UNIT-TO-UNIT INTERACTION AS A BASIS FOR SEMANTIC INTERPRETATION 
OF JAPANESE SENTENCES 
Hozumi TANAKA 
ELECTROTECHNICAL LABORATORY, 1-1-4 UMEZONO, SAKURA-MURA, NIIHARI-GUN, 
IBARAKI-KEN, JAPAN 
ABSTRACT: The notion of UNIT-to-UNIT inter- 
action is introduced to analyse dependency 
relations between words in a sentence. A UNIT 
is a basic framework for concept representation 
and is composed of many slots. After generating 
a parsed tree from an input sentence, our se- 
mantic interpretation begins traversing.the tree 
from right to left to discern the case frame in 
a stage as early as possible, since Japanese is 
a language in which verb is in the sentence- 
final and has a case frame. UNIT-to-UNIT inter- 
action, which is performed at each node of the 
parsed tree, follows a bottom-up progression. 
There are UNIT descriptions at terminal (bottom) 
nodes and the UNIT descriptions are modified or 
merged into other UNITs in the course of the 
interaction. The results of the interaction 
will be transferred to upper nodes. The inter- 
action process continues on upward until the 
top node; at this point, the semantic structure 
of the input sentence is finally obtained. The 
notion of UNIT-to-UNIT interaction is feasibly 
applicable to semantic interpretation of 
English. 
i. INTRODUCTION 
Semantic processing is very important for 
us to build a natural language (NL) under- 
standing system. It will be true that semantics 
takes precedence over syntax when human beings 
understand language. Based on this assumption, 
some NL understanding system designers have 
totally abandoned the traditional use of 
grammars for linguistic analysis. They are 
based on special procedures of semantic inter- 
pretation to build up semantic structures, and 
the result of syntactic parsing is not used. 
Such systems without grammar often lack 
formalism. 
We should not totally abandon a traditional 
use of grammars for linguistic analysis, since 
results of syntactic parsing fill the gap be- 
tween an input sentence and its semantic 
structure. We have developed Extended LINGOL 
\[13,12\] that is the extended version of Pratt's 
LINGOL \[8,9\]. Pratt's LINGOL has a very good 
formalism to merge syntactic and semantic 
information. The idea is that the result of 
syntactic parsing, a parsed tree, is considered 
as a program tree which is evaluated at the time 
of semantic interpretation. In the course of 
the interpretation, UNIT-to-UNIT interactions 
are performed. Thus a parsed tree of LINGOL 
corresponds to an analysis tree of Montague 
grammar and the evaluation phase of the parsed 
tree is analogous to the translation phase of 
Montague grammar \[3\] (See the See.6). 
Our Extended LINGOL inherits the semantic 
interpretation method from the original Pratt's 
LINGOL. After generating a parsed tree from an 
input sentence, semantic interpretation is set 
out. The parsed tree is composed of context- 
free rules to each of which a LISP program is 
attached. In other words, at each node of the 
tree, there is a program for making semantic 
interpretation. As will be explained in the 
Sec. 5.,UNIT-to-UNIT interaction will take 
place at each node of the program tree. The 
interaction process continues on upward until 
the top node, at which it stops getting the 
results of the semantic interpretation. 
2. 4SEM>-PROGRAM TREE AND PARSED TREE 
Our Extended LINGOL produces a parsed tree 
using both grammar and dictionary. The format 
of our grammatical rule is: 
~<left> <right> i<advice> <cog>~ <sem>~. 
The left-right pair represents a 
context-free rule in the form of A--> B or 
A > B C. The <advice>, which is introduced 
