9 Haberstroh␈↓↓7␈↓).␈α8Spoken
04400 conversations␈α∨in␈α∨English␈α∨are␈α∨probably␈α∨better␈α∨than␈α≡50%
04500 redundant(Carroll␈↓↓1␈↓).␈α≡Words␈α≡can␈α≡be␈α≡garbled␈α≡and␈α≥listeners
04600 nonetheless␈α∂get␈α∂the␈α⊂gist␈α∂or␈α∂drift␈α∂of␈α⊂what␈α∂is␈α∂being␈α⊂said.␈α∂ They
04700 see␈α the␈α "pattern"␈α and␈α thus␈α can␈α supply␈α much␈α of␈α what␈α is␈α missing.
04800
04900 To␈α"approximate␈α!such␈α"human␈α"achievements␈α!we
05000 require␈α_a␈α_new␈α_perspective␈α_and␈α_a␈α_practical␈α_method␈α_which
05100 differs␈α≥from␈α≥that␈α≡of␈α≥current␈α≥linguistic␈α≡approaches.␈α≥ This
05200 alternate␈α∨approach␈α∨should␈α∨incorporate␈α∨those␈α∨aspects␈α∨of
05300 parsers␈α⊃which␈α⊃have␈α⊃been␈α∩found␈α⊃to␈α⊃work␈α⊃well,␈α∩e.g.,␈α⊃detecting
05400 embedded␈α⊂clauses.␈α⊂ Also␈α⊂individualistic␈α⊂features␈α∂characteristic
05500 of␈α an␈α∨idiolect␈α should␈α∨have␈α dominant␈α emphasis.␈α∨Parsers
05600 represent␈α∞complex␈α∞and␈α∞refined␈α∞algorithms.␈α∞ While␈α∞on␈α∞one␈αhand
05700 they␈α≡subject␈α≡a␈α∨sentence␈α≡to␈α≡a␈α≡detailed␈α∨and␈α≡sometimes
05800 overkilling␈α≡analysis,␈α≥on␈α≡the␈α≥other,␈α≡they␈α≥are␈α≡finicky␈α≥and
05900 oversensitive.␈α⊗ A␈α⊗conventional␈α⊗parser␈α⊗may␈α⊗simply␈α⊗halt␈α⊗if␈α⊗a
00100 word␈α∂in␈α⊂the␈α∂input␈α∂sentence␈α⊂is␈α∂not␈α∂present␈α⊂in␈α∂its␈α⊂dictionary.␈α∂It
00200 ␈↓finds␈α∪ungrammatical␈α∩expressions␈α∪such␈α∩as␈α∪double␈α∩prepositions
00300 (`Do␈α~you␈α~want␈α~to␈α≠get␈α~out␈α~of␈α~from␈α~the␈α≠hospital?')␈α~quite
00400 confusing.␈α_ Parsers␈α→constitute␈α_a␈α→tight␈α_conjunction␈α→of␈α_tests
00500 rather␈α⊂than␈α∂a␈α⊂loose␈α⊂disjunction.␈α∂ As␈α⊂more␈α∂and␈α⊂more␈α⊂tests␈α∂are
00600 added␈α∂to␈α∞the␈α∂conjunction,␈α∂the␈α∞parser␈α∂behaves␈α∞like␈α∂a␈α∂finer␈α∞and
00700 finer␈α≡filter␈α≡which␈α∨makes␈α≡it␈α≡increasingly␈α≡difficult␈α∨for␈α≡an
00800 expression␈α⊂to␈α∂pass␈α⊂through.␈α∂ Parsers␈α⊂do␈α∂ not␈α⊂ allow␈α⊂ for␈α∂ the
00900 exclusions␈α typical␈α of␈α everyday␈α human␈α dialogues.
01000
01100 Finally,␈α∂it␈α∂is␈α⊂difficult␈α∂to␈α∂keep␈α∂consistent␈α⊂a␈α∂dictionary
01200 of␈α≡over␈α≡500␈α≡multiple-sense␈α≡words␈α≡classified␈α≡by␈α≡binary
01300 semantic␈α∂features␈α∂or␈α∂rules.␈α⊂For␈α∂example,␈α∂suppose␈α∂a␈α⊂noun␈α∂(Ni)
01400 is␈α∀used␈α∀by␈α∀some␈α∀verbs␈α∪as␈α∀a␈α∀direct␈α∀object␈α∀in␈α∀the␈α∪semantic
01500 sense␈α⊂of␈α⊃a␈α⊂physical␈α⊃object.␈α⊂Then␈α⊂it␈α⊃is␈α⊂noticed␈α⊃that␈α⊂Ni␈α⊃is␈α⊂also
01600 used␈α∞by␈α∞other␈α∞verbs␈α∞in␈α∞the␈α∞sense␈α∞of␈α∞a␈α∞location␈α∞so␈α∞`location'␈αis
01700 added␈α⊗to␈α⊗Ni's␈α⊗list␈α⊗of␈α⊗semantic␈α⊗features.␈α⊗ Now␈α↔Ni␈α⊗suddenly
01800 qualifies␈αas␈α a␈αdirect␈α object␈αfor␈αa␈α lot␈αof␈α other␈αverbs.␈α But␈α some
01900 of␈α_the␈α_resultant␈α_combinations␈α_make␈α_no␈α_sense␈α_even␈α_in␈α_an
02000 idiolect.␈α⊃If␈α⊂a␈α⊃special␈α⊂feature␈α⊃is␈α⊂then␈α⊃created␈α⊂for␈α⊃Ni,␈α⊃then␈α⊂one
02100 loses␈α_the␈α↔power␈α_of␈α_general␈α↔classes␈α_of␈α_semantic␈α↔features.
02200 Adding␈α∞a␈α∞single␈α∞semantic␈α∞feature␈α∞can␈α∞result␈α∞in␈α∞the␈αpropagation
02300 of␈α∃hidden␈α∀inconsistencies␈α∃and␈α∀unwanted␈α∃side-effect..␈α∃as␈α∀the
02400 dictionary␈α∞grows␈α∂it␈α∞becomes␈α∂increasingly␈α∞unstable␈α∂and␈α∞difficult
02500 to␈α control.
02600
02700 Early␈α"attempts␈α!to␈α"develop␈α"a␈α!pattern-matching
02800 approach␈α)using␈α)special-purpose␈α)heuristics␈α*have␈α)been
02900 described␈α↔by␈α⊗Colby,␈α↔Watt␈α⊗and␈α↔Gilbert,␈↓↓2␈↓␈α↔Weizenbaum␈↓↓11␈↓␈α⊗and
03000 Colby␈α∩and␈α∩Enea.␈↓↓3␈↓␈α∩The␈α⊃limitations␈α∩of␈α∩these␈α∩attempts␈α∩are␈α⊃well
03100 known␈α∪to␈α∪workers␈α∪in␈α∪artificial␈α∪intelligence.␈α∪The␈α∪man-machine
03200 conversations␈α∪of␈α∀such␈α∪programs␈α∪soon␈α∀becomes␈α∪impoverished
03300 and␈α∩boring.␈α∩Such␈α⊃primitive␈α∩context-restricted␈α∩programs␈α⊃often
03400 grasp␈α∪a␈α∩topic␈α∪well␈α∪enough␈α∩but␈α∪too␈α∩often␈α∪do␈α∪not␈α∩understand
03500 quite␈α∃what␈α∃is␈α∃being␈α∃said␈α∃about␈α∃the␈α∃topic,␈α∃with␈α∃amusing␈α∃or
03600 disastrous␈α consequences.␈α This␈αshortcoming␈α is␈α a␈αconsequence␈α of
03700 the␈α⊂limitations␈α⊂of␈α⊂a␈α⊂pattern-␈α⊂matching␈α⊂approach␈α⊂lacking␈α⊂a␈α⊂rich
03800 conceptual␈α∩structure␈α∩into␈α∪which␈α∩the␈α∩pattern␈α∪abstracted␈α∩from
03900 the␈α∀input␈α∃can␈α∀be␈α∀matched␈α∃for␈α∀inferencing.␈α∃ These␈α∀programs
04000 also␈α∀lack␈α∀a␈α∀subroutine␈α∀structure,␈α∀both␈α∀pattern␈α∀directed␈α∀and
04100 specific,␈α desirable␈α for␈α generalizations␈α and␈α further␈α analysis.
