Querying Temporal Databases Using Controlled Natural 
Language* 
Rani Nelken Nissiln Francez 
Computer Science Department 
The Technion 
Haifa 32000, Israel 
Abstract 
Recent years have shown a surge in interest in 
temporal database systems, which allow users to 
store time-dependent intbrmation. We present 
a novel controlled natural  interface to 
temporal databases, based on translating nat- 
ural  questions into SQL/Telnporal, a 
temporal database query . The syn- 
tactic analysis is done using the Type-Logical 
Grammar framework, highlighting its utility not; 
only as a theoretical fralnework but also as a 
practical tool. The semantic analysis is done 
using a novel theory of the semantics of tempo- 
ral questions, focusing on the role of temporal 
preposition phrases rather than the more tradi- 
tional focus on tense and aspect. Our transla- 
tion method is considerably simpler than pre- 
vious attempts in this direction. We present a 
prototype software implementation. 
1 Introduction 
Traditionally, database management systems 
were designed to store snapshot information, 
valid at a particular moment of time (state). 
Itowever, many applications require handling 
dynamic time-dependent information, pertain- 
ing not only to the present, but also to the 
past and future. Adding temporal support 
to databases has proved to be a surprisingly 
* This work was carried out as part of the research 
project "Semantics of Natural Language Temporal Ques- 
tions and Interfaces to Temporal Database Systems" 
sponsored by the Fund for interdisciplinary researdl, ad- 
ministered by the Israeli Academy of Science. We thank 
Michael BShlen, Bob Carpenter and Andreas Steiner for 
each allowing us to incorporate their software within our 
own. We also thank Yoad Winter and the anonymous 
referees for helpful comments on a previous version of 
this paper. The work of the second author was partially 
supported by the fund for the promotion of l'eseardl in 
the Tcchniou. 
thorny issue (Tansel et el., 1993). A re- 
cent drive to consolidate research efforts has 
led to the design of a consensus temporal 
data model and associated temporal database 
query , SQL/Temporal, an extension 
of the popular Structured Query Language 
(SQL) (Snodgrass, 2000). 1 SQL/¢J)enlpora\] rep- 
resents a significant improvement over standard 
SQL in allowing programmers to express tem- 
poral queries (Snodgrass, 2000). Since tempo- 
ral database (TI)B) implementations are still 
in their infancy (Boehlen, 1995), there is lit- 
tle practical experience with SQL/Temporal; let 
alone experience of non-expert users. However, 
it; is our belief that such users are bound to find 
the expression of complex temporal queries in 
SQL/Temporal to be extremely difficult. 
In an attempt to counter this problem, we 
present a translation method from controlled 
natural  (NL) to SQL/Temporal. Fol- 
lowing the standard pipeline architecture of 
such methods, translation is done via an inter- 
mediate meaning representation  illus- 
