Exploiting Paraphrases in a Question Answering System
Fabio Rinaldi, James Dowdall,
Kaarel Kaljurand, Michael Hess
Institute of Computational Linguistics,
University of Z urich
Winterthurerstrasse 190
CH-8057 Z urich, Switzerland
frinaldi,dowdall,kalju,hessg
@ifi.unizh.ch
Diego Moll a
Centre for Language Technology,
Macquarie University,
Sydney NSW 2109, Australia
fdiegog@ics.mq.edu.au
Abstract
We present a Question Answering system
for technical domains which makes an in-
telligent use of paraphrases to increase the
likelihood of  nding the answer to the user’s
question. The system implements a simple
and e cient logic representation of ques-
tions and answers that maps paraphrases
to the same underlying semantic represen-
tation. Further, paraphrases of technical
terminology are dealt with by a separate
process that detects surface variants.
1 Introduction
The problem of paraphrases conceals a number of
di erent linguistic problems, which in our opinion
need to be treated in separate ways. In fact, para-
phrases can happen at various levels in language. Us-
ing the examples provided in the call for papers for
this workshop, we would like to attempt a simple
classi cation, without any pretense of being exhaus-
tive:
1. Lexical synonymy.
Example: article, paper, publication
2. Morpho-syntactic variants.
a) Oswald killed Kennedy. / Kennedy was killed
by Oswald.
b) Edison invented the light bulb. / Edison’s
invention of the light bulb.
while (a) is purely syntactical (active vs pas-
sive), (b) involves a nominalisation.
3. PP-attachment.
a plant in Alabama / the Alabama plant
4. Comparatives vs superlatives.
be better than anybody else / be the best
5. Subordinate clauses vs separate sentences linked
by anaphoric pronouns.
The tree healed its wounds by growing new bark.
/ The tree healed its wounds. It grew new bark.
6. Inference.
The stapler costs $10. / The price of the stapler
is $10.
Where is Thimphu located? / Thimphu is capi-
tal of what country?
Of course combinations of the di erent types are
possible, e.g. Oswald killed Kennedy / Kennedy
was assassinated by Oswald is a combination of (1)
and (2).
Di erent types of knowledge and di erent linguis-
tic resources are needed to deal with each of the
above types. While type (1) can be dealt with us-
ing a resource such as WordNet (Fellbaum, 1998),
type (2) needs e ective parsing and mapping of syn-
tactic structures into a common deeper structure,
possibly using a repository of nominalisations like
NOMLEX (Meyers et al., 1998). More complex
approaches are needed for the other types, up to
type (6) where generic world knowledge is required,
for instance to know that being a capital of a country
implies being located in that country. 1 Such world
knowledge could be expressed in the form of axioms,
like the following:
(X costs Y) iff (the price of X is Y)
In this paper we focus on the role of paraphrases
in a Question Answering (QA) system targeted at
1Note that the reverse is not true, and therefore this
is not a perfect paraphrase.
technical manuals. Technical documentation is char-
acterised by vast amounts of domain-speci c termi-
nology, which needs to be exploited for providing in-
telligent access to the information contained in the
manuals (Rinaldi et al., 2002b). The approach taken
by QA systems is to allow a user to ask a query (for-
mulated in natural language) and have the system
search a background collection of documents in order
to locate an answer. The  eld of Question Answer-
ing has  ourished in recent years2, in part, due to
the QA track of the TREC competitions (Voorhees
and Harman, 2001). These competitions evaluate
systems over a common data set allowing develop-
ers to benchmark performance in relation to other
competitors.
It is a common assumption that technical termi-
nology is subject to strict controls and cannot vary
within a given editing process. However this assump-
tion proves all too often to be incorrect. Unless edi-
tors are making use of a terminology control system
that forces them to use a speci c version of a term,
they will naturally tend to use various paraphrases
to refer to the intended domain concept. Besides in
a query a user could use an arbitrary paraphrases of
the target term, which might happen to be one of
those used in the manual itself or might happen to
be a novel one.
We describe some potential solutions to this prob-
lem, taking our Question Answering system as an ex-
ample. We show which bene ts our approach based
on paraphrases bring to the system. So far two dif-
ferent domains have been targeted by the system.