into our Extended LINGOL, is an arbitrary LISP 
program for controlling parsing process \[13\]. 
The role of <cog> and <sem> is the same as 
that of Pratt's LINGOL \[9\]. The <sem> is any 
LISP program to perform semantic interpretation. 
The Pratt's LINGOL offers us a flexible method 
of semantic interpretation. 
In order to understand UNIT-to-UNIT inter- 
action, we will briefly illustrate the inter- 
pretation method. By means of <sem> attached to 
each (augmented) context-free rule, we can 
obtain a <sem>-program tree from a parsed tree. 
Consider the following very simple example. 
The input sentence is "iOKGNOOMOSANOMIZU (water 
of i0 kg)',, grammatical rules are: 
(NP (NLNK NP) (...) <S-exprl>) 
(NLNK (NOUN NO) (...) <S-expr2>) 
(NP NOUN (...) <S-expr3>) 
(NOUN (NPOS2 NOUN) (...) <S-expr4>) 
(NPOS2 (QUANT NO) (...) <S-exprS>) 
(QUANT (NUMB UNIT) (...) <S-expr6>) 
and dictionary entries are: 
383-- 
Flg,I MP 
I 
NLHK ---Z ....... ....... NP ! 
NOUN ................ NO 
NPOS2 ............. NOUN 
QUAMT .......... NO 
NUMB ..... UNIT ! 
tO KQ NO OMOgA 
! 
! NoUN 
t ! 
NO MIZU 
Ft~.E(a) (~-ex~ri> 
(S-exprS> .......... (S-exp.3} 
(S-exit4) .... (S-mxprB) <S-~xp~le) 
<S-exprS> .... (S-exp~9) ! 
<S-expr6) .... {S-exprS) l 
lO ........ ($-exprT) 
Ft~,~(b) 
<S-exprl> ! 
(~-e×pr2> ! 
!(S-exp.4) ! 
t 
!($-exprS> ! 
! !<S-expr6) ! 
! B + ............ + 
t t ~i@, ! 
! ! !<$-exp~7) t 
t ! ~KQ ! +--+ ............ + 
! !<S-exprS) I 
+--+ ........ - ...... ¢ 
!(S-exprg) ! 
!OMOSA ! 
+--+ ............. ~ .... 4 
!(~-expr8) ! !NO ! 
+---+ ..................... ¢ 
\](S-expr3> ! ! ! 
! + ..................... 
! !<~-expri~> ! 
(KG UNIT (...) <S-expr7>) 
(NO NO (...) <S-expr8>) 
(OMOSA NOUN (...) <S-expr9>) 
(MIZU NOUN (...) <S-exprl0>). 
Fig.l is a parsed tree of 
"iOKGNOOMOSANOMIZU." Fig.2(a) is a <sem>- 
program tree obtained from Fig.l. Fig.2(b) is 
the nesting structure of the <sem>-program tree 
of Fig.2(a) which defines the scope of 
variables. At lower nodes, we see values of 
variables at upper nodes. For instance, fro~ 
<S-exprA>, one can refer to the value of 
variables in <S-expr2> and <S-exprl>. 
Semantic interpretation begins with the 
evaluation of S-expression at the top node. 
There are several built-in functions two of 
which are LG and RG, which are evaluated at each 
<sem>-tree node with left and right branches. 
The evaluation sequence of LG and RG determines 
the evaluation sequence of S-expressions at 
one-level-lower nodes. For example, in 
<S-exprl>, if the evaluation of RG precedes 
that of LG, <S-expr3> is evaluated and then 
<S-expr2>. The result of RG evaluation becomes 
equal to that of <S-expr3> evaluation. 
Usually, at each node of the <sem>-program tree, 
UNIT-to-UNIT interaction takes place and the 
results of the interaction are transferred to 
one-level-upper node. As will be explained 
before, the role of a parsed tree is similar to 
that of an analysis tree of Montague grammar. 
3. UNIT DESCRIPTION 
A UNIT is a basic framework for concept re- 
presentation and is composed of many slots. 
Our UNIT description incorporates some useful 
features from KRL \[i\] which was developed by 
Bobrow and Winograd. Fig.3(a) is an example of 
our UNIT descriptions. 
(MIZU unit ... 
(self (a EKITAI)) 
(sf +natural) 
• ...,... ....... °...,) 
(EKITAI unit ... 
(self (a BUSSITU)) 
(sf +natural) 
.,,..........,....°,) 
(BUSSITU unit ... 
(self ...) 
(sf) 
(OMOSA ((%value (a OMOSA)) 
............. 
(OMOSA unit ... 
(self ...) 
(sf ...) 
(VALUE ((%value (a OMOSA)) 
<action-l>) 
., .,o.°... .... .,....) 
Fig.3(a) 
(KG unit ... 
(self (a OMOSA)) 
(sf %unit) ...... ) 
-NO) ) 
-NO) 
--384 