04200
04300 The␈α⊃strength␈α⊃of␈α⊃these␈α⊃pattern␈α⊃matching␈α⊂approaches
04400 lies␈α∞in␈αtheir␈α∞ability␈αto␈α∞ignore␈αsome␈α∞of␈αthe␈α∞input.␈α They␈α∞look␈αfor
04500 patterns,␈α∃which␈α∃means␈α∃the␈α∀emphasis␈α∃of␈α∃some␈α∃detail␈α∃to␈α∀the
04600 exclusion␈α∞of␈α∞other␈α∞detail.␈α∞ Thus␈α∞they␈α∞can␈α∞get␈α∞something␈α∞out␈α∞of
04700 nearly␈α every␈α sentence--␈α sometimes␈α more,␈α sometimes␈α less.
04800
04900 An␈α*interesting␈α)pattern-matching␈α*approach␈α)for
05000 machine␈α≤translation␈α≤has␈α≤been␈α≤developed␈α≤by␈α≥Wilks.␈↓↓12␈↓␈α≤His
05100 program␈α∂constructs␈α∂a␈α∂pattern␈α∞from␈α∂English␈α∂text␈α∂input␈α∂which␈α∞is
05200 matched␈α∪against␈α∪templates␈α∪in␈α∪an␈α∪interlingual␈α∪data␈α∪base␈α∪from
05300 which,in␈α⊗turn,␈α∃ French␈α⊗output␈α⊗is␈α∃generated␈α⊗without␈α⊗using␈α∃a
05400 generative␈α grammar.
05500
05600 In␈α$the␈α%course␈α$of␈α$constructing␈α%an␈α$interactive
05700 simulation␈α∀of␈α∀paranoia␈α∀we␈α∀were␈α∀faced␈α∀with␈α∀the␈α∀problem␈α∀of
05800 dealing␈α with␈αunedited␈α and␈α unrestricted␈αnatural␈α language␈α as␈αit␈α is
05900 used␈α$in␈α$the␈α#doctor-patient␈α$situation␈α$of␈α$a␈α#psychiatric
00100 interview.(Colby,␈α⊃Hilf,␈α⊃Weber,␈α⊃and␈α⊃Kraemer,␈↓↓4␈↓␈α⊃Colby␈α⊃and␈α⊂Hilf␈↓↓5␈↓).
00200 ␈↓This␈α+domain␈α+of␈α+discourse␈α+admittedly␈α,contains␈α+many
00300 psychiatrically␈α∩stereotyped␈α⊃expressions␈α∩and␈α⊃is␈α∩constrained␈α⊃in
00400 topics␈α∃(Newton`s␈α∃laws␈α∃are␈α∃rarely␈α∃discussed).␈α∃But␈α∃it␈α⊗is␈α∃rich
00500 enough␈α∪in␈α∪verbal␈α∀behavior␈α∪to␈α∪be␈α∀a␈α∪challenge␈α∪to␈α∀a␈α∪language
00600 understanding␈α∂algorithm␈α∂since␈α∂a␈α∂variety␈α∂of␈α∂human␈α∂experiences
00700 are␈αdiscussed␈α domain␈αincluding␈α the␈αinterpersonal␈αrelation␈α which
00800 develops␈α⊃between␈α⊃the␈α⊃interview␈α⊃participants.␈α⊃ A␈α⊃look␈α∩at␈α⊃the
00900 contents␈α∞of␈α∞a␈α∞thesaurus␈α∞reveals␈α∞that␈α∞words␈α∞relating␈α∞to␈α∞people
01000 and␈α their␈α interrelations␈α make␈α up␈α roughly␈α 70%␈α of␈α English␈α words.
01100
01200 The␈α∩diagnosis␈α∩of␈α∩paranoia␈α∩is␈α∩made␈α∪by␈α∩psychiatrists
01300 relying␈α_mainly␈α↔on␈α_the␈α_verbal␈α↔behavior␈α_of␈α_the␈α↔interviewed
01400 patient.␈α∞ If␈αa␈α∞paranoid␈α∞model␈αis␈α∞to␈αexhibit␈α∞paranoid␈α∞behavior␈αin
01500 a␈α≠psychiatric␈α≠interview,␈α≠ it␈α≠must␈α≠be␈α≠capable␈α≠of␈α~handling
01600 dialogues␈α⊗typical␈α↔of␈α⊗the␈α↔doctor-patient␈α⊗context.␈α↔ Since␈α⊗the
01700 model␈α∩can␈α∪communicate␈α∩only␈α∩through␈α∪teletyped␈α∩messages,the
01800 vis-a-vis␈α≠aspects␈α≠of␈α≠the␈α≠usual␈α≠psychiatric␈α≤interview␈α≠are
01900 absent.␈α_ Therefore␈α_the␈α_model␈α↔must␈α_be␈α_able␈α_to␈α_deal␈α↔with
02000 unedited␈α⊗typewritten␈α↔natural␈α⊗language␈α↔input␈α⊗and␈α↔to␈α⊗output
02100 replies␈α∂which␈α∂are␈α∂indicative␈α∞of␈α∂an␈α∂underlying␈α∂paranoid␈α∞thought
02200 process␈α during␈α the␈α episode␈α of␈α a␈α psychiatric␈α interview.
02300
02400 In␈α≠an␈α≠interview␈α≤there␈α≠is␈α≠always␈α≠a␈α≤who␈α≠saying
02500 something␈α to␈αa␈α whom␈α with␈αdefinite␈α intentions␈αand␈α expectations.
02600 There␈α∩are␈α∩two␈α∪situations␈α∩to␈α∩be␈α∪taken␈α∩into␈α∩account,␈α∪the␈α∩one
02700 being␈α~talked␈α~about␈α~and␈α→the␈α~one␈α~the␈α~participants␈α~are␈α→in.
