Automatic Construction of Nominal Case Frames and
its Application to Indirect Anaphora Resolution
Ryohei Sasano, Daisuke Kawahara and Sadao Kurohashi
Graduate School of Information Science and Technology, University of Tokyo
{ryohei,kawahara,kuro}@kc.t.u-tokyo.ac.jp
Abstract
This paper proposes a method to auto-
matically construct Japanese nominal case
frames. The point of our method is the in-
tegrated use of a dictionary and example
phrases from large corpora. To examine the
practical usefulness of the constructed nom-
inal case frames, we also built a system of
indirect anaphora resolution based on the
case frames. The constructed case frames
were evaluated by hand, and were confirmed
to be good quality. Experimental results of
indirect anaphora resolution also indicated
the effectiveness of our approach.
1 Introduction
What is represented in a text has originally
a network structure, in which several concepts
have tight relations with each other. However,
because of the linear constraint of texts, most
of them disappear in the normal form of texts.
Automatic reproduction of such relations can be
regarded as the first step of “text understand-
ing”, and surely benefits NLP applications such
as machine translation, automatic abstraction,
and question answering.
One of such latent relationship is indirect
anaphora, functional anaphora, or bridging ref-
erence, such as the following examples.
(1) I bought a ticket. The price was 20 dollars.
(2) There was a house. The roof was white.
Here, “the price” means “the price of a ticket”
and “the roof” means “the roof of a house.”
Most nouns have their indispensable or req-
uisite entities: “price” is a price of some goods
or service, “roof” is a roof of some building,
“coach” is a coach of some sport, and “virus”
is a virus causing some disease. The relation
between a noun and its indispensable entity is
parallel to that between a verb and its argu-
ments or obligatory cases. In this paper, we call
indispensable entities of nouns obligatory cases.
Indirect anaphora resolution needs a compre-
hensive information or dictionary of obligatory
cases of nouns.
In case of verbs, syntactic structures such as
subject/object/PP in English or case markers
such as ga, wo, ni in Japanese can be utilized
as a strong clue to distinguish several obliga-
tory cases and adjuncts (and adverbs), which
makes it feasible to construct case frames from
large corpora automatically (Briscoe and Car-
roll, 1997; Kawahara and Kurohashi, 2002).
(Kawahara and Kurohashi, 2004) then utilized
the automatically constructed case frames to
Japanese zero pronoun resolution.
On the other hand, in case of nouns, obliga-
tory cases of noun Nh appear, in most cases, in
the single form of noun phrase “Nh of Nm” in
English, or “Nm no Nh” in Japanese. This sin-
gle form can express several obligatory cases,
and furthermore optional cases, for example,
“rugby no coach” (obligatory case concerning
what sport), “club no coach” (obligatory case
concerning which institution), and “kyonen ‘last
year’ no coach” (optional case). Therefore, the
key issue to construct nominal case frames is to
analyze “Nh of Nm” or “Nm no Nh” phrases to
distinguish obligatory case examples and others.
Work which addressed indirect anaphora in
English texts so far restricts relationships to a
small, relatively well-defined set, mainly part-of
relation like the above example (2), and utilized
hand-crafted heuristic rules or hand-crafted lex-
ical knowledge such as WordNet (Hahn et al.,
1996; Vieira and Poesio, 2000; Strube and
Hahn, 1999). (Poesio et al., 2002) proposed
a method of acquiring lexical knowledge from
“Nh of Nm” phrases, but again concentrated on
part-of relation.
In case of Japanese text analysis, (Murata et
al., 1999) proposed a method of utilizing “Nm
no Nh” phrases for indirect anaphora resolution
of diverse relationships. However, they basically
used all “Nm no Nh” phrases from corpora, just
excluding some pre-fixed stop words. They con-
fessed that an accurate analysis of “Nm no Nh”
phrases is necessary for the further improve-
ment of indirect anaphora resolution.
As a response to these problems and follow-
ing the work in (Kurohashi and Sakai, 1999), we
propose a method to construct Japanese nom-
inal case frames from large corpora, based on
an accurate analysis of “Nm no Nh” phrases
using an ordinary dictionary and a thesaurus.