V×(~IZU(×) ---> EKITAI(x)) I Vx(~ater(x) ---> llquld(x)) 
Vx(EKITAI(x) ---> BUSSITU(x)) I Vx(llqutd(x) ---) materlal(x)) 
Vx(HAKO(x) ---> KOTQI(x)) 1 Vx(box(x) ---> solid(x)) 
.................. I .................... Bg dl~jolntneee 
of same level UNIT:) 
Vx(EKITAI(x) ---> ~KOTAI(x)) I Vx(tiqUid(x) ---) ~otld(x)) 
For eeman~Ic tea£uree In "el'' 
~x(+na~ural(x) ---) +\[iVlng(x) V -\[IVlng(x)) 
~x(-natural(x) ---) -living(x)) 
...... , , ........ , , , , , Vx(+llving(x) --->~-tlulng(x)) 
,.,'4.,,. ...... ~,,. 
Fiff.3(lo) 
The "self" slot which is present in each 
UNIT description enables us to understand the 
UNIT framework as a whole. As in KRL, the 
"self" slot is used for the hierarchical 
organization of UNITs and enables all infor- 
mation (slots) to transfer from superordinate 
UNIT to subordinate UNITs. For example, the 
"self" slot in MIZU (water) UNIT indicates that 
the superordinate UNIT of MIZU (water) is EKITAI 
(liquid). 
Our UNIT descriptions are slightly differ- 
ent from KRL descriptions. Semantic features 
are incorporated into each special slot named 
"sf". The "sf" slot in MIZU (water) indicates 
that the semantic feature of MIZU is \[+natural\]. 
In order to express gross semantics of a 
UNIT description, we can use logical expressions 
of first-order predicate calculus. For example, 
gross semantics of MIZU UNIT is expressed as: 
<precondition> .. > <action>. 
As will be explained later, <precondition> and 
<action> act as though they are to-fill and 
when-filled method of KRL \[i\]. 
Two UNITs, which relate to each other in 
UNIT-to-UNIT interaction, are cilled FILLER and 
ORIGIN. During the interaction, FILLER must 
satisfy some slot of ORIGIN. <Precondition> 
specifies conditions of FILLER filled in the 
slot. 
\[A\] <PRECONDITION> 
In order to satisfy some slot of ORIGIN, 
FILLER has to satisfy the <percondition>, which 
specifies not only semantics of FILLER but also 
Japanese surface cases that can follow FILLER in 
the sentence. <Precondition> is divided into 
two parts, <f-constraint> and <J-case>: 
> MIZU (UNIT) 
> +natural (UNIT). 
Hierarchical organization of UNITs is expressed 
as a set of logical entailments \[I0\]. For 
example, from Fig. 3(a) we will have Fig.3(b). 
We can regard Fig.3(b) as a set of axioms 
which is used in performing UNIT-to-UNIT 
interactions. The details will be explained in 
the Sec.5. 
<precondition>::= i<f-constraint> ° <J-case>\[. 
Semantics of FILLER is expressed by 
<f-constraint>. On the other hand, Japanese 
surface cases, which can follow FILLER in a 
sentence, are specified in <J-case>. 
For example, in BUSSITU (material) UNIT of 
Fig.3(a), there is an <unsatisfied> slot: 
(OMOSA ((%value (a OMOSA)) - -NO)). 
4. ORDINARY SLOT 
Most UNITs include a block of ordinary 
slots which are classified into two categories, 
<satisfied> and <unsatisfied>: 
<ordinary-slot>::= <satisfied>l<unsatisfied>. 
The format of two ordinary slots is: 
<satisfied>::=i<slot-name>=<value>~ 
<unsatisfied>::=~<slot-name> 
<precondition> (<action>)~. 
As a proeedual attachment \[i\], we use a pro- 
duction rule \[2,7\]. A pair of <precondition> 
and <action> expresses a production rule in the 
form of: 
The <slot-name> and <precondition> are OMOSA 
(weight) and ((%value (a OMOSA)) -NO), 
respectively. The <f-constraint> and <J-case> 
are (%value (a OMOSA)) and (- -NO), 
respectively. The <f-constraint> is expressed 
as follows: 
> %value(FILLER)A OMOSA(FILLER). 
(Note that logical "and" is always omitted in 
the description of <f-constraint>.) It is 
possible to describe any we~l-formed formula by 
using @OR and @NOT in <f-constraint>. For 
example, (%value (@OR (a WEIGHT) (a VOLUME))) is 