An initial application aims at answering questions
about the Unix man pages (Moll a et al., 2000a; Moll a
et al., 2000b). A more complex application targets
the Aircraft Maintenance Manual (AMM) of the Air-
bus A320 (Rinaldi et al., 2002b). Recently we have
started new work, using the Linux HOWTOs as a
new target domain.
In dealing with these domains we have identi ed
two major obstacles for a QA system, which we can
summarise as follows:
 The Parsing Problem
 The Paraphrase Problem
The Parsing Problem consists in the increased
di culty of parsing text in a technical domain due to
domain-speci c sublanguage. Various types of multi
word expressions characterise these domains, in par-
ticular referring to speci c concepts like tools, parts
or procedures. These multi word expressions might
2Although early work in AI already touched upon the
topic, e.g. (Woods, 1977).
include lexical items which are either unknown to
a generic lexicon (e.g. coax cable) or have a spe-
ci c meaning unique to this domain. Abbreviations
and acronyms are another common source of incon-
sistencies. In such cases the parser might either
fail to identify the compound as a phrase and con-
sequently fail to parse the sentence including such
items. Alternatively the parser might attempt to
‘guess’ their lexical category (in the set of open class
categories), leading to an exponential growth of the
number of possible syntactic parses. Not only the in-
ternal structure of the compound can be multi-way
ambiguous, even the boundaries of the compounds
might be di cult to detect and the parsers might
try odd combinations of the tokens belonging to the
compounds with neighbouring tokens.
The Paraphrase Problem resides in the imper-
fect knowledge of users of the systems, who can-
not be expected to be completely familiar with the
domain terminology. Even experienced users, who
know very well the domain, might not remember the
exact wording of a compound and use a paraphrase
to refer to the underlying domain concept. Besides
even in the manual itself, unless the editors have been
forced to use some strict terminology control system,
various paraphrases of the same compound will ap-
pear, and they need to be identi ed as co-referent.
However, it is not enough to identify all paraphrases
within the manual, novel paraphrases might be cre-
ated by the users each time they query the system.
In the rest of this paper we describe  rst our Ques-
tion Answering System (in Section 2) and brie y
show how we solved the  rst of the two problems
described above. Then, in Section 3 we show in de-
tail how the system is capable of coping with the
Paraphrase Problem. Finally in Section 4 we discuss
some related work.
2 A Question Answering System for
Technical Domains
Over the past few years our research group has devel-
oped an Answer Extraction system (ExtrAns) that
works by transforming documents and queries into a
semantic representation called Minimal Logical Form
(MLF) (Moll a et al., 2000a) and derives the answers
by logical proof from the documents. A full linguis-
tic (syntactic and semantic) analysis, complete with
lexical alternations (synonyms and hyponyms) is per-
formed. While documents are processed in an o -line
stage, the query is processed on-line.
Two real world applications have so far been im-
plemented with the same underlying technology. The
original ExtrAns system (Moll a et al., 2000b) is used
///// a.d electrical coax cable.n4 connects.v062 the.d external antenna.n1 to.o the.d ANT connection.n1 /////
-Wd
 Dsu  Ss
-MVp
-Os
 Ds
-Js
 Ds
RW
Figure 1: An Example of LG Output
to extract answers to arbitrary user queries over the
Unix documentation  les (\man pages"). A set of
500+ unedited man pages has been used for this ap-
plication. An on-line demo of ExtrAns can be found
at the project web page.3
 Knowledge 
Base
Document
Linguistic
Analysis
Term
processing
Figure 2: O -line
Processing of Docu-
ments
More recently we tackled
a di erent domain, the Air-
plane Maintenance Manu-
als (AMM) of the Air-
bus A320 (Rinaldi et al.,
2002b), which o ered the
additional challenges of an
SGML-based format and a
much larger size (120MB).4
Despite being developed
initially for a speci c do-
main, ExtrAns has demon-
strated a high level of do-
main independence.
As we work on relatively
small volumes of data we
can a ord to process (in
an o -line stage) all the
documents in our collection
rather than just a few se-
lected paragraphs (see Fig-
ure 2). Clearly in some sit-
uations (e.g. processing in-
coming news) such an ap-
proach might not be fea-
sible and paragraph index-
ing techniques would need
to be used. Our current ap-
proach is particularly tar-
geted to small and medium sized collections.