To examine the practical usefulness of the con-
structed nominal case frames, we also built a
system of indirect anaphora resolution based on
the case frames.
2 Semantic Feature Dictionary
First of all, we briefly introduce NTT Seman-
tic Feature Dictionary employed in this paper.
NTT Semantic Feature Dictionary consists of a
semantic feature tree, whose 3,000 nodes are se-
mantic features, and a nominal dictionary con-
taining about 300,000 nouns, each of which is
given one or more appropriate semantic fea-
tures.
The main purpose of using this dictionary is
to calculate the similarity between two words.
Suppose the word x and y have a semantic fea-
ture sx and sy, respectively, their depth is dx
and dy in the semantic tree, and the depth of
their lowest (most specific) common node is dc,
the similarity between x and y, sim(x,y), is cal-
culated as follows:
sim(x,y) = (dc ×2)/(dx +dy).
If sx and sy are the same, the similarity is 1.0,
the maximum score based on this criteria.
We also use this dictionary to specify seman-
tic category of words, such as human, time and
place.
3 Semantic Analysis of Japanese
Noun Phrases Nm no Nh
In many cases, obligatory cases of nouns are
described in an ordinary dictionary for human
being. For example, a Japanese dictionary for
children, Reikai Shougaku Kokugojiten, or RSK
(Tajika, 1997), gives the definitions of the word
coach and virus as follows1:
coach a person who teaches technique in some
sport
virus a living thing even smaller than bacte-
ria which causes infectious disease like in-
fluenza
1Although our method handles Japanese noun
phrases by using Japanese definition sentences, in this
paper we use their English translations for the explana-
tion. In some sense, the essential point of our method is
language-independent.
Based on such an observation, (Kurohashi
and Sakai, 1999) proposed a semantic analy-
sis method of “Nm no Nh”, consisting of the
two modules: dictionary-based analysis (abbre-
viated to DBA hereafter) and semantic feature-
based analysis (abbreviated to SBA hereafter).
This section briefly introduces their method.
3.1 Dictionary-based analysis
Obligatory case information of nouns in an ordi-
nary dictionary can be utilized to solve the dif-
ficult problem in the semantic analysis of “Nm
no Nh” phrases. In other words, we can say the
problem disappears.
For example, “rugby no coach” can be inter-
preted by the definition of coach as follows: the
dictionary describes that the noun coach has an
obligatory case sport, and the phrase “rugby no
coach” specifies that the sport is rugby. That is,
the interpretation of the phrase can be regarded
as matching rugby in the phrase to some sport
in the coach definition. “Kaze ‘cold’ no virus”
is also easily interpreted based on the definition
of virus, linking kaze ‘cold’ to infectious disease.
Dictionary-based analysis (DBA) tries to find
a correspondence between Nm and an obliga-
tory case of Nh by utilizing RSK and NTT Se-
mantic Feature Dictionary, by the following pro-
cess:
1. Look up Nh in RSK and obtain the defini-
tion sentences of Nh.
2. For each word w in the definition sentences
other than the genus words, do the follow-
ing steps:
2.1. When w is a noun which shows an
obligatory case explicitly, like kotog-
ara ‘thing’, monogoto ‘matter’, nanika
‘something’, and Nm does not have a
semantic feature of human or time,
give 0.8 to their correspondence2.
2.2. When w is other noun, calculate the
similarity between Nm and w by us-
ing NTT Semantic Feature Dictionary,
and give the similarity score to their
correspondence.
3. Finally, if the best correspondence score is
0.75 or more, DBA outputs the best corre-
spondence, which can be an obligatory case
of the input; if not, DBA outputs nothing.
2For the present, parameters in the algorithm were
given empirically, not optimized by a learning method.
Table 1: Examples of rules for semantic feature-based analysis.