trated in Figure 1. NL questions are first parsed 
using a grammar in the Type Logical Gram- 
mar (TLG) framework (Carpenter, 1998; Mor- 
rill, 1.998). Simultaneously with parsing, the NL 
question is translated into a formula of a for- 
mal  called Lane, (Toman, 11.996), based 
on the interval operators of (Allen, 1983). The 
translation is based on an independently moti- 
vated novel semantics of sentences modified by 
temporal Preposition Ph, rases (PPs) (Pratt and 
Francez 1997; 2000, Nelken and Francez 1999). 
The constructed LAno. formula is then trans- 
tThrough continued design, SQL/Temporal has 
evolved from predecessor versions named TSQL2 (Snod- 
grass, 1.995) and ATSQL2 (Boehlen et el., 1996). It is 
expected to be incorporated within the new version of 
SQL named SQL3. 
1076 
la,ted into an SQl,/Teml)oral query, which is 
subsequently submitted to a. l)rototyl)e TI)B im- 
plementation for eva.luation. Finally, the TI)B's 
answer is 1)resented to the user. 
Figure 1: The tr~nslation pipeline 
We have implemented this method as a pro- 
totype software tool, called Q IYEI~TY, (allnost) 
an acronym lbr "Querying with 15'nglish, of l~cla- 
rio'hal Temporal Databases". Parsing and trans- 
la.tion to I,Alle,, is done using the TLG Theo- 
rem Prover (C~u:penter, 1999). The trmlslation 
from \['allen to SQL/Temporal is done using an 
adaptation of a tempera.1 logic (Ti,) to Aq'SQL2 
translator of (Boehlen et al., 1996). The result- 
lag query is submitted to a prototype TI)B im- 
plementation, calle(l Timel)\]} (Steiner, 1{)97). 
T'he different modules are COul)led into an inte- 
grated system iml)lemented in Sicstus 1)rolog on 
a. UNIX platform with a WW\¥-l)ased gral)hi-- 
caJ fi'ont-end. We discuss some of the directions 
required in order to turn the system fi:om a. re- 
search prototype to a working tool. 
2 Related work 
There is voluminous litera.ture on the design 
of Nil, interfaces to general (non-tenq)oral) 
databa.ses (see (lYrrault and Grosz, \]988; 
Col)estake and Jones, 1990; Androutsopoulos et 
al, 1995) for surveys) and by now their main 
a,:lva.ntages and disadvantages are well under- 
stood. Much less work has been devoted to 
the design of Nil, interfa.ces to TDBs (C'lifford, 
1990; Hinrichs, 1988) or other computer sys- 
t(.'nls involving a temporal dimension (Crouch 
and l)ulnm.n, 71993). ()f lYa, rticular relevance 
is (Androutsopoulos, 1996), who presents a lin- 
guistically motivated translation method fi'om 
NL queries into TSQL2 using a.n ltl)SG (Pol- 
la, rd and Sag, 1.994) grammar and a TL as an 
intermediate representation . Our al)- 
1)l:oach shares many characte.ristics with (An- 
d routsopoulos, 1996), but there also important 
difl'erences, which we point out throughout the 
paper. 
\¥e I)egin our presentation of the. translation 
method with a brief overview of the T\])I}, as its 
structure determines many of the deign choices 
ta.lw.n in devising the translation method. 
3 The TDB 
A TDB is a two-sorted tirst-oMer structure. 