In an initial phase all multi-word expressions
from the domain are collected and structured in
an external resource, which we will refer to as the
TermBase (Rinaldi et al., 2003; Dowdall et al., 2003).
The document sentences (and user queries) are syn-
tactically processed with the Link Grammar (LG)
parser (Sleator and Temperley, 1993) which uses a
3http://www.ifi.unizh.ch/cl/extrans/
4Still considerably smaller than the size of the docu-
ment collections used for TREC
grammar with a wide coverage of English and has
a robust treatment of ungrammatical sentences and
unknown words. The multi-word terms from the the-
saurus are identi ed and passed to the parser as sin-
gle tokens. This prevents (futile) analysis of the in-
ternal structure of terms (see Figure 1), simplifying
parsing by 46%. This solves the  rst of the problems
that we have identi ed in the introduction (\The
Parsing Problem").
In later stages of processing, a corpus-based ap-
proach (Brill and Resnik, 1994) is used to deal with
ambiguities that cannot be solved with syntactic in-
formation only, in particular attachments of preposi-
tional phrases, gerunds and in nitive constructions.
ExtrAns adopts an anaphora resolution algorithm
(Moll a et al., 2003) that is based on Lappin and Le-
ass’ approach (Lappin and Leass, 1994). The original
algorithm, which was applied to the syntactic struc-
tures generated by McCord’s Slot Grammar (Mc-
Cord et al., 1992), has been ported to the output of
Link Grammar. So far the resolution is restricted to
sentence-internal pronouns but the same algorithm
can be applied to sentence-external pronouns too.
A lexicon of nominalisations based on NOMLEX
(Meyers et al., 1998) is used for the most important
cases. The main problem here is that the semantic
relationship between the base words (mostly, but not
exclusively, verbs) and the derived words (mostly,
but not exclusively, nouns) is not su ciently sys-
tematic to allow a derivation lexicon to be compiled
automatically. Only in relatively rare cases is the
relationship as simple as with to edit <a text> $
editor of <a text> / <text> editor, as the e ort
that went into building resources such as NOMLEX
also shows.
User queries are processed on-line and converted
into MLFs (possibly expanded by synonyms) and
proved by refutation over the document knowledge
base (see Figure 3). Pointers to the original text at-
tached to the retrieved logical forms allow the system
to identify and highlight those words in the retrieved
sentence that contribute most to that particular an-
swer. When the user clicks on one of the answers
provided, the corresponding document will be dis-
played with the relevant passages highlighted.
 Knowledge 
Base
ANSWERSQuery
Document
Linguistic
Analysis
Paraphrase
Identification
Figure 3: On-line Processing of Queries
The meaning of the documents and of the queries
produced by ExtrAns is expressed by means of Mini-
mal Logical Forms (MLFs). The MLFs are designed
so that they can be found for any sentence (using
robust approaches to treat very complex or ungram-
matical sentences), and they are optimized for NLP
tasks that involve the semantic comparison of sen-
tences, such as Answer Extraction.
The expressivity of the MLFs is minimal in the
sense that the main syntactic dependencies between
the words are used to express verb-argument rela-
tions, and modi er and adjunct relations. However,
complex quanti cation, tense and aspect, temporal
relations, plurality, and modality are not expressed.
One of the e ects of this kind of underspeci cation
is that several natural language queries, although
slightly di erent in meaning, produce the same logi-
cal form.
The main feature of the MLFs is the use of rei -
cation (the expression of abstract concepts as con-
crete objects) to achieve  at expressions (Moll a et
al., 2000b). The MLFs are expressed as conjunc-
tions of predicates with all the variables existentially
bound with wide scope. For example, the MLF of
the sentence \cp will quickly copy the  les" is:
(1) holds(e4), object(cp,o1,[x1]),
object(s command,o2,[x1]),
evt(s copy,e4,[x1,x6]),
object(s file,o3,[x6]),
prop(quickly,p3,[e4]).