1. Nm:human, Nh:relative → <obligatory case(relative)> e.g. kare ‘he’ no oba ‘aunt’
2. Nm:human, Nh:human → <modification(apposition)> e.g. gakusei ‘student’ no kare ‘he’
3. Nm:organization, Nh:human → <belonging> e.g. gakkou ‘school’ no seito ‘student’
4. Nm:agent, Nh:event → <agent> e.g. watashi ‘I’ no chousa ‘study’
5. Nm:material, Nh:concrete → <modification(material)> e.g. ki ‘wood’ no hako ‘box’
6. Nm:time, Nh:∗ → <time> e.g. aki ‘autumn’ no hatake ‘field’
7. Nm:color, quantity, or figure, Nh:∗ → <modification> e.g. gray no seihuku ‘uniform’
8. Nm:∗, Nh:quantity → <obligatory case(attribute)> e.g. hei ‘wall’ no takasa ‘height’
9. Nm:∗, Nh:position → <obligatory case(position)> e.g. tsukue ‘desk’ no migi ‘right’
10. Nm:agent, Nh:∗ → <possession> e.g. watashi ‘I’ no kuruma ‘car’
11. Nm:place or position, Nh:∗ → <place> e.g. Kyoto no mise ‘store’
‘∗’ meets any noun.
In case of the phrase “rugby no coach”, “tech-
nique” and “sport” in the definition sentences
are checked: the similarity between “technique”
and “rugby” is calculated to be 0.21, and the
similarity between “sport” and “rugby” is cal-
culated to be 1.0. Therefore, DBA outputs
“sport”.
3.2 Semantic feature-based analysis
Since diverse relations in “Nm no Nh” are han-
dled by DBA, the remaining relations can be
detected by simple rules checking the semantic
features of Nm and/or Nh.
Table 1 shows examples of the rules. For ex-
ample, the rule 1 means that if Nm has a seman-
tic feature human and Nh relative, <obliga-
tory case> relation is assigned to the phrase.
The rules 1, 2, 8 and 9 are for certain oblig-
atory cases. We use these rules because these
relationscanbeanalyzedmoreaccuratelybyus-
ing explicit semantic features, rather than based
on a dictionary.
3.3 Integration of two analyses
Usually, either DBA or SBA outputs some re-
lation. When both DBA and SBA output some
relations, the results are integrated (basically, if
DBA correspondence score is higher than 0.8,
DBA result is selected; if not, SBA result is se-
lected). In rare cases, neither analysis outputs
any relations, which means analysis failure.
4 Automatic Construction of
Nominal Case Frames
4.1 Collection and analysis of Nm no Nh
Syntactically unambiguous noun phrases “Nm
no Nh” are collected from the automatic parse
results of large corpora, and they are analyzed
using the method described in the previous sec-
tion.
Table 2: Preliminary case frames for hisashi
‘eaves/visor’.
DBA result
1. a roof that stick out above the window of
a house.
[house] hall:2, balcony:1, building:1, ···
[window] window:2, ceiling:1, counter:1, ···
2. the fore piece of a cap.
[cap] cap:8, helmet:1, ···
SBA result
<place> parking:3, store:3, shop:2, ···
<mod.> concrete:1, metal:1, silver:1, ···
No semantic analysis result
<other> part:1, light:1, phone:1, ···
By just collecting the analysis results of each
head word Nh, we can obtain its preliminary
case frames. Table 2 shows preliminary case
frames for hisashi ‘eaves/visor’. The upper part
of the table shows the results by DBA. The line
starting with “[house]” denotes a group of anal-
ysis results corresponding to the word “house”
in the first definition sentence. For example,
“hall no hisashi” occurs twice in the corpora,
and they were analyzed by DBA to correspond
to “house.”
The middle part of the table shows the results
by SBA. Noun phrases that have no semantic
analysisresult(analysisfailure)arebundledand
named <other>, as shown in the last part of the
table.
A case frame should be constructed for each
meaning (definition) of Nh, and groups start-
ing with “[...]” or “<...>” in Table 2 are possi-
ble case slots. The problem is how to arrange
the analysis results of DBA and SBA and how
to distinguish obligatory cases and others. The
following sections explain how to handle these
problems.
