Báo cáo khoa học: "Dialog Control in a Natural Language System" docx

Knowledge about the state of the dialog is represented in a dedicated language and changes of this state are described by a compact set of rules.. the dialog, at least to the extent that

Trang 1

D i a l o g C o n t r o l in a N a t u r a l L a n g u a g e S y s t e m 1

Michael Gerlach Helmut Horacek Universit~t Hamburg Fachbereich Informatik Projektgruppe WISBER

ABSTRACT

In this paper a method for controlling

the dialog in a natural language (NL)

system is presented It provides a deep

modeling of information processing

based on time dependent propositional

attitudes of the interacting agents

Knowledge about the state of the dialog

is represented in a dedicated language

and changes of this state are described

by a compact set of rules An appropri-

ate organization of rule application is

introduced including the initiation of

an adequate system reaction Finally

the application of the method in an NL

consultation system is outlined

INTRODUCTION

The solution of complex problems fre-

quently requires cooperation of multi-

ple agents A great deal of interaction is

needed to identify suitable tasks whose

completion contributes to attaining a

common goal and to organize those

tasks appropriately In particular, this

involves carrying out communicative

subtasks including the t r a n s f e r of

knowledge, the adjustment of beliefs,

expressing wants and pursuing their

satisfaction, all of which is motivated

by the intentions of the interacting

agents [Werner 88] An ambitious dia-

log system (be it an interface, a mani-

pulation system, or a consultation sys-

tem) which is intended to exhibit (some

of) these capabilities should therefore

consider these intentions in processing

1 The work described in this paper is part

of the joint project WISBER, which is sup-

ported by the German Federal Ministery for

Research and Technology under grant ITW-

8502 The partners in the project are: Nixdorf

Computer AG, SCS Orbit GmbH, Siemens

AG, the University of Hamburg, and the Uni-

versity of SaarbrOcken

the dialog, at least to the extent that is required for the particular type of the dialog and the domain of application

A considerable amount of work in current AI research is concerned with in- ferring intentions from utterances (e.g., [Allen 83], [Carberry 83], [Grosz, Sidner 86]) or planning speech acts serving certain goals (e.g., [Appelt 85]), but only

a few uniform approaches to both aspects have been presented

Most approaches to dialog control described in the literature offer either rigid action schemata that enable the simulation of the desired behavior on the surface (but lack the necessary degree of flexibility, e g., [Metzing 79]), or descriptive methods which may also in- clude possible alternatives for the con- tinuation of the dialog, but without expressing criteria to ~aide an adequate choice among them (e g., [Me~ing et al 87], [Bergmann, Gerlach 87])

Modeling of beliefs and intentions (i.e.,

of propositional attitudes) of the agents involved is found only in the ARGOT system [Litman, Allen 84] This approach behaves sufficiently well in se- veral isolated situations, but it fails to demonstrate a continuously adequate behavior in the course of a complete dialog An elaborated theoretical frame- work is provided in [Cohen, Perrault 79] but they explicitly exclude the dele- tion of propositional attitudes Hence, they cannot explain w h a t h a p p e n s when a want has been satisfied

In our approach we have enhanced the propositional attitudes by associating them with time intervals expressing their time of validity This enables us to represent the knowledge about the actual state of the dialog (and also about past states) seen from the point of view

of a certain a g e n t and to express changes in the propositional attitudes

Trang 2

occurring in the course of the dialog

and to calculate their effect This deep

modeling is the essential resource for

controling the progress of the conversa-

tion in a p p r o a c h i n g its overall goal,

and, in particular, for determining the

n e x t subgoal in the conversation which

m a n i f e s t s itself in a system utterance

We have applied our method in the NL

consultation s y s t e m W I S B E R ([Hora-

cek et al 88], [Sprenger, Gerlach 88])

which is able to participate in a dialog

in the domain of financial investment

R E P R E S E N T I N G

P R O P O S I T I O N A L A T T I T U D E S

Knowledge about the state of the dialog

is represented as a set of propositional

attitudes The following three types of

p r o p o s i t i o n a l a t t i t u d e s of a n a g e n t

towards a proposition p form a basic

repertoire :