02800 Sometimes␈α↔the␈α↔latter␈α⊗becomes␈α↔the␈α↔former.␈α↔ Participants␈α⊗in
02900 dialogues␈α∪have␈α∪intentions␈α∀and␈α∪dialogue␈α∪algorithms␈α∀must␈α∪take
03000 this␈α⊃into␈α⊃account.␈α⊂ The␈α⊃doctor's␈α⊃intention␈α⊂is␈α⊃to␈α⊃gather␈α⊂certain
03100 kinds␈α∀of␈α∪information␈α∀while␈α∪the␈α∀patient's␈α∪intention␈α∀is␈α∀to␈α∪give
03200 information␈α∂and␈α∂get␈α∂help.␈α∂ A␈α∂job␈α∞is␈α∂to␈α∂be␈α∂done;␈α∂it␈α∂is␈α∂not␈α∞small
03300 talk.␈α∩ Our␈α∩working␈α∩hypothesis␈α∩is␈α∩that␈α∩each␈α∩participant␈α∩in␈α∩the
03400 dialogue␈α%understands␈α%the␈α%other␈α%by␈α&matching␈α%selected
03500 idiosyncratically-␈α≤significant␈α≤patterns␈α≤in␈α≤the␈α≥input␈α≤against
03600 conceptual␈α≡patterns␈α≥which␈α≡contain␈α≥information␈α≡about␈α≥the
03700 situation␈α$or␈α$event␈α$being␈α$described␈α$linguistically.␈α# This
03800 understanding␈α"is␈α"communicated␈α"reciprocally␈α#by␈α"linguistic
03900 responses␈α(judged␈α)appropriate␈α(to␈α(the␈α)intentions␈α(and
04000 expectations␈α∩of␈α∪the␈α∩participants␈α∩and␈α∪to␈α∩the␈α∪requirements␈α∩of
04100 the␈α∞situation.␈α∂In␈α∞this␈α∂paper␈α∞we␈α∞shall␈α∂describe␈α∞only␈α∂our␈α∞current
04200 input-analyzing␈α_processes␈α_used␈α_to␈α_extract␈α_a␈α_pattern␈α_from
04300 natural␈α↔language␈α↔input.␈α_ In␈α↔a␈α↔later␈α↔communication␈α_we␈α↔shall
04400 describe␈α"the␈α"inferential␈α!processes␈α"carried␈α"out␈α"at␈α!the
04500 conceptual␈αlevel␈αonce␈αa␈α pattern␈αhas␈αbeen␈αreceived␈αby␈α memory
04600 from␈α the␈α input-analysing␈α processes.
04700
04800 Studies␈α↔of␈α↔our␈α↔1971␈α↔model␈α↔of␈α_paranoia␈α↔(PARRY)
04900 indicated␈α∞that␈α∞about␈α∞thirty␈α∞percent␈α∞of␈α∞the␈α∞sentences␈α∞were␈αnot
05000 understood␈α∩at␈α∪all␈α∩,␈α∩that␈α∪is,␈α∩no␈α∩concept␈α∪in␈α∩the␈α∪sentence␈α∩was
05100 recognized.␈α_ In␈α_a␈α_somewhat␈α↔larger␈α_number␈α_of␈α_cases␈α↔some
05200 concepts,␈α∂but␈α∂not␈α∂all,␈α⊂ were␈α∂recognized.␈α∂ In␈α∂many␈α⊂cases␈α∂these
05300 partially␈α∂recognized␈α⊂sentences␈α∂lead␈α⊂to␈α∂a␈α⊂partial␈α∂understanding
05400 that␈α∞was␈α∞sufficient␈α∂to␈α∞gather␈α∞the␈α∂intention␈α∞of␈α∞the␈α∂speaker␈α∞and
05500 thus␈α≠lead␈α≠to␈α≠output␈α≠an␈α≠appropriate␈α≠response.␈α~ However,
05600 misunderstandings␈α occurred␈α too␈α often.␈α For␈α example:
00100 ␈↓∞ DOCTOR: How old is your mother ?
00200
00300 PARRY: Twenty-eight
00400
00500 ␈↓PARRY␈α⊃has␈α⊃interpreted␈α⊃the␈α∩question␈α⊃as␈α⊃referring␈α⊃to␈α∩his␈α⊃own
00600 age␈αand␈α∞answered␈αby␈αgiving␈α∞his␈αage.␈α The␈α∞purpose␈αof␈α∞our␈αnew
00700 language␈α⊂analysis␈α⊂system␈α⊂is␈α⊃to␈α⊂significantly␈α⊂raise␈α⊂the␈α⊃level␈α⊂of
00800 understanding␈α∞by␈α∂preventing␈α∞such␈α∞misunderstandings␈α∂while␈α∞not
00900 restricting␈α∪what␈α∩can␈α∪be␈α∪said␈α∩to␈α∪PARRY.␈α∩ We␈α∪do␈α∪not␈α∩expect
01000 complete␈α∃under-␈α∀standing␈α∃from␈α∃this␈α∀system␈α∃--␈α∃even␈α∀native
01100 speakers␈α∪of␈α∪the␈α∪language␈α∩do␈α∪not␈α∪completely␈α∪understand␈α∩the
01200 language.
01300
01400 By␈α`understanding'␈αwe␈αmean␈α the␈αsystem␈αcan␈αdo␈α some
01500 or␈α all␈α of␈α the␈α following:
01600
01700 1)␈α+Determine␈α*the␈α+intention␈α+of␈α*the
01800 interviewer␈α1in␈α0making␈α1a␈α0particular
01900 utterance.
02000
02100 2)␈α_Make␈α_common␈α_logical␈α_deductions␈α_that
02200 follow␈α from␈α the␈α interviewers␈α utterance
02300
02400 3)␈α∞Form␈α∞an␈α∞idioletic␈α∂internal␈α∞representation
02500 of␈α∂the␈α∂utterance␈α∂so␈α∂that␈α∂questions␈α∂may␈α∂be
02600 answered,␈α∀commands␈α∀carried␈α∀out,␈α∃or␈α∀data
02700 added␈α to␈α memory.
02800
02900 4)␈α∂Determine␈α∂references␈α∂for␈α∂pronouns,␈α∂and
03000 other␈α anaphora.
03100
03200 5)␈α∃Deduce␈α∀the␈α∃tone␈α∀of␈α∃the␈α∀utterance,i.e.,
03300 hostile,␈α insulting...
03400
03500 6)␈α!Classify␈α"the␈α!input␈α!as␈α"a␈α!question,
03600 rejoinder,command,␈α ...
03700
03800 ␈↓ The␈α∞approach␈α∂we␈α∞are␈α∞taking␈α∂consists␈α∞of␈α∂merging␈α∞the
03900 best␈α⊃features␈α∩of␈α⊃pattern␈α∩directed␈α⊃systems␈α∩such␈α⊃as␈α∩the␈α⊃MAD
04000 DOCTOR,␈↓↓2␈↓␈α⊃ELIZA␈↓↓11␈↓␈α⊃and␈α⊂parsing␈α⊃directed␈α⊃systems␈α⊃for␈α⊂example,
04100 Winograd,␈↓↓13␈↓␈α⊃Woods.␈↓↓14␈↓.␈α⊃ By␈α⊂merging␈α⊃the␈α⊃BNF␈α⊃phrase␈α⊂structure
04200 approach␈α≤ot␈α≤analyzing␈α≤English␈α≤with␈α≤the␈α≤pattern␈α≠matching
04300 approach,␈αwith␈αits␈α attendant␈αemphasis␈αof␈α some␈αconcepts␈αto␈α the
04400 exclusion␈α↔of␈α⊗others.␈α↔ The␈α⊗programs␈α↔to␈α⊗accomplish␈α↔this␈α⊗are
04500 written␈α∂in␈α∂MLISP2,␈α⊂ an␈α∂extensible␈α∂version␈α∂of␈α⊂the␈α∂programming
04600 language␈α∃MLISP,␈↓↓6,10␈↓␈α∀and␈α∃uses␈α∃an␈α∀interpreted␈α∃version␈α∃of␈α∀the
04700 pattern␈α⊗matcher␈α∃designed␈α⊗for␈α∃a␈α⊗new␈α⊗programming␈α∃language
04800 LISP70.
04900
05000 The␈α∩following␈α∩is␈α∩a␈α⊃basic␈α∩description␈α∩of␈α∩the␈α⊃pattern
05100 matcher.␈α⊃ We␈α⊃shall␈α⊃illustrate␈α⊃its␈α⊃operation␈α⊃using␈α∩examples␈α⊃of
05200 problems␈α common␈α to␈α teletyped␈α psychiatric␈α dialogues.