Table 3: Threshold to select obligatory slots.
type of case slots threshold of probability
analyzed by DBA 0.5% (1/200)
<obligatory case> 2.5% (1/40)
<belonging> 2.5% (1/40)
<possessive> 5% (1/20)
<agent> 5% (1/20)
<place> 5% (1/20)
<other> 10% (1/10)
<modification> not used
<time> not used
Probability = (# of Nm no Nh) / (# of Nh)
4.2 Case slot clustering
One obligatory case might be separated in pre-
liminary case frames, since the definition sen-
tence is sometimes too specific or too detailed.
For example, in the case of hisashi ‘eaves/visor’
in Table 2, [house], [window], and <place>
have very similar examples that mean building
or part of building. Therefore, case slots are
merged if similarity of two case slots is more
than 0.5 (case slots in different definition sen-
tences are not merged in any case). Similarity
of two case slots is the average of top 25% sim-
ilarities of all possible pairs of examples.
In the case of Table 2, the similarity between
[house] and [window] is 0.80, and that between
[house] and <place> is 0.67, so that these three
case slots are merged into one case slot.
4.3 Obligatory case selection
Preliminary case frames contain both obliga-
tory cases and optional cases for the head word.
Since we can expect that an obligatory case
co-occurs with the head word in the form of
noun phrase frequently, we can take frequent
case slots as obligatory case of the head word.
However, we have to be careful to set up
the frequency thresholds, because case slots de-
tected by DBA or <obligatory case> by SBA
are more likely to be obligatory; on the other
hand case slots of <modification> or <time>
should be always optional. Considering these
tendencies, we set thresholds for obligatory
cases as shown in Table 3.
In the case of hisashi ‘eaves/visor’ in Table 2,
[house-window]-<place> slot and [cap] slot are
chosen as the obligatory cases.
4.4 Case frame construction for each
meaning
Case slots that are derived from each definition
sentence constitute a case frame.
If a case slot of <obligatory case> by SBA
or <other> is not merged into case slots in def-
inition sentences, it can be considered that it
indicates a meaning of Nh which is not covered
in the dictionary. Therefore, such a case slot
constitutes an independent case frame.
On the other hand, when other case slots by
SBA such as <belonging> and <possessive>
are remaining, we have to treat them differently.
The reason why they are remaining is that they
are not always described in the definition sen-
tences, but their frequent occurrences indicate
they are obligatory cases. Therefore, we add
these case slots to the case frames derived from
definition sentences.
Table 4 shows several examples of resul-
tant case frames. Hyoujou ‘expression’ has a
case frame containing two case slots. Hisashi
‘eaves/visor’ has two case frames according to
the two definition sentences. In case of hiki-
dashi ‘drawer’, the first case frame corresponds
to the definition given in the dictionary, and
the second case frame was constructed from the
<other> case slot, which is actually another
sense of hikidashi, missed in the dictionary. In
case of coach, <possessive> is added to the case
frame which was made from the definition, pro-
ducing a reasonable case frame for the word.
4.5 Point of nominal case frame
construction
The point of our method is the integrated
use of a dictionary and example phrases from
large corpora. Although dictionary definition
sentences are informative resource to indicate
obligatory cases of nouns, it is difficult to do
indirect anaphora resolution by using a dictio-
nary as it is, because all nouns in a definition
sentence are not an obligatory case, and only
the frequency information of noun phrases tells
us which is the obligatory case. Furthermore,
sometimes a definition is too specific or detailed,
and the example phrases can adjust it properly,
as in the example of hisashi in Table 2.
On the other hand, a simple method that
just collects and clusters “Nm no Nh” phrases
(based on some similarity measure of nouns)
can not construct comprehensive nominal case
frames, because of polysemy and multiple oblig-
atory cases. We can see that dictionary defini-
tion can guide the clustering properly even for
such difficult cases.
Table 4: Examples of nominal case frames.
case slot examples
hisashi:1 ‘eaves/visor’ (the edges of a roof that stick out above the window of a house etc.)
[house, window] parking, store, hall, ···
hisashi:2 ‘eaves/visor’ (the fore piece of a cap.)
[cap] cap, helmet, ···
hyoujou ‘expression’ (to express one’s feelings on the face or by gestures.)
[one] people, person, citizen, ···
[feelings] relief, margin, ···
hikidashi:1 ‘drawer’ (a boxlike container in a desk or a chest.)