The domain consists of a Data Do'main, D, and 
a Temporal Domain of intervals, 7'/, detined 
as follows. Let Tp 1)e a. set (of time points) 
with a discrete linear order without endl)oints, 
<._ .7'1 is defined as the set of pairs : 7) = 
Z,)la _< b c Tu{-oo, oo}}. a rdational 
database schema is a set, of single-sorted pred- 
icate symbols (H,1,..., l~k). Given a relational 
database schelna p, a Tl)l} schema p' is the sol; 
3/ of two-sorted predicate sylnbols (l£11,..., \]~), 
where the sort of /¢~ is D aritv(l~i) x TI. A 
TI)13 instance of schema f is a set of relations 
R\[ C l) arity(l{i) X 7), where each R~ is finite. 
For instance, assume a database schema p 
consisting era single binary predicate symbol 
'work, storing lbr each elnployee the department 
ill which she is employed. The Tl)l} schema 
p' consists of the relation wo'rk', called a valid- 
ti'mc state table, which adds a. temporal argu- 
nlent 1;o the original relation, caJled the valid- 
time of the table. '\['he temporal argument can 
be used to store the history (and 1)erhaps even 
future plans) of departments in which eml)loyees 
are employed. Following a. suggestion of (An- 
droutsopoulos, 1996), we also include relations 
mal)l)ing names of ca.lendricat items to tempo- 
ral intervals in p'. For instance, we store a. rela- 
tion year / ma.l)l)ing the year 2000 to the interval 
\[71. \].2(100-3 J. 12.2000\] (which in turn is n lappe(t 
to an element of 7)). 
We now describe the translation process. 
4 The translation process 
The translation 1)recess a.ccepts input NL ques- 
tions in a controlled subset of Nil,. Restrict- 
ing inl)ut  in this way enables e\[l'ective 
processing of a sufficiently rich fragment while 
avoiding many of the well-know problems of un- 
restricted NI,. We use a formal grammar in the 
TLG framework. Our grammar is specially de- 
signed for use with ~ particular TDB schema. 
Future work will allow easier configuration of 
the grammar with respect to the schema. 
Our grammar is based on work on all inde- 
pendently motivated theory of the semantics of 
telnt)orality. Most of the research in this tield 
(see (Steedman, 1997) for a survey) has rocllsed 
on the issues of tense a.nd aspect. \~e ha.ndle 
tense, but purposefully not aspect, which plays 
1077 
an import,~nt role in (Androutsol)oulos , 1996). 
Aspect, which is used to retlect speal~ers' tempo- 
ral viewpoint with respect to reported situations 
is an imi)ortant facet of NL temporality. How- 
ever, its relevance to TDBs is questionable, as it 
is unlikely that a realistic TDI3 would actually 
encode such subjective viewpoints. Moreover, 
handling aspect requires postulating a more 
complex data model. For instance, (Androut- 
sopoulos, 1996) augments tile TDB model with 
event-like "occurrence klentifiers", and adds a.n 
additional argument to temt)oral relations indi- 
cating whether a given event has cuhninated or 
not. While such devices may perhaps be linguis- 
tically justified, it is unclear whether the TDI3 
community would adopt such augmentations of 