expressed as: 
--> %value(FILLER) ^ (WEIGHT(FILLER) V 
VOLUMECFILLER)). 
--385-- 
The <J-case> of (- -NO) describes what a 
Japanese surface case is allowed to follow 
FILLER in ~ sentence. "-" indicates none of 
Japanese surface cases should follow FILLER, and 
"-NO" indicates that Japanese surface case NO 
("of") should follow FILLER. 
\[B\] <ACTION> 
After FILLER satisfies the <precondition> 
of some ordinary slot, the <action> which is 
any LISP program is activated. Typical effects 
of <action> are: 
(i) Modification of UNITs and slots 
(2) Creation of new UNITs and new slots 
(3) Deletion of UNITs and slots. 
If no <action> is specified, the <unsatisfied> 
slot becomes <satisfied> slot, whose <value> 
becomes FILLER's name but the <slot-name> 
remains unchanged. 
5. UNIT-to-UNIT INTERACTION 
As explained before, UNIT-to-UNIT 
interaction usually occurs at each node of 
<sem>-program tree. In other word, the 
structure of <sem>-program tree determines what 
UNITs should be interacted to each ~ther. For 
example, at <S-expr4> of Fig.2(b), both UNITs 
of "IOKG" and "OMOSA (weight)" are related by 
UNIT-to-UNIT interaction. 
Two UNITs, which relate to each other in 
UNIT-to-UNIT interaction, are called FILLER and 
ORIGIN. During the interaction, FILLER must 
satisfy some <unsatisfied> slot of ORIGIN. If 
it is impossible to find out any satisfiable 
slot in ORIGIN, superordinate UNITs of ORIGIN 
will be retrieved through "self" until some 
satisfiable slot will be found. The satis- 
fiability is determined by FILLER and 
<precondition> in an ordinary slot of ORIGIN. 
At first, a surface case which follows 
FILLER is checked by using <J-case> in 
<precondition>. If this checking succeeds, then 
the semantics of FILLER is checked by using 
<f-constraint> in <precondition>. These 
checkings are expressed as follows: 
Given the semantics of FILLER and a 
set of axioms as shown in Fig.3(b), then 
examine whether <f-constraint> hold or not. 
Let us consider two simple examples. As 
explained before, at <S-expr4> of Fig.2(b), the 
following two UNITs are interacted to each 
other: 
FILLER: 
((i0 KG) unit ... 
(self (a OMOSA)) ...) 
ORIGIN: 
(OMOSA unit ... 
(VALUE ((%value (a OMOSA)) - -NO) 
<action-l>) 
..... .,.......o...). 
In this case, if a Japanese surface case of 
NO ("of") follows FILLER, then FILLER can satis- 
fy VALUE-slot of (VALUE ((%value (a OMOSA)) - 
-NO) <action-l>), since the semantics of FILLER 
is: 
> OMOSA(FILLER) ; 
+ %value(FILLER). 
and it is easy to show that the following 
<f-constraint> holds: 
• > OMOSA(FILLER) A %value(FILLER). 
If the VALUE-slot is satisfied by FILLER, 
<action-l> will be activated to make further 
semantic interpretation if necessary. Let us 
consider another example: 
FILLER: 
(MIZU unit ... 
(self (a EKITAI)) 
(sf +natural) 
......,..°.°.*..) 
ORIGIN: 
(SOSOGU unit ... 
(self ...) 
°......,. 
(THEME ((a EKITAI) -WO)) 
... ..................... ) 
where the words, SOSOGU, EKITAI and MIZU in 
Japanese are POUR, LIQUID and WATER in English, 
respectively. 
If a Japanese surface case of WO follows 
FILLER, the slot (THEME ((a EKITAI) -WO)) is 
satisfied by FILLER, because the semantics of 
FILLER is: 
---> MIZU(FILLER) ; 
---~ +natural(FILLER) 
and from a set of axioms as shown in Fig.3(b), 
we get 
~x (MIZU (x) ~> EKITAI (x)). 
It is easy to show that the following 
<f-constraint> holds: 
> EKITAI(FILLER). 
As the result of the interactions, the slot 
of (THEME ((a EKITAI) -WO)) becomes a 
<satisfied> slot of (THEME = MIZU). 
6. SIMPLE EXAMPLE OF SEMANTIC INTERPRETATION 
BY UNIT-to-UNIT INTERACTION 
Let us trace semantic interpretation 
- 386 