[desk, chest] desk, chest, dresser, ···
hikidashi:2 ‘drawer’ <other> credit, fund, saving, ···
coach (a person who teaches technique in some sport.)
[sport] baseball, swimming, ···
<belonging> team, club, ···
kabushiki ‘stock’ (the total value of a company’s shares.)
[company] company, corporation, ···
5 Indirect Anaphora Resolution
To examine the practical usefulness of the con-
structed nominal case frames, we built a pre-
liminary system of indirect anaphora resolution
based on the case frames.
An input sentence is parsed using the
Japanese parser, KNP (Kurohashi and Nagao,
1994). Then, from the beginning of the sen-
tence, each noun x is analyzed. When x has
more than one case frame, the process of an-
tecedent estimation (stated in the next para-
graph) is performed for each case frame, and the
case frame with the highest similarity score (de-
scribed below) and assignments of antecedents
to the case frame are selected as a final result.
For each case slot of the target case frame of
x, its antecedent is estimated. A possible an-
tecedent y in the target sentence and the previ-
ous two sentences is checked. This is done one
by one, from the syntactically closer y. If the
similarity of y to the case slot is equal to or
greater than a threshold α (currently 0.95), it
is assigned to the case slot.
The similarity between y and a case slot is
defined as the highest similarity between y and
an example in the case slot.
For instance, let us consider the sentence
shown in Figure 1. soccer, at the beginning of
the sentence, has no case frame, and is consid-
ered to have no obligatory case.
For the second noun ticket, soccer, which is
a nominal modifier of ticket, is examined first.
The similarity between soccer and the examples
of the case slot [theater, transport] exceeds the
soccer-no
ticket-ga
takai
nedan-de
urareteita.
expensive
price
be sold
case slot examples result
ticket [theater, transport] stage, game,··· soccer
nedan [things] thing, ticket,··· ticket
ticket a printed piece of paper which shows that you have
paid to enter a theater or use a transport
nedan the amount of money for which things are sold or
bought
Figure 1: Indirect anaphora resolution example.
threshold α, and soccer is assigned to [theater,
transport].
Lastly, for nedan ‘price’, its possible an-
tecedents are ticket and soccer. ticket, which
is the closest from nedan, is checked first. The
similarity between ticket and the examples of
the case slot [things] exceeds the threshold α,
and ticket is judged as the antecedent of nedan.
6 Experiments
We evaluated the automatically constructed
nominal case frames, and conducted an experi-
ment of indirect anaphora resolution.
6.1 Evaluation of case frames
We constructed nominal case frames from news-
paper articles in 25 years (12 years of Mainichi
newspaper and 13 years of Nihonkeizai newspa-
per). These newspaper corpora consist of about
Table 5: Evaluation result of case frames.
precision recall F
58/70 (0.829) 58/68 (0.853) 0.841
25,000,000 sentences, and 10,000,000 “Nm no
Nh” noun phrases were extracted from them.
The result consists of 17,000 nouns, the average
number of case frames for a noun is 1.06, and
the average number of case slots for a case frame
is 1.09.
We randomly selected 100 nouns that occur
more than 10,000 times in the corpora, and cre-
ated gold standard case frames by hand. For
each test noun, possible case frames were con-
sidered, and for each case frame, obligatory case
slots were given manually. As a result, 68 case
frames for 65 test nouns were created, and 35
test nouns have no case frames.
We evaluated automatically constructed case
frames for these test nouns against the gold
standard case frames. A case frame which has
the same case slots with the gold standard is
judged as correct. The evaluation result is
shown in Table 5: the system output 70 case
frames, and out of them, 58 case frames were
judged as correct.
The recall was deteriorated by the highly re-
stricted conditions in the example collection.
For instance, maker does not have obligatory
case slot for its products. This is because maker
is usually used in the form of compound noun
phrase, “products maker”, and there are few
occurrences of “products no maker”. To ad-
dress this problem, not only “Nm no Nh” but
also “Nm Nh” (compound noun phrase) and
“Nm ni-kansuru ‘in terms of’ Nh” should be
collected.