05300
05400 ␈↓ ↓j␈↓
05500 PATTERN MATCHING␈↓
05600
05700 ␈↓ Pattern␈α∪directed␈α∀computation␈α∪involves␈α∪two␈α∀kind␈α∪of
05800 operations␈α5on␈α4data␈α5structures:␈α5decomposition␈α4and
05900 recomposition.␈α∀ Decomposition␈α∀breaks␈α∀down␈α∀an␈α∃input␈α∀stream
06000 into␈α components␈αunder␈α the␈α direction␈αof␈α a␈αdecompostion␈α pattern
00100 ("dec").␈α∪ The␈α∩inverse␈α∪operation,␈α∩recomposition,␈α∪constructs␈α∩an
00200 ␈↓output␈α⊂stream␈α∂under␈α⊂the␈α⊂direction␈α∂of␈α⊂a␈α⊂recomposition␈α∂pattern
00300 ("rec").
00400
00500 A␈α rewrite␈α rule␈α is␈α of␈α the␈α form:
00600
00700 ␈↓∞ dec → rec
00800
00900
01000 ␈↓It␈α∞defines␈αa␈α∞partial␈α∞function␈αon␈α∞streams␈α∞as␈αfollows:␈α∞if␈α∞the␈αinput
01100 stream␈α∂matches␈α∂the␈α∞dec,␈α∂then␈α∂the␈α∞output␈α∂stream␈α∂is␈α∞generated
01200 by␈α∃the␈α∀rec.␈α∃The␈α∀following␈α∃rule␈α∀(given␈α∃as␈α∀an␈α∃example␈α∀only)
01300 could␈α be␈α part␈α of␈α a␈α question␈α answering␈α function:
01400
01500 ␈↓∞ How are you ? → Very well and you ?
01600
01700 ␈↓If␈α the␈α input␈α stream␈α consists␈α of␈α the␈α four␈α tokens:
01800
01900 ␈↓∞ How are you ?
02000
02100 ␈↓the␈α output␈α stream␈α will␈α consist␈α of␈α the␈α five␈α tokens:
02200
02300 ␈↓∞ Very well and you ?
02400
02500 ␈↓␈↓
02600 REWRITE FUNCTIONS ␈↓
02700
02800 ␈↓ A␈α∨rewrite␈α≡rule␈α∨defines␈α≡a␈α∨partial␈α∨function,␈α≡for
02900 example,␈α↔the␈α↔mapping␈α↔of␈α↔some␈α↔particular␈α↔token␈α↔into␈α⊗some
03000 other␈α↔particular␈α↔token.␈α↔ A␈α↔broader␈α↔partial␈α↔function␈α↔can␈α⊗be
03100 defined␈α⊗as␈α⊗the␈α⊗union␈α⊗of␈α⊗several␈α⊗rewrite␈α⊗rules.␈α⊗ A␈α∃rewrite
03200 function␈α definition␈α is␈α of␈α the␈α form:
03300
03400 ␈↓∞ RULES OF <name> =
03500 dec1 → rec1,
03600 dec2 → rec2,
03700 ...
03800 decn → recn;
03900
04000 ␈↓␈↓
04100 VARIABLES ␈↓
04200
04300 ␈↓ A␈α⊃function␈α⊃is␈α⊃difficult␈α⊃to␈α⊃define␈α⊃if␈α⊃every␈α∩case␈α⊃must
04400 be␈α∩enumerated.␈α∩ Therefore,␈α∪rewrite␈α∩rules␈α∩allow␈α∪variables␈α∩to
04500 appear␈α⊂in␈α⊂patterns.␈α⊂ The␈α⊂value␈α⊂of␈α⊂a␈α⊂variable␈α⊂can␈α⊂be␈α⊃either␈α⊂a
04600 list␈α or␈α an␈α atom.␈α In␈α this␈α paper␈α the␈α notation:
04700
04800 ␈↓∞ :X
04900
05000 ␈↓where␈α∀X␈α∪ia␈α∀any␈α∪identifier,␈α∀will␈α∪denote␈α∀the␈α∪variable␈α∀X.␈α∪ The
05100 variables␈α∩of␈α∩each␈α∪rule␈α∩are␈α∩distinct␈α∪from␈α∩the␈α∩variables␈α∪of␈α∩all
05200 other␈α rules,␈α even␈α if␈α their␈α names␈α are␈α the␈α same.
05300
05400 The␈αfollowing␈αdefinition␈αhas␈αonly␈αthree␈αrewrite␈αrules,
05500 but␈α handles␈α an␈α unlimited␈α number␈α of␈α input␈α streams:
00100 ␈↓∞ RULES OF REPLY=
00200 HOW ARE YOU '? → VERY WELL '?
00300 HOW IS :X → I HAVEN''T SEEN :X ',
00400 LATELY'.,
00500 DID :X GO TO :Y '? →
00600 WHY DON''T YOU ASK :X YOURSELF'?;
00700
00800 ␈↓A␈α∞variable␈αcan␈α∞appear␈α∞more␈αthan␈α∞once␈α∞in␈αa␈α∞single␈α∞dec␈αpattern,
00900 but␈α≡it␈α≡must␈α≡match␈α≥identical␈α≡items␈α≡at␈α≡each␈α≥appearance.
01000 Example:
01100
01200 ␈↓∞ RULES OF EQUAL =
01300 (EQUAL :X :X) → TRUE;
01400
01500 ␈↓␈↓
01600 ELLIPSIS ␈↓
01700
01800 ␈↓ To␈α↔make␈α⊗patterns␈α↔easier␈α⊗to␈α↔read␈α⊗and␈α↔write,␈α⊗the
01900 ellipsis␈α_symbol␈α→...␈α_can␈α→be␈α_used␈α_to␈α→stand␈α_for␈α→an␈α_unnamed
02000 variable.␈α Thus:
02100
02200
02300 ␈↓∞ IS ... COMING → NO, ... COULD NOT MAKE IT.
02400
02500 ␈↓If␈α an␈α ellipsis␈α (...)␈α occurs␈αseveral␈α times␈α on␈α a␈α side,␈α it␈αdesignates␈α a
02600 different␈α~variable␈α~each␈α~time.␈α~ The␈α~n'th␈α~ellipsis␈α~in␈α≠a␈α~dec
02700 designates␈α the␈α same␈α variable␈α as␈α the␈α n'th␈α ellipsis␈α in␈α the␈α rec.
02800
02900 Ellipsis␈α_is␈α_one␈α_of␈α_the␈α_principle␈α_ideas␈α→of␈α_pattern
03000 matching.␈α It␈αpermits␈αimprecise␈αmathching;␈αthat␈αis,␈αthe␈α emphasis
03100 or␈α ignoring␈α of␈α items.
03200
03300 ␈↓
03400 AUTOMATIC ORDERING OF RULES ␈↓
03500
03600 ␈↓ The␈α⊂order␈α⊂of␈α⊂rules␈α⊂in␈α⊂a␈α⊂function␈α⊂definition␈α⊂does␈α∂not
03700 specify␈α⊃the␈α⊃order␈α⊃in␈α⊃which␈α⊃the␈α⊃system␈α⊃will␈α⊃attempt␈α⊃to␈α⊃apply
03800 them.␈α∂ This␈α∂ordering␈α∂operation␈α∞is␈α∂handled␈α∂by␈α∂a␈α∂special␈α∞system
03900 ordering␈α function.␈α Consider␈α the␈α rewrite␈α function:
04000
04100 ␈↓∞ RULES OF REPLY =
04200 I SEE :X → SO WHAT '?,
04300 I SEE ANN → WOW '!;
04400
04500 ␈↓Both␈α rules␈α would␈α match:
04600
04700 ␈↓∞ I SEE ANN
04800
04900 ␈↓In␈α∞such␈α∂cases␈α∞the␈α∂more␈α∞specific␈α∂rule␈α∞takes␈α∂precedence.␈α∞ Thus,
05000 given:
05100
05200 ␈↓∞ I SEE ANN
05300
05400 ␈↓as␈α the␈α input␈α stream␈α ,␈α the␈α output␈α stream␈α would␈α be:
05500
05600 ␈↓∞ WOW !