the model. 
Instead, following (Pratt and Vrancez, 2000; 
Nelken and Francez, 1999) our focus is on sen- 
tences modified by temporal PPs. These PPs 
are analyzed as variants of standard general- 
ized quantifiers (Barwise and Cooper, 1981), 
in which qnantification is over: time. Using 
this framework, we handle questions that re- 
fer explicitly to the temporal (timension (e.g. 
When/during which year ...) as well as ques- 
tions in which temporality is hnplied by the 
TI)B context (e.g. Did Mary work in marketing?, 
Which employees worked in marketing?). We ban- 
(lie both clausal a.nd phrasal teml)ora\] Pl)s (e.g. 
after John worked in R&D, during every year). J\n 
important strength of this semantic theory is 
that it; allows for arbitrary iteration of PPs (e.g. 
one month during every year until 1992). In ad- 
dition, our grammar also handles qua.ntific~tion 
over individuals (e.g. some employee), coordina- 
tion and negation. 
input questions are parsed using a lexicalized 
type-logical grammar. Lexical items are ass(> 
elated with a syntactic category and a higher- 
order lambda-term representing its semantics. 
Taking advantage of 'I'LG's elegantly tight cou- 
pling of syntax and semantics, parsing and con- 
struction of a semantic representation in the 
form of an LAlle n Ibrmula proceed silnultane- 
onsly, in a bottom-up fashion. \¥e have found 
using TLG to be advantageous over a feature- 
structure based formalism (such as IIPSG as 
in (A ndroutsopoulos, 1996)), since formula con- 
struction is an integral part of the l)arslng and 
does not require complex ad-hoc manipulations 
of feature structures. 
Using a particular grammar helps reduce 
some of the ambiguity inherent in unrestricted 
NL. For instance, whereas in general a preposi- 
tion such as at is ambiguous between a tempo- 
ral and a locative interpretation, the choice of 
the coml)lement NP relative to a given schema- 
induced grammar deterministically fixes the in- 
terpretation. As another example, whereas it- 
erating several temporal PPs (e.g. during some 
month every year) opens up exponential seep- 
ing possibilities, some choices are eliminated by 
world knowledge, which is encoded in the gram- 
mar (e.g. every year must have higher scope than 
some month since months are included in years 
and not vice-versa). In cases of remaining ambi- 
guity, Cite user is presented with all Cite distinct 
possibilities. Future work will allow the user: to 
lnalce informed choices between different possi- 
ble readings, e.g. by presenting him with NL 
l)ara.phrases of the alternatives. 
\¥e translate NI, questions into \],All(',,,- if'ire 
main reason for: not translating directly to 
SQl,/Temporal is that the \]atter is not closed 
\['or sub-formulae, i.e. a sub-formula of a well- 
formed query is not necessarily well-formed. 
Since LAne, "is closed for" sub-formulae, composi- 
tionally constructing formulae while parsing in 
a bottom-u 1) fashion becomes much easier. 
I, An~, is defined as follows (Toman, 1996). 