In other words, there is an entity x1 which rep-
resents an object of type cp and of type command,
there is an entity x6 (a  le), there is an entity e4,
which represents a copying event where the  rst ar-
gument is x1 and the second argument is x6, there
is an entity p3 which states that e4 is done quickly,
and the event e4, that is, the copying, holds. The
entities o1, o2, o3, e4, and p3 are the result of rei -
cation. The rei cation of the event, e4, has been used
to express that the event is done quickly. The other
entities are not used in this MLF, but other more
complex sentences may need to refer to the rei ca-
tion of properties (adjective-modifying adverbs) or
object predicates (non-intersective adjectives such as
the alleged suspect).
ExtrAns  nds the answers to the questions by
forming the MLFs of the questions and then run-
ning Prolog’s default resolution mechanism to  nd
those MLFs that can prove the question. When no
direct proof for the user query is found, the system
is capable of relaxing the proof criteria in a stepwise
manner. First, hyponyms of the query terms will be
added as disjunctions in the logical form of the ques-
tion, thus making it more general but still logically
correct. If that fails, the system will attempt approx-
imate matching, in which the sentence (or sentences)
with the highest overlap of predicates with the query
is retrieved. The (partially) matching sentences are
scored and the best  ts are returned. In the case
that this method  nds too many answers because
the overlap is too low, the system will attempt key-
word matching, in which syntactic criteria are aban-
doned and only information about word classes is
used. This last step corresponds approximately to a
traditional passage-retrieval methodology with con-
sideration of the POS tags.
3 Dealing with Paraphrases
The system is capable of dealing with paraphrases
at two di erent levels. On the phrase level, di er-
ent surface realizations (terms) which refer to the
same domain concept will be mapped into a com-
mon identi er (synset identi er). On the sentence
level, paraphrases which involve a (simple) syntactic
transformation will be dealt with by mapping them
into the same logical form. In this section we will
describe these two approaches and discuss ways to
cope with complex types of parapharases.
3.1 Identifying Terminological Paraphrases
During the construction of the MLFs, thesaurus
terms are replaced by their synset identi ers. This
results in an implicit ‘terminological normalization’
for the domain. The bene t to the QA process is
an assurance that a query and answer need not in-
volve exactly the same surface realization of a term.
Utilizing the synsets in the semantic representation
means that when the query includes a term, ExtrAns
returns sentences that logically answer the query, in-
Fastr
Term
Extraction
Hyponymy
Thesaurus ExtrAns
Document
Figure 4: Term Processing
volving any known paraphrase of that term.
For example, the logical form of the query Where
are the stowage compartments installed? is trans-
lated internally into the Horn query (2).
(2) evt(install,A,[B,C]),
object(D,E,[B]),
object(s stowage compartment,G,[C])
This means that a term (belonging to the same
synset as stowage compartment) is involved in an in-
stall event with an anonymous object. If there is
an MLF from the document that can match exam-
ple (2), then it is selected as a candidate answer and
the sentence it originates from is shown to the user.
The process of terminological variation is well
investigated (Ibekwe-SanJuan and Dubois, 2002;
Daille et al., 1996; Ibekwe-Sanjuan, 1998). The
primary focus has been to use linguistically based
variation to expand existing term sets through cor-
pus investigation or to produce domain representa-
tions. A subset of such variations identi es terms
which are strictly synonymous. ExtrAns gathers
these morpho-syntactic variations into synsets. The
sets are augmented with terms exhibiting three
weaker synonymy relations described by Hamon &
Nazarenko (2001). These synsets are organized into
a hyponymy (isa) hierarchy, a small example of which
can be seen in Figure 5. Figure 4 shows a schematic
representation of this process.
The  rst stage is to normalize any terms that con-
tain punctuation by creating a punctuation free ver-
sion and recording the fact that that the two are
strictly synonymous. Further processing is involved
in terms containing brackets to determine if the
bracketed token is an acronym or simply optional. In
the former case an acronym-free term is created and
the acronym is stored as a synonym of the remain-
ing tokens which contain it as a regular expression.
So evac is synonymous with evacuation and ohsc is
synonymous with overhead stowage compartment. In
cases such as emergency (hard landings) the brack-
eted tokens can not be interpreted as acronyms and
so are not removed.