05700
05800 ␈↓but␈α given:
05900
06000 ␈↓∞ I SEE STARS
06100
06200 ␈↓the␈α output␈α stream␈α would␈α be:
00100 ␈↓∞ SO WHAT ?
00200
00300 ␈↓A␈αliteral␈αis␈α∞more␈αspecific␈αthan␈α∞a␈αvariable.␈αA␈α∞variable␈αappearing
00400 for␈αthe␈α∞second␈αtime␈α∞is␈αmore␈αspecific␈α∞than␈αa␈α∞variable␈αappearing
00500 for␈α∪the␈α∩first␈α∪time␈α∩in␈α∪a␈α∩dec.␈α∪ This␈α∩is␈α∪so␈α∩because␈α∪the␈α∩second
00600 occurence␈αof␈αthe␈αvariable␈αmust␈αmatch␈αthe␈αsame␈αpattern␈αas␈αthe
00700 first␈α∪occurence.␈α∩The␈α∪precedence␈α∩function␈α∪is␈α∩itself␈α∪written␈α∩in
00800 rewrites␈αand␈α so␈αis␈α both␈αextendable␈α and␈αchangable␈α by␈αthe␈α user.
00900 Currently␈α precedence␈α∨is␈α calculated␈α by␈α∨a␈α left␈α to␈α∨right
01000 application␈α⊗of␈α⊗the␈α⊗above␈α⊗criteria.␈α⊗ Therefore,␈α↔the␈α⊗following
01100 function␈α defines␈α the␈α LISP␈α function␈α EQUAL:
01200
01300 ␈↓∞ RULES OF EQUAL =
01400 (EQUAL :X :X) → T,
01500 (EQUAL :X :Y) → NIL;
01600
01700 ␈↓␈↓
01800 SEGMENTS ␈↓
01900
02000 ␈↓ Sometimes␈α⊗ it␈α↔ is␈α⊗ desirable␈α↔ for␈α⊗ a␈α↔ variable␈α⊗ to
02100 match␈α an␈α indeterminate␈α number␈α of␈α items.␈α This␈α is␈α notated:
02200
02300 ␈↓∞ ::X
02400
02500 ␈↓Use␈α∞of␈αthe␈α∞double-colon␈α∞("::")␈αmeans␈α∞that␈αthe␈α∞variable␈α∞(e.g.,␈αX)
02600 will␈α match␈α zero␈α or␈α more␈α items.␈α Example:
02700
02800 ␈↓∞ RULES OF APPEND=
02900 (APPEND (::X)(::Y)) → (::X ::Y);
03000
03100 ␈↓or␈α if␈α the␈α input␈α stream␈α were:
03200
03300 ␈↓∞ (APPEND (A B) (C D E))
03400
03500 ␈↓the␈α output␈α stream␈α would␈α be:
03600
03700 ␈↓∞ (A B C D E)
03800
03900 ␈↓For␈α increased␈α readability␈α the␈α rule␈α could␈α also␈α be␈α written:
04000
04100 ␈↓∞ RULES OF APPEND =
04200 (APPEND (...) (...)) → (... ...);
04300
04400 ␈↓Another␈α example:
04500
04600 ␈↓∞ RULES OF REPLY =
04700 WHERE DID ::X GO →
04800 ::X WENT HOME '.;
04900
05000 ␈↓Therefore,
05100
05200 ␈↓∞ WHERE DID THE CARPENTER GO →
05300 THE CARPENTER WENT HOME.
05400
05500 ␈↓␈↓
05600 APPLICATION ␈↓
05700
05800 ␈↓ One␈α∂of␈α∂the␈α∞main␈α∂deficiencies␈α∂of␈α∞the␈α∂system␈α∂in␈α∞which
05900 the␈α⊂MAD␈α⊂DOCTOR␈α⊂was␈α⊃programmed␈α⊂was␈α⊂its␈α⊂lack␈α⊃of␈α⊂adequat`α@
06000 subroutining␈α∩capability.␈α∩ Subroutines␈α∩may␈α∩be␈α∩indicated␈α∩in␈α∩the
06100 rewrite␈α system␈α as␈α follows:
00100 ␈↓∞ RULES OF LAST =
00200 ␈↓∞ () → (),
00300 (:X) → :X,
00400 (:X ...) → <LAST (...)>;
00500
00600 ␈↓The␈α⊂"<>"␈α⊃surrounding␈α⊂a␈α⊃pattern␈α⊂means␈α⊃that␈α⊂the␈α⊃current␈α⊂input
00700 stream␈α⊃is␈α⊃to␈α⊃be␈α⊃pushed␈α⊂down,␈α⊃that␈α⊃the␈α⊃function␈α⊃indicated␈α⊂by
00800 the␈α∂firtoken␈α∂within␈α∂the␈α⊂brackets␈α∂is␈α∂to␈α∂be␈α∂entered␈α⊂with␈α∂the
00900 rest␈αof␈αthe␈αpattern␈αappended␈αto␈αthe␈αfront␈αof␈αthe␈α∞input␈αstream,
01000 and␈α∂that␈α⊂the␈α∂output␈α∂stream␈α⊂is␈α∂to␈α∂be␈α⊂placed␈α∂into␈α⊂the␈α∂restored
01100 current␈α↔input␈α_stream.␈α↔Note␈α↔that␈α_MLISP2␈α↔functions␈α_may␈α↔be
01200 called␈α as␈α well␈α as␈α rewrite␈α functions.
01300
01400 ␈↓
01500 GOALS ␈↓
01600
01700 ␈↓ To␈α→gain␈α→the␈α~advantage␈α→of␈α→goal␈α~directed␈α→pattern
01800 matching␈α∂and␈α⊂computing,␈α∂as␈α⊂well␈α∂as␈α∂the␈α⊂full␈α∂power␈α⊂of␈α∂context
01900 sensitive␈α grammars,␈α the␈α following␈α form␈α may␈α be␈α used:
02000
02100 ␈↓∞ RULES OF PREPOSITIONAL_PHRASE =
02200 <PREPOSITION>:P <NOUN_PHRASE>:N
02300 → (PREP_PH :P :N);
02400
02500 ␈↓The␈α∃identifer␈α∃between␈α∃the␈α∃angled␈α∃brackets␈α∃("<>")␈α∃names␈α∀a
02600 rewrite␈α∞function␈α∞the␈α∞rules␈α∞of␈α∞which␈α∞are␈α∞to␈α∞be␈α∞matched␈α∞against
02700 the␈α∞input␈α∂stream.␈α∞ When␈α∂a␈α∞match␈α∞occurs␈α∂the␈α∞output␈α∂stream␈α∞of
02800 the␈α goal␈α will␈α be␈α bound␈α to␈α the␈α associated␈α variable.␈α Example:
02900
03000 ␈↓∞ RULES OF PREPOSITIONAL_PHRASE =
03100 <PREPOSITION>:P <NOUN_PHRASE>:N
03200 → (PREP_PH :P :N);
03300
03400 RULES OF NOUN_PHRASE =
03500 TOWN → (NOUN_PH TOWN),
03600 PALO ALTO → (NOUN_PH PALO_ALTO);
03700
03800 RULES OF PREPOSITON =
03900 IN → IN,
04000 ON → ON;
04100
04200 ␈↓and␈α the␈α input␈α stream:
04300
04400 ␈↓∞ IN PALO ALTO
04500
04600 ␈↓the␈α output␈α stream␈α would␈α be:
04700
04800 ␈↓∞ (PREP_PH IN (NOUN_PH PALO_ALTO))
04900
05000 ␈↓␈↓
05100 OPTIONALS ␈↓
05200
05300 ␈↓ Many␈α≠other␈α~shorthands␈α≠exist␈α~to␈α≠simplify␈α~writing
05400 rules.␈α⊃One␈α⊃useful␈α⊃feature␈α⊃that␈α⊃will␈α⊃be␈α⊃mentioned␈α⊃here␈α⊃is␈α⊂the
05500 optional.