KNOW : The a g e n t is sure t h a t p is true

This does not imply t h a t p is really true

since the system h a s no m e a n s to find

o u t the r e a l s t a t e of t h e world As-

s u m i n g t h a t the user of a dialog system

obeys the sincerity condition (i.e., al-

w a y s telling the t r u t h , c.f [Grice 75])

a n assertion uttered by the user implies

t h a t the user knows the content of t h a t

assertion

BELIEVE : The a g e n t believes, b u t is not

sure, t h a t p is true, or he/she assumes p

without sufficient evidence

WANT : The a g e n t w a n t s p to be true

Propositional attitudes are represented

in our s e m a n t i c r e p r e s e n t a t i o n lan-

g u a g e IRS, which is used by all system

components involved in semantic-prag-

m a t i c processing IRS is based on predi-

cate calculus, and contains a rich collec-

tion of additional features required by

N L processing (see [Bergmann et al 87]

for detailed information) A propositio-

nal attitude is written as

( < t y p e > < a g e n t > < p r o p > < t i m e > ) :

• < t y p e > is a n e l e m e n t of the set:

• The two agents r e l e v a n t in a dialog

system are the USER and the SYSTEM

I n a d d i t i o n , we u s e t h e n o t i o n

' m u t u a l k n o w l e d g e ' I n f o r m a l l y ,

this m e a n s t h a t both the user and

the s y s t e m k n o w t h a t < p r o p > is true, and t h a t each knows t h a t the other knows, r e c u r s i v e l y We will use t h e n o t a t i o n (KNOW MUTUAL

< prop > .) to express t h a t the proposition < prop > is m u t u a l l y k n o w n

by the user and the system

• < prop > is an IRS formula denoting the proposition the attitude is about

It m a y a g a i n be a propositional attitude, as in (WANT USER (KNOW USER x .) ) which m e a n s that the user wants to k n o w x T h e proposition

m a y also contain the meta-predi-

cates RELATED and AUGMENT: (RELATED

x) m e a n s 'something which is related

to the individual x', i.e., it m u s t be possible to establish a chain of l i n k s connecting the i n d i v i d u a l a n d t h e proposition In t h i s g e n e r a l f o r m

assumptions about the user's competence For a more intensive application, however, f u r t h e r c o n d i t i o n s

m u s t be put on the connecting links

(AUGMENT 0 m e a n s 'something more

specific t h a n the formula f', i.e., a t least one of the v a r i a b l e s m u s t be

q u a n t i f i e d or c a t e g o r i z e d m o r e precisely or additional propositions

m u s t be associated These meta-predicates are used by the dialog control rules as a very compact w a y of expressing general properties of propositions

• Propositional attitudes as a n y o t h e r states hold d u r i n g a period of time

In W I S B E R we use A l l e n ' s t i m e logic [Allen 84] to r e p r e s e n t such

t e m p o r a l i n f o r m a t i o n [Poesio 88]

< t i m e > m u s t be a n i n d i v i d u a l of type TIME-INTERVAL In t h i s p a p e r , however, for the sake of brevity we will use almost exclusively the special constants NOW, PAST a n d FUTURE, denoting time i n t e r v a l s w h i c h a r e asserted to be during, before or a f t e r the current time

I N F E R E N C E R U L E S

A s n e w information is provided by the user and inferences are m a d e by the system, the set of propositional attitudes to be represented in the system will evolve While the semantic-pragmatic analysis of user utterances ex- ploits linguistic features to derive the

Trang 3

attitudes expressed by the u t t e r a n c e s

(c.f [Gerlach, Sprenger 88]), the dialog

control c o m p o n e n t i n t e r p r e t s r u l e s

which embody knowledge about know-

ing and w a n t i n g as well as about the

domain of discourse These rules de-

scribe communicative as well as r/on-

c o m m u n i c a t i v e actions, a n d specify

how new propositional attitudes can be

derived Rules about the domain of dis-

course express state changes including

the involved action The related states

a n d the t r i g g e r i n g action are associated

with time-intervals so t h a t the correct

temporal sequence can be derived

Both classes of rules are represented in

a uniform formalism based on the sche-

m a p r e c o n d i t i o n - a c t i o n - effect:

• The p r e c o n d i t i o n consists of patterns

of propositional attitudes or s t a t e s

in the domain of discourse The pat-

terns m a y contain temporal restric-

tions as well as the meta-predicates

m e n t i o n e d above A p r e c o n d i t i o n

m a y also contain a rule description,

e.g., to express t h a t an a g e n t knows

a rule

el of communication (in the case of

speech act triggering rules) or on the

level of the domain (actions the dia-

log is about) However, there are al-

so pure inference rules in the dialog

control module; their action p a r t is

void

• The e f f e c t of a rule is a set of descrip-

tions of states of the world and pro-

positional a t t i t u d e s which are in-

s t a n t i a t e d when applying the rule

yielding new entries in the system's

knowledge base We do not delete

propositional attitudes or other pro-

OSitions, i.e., the s y s t e m will not

rget them, b u t we can m a r k the

time interval associated with an en-

t r y as being 'finished' Thus we can

express t h a t the e n t r y is no longer

valid, and it will no longer match a

p a t t e r n with the time of v a l i d i t y

restricted to NOW

C O N T R O L S T R U C T U R E

So far, we have only discussed how the

a c t u a l s t a t e of the d i a l o g (from t h e

point of view o f a c e r t a i n a g e n t ) can be

represented and how c h a n g e s in this

state can be described We still need a method to determine and c a r r y out the relevant changes, given a certain state

of the dialog, after i n t e r p r e t i n g a user utterance (i.e., to decide which dialog rules m a y be tried and in which order) For reasons of simplicity we have divid-

ed the set of rules into t h r e e s u b s e t s each of them being responsible for accomplishing a specific subtask, namely:

• gaining additional i n f o r m a t i o n in- ferable from the i n t e r r e l a t i o n bet-

w e e n r e c e n t i n f o r m a t i o n c o m i n g from the last user utterance a n d the actual dialog context The combina- tion of new and old information m a y ,

e g., change the degree of c e r t a i n t y

of some proposition, i e., t e r m i n a t e

an (uncertain) BELIEVE state a n d cre- ate a (certain) KNOW state with identical propositional content (the consistency m a i n t e n a n c e rule package)

• p u r s u i n g a global (cognitive or ma- nipulative) goal; this m a y be done either by t r y i n g to satisfy this goal directly, or indirectly by substitut- ing a more adequate goal for it a n d

p u r s u i n g this new goal In particular, a goal substitution is u r g e n t l y needed in case the original goal is unsatisfiable (for the system), b u t a promising a l t e r n a t i v e is a v a i l a b l e (the goal pursuit rule package)

• p u r s u i n g a communicative subgoal

I f a goal can not (yet) be accomplished due to lack of information, this leads to the creation of a WANT

c o n c e r n i n g k n o w l e d g e a b o u t t h e missing information W h e n a goal

h a s been accomplished or a signi- ficant difference in the beliefs of the user and the system h a s been disco- vered, the system WANTS the user to

be informed about that All this is done in the p h a s e concerned w i t h cognitive goals Once such a WANT is created, it can be associated with a n appropriate speech act, provided the competent dialog p a r t n e r (be it the user or an external expert) is deter-

m i n e d (the speech a c t t r i g g e r i n ~ rule package)

There is a certain l i n e a r d e p e n d e n c y between these subtasks Therefore the respective rule packages are applied in

a suitable (sequential) order, w h e r e a s those rules belonging to the same pack-