The synonymy relations are identi ed using the
terminology tool Fastr (Jacquemin, 2001). Every to-
ken of each term is associated with its part-of-speech,
its morphological root, and its synonyms. Phrasal
rules represent the manner in which tokens combine
to form multi-token terms, and feature-value pairs
carry the token speci c information. Metarules li-
cense the relation between two terms by constrain-
ing their phrase structures in conjunction with the
morphological and semantic information on the indi-
vidual tokens.
The metarules can identify simple paraphrases
that result from morpho-syntactic variation (cargo
compartment door  ! doors of the cargo compart-
ment), terms with synonymous heads (electrical ca-
ble  ! electrical line), terms with synonymous mod-
i ers (fastener strip  ! attachment strip) and both
(functional test  ! operational check). For a de-
scription of the frequency and range of types of vari-
ation present in the AMM see Rinaldi et al. (2002a).
3.2 Identifying Syntactic Paraphrases
An important e ect of using a simpli ed semantic-
based representation such as the Minimal Logical
Forms is that various types of syntactic variations
are automatically captured by a common representa-
tion. This ensures that many potential paraphrases
in a user query can map to the same answer into the
manual.
For example the question shown in Figure 6 can
be answered thanks to the combination of two fac-
tors. On the lexical level ExtrAns knows that APU
is an abbreviation of Auxiliary Power Unit, while on
the syntactic level the active and passive voices (sup-
plies vs supplied with) map into the same underlying
representation (the same MLF).
Another type of paraphrase which can be detected
at this level is the kind that was classi ed as type (3)
in the introduction. For example the question: Is
the sensor connected to the APU ECB?, can locate
the answer This sensor is connected to the Elec-
tronic Control Box (ECB) of the APU. This has been
achieved by introducing meaning postulates that op-
erate at the level of the MLFs (such as \any predicate
that a ects an object will also a ect the of -modi ers
of that object").
3.3 Weaker Types of Paraphrases
When the thesaurus de nition of terminological syn-
onymy fails to locate an answer from the docu-
ment collection, ExtrAns explores weaker types of
paraphrases, where the equivalence between the two
terms might not be complete.
TERM
doors of the cargo compartment
cargo compartment door
cargo comparment doors
cargo-compartment door
emergency ( hard landings )
emergency hard landings
emergency hard landing
emergency evacuation (evac)
emergency evacuation
evacuation
evac
electrical cable
electrical line
fastner strip
attachment strip
functional test
operational check
door functional teststowage compartment
overhead stowage compartment
OHSC
1
2
3
5
6
7
10
9
8
11
Figure 5: A Sample of the TermBase
Figure 6: Active vs Passive Voice
First, ExtrAns makes use of the hyponymy rela-
tions, which can be considered as sort of unidirec-
tional paraphrases. Instead of looking for synset
members, the query is reformulated to included hy-
ponyms and hyperonyms of the terms:
(3) (object(s stowage compartment,A,[B]);
object(s overhead stowage compartment,A,[B])),
evt(install,C,[D,B]),
object(E,F,[D|G])
Now the alternative objects are in a logical OR rela-
tion. This query  nds the answer in Figure 7 (where
stowage compartment is a hyperonym of overhead
stowage compartment).
We have implemented a very simple ad-hoc algo-
rithm to determine lexical hyponymy between terms.
Term A is a hyponym of term B if (i) A has more to-
kens than B, (ii) all the tokens of B are present in A,
and (iii) both terms have the same head. There are
three provisions. First, ignore terms with dashes and
brackets as cargo compartment is not a hyponym of
cargo - compartment and this relation (synonymy) is
already known from the normalisation process. Sec-
ond, compare lemmatised versions of the terms to
capture that stowage compartment is a hyperonym
of overhead stowage compartments. Finally, the head
of a term is the rightmost non-symbol token (i.e. a
word) which can be determined from the part-of-
speech tags. This hyponymy relation is compara-
ble to the insertion variations de ned by Daille et
al. (1996).
The expressivity of the MLF can further be ex-
panded through the use of meaning postulates of the
type: \If x is installed in y, then x is in y". This
ensures that the query Where are the equipment and
furnishings? extracts the answer The equipment and
furnishings are installed in the cockpit.