05600
05700 ␈↓∞ RULES OF AUXILARY_PHRASE =
05800 <AUXILARY>:A [<NEGATIVE>:N]:N1 →
05900 (AUX_PH :A [:N]:N1 );
06000
06100 ␈↓If␈α⊗the␈α⊗optional␈α⊗pattern,␈α⊗enclosed␈α⊗in␈α⊗square␈α↔brackets␈α⊗("[]"),
00100 occurs␈α∞in␈α∞the␈α∞input␈α∞stream␈α∞it␈α∂will␈α∞be␈α∞bound␈α∞to␈α∞:N.␈α∞ :N1␈α∂will␈α∞be
00200 ␈↓bound␈α⊃to␈α⊃2.␈α∩ If␈α⊃the␈α⊃<NEGATIVE>␈α⊃does␈α∩not␈α⊃occur,␈α⊃:N1␈α∩will␈α⊃be
00300 bound␈α to␈α 1.␈α On␈α the␈α rec␈α side␈αof␈α the␈α rules␈α if␈α :N1␈α is␈α 2␈α then␈α:N␈α will
00400 be␈α∀placed␈α∀in␈α∀the␈α∀output␈α∀stream.␈α∀ If␈α∀it␈α∀is␈α∀1␈α∀then␈α∀nothing␈α∪is
00500 placed␈α∞in␈α∞the␈α∞output␈α∞stream␈αat␈α∞that␈α∞point.␈α∞ Example,␈α∞given␈αthe
00600 rule␈α above:
00700
00800 ␈↓∞ DO → (AUX_PH DO)
00900 DO NOT → (AUX_PH DO NOT)
01000
01100 ␈↓␈↓
01200 MORE EXAMPLES ␈↓
01300
01400 ␈↓ We␈α have␈αcollected␈α a␈α large␈αnumber␈α of␈αdialogues␈α using
01500 our␈α∩previous␈α⊃program␈α∩PARRY.␈α⊃ These␈α∩dialogues␈α⊃form␈α∩a␈α⊃large
01600 body␈α∞of␈αexamples␈α∞of␈αthe␈α∞kind␈αof␈α∞English␈αwhich␈α∞we␈α∞can␈αexpect.
01700 Martin␈α∨Frost,␈α∨a␈α≡graduate␈α∨student␈α∨in␈α∨Computer␈α≡Science,
01800 Stanford␈α≥University,␈α≡ has␈α≥written␈α≡a␈α≥keyword␈α≡in␈α≥context
01900 program␈α∀which␈α∀enables␈α∀us␈α∪to␈α∀isolate␈α∀examples␈α∀centered␈α∪on
02000 particular␈α⊗words␈α⊗so␈α⊗that␈α∃uses␈α⊗of␈α⊗thoses␈α⊗words␈α⊗in␈α∃context
02100 become␈α∩more␈α∪apparent.␈α∩ Our␈α∪general␈α∩approach␈α∪is␈α∩to␈α∪build␈α∩a
02200 system␈α∀which␈α∀can␈α∀produce␈α∀desired␈α∀intreptations␈α∀from␈α∀these
02300 examples␈α∂and␈α∞to␈α∂incrementally␈α∂add␈α∞to␈α∂the␈α∞rules␈α∂in␈α∂the␈α∞system
02400 as␈α~new␈α~cases␈α→are␈α~discovered␈α~during␈α→the␈α~running␈α~of␈α→the
02500 program.
02600
02700 ␈↓ Following␈α!are␈α!some␈α examples␈α!of␈α!commonly␈α occuring
02800 situations␈α∞and␈αexamples␈α∞of␈α∞the␈αkind␈α∞of␈α∞rules␈αwe␈α∞use␈α∞to␈αhandle
02900 them.
03000
03100 ␈↓
03200 QUESTION INTRODUCER ␈↓
03300
03400 ␈↓ In␈α∃doctor-patient␈α⊗dialogues␈α∃it␈α⊗is␈α∃quite␈α⊗common␈α∃to
03500 introduce␈αa␈α∞question␈αby␈α∞the␈αuse␈αof␈α∞a␈αcommand.␈α∞ The␈α"question
03600 introducer"␈α↔is␈α⊗followed␈α↔by␈α⊗either␈α↔a␈α⊗<NOUN_PHRASE>␈α↔or␈α⊗a
03700 <DECLARATIVE_SENTENCE>.␈α For␈α example,
03800
03900 ␈↓∞ COULD YOU TELL ME YOUR NAME?
04000
04100 ␈↓Rather␈α⊃than␈α⊃attempt␈α⊂a␈α⊃literal␈α⊃analysis␈α⊂of␈α⊃this␈α⊃question,␈α⊂which
04200 might␈α lead␈α to␈α the␈α interpretation:
04300
04400 ␈↓∞ DO YOU HAVE THE ABILITY TO SPEAK YOUR NAME TO ME?
04500
04600 ␈↓we␈α utilize␈α rules␈α like:
04700
04800 ␈↓∞ RULES OF SENTENCE =
04900 <QUESTION_INTRODUCER>:Q <NOUN_PHRASE>:N
05000 → (IS :N '*'?'* );
05100
05200 RULES OF QUESTION_INTRODUCER =
05300 COULD YOU TELL ME → ,
05400 WOULD YOU TELL ME → ,
05500 PLEASE TELL ME → ;
05600
05700 ␈↓ Although␈α⊂it␈α⊂is␈α⊂conceivable␈α∂that␈α⊂there␈α⊂are␈α⊂an␈α∂infinite
05800 number␈α∩of␈α∩ways␈α∩to␈α∩introduce␈α∩a␈α∩question␈α∩in␈α∩this␈α∩manner,␈α∩we
05900 have␈α⊂found␈α⊂only␈α⊂about␈α⊃six␈α⊂literal␈α⊂strings␈α⊂are␈α⊂actually␈α⊃used␈α⊂in
06000 our␈α⊂data␈α⊂base␈α⊂of␈α⊂dialogues.␈α∂ When␈α⊂we␈α⊂discover␈α⊂a␈α⊂new␈α∂string
06100 we␈αincrementally␈αadd␈α a␈αrule.␈α When␈α we␈αhave␈αenough␈α examples
00100 to␈α∃dectect␈α∃a␈α⊗more␈α∃general␈α∃form␈α⊗we␈α∃replace␈α∃the␈α⊗rules␈α∃for
00200 ␈↓<QUESTION_INTRODUCER>␈α~by␈α~a␈α~more␈α~elegant␈α~and␈α→general
00300 formulation.␈α⊗ This␈α⊗approach␈α⊗allows␈α⊗us␈α⊗to␈α↔process␈α⊗dialogues
00400 before␈α⊂we␈α⊂have␈α⊂a␈α⊂complete␈α⊂analysis␈α⊂of␈α⊂all␈α⊃possible␈α⊂sentence
00500 constructions,␈α∪and␈α∪it␈α∪allows␈α∪us␈α∪to␈α∪build␈α∪a␈α∀language␈α∪analyzer
00600 based␈α on␈α actually␈α occurring␈α forms.