Trang 4

age m a y be applied in a n y order (there

exist no interrelations w i t h i n a single

rule package) This simple forward in-

ferencing works correctly and with an

acceptable performance for the actual

coverage and degree of complexity of

the system

A sequence c o n s i s t i n g of these three

subtasks forms a (cognitive) processing

cycle of the s y s t e m from r e c e i v i n g a

user message to i n i t i a t i n g an adequate

reply This procedure is repeated until

there is evidence t h a t the goal of the

conversation h a s been accomplished (as

indicated by knowledge a n d assump-

tions about the user's WANTS) o r that

the user wants to finish the dialog In

either case the system closes the dialog

A P P L I C A T I O N IN A

C O N S U L T A T I O N SYSTEM

In this section we present the applica-

tion of our method in the NL consul-

ration system WISBER involving rath-

er complex interaction with subdialogs,

requests for explanation, recommenda-

tions, a n d a d j u s t m e n t of proposals

However, it is possible to i n t r o d u c e

some simplifications typical for consul-

ration dialogs These are u r g e n t l y need-

ed in order to reduce the otherwise ex-

cessive a m o u n t of complexity In parti-

cular, we assume t h a t the user does not

lie a n d take h i s / h e r assertions about

real world events as true (the sincerity

condition) Moreover, we t a k e it for

g r a n t e d t h a t the user is h i g h l y interest-

ed in a consultation dialog and, there-

fore, will pay attention to the conversa-

tion on the screen so t h a t it can be rea-

sonably a s s u m e d t h a t he/she is f u l l y

aware of all utterances occurring in the

course of the dialog

Based on these (implicit) expectations,

the following (simplified) assumptions

(1) a n d (2) represent the starting point

for a consultation dialog:

(1) (BELIEVE SYSTEM

(WANT USER

((EXIST X (STATE X))

(HAS-EXPERIENCER X USER)) NOW) NOW)

(2) (BELIEVE SYSTEM

(KNOW USER

(HAS-EXPERIENCER Y USER)))

NOW) NOW)

They express t h a t the user knows some-

t h i n g t h a t 'has to do' (expressed by the

m e t a - p r e d i c a t e RELATED) w i t h s t a t e s (STATE Y) concerning h i m / h e r s e l f a n d

t h a t he/she w a n t s to a c h i e v e a state (STATE X) In assumption 1, (STATE X) is in fact specialized for a consultation system as a real world state (instead of a

m e n t a l state which is the g e n e r a l assumption in a n y dialog system) T h i s state can still be made more concrete when the domain of application is t a k e n into account:

In WISBER, we assume t h a t the u s e r wants his/her money 'to be i n v e s t e d ' The second a s s u m p t i o n e x p r e s s e s (a part of) the competence of the user T h i s

is not of particular importance for m a n y other types of dialog systems In a consultation system, however, this is t h e basis for addressing the user in order to ask him]her to m a k e his/her i n t e n t i o n s more precise In the course of the dialog these assumptions are supposed to be confirmed and, moreover, t h e i r content

is expected to become more precise

In the subsequent p a r a g r a p h s we out- line the processing behavior of the system by e x p l a i n i n g the application a n d the effect of some of the most i m p o r t a n t dialog rules (at least one of each of the three packages introduced in the previ- ous section), thus giving an impression

of the system's coverage In the r u l e s presented below, v a r i a b l e s are s u i t a b l y quantified as they appear for the first time in the precondition In s u b s e q u e n t appearences they are referred to l i k e constants The interpretation of the special constants denoting t i m e - i n t e r v a l s depends on whether they occur on the left or on the r i g h t side of a rule: in the precondition the associated state/event

m u s t hold/occur during PAST, FUTURE or overlaps NOW; in the effect the state/ event is associated with a time-interval that starts at the reference time-interval

In a consultation dialog, the user's wants m a y not always express a direct request for information, but rather re- fer to events and states in the real world F r o m such user wants the system m u s t derive requests for knowledge useful w h e n attempting to satisfy them.2 Consequently the task of infer-

Trang 5

(KNOW MUTUAL

(WANT USER

(EXIST A (ACTION A)) NOW) NOW)

A

(KNOW SYSTEM

(UNIQUE R

(AND (RULE R) (HAS-ACTION R A) (HAS-PRECONDITION R (EXIST 51 (STATE 51))) (HAS-EFFECT R

(EXIST $2 (STATE 52))))) NOW)

=~

(KNOW MUTUAL (WANT USER

51 NOW) NOW)

A (KNOW MUTUAL

R NOW)