4 Related Work
The importance of detecting paraphrasing in Ques-
tion Answering has been shown dramatically in
TREC9 by the Falcon system (Harabagiu et al.,
2001), which made use of an ad-hoc module capable
of caching answers and detecting question similar-
ity. As in that particular evaluation the organisers
deliberately used a set of paraphrases of the same
questions, such approach certainly helped in boost-
ing the performance of the system. In an environ-
ment where the same question (in di erent formula-
tions) is likely to be repeated a number of times, a
module capable of detecting paraphrases can signif-
icantly improve the performance of a Question An-
Figure 7: Overhead stowage compartment is a Hyponym of Stowage compartment
swering system.
Another example of application of paraphrases for
Question Answering is given in (Murata and Isahara,
2001), which further argues for the importance of
paraphrases for other applications such Summarisa-
tion, error correction and speech generation.
Our approach for the acquisition of terminological
paraphrases might have some points in common with
the approach described in (Terada and Tokunaga,
2001). The motivation that they bring forward for
the necessity of identifying abbreviations is related to
the problem that we have called \the Parsing Prob-
lem".
A very di erent approach to paraphrases is taken
in (Takahashi et al., 2001) where they formulate the
problem as a special case of Machine Translation,
where the source and target language are the same
but special rules, based on di erent parameters, li-
cense di erent types of surface realizations.
Hamon & Nazarenko (2001) explore the termino-
logical needs of consulting systems. This type of IR
guides the user in query/keyword expansion or pro-
poses various levels of access into the document base
on the original query. A method of generating three
types of synonymy relations is investigated using gen-
eral language and domain speci c dictionaries.
5 Conclusion
Automatic recognition of paraphrases is an e ec-
tive technique to ease the information access bur-
den in a technical domain. We have presented some
techniques that we have adopted in a Question An-
swering system for dealing with paraphrases. These
techniques range from the detection of lexical para-
phrases and terminology variants, to the use of a
simpli ed logical form that provides the same repre-
sentation for morpho-syntactic paraphrases, and the
use of meaning postulates for paraphrases that re-
quire inferences.

References
Eric Brill and Philip Resnik. 1994. A rule-based
approach to prepositional phrase attachment dis-
ambiguation. In Proc. COLING ’94, volume 2,
pages 998{1004, Kyoto, Japan.
Beatrice Daille, Benot Habert, Christian Jacquemin,
and Jean Royaut e. 1996. Empirical observation of
term variations and principles for their description.
Terminology, 3(2):197{258.
James Dowdall, Fabio Rinaldi, Fidelia Ibekwe-
SanJuan, and Eric SanJuan. 2003. Complex
structuring of term variants for Question Answer-
ing. In Proc. ACL-2003 Workshop on Multiword
Expressions, Sapporo, Japan.
Christiane Fellbaum 1998. WordNet: an electronic
lexical database. MIT Press, Cambridge, MA.
Thierry Hamon and Adeline Nazarenko. 2001. De-
tection of synonymy links between terms: Experi-
ment and results. In Didier Bourigault, Christian
Jacquemin, and Marie-Claude L’Homme, editors,
Recent Advances in Computational Terminology,
pages 185{208. John Benjamins Publishing Com-
pany.
Sanda Harabagiu, Dan Moldovan, Marius Pa sca,
Rada Mihalcea, Mihai Surdeanu, Razvan Bunescu,
Roxana G^ rju, Vasile Rus, and Paul Morarescu.
2001. Falcon: Boosting knowledge for answer
engines. In Voorhees and Harman (Voorhees and
Harman, 2001).
Fidelia Ibekwe-SanJuan and Cyrille Dubois. 2002.
Can Syntactic Variations Highlight Semantic
Links Between Domain Topics? In Proceedings
of the 6th International Conference on Terminol-
ogy and Knowledge Engineering (TKE02), pages
57{64, Nancy, August.
Fidelia Ibekwe-Sanjuan. 1998. Terminological Vari-
ation, a Means of Identifying Research Topics from
Texts. In Proceedings of COLING-ACL, pages
571{577, Quebec,Canada, August.