00700
00800 Notice␈α∀that␈α∀it␈α∀is␈α∪possible␈α∀to␈α∀make␈α∀more␈α∀than␈α∪one
00900 analysis␈α∪of␈α∀any␈α∪given␈α∀sentence␈α∪depending␈α∀on␈α∪what␈α∀is␈α∪being
01000 looked␈α⊗for.␈α↔A␈α⊗poet␈α↔might␈α⊗be␈α⊗interested␈α↔in␈α⊗the␈α↔number␈α⊗of
01100 syllables␈α→per␈α→word␈α→and␈α~the␈α→patterns␈α→of␈α→stress.␈α~ A␈α→"full"
01200 analysis␈α∪of␈α∪English␈α∀must␈α∪allow␈α∪for␈α∀this␈α∪possibility,␈α∪ but␈α∀it␈α∪it
01300 clearly␈αfoolish␈αto␈α produce␈αthis␈αkind␈α of␈αanalysis␈αfor␈αPARRY.␈α Our
01400 analysis␈α⊃will␈α⊂be␈α⊃partial␈α⊂and␈α⊃idiosyncratic␈α⊂to␈α⊃the␈α⊂needs␈α⊃of␈α⊂our
01500 program.␈α This␈α is␈α what␈α is␈α meant␈α by␈α idiolectic.
01600
01700 ␈↓
01800 FILLERS ␈↓
01900
02000 ␈↓ It␈α⊗is␈α↔quite␈α⊗common␈α↔for␈α⊗interviewers␈α↔to␈α⊗introduce
02100 words␈α∩of␈α∩little␈α∩significance␈α∩to␈α∩PARRY␈α∩into␈α∩the␈α∩sentence.␈α⊃For
02200 example:
02300
02400 ␈↓∞ WELL, WHAT IS YOUR NAME?
02500
02600 ␈↓The␈α∀"well"␈α∀in␈α∀this␈α∀sentence␈α∀serves␈α∀no␈α∀purpose␈α∃in␈α∀PARRY's
02700 analysis,␈α∞although␈α∞it␈α∞might␈α∞to␈α∞a␈α∞linguist␈α∞interested␈α∂in␈α∞hesitation
02800 phenomena.␈α∞ These␈α∞fillers␈α∞can␈α∞be␈α∞ignored.␈α∞ The␈α∞following␈αrules
02900 accomplish␈α this:
03000
03100 ␈↓∞ RULES OF SENTENCE =
03200 <FILLERS>:F <SENTENCE>:S → :S;
03300
03400 RULES OF FILLERS =
03500 WELL → ,
03600 OK → ;
03700
03800 ␈↓␈↓
03900 PUNCTUATION ␈↓
04000
04100 ␈↓ Interviewers␈α∂use␈α∂little␈α∂intra-sentence␈α∂punctuation␈α∞in
04200 talking␈α∪to␈α∪PARRY.␈α∪ When␈α∪it␈α∪is␈α∪used␈α∪it␈α∪is␈α∪often␈α∀to␈α∪seperate
04300 phrases␈α that␈α might␈α otherwise␈α be␈α ambiguous.␈α Example:
04400
04500 ␈↓∞ WHY WEREN'T YOU VERY CLOSE, FRANK
04600
04700 ␈↓Here␈α∪the␈α∪comma␈α∪clearly␈α∩puts␈α∪"CLOSE"␈α∪in␈α∪a␈α∪different␈α∩phrase
04800 from␈α⊃"FRANK".␈α⊃Punctuation,␈α⊃ when␈α⊃used␈α⊃in␈α⊃PARRY's␈α⊃rules,␈α⊂ is
04900 generally␈α∀enclosed␈α∪in␈α∀optional␈α∀brackets␈α∪("[]").␈α∀ This␈α∀has␈α∪the
05000 effect␈α of␈α seperating␈α phrases␈αwhen␈α punctuation␈α is␈α used,␈αbut␈α not
05100 requiring␈α full␈α punctuation␈α for␈α the␈α system␈α to␈α work.␈α Example:
05200
05300 ␈↓∞ RULES OF SENTENCE =
05400 <SENTENCE>:S1 [',]:C <SENTENCE_CONNECTOR>:SC
05500 <SENTENCE>:S2
05600 → (CONUNCTION :SC :S1 :S2);
05700
05800
05900 ␈↓␈↓
06000 CLICHES AND IDIOMS ␈↓
06100
06200 ␈↓ The␈α(English␈α)we␈α(encounter␈α)in␈α(doctor-patient
00100 dialogues␈α∞is␈α∞made␈α∂up␈α∞of␈α∞a␈α∞great␈α∂number␈α∞of␈α∞cliches␈α∂and␈α∞idioms,
00200 ␈↓therefore␈α⊃we␈α∩anticipate␈α⊃a␈α∩large␈α⊃number␈α∩of␈α⊃rules␈α∩devoted␈α⊃to
00300 them.␈α For␈α example:
00400
00500 ␈↓∞ RULES OF TIME_PHRASES =
00600 A COUPLE OF <TIME_UNIT>:T AGO
00700 → (TIME (RELATIVE PAST)(REF PRESENT) :T);
00800
00900 RULES OF TIME_UNIT =
01000 SECONDS → (WITHIN CONVERSATION),
01100 MOMENTS → (WITHIN CONVERSATION),
01200 DAYS → (BEFORE CONVERSATION DAYS);
01300
01400 ␈↓␈↓
01500 REPRESENTATION CORRECTION ␈↓
01600
01700 ␈↓ Intermediate␈α∪results␈α∪are␈α∩often␈α∪produced␈α∪which␈α∩are
01800 misleading␈α∪in␈α∪meaning␈α∪or␈α∪are␈α∪in␈α∪the␈α∪wrong␈α∪form␈α∀for␈α∪further
01900 processing.␈α∂We,␈α∂therefore,␈α∞incorporate␈α∂at␈α∂various␈α∂points␈α∞rules
02000 which␈α≡detect␈α≥certain␈α≡undesired␈α≥intermediate␈α≡results␈α≥and
02100 convert␈α them␈α to␈α the␈α desired␈α form.␈α Example:
02200
02300 ␈↓∞ RULES OF CORRECT_FORM =
02400 (QUESTION ... (SENTENCE ...)) →
02500 (QUESTION ... ...);
02600
02700 ␈↓␈↓
02800 UNKNOWN WORDS ␈↓
02900
03000 ␈↓ Rules␈α⊃can␈α⊃be␈α∩derived␈α⊃to␈α⊃handle␈α⊃words␈α∩which␈α⊃were
03100 previously␈α unknown␈α to␈α the␈α system.␈α For␈α example:
03200
03300 ␈↓∞ RULES OF UNKNOWN_WORD =
03400 DR'. :X → <NEW_WORD NAME :X>,
03500 THE :X <VERB_PHRASE>:V →
03600 <NEW_WORD NOUN :X>,
03700 I :X YOU → <NEW_WORD VERB :X>;
03800
03900 ␈↓Here␈α "NEW_WORD"␈α is␈αa␈α function␈α which␈αadds␈α new␈α words␈αto␈α the
04000 dictionary.
04100
04200 ␈↓
04300 CONCLUSION ␈↓
04400
04500 ␈↓ We␈α∞are␈αfaced␈α∞with␈α∞the␈αproblems␈α∞of␈α∞natural␈αlanguage
04600 being␈α∩used␈α∩to␈α∩interview␈α∩people␈α∩in␈α∩a␈α∪doctor-patient␈α∩context.