I,et p be the database schema (1~,..., I~t,.). l,et: 
L ::=/~,~(x, I)ILA LI~LI~x.LI~I.LIx = yllc~J 
where x,y are variables over D, x is a vector 
of such variables, l, J are varlet)los or constants 
over T\], and (r is one of the operators: precedes, 
meets, overlaps, equals, contains. LAlien is de- 
fined as tile set of' \[brmulae p E L that con- 
Lain at most one free variable over TI. The 
answer to a formula 90 relative to a TDB D is 
{x, ll> ~ :(x, :)}. 
To illustrate, consider the NL question: 
During which years did Mary work in marketing? 
'File I,Alle n representation for" it is constructed in 
a bottom-up nla, nnel'. The meaning representa- 
tion of the main clause Mary worked in marketing 
is constructed as: 
AJ _ /) 
In this formula, I denotes a Reichenbachian- 
like reference time, J denotes a time interval 
1078 
(I,ring which Mary worked in nia,rl(eA:\]nlg, which 
is loca,l;od in the l)a,sl; (l;he conl;ribution ol7 the 
t, onse) a, nd is hicluded wil;hin 1. 
The me~miug o\[" the \['ull (lll(~,~lJon \]8 COil- 
structed 1)y a,l)t)lying th(; Inca, sing o\[' the \]lll;or- 
rclga,1;ive ton~l)oi:a,1 1)1 ) during which year i;o the 
lllea,ning o\[' Lho c.\[a, uso. \¥il;\]io.t going \]lifO (lc- 
ta, ils, l;ho r('~s, lt is: 
A./ Cpa,,~tAJ C\]) 
The effect of applying the 1)1 ~ is that 1;lie wu'i- 
able / is now 1)ol;h free ~ul(l rest;rioted to 1)e {\]lo 
time of a, yea, r. The a,liswer 1;o l:li(; I'orin.la, is 
~lie so{ o\[" ilitoi:vaJ,~ \] 1;ha,t a, re. ye4~,i',% a,n(I (lur-- 
\]ilg whi(:h 1;\]ioro is a,n iill;orva,l ,\] COliLaiil0d ill 
tlle~ l)asl;~ (l.ring whi(:h ~l/\]il.rry work(,(l ill llla,rl,:ol,- 
ing. VVo allow \]l;ol:a,1;od I)l)s 1;o a,l)l)iy ili a, siniila,r 
Ill a,lillor. 
Not (;vory I~Allc n 1'O1'I111111"/, (:orrosl)Oi~(l~ 1;o ;/,11 
eva hlal)lo ,qQl~/'thiilpora\] query, in 1)arlJc, ula,r~ 
fornuila,e niigJll; \]ia,vo a,li hl:finil;e a lis\vor. \]"orinu- 
la,e tha,t a,r(; safe \[rOlll this a, nd \]'dato(l l)rol) - 
loins a, re t(;ruied domai'n,-i~zd<lJ<'~M¢'n,~ ((Ml(lur 
a,n(l Tol)or ~ \]99\]; AI)it(q)o,t (;t a l., 19!)5). I)o- 
nla,in in(let)en(lonce is a,n un(l(wida,1)lu sonlanl;i(: 
l ) I:O\[)Ol:i;y. 11 ow(;v(;r; w(; i ln pose cerl,ai li ~yllta(:l,ic 
r('M,l:ict;iOliS on g(;nera,lx;d l'ornlulao l,ha,t oitSlli'O 
it,. '\]'heso l:(~slw\]c!,\]OllS :-tISO nim l)lil 7, l\[IO l,i:ansla,- 
1;ion task l'roin I,A,,~I, to ,gQl;/Toml)or~d. This 
tra, nslalsioli is I)ased Oil a nlo(lili('a,tion of the 
Lra, llshl,l;or \[1"()111 \[irM;-or(l(H" TI, over t\]lll(~-I)O\]\]ll,,~ 
1;o A'I'SQI,? of (l/oehlen el; a l., 71996). 
Tho syiita,ctica,l\]y restrh:te(l version o1' l,Alle. 
we use has l;ho uni(luO a,(Iva, nl;a,ge o1" I)ohig very 
clos(; both 1;o the langllag(; IlSO(I \]11 (N(2lkCll 
a, nd IPra,ncez, 1999) Oil tli(~ on(; hand a li(\] i;o 
SQl,/'tbml)orM 011 tho, oLh('r. The S(HlllqJll;\](:g 
o\[' NI, Lonlpora\] exl)rossions is o\['ten expressed 
using (;xl)licit i:el'(;r(;llCO Lo intorwds. IAk(;wis(;, 
SQL/'lhml)ol:a,1 has Alhm-sl,ylo Ol)orators over 