process by UNIT-to-UNIT interaction, provided 
that the input sentence is "IOKGNOOMOSANOMIZU." 
The parsed tree and its <sem>-program tree are 
shown in Fig.l, Fig.2(a) and Fig.2(b). 
Depending on the evaluation sequence of LG and 
RG, we can traverse <sem>-program tree in any 
order (see the Sec.2). Suppose <sem>-program of 
Fig.2(a) is traversed as shown by the arrow of 
Fig.4. 
Ftq.4. \[g\] I@ KG HO OMOSA HO MIZU 
<5-exprl> 
i@ KQ NO OMOSA NO ! \[/3 MIZO <5-expr2> .......... <S-expr3) 
/ , i 
• IZU . , \[e\] i@ KG HO OMOSA I NO I 
<5-expr4> .... {$-~xprB)}~ 
i@ KG HO ! Ed\] OMOSA~ Ec\] 
<S-~xprS> .... (S-exprg)¢ 
\[b\] I@ KG ! NO <5-ex~r6) .... <S-exprB) 
! \[a\] KQ 
...... <S-expr . 
Trace of UNIT-to-UNIT interaction becomes as 
follows: 
\[a\] KG ~ Create KG UNIT 
(KG unit ... at <S-expr7>. 
(self (a OMOSA)) 
(sf %unit) ...... ) 
\[b\] i0 KG ~ At <S-expr6>, 
((i0 KG) unit ... i0 and \[a\] are 
(self (a OMOSA)) merged. 
(sf %value) ..... ) 
\[c\] 
\[d\] 
i0 KG NO. > \[b\] and NO At <S-expr5>. 
OMOSA ~ Create OMOSA 
((N000001 . OMOSA) UNIT at <S-exprg>. 
unit ... 
(self ...) 
(sf ...) 
.,°.... ° 
(VALUE ((%value (a OMOSA)) - -NO) 
<action-l>) 
.°.,....,,°,..°.) 
Eel i0 KG NO OMOSA___~ 
((i0 KG) unit ... 
(self (a OMOSA)) 
(sf %value) ...) 
At <S-expr4>, 
UNIT-to-UNIT 
interaction occurs 
between FILLER \[b\] 
and ORIGIN \[d\]. 
VALUE-slot of \[d\] 
is satisfied by 
FILLER and 
<action-l> is 
activated to 
remove UNIT \[d\], 
since OMOSA is 
redundant in this 
case. 
\[f\] MIZU ~_--> 
((NO00002 . MIZU) 
unit ... 
(self (a EKITAI)) 
(sf +natural) 
................. 
Create MIZU UNIT 
at <S-exprl0> and 
send it to 
<S-expr3>. 
\[g\] i0 KG NO OMOSA NO MIZU 
.__~> ((N000002 . MIZU) 
unit ... 
(self 
(a EKITAI with 
(OMOSA = 
(i0 KG)))) 
(sf +natural) 
............... 
At <S-exprl>, 
UNIT-to-UNIT 
interaction occurs 
between FILLER \[b\] 
and ORIGIN El). 
There is no 
<unsatisfied> slot 
in MIZU UNIT, so 
superordinate 
UNITs are re- 
trieved and a 
OMOSA-slot is 
found in BUSSITU 
UNIT (see 
Fig.3(a)). 
7. CONCLUSION 
In order to explain a basic notion of 
UNIT-to-UNIT interaction, we showed a very 
simple example in the Sec.6. Based on the idea, 
we have implemented a semantic interpretation 
system called EXPLUS \[12,14\]. Our experiments 
by EXPLUS proved that EXPLUS can extract 
semantic strucures from rather complicated 
Japanese sentences. Sato uses in turn the 
semantic structures as a source to generate 
Japanese sentences\[ll\]. However, we have needed 
more refinements for UNIT description. For 
example, we have augmented following features to 
the UNIT description described in the Sec.3. 
(a) Incorporation of arbitrary LISP 
programs in <precondition> through which we 
can specify FILLER's semantics in any level 
of details; 
(b) A special slot "part-of" to organize 
part-whole relations \[5\]; 
(c) "Without" and "selector" descriptions 
to exclude undesirable slots in the 
superordinate UNITs. 
They are related to the problems of 
knowledge representation. The details of (a)- 
(c) will be explained in \[14\]. In \[14\], more 
complex examples of UNIT-to-UNIT interaction 
will be explained. 
From our experience, we believe that 
UNIT-to-UNIT interaction gives us a reasonable 
framework for semantic interpretation, and will 
be feasibly applicable to other kinds of 
languages such as Egnlish. 
387 
ACKNOWLEDGEMENT: The author is grateful to 
Dr. Fuchi, Head of the Pattern Information 
Division of Electrotechnical Laboratory, for his 
patient encouragement of this study. Thanks are 
also due to all members of the Machine Inference 
Section of Electrotechnical Laboratory, for 
their valuable discussions. 