04700 We␈α⊃have␈α⊂developed␈α⊃a␈α⊃language␈α⊂processing␈α⊃system␈α⊃which␈α⊂we
04800 believe␈α⊗is␈α⊗capable␈α∃of␈α⊗performing␈α⊗in␈α∃these␈α⊗interviews␈α⊗at␈α∃a
04900 significantly␈α≥improved␈α≥level␈α≤of␈α≥performance␈α≥compared␈α≤to
05000 systems␈α∀used␈α∀in␈α∀the␈α∀past.␈α∀ We␈α∀have␈α∃developed␈α∀techniques
05100 which␈α⊂can␈α⊃measure␈α⊂performance␈α⊂in␈α⊃comparison␈α⊂with␈α⊃the␈α⊂ideal
05200 of␈α⊗a␈α⊗real␈α↔human␈α⊗patient␈α⊗in␈α↔the␈α⊗same␈α⊗context.␈↓↓4,5,7␈↓␈α↔We␈α⊗are
05300 designing␈αour␈αsystem␈α∞with␈αthe␈αrealization␈α∞that␈αa␈αlong␈α∞period␈αof
05400 development␈α"is␈α"necessary␈α"to␈α"reach␈α"desired␈α"levels␈α!of
05500 performance.␈α⊂ This␈α⊂is␈α⊃a␈α⊂system␈α⊂that␈α⊃can␈α⊂work␈α⊂at␈α⊃a␈α⊂measured
05600 level␈α∪of␈α∪performance␈α∪and␈α∪be␈α∪improved␈α∪over␈α∪time␈α∀with␈α∪new
05700 rules␈α⊃having␈α⊃minimum␈α⊃interaction␈α⊃with␈α⊃those␈α∩already␈α⊃existing.
05800 Our␈α⊂system␈α⊂is␈α⊂designed␈α⊂so␈α∂that␈α⊂a␈α⊂complete␈α⊂analysis␈α⊂of␈α∂every
05900 word␈α or␈α phrase␈α of␈α an␈α utterance␈α is␈α not␈α neceesary.
06000
06100 ␈↓ The␈α∀basis␈α∀of␈α∀this␈α∀system␈α∀is␈α∀a␈α∀rewrite␈α∪interpreter
06200 which␈α↔will␈α↔automatically␈α↔merge␈α↔new␈α↔rules␈α↔into␈α↔the␈α_set␈α↔of
00100 already␈α⊗existing␈α∃rules␈α⊗so␈α∃that␈α⊗the␈α∃system␈α⊗will␈α⊗continue␈α∃to
00200 ␈↓handle␈α sentences␈α which␈α it␈α handled␈α in␈α the␈α past.
00300
00400 ␈↓ α>␈↓
00500 REFERENCES␈↓
00600
00700 ␈↓␈↓↓1␈↓Carroll,␈α⊃J.B.␈α⊃Language␈α⊃and␈α⊃Thought.␈α⊃Prentice-Hall,␈α⊂Englewood
00800 Cliffs,␈α New␈α Jersey,␈α p.␈α 59.
00900
01000 ␈↓↓2␈↓Colby,␈α K.M.,␈α Watt,␈α J.␈αand␈α Gilbert,␈αJ.P.␈α A␈α computer␈α method
01100 of␈α⊃psychotherapy.␈α∩Journal␈α⊃of␈α∩Nervous␈α⊃and␈α∩Mental␈α⊃Disease,
01200 142,148-152,1966.
01300
01400 ␈↓↓3␈↓Colby,K.M.␈α∨and␈α∨Enea,H.␈α∨Heuristic␈α∨methods␈α for␈α∨computer
01500 understanding␈α∩ of␈α∩natural␈α∩language␈α∩in␈α∪context␈α∩restricted
01600 on-line␈α dialogues.␈α Mathematical␈α Biosciences,1,1-25,1967.
01700
01800 ␈↓↓4␈↓Colby,␈α⊂K.M.,␈α⊃Hilf,␈α⊂F.D.,␈α⊃Weber,␈α⊂S.,␈α⊃and␈α⊂Kraener,␈α⊃H.␈α⊂Turing-like
01900 indistinguishability␈α∃tests␈α∃for␈α∀the␈α∃validation␈α∃of␈α∃a␈α∀computer
02000 simulation␈α?␈α⊂of␈α?␈α⊂paranoid␈α?␈α⊂processes.␈α?␈α⊂Artificial
02100 Intelligence,3,199-221,1972.
02200
02300 ␈↓↓5␈↓Colby,␈α!K.M.␈α!and␈α"Hilf,␈α!F.D.␈α!Multidimensional␈α"analysis␈α!in
02400 evaluating␈α≥the␈α≤adequacy␈α≥of␈α≤a␈α≥simulation␈α≥of␈α≤paparnoid
02500 processes.␈α∪ Memo␈α∀AIM-194.␈α∪Stanford␈α∀Artificial␈α∪Intelligence
02600 project,␈α Stanford␈α University.
02700
02800 ␈↓↓6␈↓Enea,␈α⊂H.␈α∂MLISP,␈α⊂Technical␈α∂report␈α⊂no.␈α∂CS-92,␈α⊂1968,␈α∂Computer
02900 Science␈α Department,␈α Stanford␈α University.
03000
03100 ␈↓↓7␈↓Rubenstein,␈α↔A.H.␈α↔and␈α↔Haberstroh,␈α⊗C.␈α↔J.,␈α↔Some␈α↔Theories␈α⊗of
03200 Organization,␈α Dorsey␈α Press,␈α Homewood␈α Ill.,1960,␈α p.␈α 232.
03300
03400 ␈↓↓8␈↓Schank,␈α∩R.C.,␈α∩Tesler,␈α∪L.␈α∩and␈α∩Weber,S.␈α∩Spinoza␈α∪ii:␈α∩Conceptual
03500 case-based␈α natural␈αlanguage␈α analysis.␈α Memo␈αAIM-109,␈α 1970,
03600 Stanford␈α Artificial␈α Intelligence␈α Project,␈α Stanford␈α University.
03700
03800 ␈↓↓9␈↓Simmons,␈α~R.F.␈α→Some␈α~semantic␈α→structures␈α~for␈α→representing
03900 English␈α&meanings.␈α&Preprint,␈α&1970,␈α'Computer␈α&Science
04000 Department,␈α University␈α of␈α Texas,␈α Austin.
04100
04200 ␈↓↓10␈↓Smith,␈α⊂D.C.,␈α⊂MLISP,␈α∂Memo␈α⊂AIM-135,␈α⊂1970,␈α⊂Stanford␈α∂Artificial
04300 Intelligence␈α Project,␈α Stanford␈α University.
04400
04500 ␈↓↓11␈↓Weizenbaum,␈α∂J.␈α∂Eliza-␈α∂a␈α∂computer␈α∂program␈α∂for␈α∂the␈α⊂study␈α∂of
04600 natural␈α-communication␈α.between␈α-man␈α.and␈α-machine.
04700 Communications␈α of␈α the␈α ACM,␈α 9,36-45,1966.
04800
04900 ␈↓↓12␈↓Wilks,␈α Y.A.␈α Understanding␈α without␈α proofs.␈α (See␈α this␈α volume).
05000
05100 ␈↓↓13␈↓Winograd,␈α⊃T.␈α⊂A␈α⊃program␈α⊂for␈α⊃understanding␈α⊃natural␈α⊂language.
05200 Cognitive␈α Psychology,3,1-191,1972.
05300
05400 ␈↓↓14␈↓Woods,␈α∨W.A.␈α∨Transition␈α∨network␈α∨grammars␈α∨for␈α≡natural
05500 language␈α≠analysis.␈α≠ Communications␈α≠of␈α≠the␈α≠ACM,13,591-
05600 606,1970.