inl;erva,ls. Androul;sot)oulos (\]996) uses a, CtlS- 
tomized 'FI, :.t,s a,ll hilx;rniodia,1;e~ \].q, ll~lla,g0, in 
whi¢\]i t(;ntl)ora,1 i:(;la,tions a,t:e encod(;(l USillg 
L(;nil)ol:a,l opora£ors ra,t;hor 1;ii++i~ oxplicii~ ro\['or- 
once Lo hiterva,\]s. We \]ia,v(; \['ound using a, syl'i- 
l;actica,lly i:osl, ri(:l;ed vel:8iOli of' I,All<;n 1;o t)e a,(i- 
v,anl;ageous, a.s il; a cl;ua,lly sire plifie;~ the tra n,~la- 
tion. 
(\]oni;in,ing oil r i)reviou,~ oxa,lllt)le> 1;lie rostllt- 
illg I~Alle. rOl:lllllla, iS StllJS0.(tllonl;iy IA'P.,llSla,l,(,(l 
into tile \['ollowing S(~\]JTcnli)oral (lu(;ry: 
NONSEQUENCED VALIDTINE 
SELECT DISTINCT aO.cl AS cl 
FRON work t AS al,year I AS aO 
WHERE VALIDTIME(aO) con%ains 
VALIDTINE(al) 
AND al.cl = ~mary' 
AND al.c2 = ~marketing' 
AND PERIOD(TIMESTANP'beginning', 
TIMESTAMP'now') contains VALIDTIME(al) 
The (ltl(;ry :a,sks for the first a,rglunon{ of tlio 
rela, tion \]nsi41,1lCO yeal J such tiia, t the refection 
\]IIM,a,IlCO work # in(:lu(los a, 1;ul)lo consisting of 
'Mary'~ 'marketing' a, nd a, va,lid time. which 
is l;oml)orally iucl.dc'(l in the wind l;ime of" the 
yea, r, as well as in the \]nl;ol:va,l sl;:4,1:l;ing a,1; the 
~\])Ogillllilig ~ O\[' (;\]1110 i/,ll(\] (ql(l\]n~ :llOW :- V\]Z. l, he~ 
past. The 'l'l)l~ rt;,~polMs 1)y returning a table 
conl,aini,<e; exactly 1;ho re(ltleslx;d year names. 
5 Conclusion 
rl~\]lO, a,ddith)n of the. tOtal)oral (limonsion to 
(la, ta,1)asc ,~ysl,cnl,<-; i,crca,ses lJie\]r power 1)ul, also 
thoir COml)lexity. 'lb increa,se the usability of 
T1)171.% we, pi'esout a, prollotyl)e coutrolled NI~ \]ll- 
terl'a,ce, to a, TI)l:f. Our soma~nl;ic focus is ou t,h(~ 
.su of l,(;ntpora,1 gun(;ra, lized qua,lH;ificrs, based 
ou (I)ra,IL a,u(I l"ra,ncez~ 20()())~ ra,ther tha, n t(ms(; 
a,m.l aSl)e(:l,. As a, rB:.ed 1)y ((.k)l)(~sl:a,l,:(~ a, ud 
Joiles, 1990), \]la,lMlin~ <lUa,utificath)ll i~ one o1" 
the areas in wli\]ch Nil, interfaces ha,vo a, pot(m- 
tia,l a,(Iva, ni,a,ge over l)otll \['orma\] la, ilgtia,ge.s mid 
~ra, l)hica,1 llsor iuterf'aces. 
In (-onq)a,rison with previous work, w(; ~-we 
able 1,o co\]lsidera, bly silnl)li\[3, tim tra,usla, Cion 
melJlo(l, l,'irst, ,sing 'FLG~ ra, thor t;lia.u a, 
f'o.~-l,|AlrO-s|;l'llC(,lll'(~ ibrma,lisin provides ~ much 
s\]nll)le.r method for ('oi).,sl;rucl;ilig s(mla, nl,i(: rol)re- 
sonl;a,tion~. ,g(~(:on(1, using LAIlcl, :+'4 a,ll iul;(;rnl(;- 
dia,te m(;a,,in<e; r(q)r(;s(mta, tion la,nB;ua,B;e yields a, 
lu,(:h lnor(; stra,iglH;\['orwazd trnllsla,tion tha, n us- 
in<e; a restricted TI. 
0.o re.st, l)e:~r in mi.d~ tha,t our iml)IOmo.t~- 
tion in at IJm prototypo stago. Turuing it into ;~ 
l)ractica,l tool would roq.iro considora, blo work, 
as i8 true o1:' most COml)a, ra, ble systems. Futuro 
work includos increa,sod NI, cov(;ra,ge, adding a, 
(lisa, nd)il4,a, tion mod.lo, ha,ndling nonlimd a, ud 
tonipol:a,I a na, i)hora, , allowing uiull,il)l(;-senl;once~ 
(lu(;rlo~, a, nd ~onora,l;\]on of" N\]~ a.ii,~\voi;s \['i'onl the 
resull;s i)resont;(;d I)y tli(; T\])I}. 
1079 