Christian Jacquemin. 2001. Spotting and Discover-
ing Terms through Natural Language Processing.
MIT Press.
Shalom Lappin and Herbert J. Leass. 1994. An algo-
rithm for pronominal anaphora resolution. Com-
putational Linguistics, 20(4):535{561.
Michael McCord, Arendse Bernth, Shalom Lap-
pin, and Wlodek Zadrozny. 1992. Natural lan-
guage processing within a slot grammar frame-
work. International Journal on Arti cial Intelli-
gence Tools, 1(2):229{277.
Adam Meyers, Catherine Macleod, Roman Yangar-
ber, Ralph Grishman, Leslie Barrett, and Ruth
Reeves. 1998. Using NOMLEX to produce
nominalization patterns for information extrac-
tion. In Proceedings: the Computational Treat-
ment of Nominals, Montreal, Canada, (Coling-
ACL98 workshop), August.
Diego Moll a, Gerold Schneider, Rolf Schwitter, and
Michael Hess. 2000a. Answer Extraction using
a Dependency Grammar in ExtrAns. Traitement
Automatique de Langues (T.A.L.), Special Issue
on Dependency Grammar, 41(1):127{156.
Diego Moll a, Rolf Schwitter, Michael Hess, and
Rachel Fournier. 2000b. Extrans, an answer ex-
traction system. T.A.L. special issue on Informa-
tion Retrieval oriented Natural Language Process-
ing.
Diego Moll a, Rolf Schwitter, Fabio Rinaldi, James
Dowdall, and Michael Hess. 2003. Anaphora res-
olution in ExtrAns. In Proceedings of the Interna-
tional Symposium on Reference Resolution and Its
Applications to Question Answering and Summa-
rization, 23{25 June, Venice, Italy.
Masaki Murata and Hitoshi Isahara. 2001. Univer-
sal model for paraphrasing - using transformation
based on a de ned criteria. In Proceedings of the
NLPRS2001 Workshop on Automatic Paraphras-
ing: Theories and Applications.
Fabio Rinaldi, James Dowdall, Michael Hess, Kaarel
Kaljurand, Mare Koit, Kadri Vider, and Neeme
Kahusk. 2002a. Terminology as Knowledge in An-
swer Extraction. In Proceedings of the 6th Interna-
tional Conference on Terminology and Knowledge
Engineering (TKE02), pages 107{113, Nancy, 28{
30 August.
Fabio Rinaldi, James Dowdall, Michael Hess, Diego
Moll a, and Rolf Schwitter. 2002b. Towards An-
swer Extraction: an application to Technical Do-
mains. In ECAI2002, European Conference on Ar-
ti cial Intelligence, Lyon, 21{26 July.
Fabio Rinaldi, James Dowdall, Michael Hess, Kaarel
Kaljurand, and Magnus Karlsson. 2003. The Role
of Technical Terminology in Question Answering.
In Proceedings of TIA-2003, Terminologie et In-
telligence Arti cielle, Strasbourg, April.
Daniel D. Sleator and Davy Temperley. 1993. Pars-
ing English with a link grammar. In Proc. Third
International Workshop on Parsing Technologies,
pages 277{292.
Tetsuro Takahashi, Tomoya Iwakura, Ryu Iida, and
Kentaro Inui. 2001. Kura: A revision-based
lexico-structural paraphrasing engine. In Proceed-
ings of the NLPRS2001 Workshop on Automatic
Paraphrasing: Theories and Applications.
Akira Terada and Takenobu Tokunaga. 2001. Au-
tomatic disabbreviation by using context informa-
tion. In Proceedings of the NLPRS2001 Workshop
on Automatic Paraphrasing: Theories and Appli-
cations.
Ellen M. Voorhees and Donna Harman, editors.
2001. Proceedings of the Ninth Text REtrieval
Conference (TREC-9), Gaithersburg, Maryland,
November 13-16, 2000.
W.A. Woods. 1977. Lunar rocks in natural English:
Explorations in Natural Language Question An-
swering. In A. Zampolli, editor, Linguistic Struc-
tures Processing, volume 5 of Fundamental Studies
in Computer Science, pages 521{569. North Hol-
land.