A (KNOW MUTUAL (WANT USER s2 NOW) NOW)

Rule 1: Inference d r a w n from a user w a n t referring to an action with u n a m b i -

guous consequences (pursuing a global goal)

r i n g communicative goals is of central

importance for the functionality of the

system

There is, however, a f u n d a m e n t a l dis-

tinction w h e t h e r the content of a w a n t

refers to a state or to an event (to be

more precise, to an action, mostly) In

the l a t t e r case some i m p o r t a n t infer-

ences can be d r a w n depending on the

d o m a i n k n o w l e d g e a b o u t t h e e n v i -

sioned action and the degree of preci-

sion expressed in its specificatiqn If,

a c c o r d i n g to t h e s y s t e m ' s d o m a i n

model, the effect of the specified action

is unambiguous, the user can be expect-

ed to be f a m i l i a r with this relation, so

he/she can be assumed to envision the

r e s u l t i n g state and, possibly, the pre-

condition as well, if it is not yet ful-

filled Thus, in principle, a plan consist-

ing of a sequence of actions could be cre-

a t e d b y a p p l i c a t i o n of s k i l l f u l r u l e

chaining

This is e x a c t l y w h a t Rule 1 asserts:

Given the m u t u a l knowledge t h a t the

user w a n t s a certain action to occur,

a n d the system's knowledge (in form of

a unique rule) about the associated pre-

condition a n d effect, the s y s t e m con-

c l u d e s t h a t t h e u s e r e n v i s i o n s t h e

r e s u l t i n g state and he/she is f a m i l i a r

with the connecting causal relation I f

the uniqueness of the rule c a n n o t be

2 Unlike other systems, e.g., UC [Wilensky et

al 84], which can directly perform some kinds

of actions required by the user, WISBER is

unable to affect any part of the real world in

the domain of application

established, sufficient evidence derived from the p a r t n e r model m i g h t be a n alternative basis to obtain a sufficient categorization of the desired e v e n t so

t h a t a unique rule is found Otherwise the u s e r h a s to be a s k e d to p r e c i s e his/her intention

Let us suppose, to give an example, t h a t the user has expressed a w a n t to invest his/her money According to W I S B E R ' s domain model, there is only one match- ing d o m a i n rule e x p r e s s i n g t h a t t h e user has to possess the m o n e y before but not after investing his/her money, and obtains, in exchange, an asset of a n equivalent value Hence Rule 1 fires The w a n t expressed by the second p a r t

of the conclusion can be i m m e d i a t e l y satisfied as a consequence of the u s e r utterance 'I have inherited 40 000 DM'

by applying Rule 5 (which will be ex- plained later) The r e m a i n d e r p a r t of the conclusion m a t c h e s a l m o s t com- pletely the precondition of Rule 2

This rule states: I f the u s e r w a n t s to achieve a goal state (G) and is informed about the w a y this can be done (he/she knows the specific RULE R and is capable

of performing the r e l e v a n t action), the system is right to assume t h a t the user

is lacking some information which in- hibits him/her from actually doing it Therefore, a w a n t of the user indicating the intention to know more about this transaction is created (expressed by the meta-predicate AUGMENT) I f the necessary capability cannot be a t t r i b u t e d to the user a consultation is impossible

If, to discuss another example, the u s e r

h a s expressed a w a n t a i m i n g a t a cer-

Trang 6

(KNOW MUTUAL

(WANT USER

(EXIST S (STATE S)) NOW) NOW)

A

(KNOW MUTUAL

(UNIQUE R

(AND (RULE R)

(HAS-EFFECT R S) (HAS-ACTION R (EXIST A (ACTION A))))) NOW)

A

(KNOW MUTUAL

(CAPABILITY USER A) NOW)

=~

(BELIEVE SYSTEM (WANT USER (KNOW USER

(AUGMENT S) FUTURE) NOW) NOW)