References 

S. Abiteboul, R. Hull, and V. Vianu. 1995. 
Foundations of Databascs. Addison-Wesley. 

J .F. Allen. 1983. Maintaining knowledge about 
temporal intervals. CACM, 26(11):832 843, 
nOV. 

I. Androutsopoulos, G. l). Ritchie, and 
P. Thanisch. 1995. Natural  inter- 
faces to databases- an introduction. Natural 
 Engineering, 1(1) :29-81. 

I. Androutsopoulos. 1996. A principled Frame- 
work for Constructing Natural Language In- 
terfaces to Temporal Databases. Ph.D. thesis, 
University of Edinburgh. 

J. Barwise and R. Cooper. 1981. Generalized 
quantifiers and natural . Linguistics 
and Philosophy, 4:159-219. 

Michael It. Boehlen, Jan Chomicki, Richard T. 
Snodgrass, and David ~\]bma.n. 1996. Query- 
ing TSQL2 databases with temporal logic. 
In lb'oceedings of the 5th h~ternational Con,- 
ference on Extending Database Technology 
(EDBT), Avignon, France. 

M. \[l. Boehlen. 1995. Temporal database 
system implementations. Unpublished 
manuscript, Department of Ma.thema.tics and 
Computer Science, Aalborg University. 

B. Carpenter. 1998. Lectures on Type-Logical 
Semantics. MIT Press. 

B. Carpenter. 1999. Type-logical gramn~ar the- 
orem prover, http://www.colloquial.com/tlg. 

J. Clifford. 1990. Fornzal ,5'emantics and Prag- 
rustics for Natural Language Querying. Cam- 
bridge University Press, Cambridge. C~m~- 
bridge Tracts in Theoretical Computer Sci- 
ence 8. 

A. Copestake and K. Sparck Jones. 1990. Nat- 
ural  interfaces to databases. The 
Knowledge Engineering Review, 5(4):225- 
249. 

R. S. Crouch and S. G. Puhmm. 1993. Time 
and modality in a natural  interface 
to a planning system. Artificial Intclligenee 
63, 63:265-304. 

A. Van Gelder and R.W. Topor. 1991. Safety 
and translation of relational calculus queries. 
ACM Transaction on Database Systems, 
1.6(2) :235-278, June. 

E. W. Hinrichs. 1988. Tense, quantifiers, and 
contexts. Computational Linguistics, 14:3 
\].~:° 

G. Morrill. 11998. Type Logical Grammar: Cat- 
egorial Logic of ,S'ign, s. Kluwer, Dordrecht. 

R. Nelken and N. Francez. 1999. A seman- 
tics tbr temporal questions. In Geert-Jan M. 
Kruijf and Richard T. Oehrle, editors, Pro- 
ceedings of Formal Grammar i999, pages 
1.3:1 142. 

C. R. Perrault and B. J. Grosz. 1988. Natural 
 interfaces. In H. E. Shrobe, editor, 
Exploring Art,ificial Intelligence, pages 133- 
172. Morgan Kaufmann Publishers Inc., San 
Mateo, California. 

C. Pollard and i. A. Sag. 1994. Ilead Driven 
Phrase StructuT~ Grammar. University of 
Chicago Press, Chicago. 

1. Pratt and N. Francez. 1997. On the se- 
mantics of temporal prepositions and preposi- 
tion phrases. Technical Report UMCS-97-4- 
2, University of Manchester, Department of' 
Computer Science. 

I. Pratt and N. Francez. 2000. Temporal prepo- 
sitions and temporal generalized qnantifiers. 
To appear in Linguistics and Philosophy. 

R. T. Snodgrass, editor. 1995. The TSQL2 
TEmporal query . Klnwer, Norwell. 
MA. 

R. T. Snodgrass. 2000. Developin, g Time- 
Oriented Database Applicatio'n,s in SQL. 
Morgan Kauflnani~ Publishers, San Fran- 
cisco, CA. 

M. Steedman. 1997. Temporality. In J. Van 
Benthem and A. Ter Meulen, editors, Hand- 
boo~: of loflic and , pages 895 938. El- 
sevier. 

A. Steiner. 1997. A Generalization Approach to 
Temporal \])ata Models and their hnplemen~ 
ration. F'h.l). thesis, Department lnformatik, 
ETIt Zurich. 

A. Tansel, J. Cliflbrd, S. Gadia, S. Jajodia, 
A. Segev, and R. Snodgrass (eds.). 1993. 
Temporal Databases: Theory, l)esign~ and 
Implementation. Database Systems and Ap- 
plications Series. Benjamin/Cummings, Red- 
wood City, CA. 

1). Toman. 1996. Point vs. Interval-based 
Query Languages for Temporal Databases. In 
Proceedings of the A CM SIGA CT-,S'IGMOD- 
,5'IGAI~,'I ~ POD,5', pages 58-67, Montreal, 
Canada, June. 