REFERENCES

\[i\] Bobrow,D.G. and Winograd,T.: "An Overview 
of KRL, a Knowledge Representation Language", 
Cognitive Science, Vol.l, No.l, 1977. 

\[2\] Davis,R. and King,J.: "An Overview of 
Production Systems", Stanford AIM-271, 0ct.1975. 

\[3\] Dowty,D.R.: "A Guide to Montague's PTQ", 
Indiana University Linguistic Club, Dec. 1978. 

\[4\] Fillmore,C.J.: "The Case for Case", in 
Bach and Harms (Eds): "Universals in Linguistic 
Theory", Holt, Rinehart and Winston, 1968. 

\[5\] Miller,G.A. and Johnson-Laird,P.N.: 
"Language and Perception", Harvard Univ. Press, 
1976. 

\[6\] Minsky,M.: "Framework for Representing 
Knowledge", in Winston (Ed.): "The Psychology of 
Computer Vision", McGrow-Hill, 1975. 

\[7\] Newell,A.: "Productions Systems: Models 
of Control Structures", in Chase,W.G. (Ed.): 
"Visual Information Processing", Academic Press, 
1973. 

\[8\] Pratt,V.R.: "A Linguistic Oriented 
Programming Language", IJCAI3, 1973, 372-381. 

\[9\] Pratt,V.R.: "LINGOL-A Progress Report", 
IJCAI4, 1975, 422-428. 

\[i0\] Reiter,R.: "On Reasoning by Default", in 
Waltz (Ed.): TINLAP2, 1978, 210-218. 

\[ii\] Sato,T.: "SGS: A System for Mechanical 
Generation of Japanese Sentences", Proc. of 
COLING80, 1980, (in this volume). 

\[12\] Tanaka,If., Sato,T. and Motoyoshi,F.: 
"EXPLUS-A Semantic Parsing System for Japanese 
Sentences", 3rd USA-JAPAN Computer Conference, 
1977, 236-240. 

\[13\] Tanaka,H., Sato,T. and Motoyoshi,F.: 
"Predictive Control Parser: Extended LINGOL", 
IJCAI-79, 1979, 868-870. 

\[14\] Tanaka,H.: "A Semantic Processing System 
for Natural Language Understanding", Research 
No.797, Electrotechnical Laboratory, July, 1979 
(in Japanese). 

\[15\] Wilks,Y.: "An Artificail Intelligence 
Approach to Machine Translation", in Schank and 
Colby (Eds.): "Computer Models of Thought and 
Language", Freeman and Company, 1973. 

\[16\] Winograd,T.: "Understanding Natural 
Language", Academic Press, 1972. 