Rule 2: Inference drawn from a user want referring to a state, given his/her ac-

q u a i n t a n c e with the associated causal relation (pursuing a global goal) rain state (e.g., 'I w a n t to have m y mon-

ey back'), the a p p l i c a t i o n of a n o t h e r

rule almost identical to Rule 1 is at-

tempted W h e n its successful applica-

tion yields the association of a unique

event, the required causal r e l a t i o n is

established Moreover, the user's fami-

l i a r i t y with this relation m u s t be deri-

v a b l e in order to follow the path indi-

cated by Rule 2 Otherwise, a w a n t of

the user would be created whose con-

tent is to find out about suitable m e a n s

to a c h i e v e the d e s i r e d s t a t e (as ex-

pressed by Rule 3, l e a d i n g to a system

reaction like, e.g., 'you m u s t dissolve

your s a v i n g s account')

It is very frequently the case t h a t the

satisfaction of a w a n t cannot immedi-

ately be achieved because the precision

of its specification is insufficient W h e n the domain-specific p r o b l e m s o l v i n g component indicates a clue about w h a t information would be helpful in this re- spect this triggers the creation of a system w a n t to get a c q u a i n t e d w i t h it

W h e n e v e r the user's u n i n f o r m e d n e s s in

a particular case is not yet proved, a n d this information falls into his/her competence area, control is passed to the ge- neration component to address a suitable question to the user (as expressed

in Rule 4)

Provided with new information hopeful-

ly obtained by the user's reply the system tries a g a i n to satisfy the (more precisely specified) user want This process

is repeated u n t i l an adequate degree of specification is achieved at some stage

(KNOW MUTUAL

(WANT USER

(EXIST G (STATE G)) NOW) NOW)

A

(KNOW SYSTEM

(EXIST R

(AND (RULE R) (HAS-EFFECT R G) (HAS-PRECONDITION R (EXIST S (STATE S))) (HAS-ACTION R (EXIST A (ACTION A))))) NOW)

A

(-= (KNOW USER R NOW))

=~

(BELIEVE SYSTEM (WANT USER (KNOW USER

R FUTURE) NOW) NOW)

quaintance with the associated causal relation (pursuing a global goal)

- 3 2 -

Trang 7

(WANT SYSTEM

(KNOW SYSTEM

X FUTURE) NOW)

A

(BELIEVE SYSTEM

(KNOW USER

(RELATED X)

NOW) NOW)

A

(-I (KNOW SYSTEM

(-1 (KNOW USER

x NOW)) NOW))

(ASK

SYSTEM

USER

x)

(KNOW MUTUAL (WANT SYSTEM (KNOW SYSTEM X FUTURE)

NOW) NOW)

A (KNOW MUTUAL (BELIEVE SYSTEM (KNOW USER

(RELA TED X) NOW)

NOW) NOW)

Rule 4: Inference drawn from the user's (assumed) competence and a system

want in this area (triggering a speech act)

In the course of the dialog each utter-

ance effects parts of the system's cur-

rent model of the user (concerning as-

sumptions or temporarily established

knowledge) Therefore, these effects are

checked in order to keep the data base

consistent Consider, for in st an c e, a

user w a n t a i m i n g at investing some

money which, after a phase of para-

meter assembling, has led to the system

proposal 'I r e c o m m e n d you to buy

bonds' a p p a r e n t l y accomplishing the

(substitued) goal of obtaining enough

information to perform the envisioned

action Consequently, the state of the

associated u s e r w a n t is s u b j ect to

change which is expressed by Rule 5

Therefore, the mutual knowledge about

the user want is modified (by closing

the associated time-interval) and the

the user's want is marked as being 'fin-

ished' and added to the (new) mutual

knowledge

However, this simplified treatment of

the satisfaction of a want includes the

restrictive assumptions that the acceptance of the proposal is (implicitly) anti- cipated, and that modifications of a want or of a proposal are not manage- able In a more elaborated version, the goal accomplishment has to be m a r k e d

as provisory If the user expresses his/her acceptance either explicitly or changes the topic (thus implicitly agreeing to the proposal), the application of Rule 5 is fully justified

Apart from the problem of the increas- ing complexity and the a m o u n t of necessary additional rules, the prelimi- nary status of our solution has m u c h to

do with problems of interpreting the

the representation of a communicative goal according to the derivation by Rule 2: The system is satisfied by finding any additional information augmenting the user's knowledge, but it is not aware of the requirement that the information must be a suitable supplement (which is recognizable by the user's confirmation only)

(KNOW MUTUAL

x NOW)

Rule 5: Inference drawn from a (mutually known) user want which the user

k n o w s to be accomplished (pursuing consistency maintenance)

Trang 8

F U T U R E R E S E A R C H

The method described in this paper is

fully implemented and integrated in

the complete NL system WISBER A re-

latively small set of rules has proved

sufficient to guide basic consultation di-

alogs Currently we are extending the

set of dialog control rules to perform

more complex dialogs Our special in-

terest lies on clarification dialogs to

handle misconceptions and inconsisten-

cies The first steps towards handling

inconsistent user goals will be an expli-

cit representation of the interrelation-

ships holding between propositional at-

titudes, e.g., goals being simultaneous

or competing, or one goal being a re-

finement of another goal A major ques-

tion will be specifying the operations

necessary to recognize those interrela-

tionships working on the semantic re-

presentation of the propositional con-

tents As our set of rules grows, a more

sophisticated control mechanism will

become necessary, structuring the deri-

vation process and employing both for-

ward and backward reasoning

R E F E R E N C E S

Allen 83

wick, R.C (eds.): Computational Models of Dis-

course, MIT Press, 1983, pp 107-166

Allen 84

1984, pp 123-154

Appelt 85

Appelt, D.E.: P l a n n i n g E n g l i s h Sentences

Cambridge University Press, 1985

Bergmann, Gerlaeh 88

B e r g m a n n , H., Gerlach, M.: S e m a n t i s c h -

pragmatische Verarbeitung von ~,uflerungen

im nattlrlichsprachlichen Beratungssystem

W I S B E R In: Brauer, W., Wahlster, W (eds.):

Wissensbasierte Systeme - GI-Kongress 1987

Springer Verlag, Berlin, 1987, pp 318-327

Bergmann et al 87

Bergmann, H., Fliegner, M., Gerlach, M.,

14, Universi~t Hamburg, 1987

Carberry 83

of the AAAI-83, Washington, D.C., 1983, pp

59-63

Cohen, Perrault 79

Science 3, 1979, pp 177-212

Gerlach, S p r e n g e r 88

tion of Pragmatic Clues: Connectives, Modal

COLING-88, Budapest, 1988, pp 191-195

Morgan (ed.): Syntax and Semantics, Vol 3: Speech Acts, Academic Press, New York, 1975,

pp 41-58

Grosz, S i d n e r 86

tational Linguistics 12 (3), 1986, pp 175-204

H o r a c e k et al 88 Horacek, H., Bergmann, H., Block, R., Fliegner,

Meaning to Meaning - a Walk through WIS

genz - GWAI-88, Springer Verlag, Berlin, 1988,

pp 118-129

Litman, Allen 84

COLING'84, Stanford, pp 302-311

MeBing, et al 87 Mefling, J., Liermann I., Schachter-Radig M.-J.:

HandIungsschemata in Beratungsdialogen -

A m Gespr(zchsgegenstand orientierte Dialog-

jekt, Dezember 1987, SCS Organisationsbera- tung und Informationstechnik GmbH, Ham- burg

muster und Dialogprozesse H e l m u t Buske Verlag, Hamburg, 1979

Poesio 88

pp, 247-252

Sprenger, Gerlach 88

Sprenger, M., Gerlach, M.: Expectations and Propositional Attitudes - Pragmatic Issues in WISBER In: Proc of the International Com- puter Science Conference '88, ttong Kong, 1988

Werner 88

tion and Cooperation for Multiagent Planning

In: Theoretical Aspects of Reasoning about Knowledge, Proceedings of the 1988 Confer- ence, Morgan Kaufman Publishers, Los Altos,

1988, pp 129-143

Wilensky et al, 84

munications of the ACM, Vol 27, No 6, pp 574-

593

- 3 4 -

Định dạng
Số trang	8
Dung lượng	746,88 KB