6.2 Experimental results of indirect
anaphora resolution
We conducted a preliminary experiment of
our indirect anaphora resolution system using
“Relevance-tagged corpus” (Kawahara et al.,
2002). This corpus consists of Japanese news-
paper articles, and has relevance tags, including
antecedents of indirect anaphors.
We prepared a small test corpus that con-
sists of randomly selected 10 articles. The test
corpus contains 217 nouns. Out of them, 106
nouns are indirect anaphors, and have 108 an-
tecedents, which is because two nouns have dou-
ble antecedents. 49 antecedents directly depend
on their anaphors, and 59 do not. For 91 an-
tecedents out of 108, a case frame of its anaphor
Table 6: Experimental results of indirect
anaphora resolution.
precision recall F
w dep. 40/46 (0.870) 40/59 (0.678) 0.762
w/o dep. 31/61 (0.508) 31/49 (0.633) 0.564
total 71/107 (0.664) 71/108 (0.657) 0.660
includes the antecedent itself or its similar word
(the similarity exceeds the threshold, 0.95). Ac-
cordingly, the upper bound of the recall of our
case-frame-based anaphora resolution is 84.3%
(91/108).
We ran the system on the test corpus, and
compared the system output and the corpus an-
notation. Table 6 shows the experimental re-
sults. Inthistable, “wdep.” (withdependency)
is the evaluation of the antecedents that directly
depend on their anaphors. “w/o dep.” (with-
out dependency) is the case of the antecedents
that do not directly depend on their anaphors.
Although the analysis of “w dep.” is intrinsi-
cally easier than that of “w/o dep.”, the recall
of “w dep.” was not much higher than that
of “w/o dep.”. The low recall score of “w dep.”
wascausedbynonexistenceofcaseframeswhich
include the antecedent itself or its similar word.
The antecedents that directly depend on their
anaphors were often a part of compound noun
phrases, such as “products maker”, which are
not covered by our examples collection.
Major errors in the analyses of the an-
tecedents that do not directly depend on their
anaphors were caused by the following reasons.
Specific/generic usages of nouns
Some erroneous system outputs were caused by
nouns that have both specific and generic us-
ages.
(3) kogaisya-no
subsidiary
kabushiki-wo
stock
baikyaku-shita.
sell
(φ sold the stock of the subsidiary.)
In this case, kogaisya ‘subsidiary’ is an oblig-
atory information for kabushiki ‘stock’, which is
specifically used. kogaisya matches the [kaisya
‘company’] case slot in Table 4.
However, kabushiki ‘stock’inthefollowingex-
ampleisusedgenerically, anddoesnotneedspe-
cific company information.
(4)kabushiki
stock
souba-no
price
oshiage
rise
youin-to naru.
factor become
(φ become the rise factor of the stock prices.)
Since the current system cannot judge generic
or specific nouns, an antecedent which corre-
sponds to [kaisha ‘company’] is incorrectly esti-
mated.
Beyond selectional restriction of case
frames
Selectional restriction based on the case frames
usually worked well, but did not work to distin-
guish candidates both of which belong to Hu-
man or Organization.
(5) Bush bei
American
seiken-wa
administration
Russia-tono
... Bush daitouryou-ga
president
shutyou-shita.
claim
(Bush American administration ... with
Russia ... President Bush claimed ...)
In this example, daitouryou ‘president’ re-
quires an obligatory case kuni ‘nation’. The sys-
tem estimates its antecedent as Russia, though
the correct answer is bei ‘America’. This is be-
cause Russia is closer than beikoku. This prob-
lem is somehow related to world knowledge, but
if the system can carefully exploit the context,
it might be able to find the correct answer from
“Bush bei seiken” ‘Bush American administra-
tion’.
7 Conclusion
This paper has first proposed an automatic
construction method of Japanese nominal case
frames. This method is based on semantic anal-
ysis of noun phrases “Nm no Nh” ‘Nh of Nm’.
To examine the practical usefulness of the con-
structed nominal case frames, we built a pre-
liminary system of indirect anaphora resolution
based on the case frames. The evaluation indi-
cated the good quality of the constructed case
frames. On the other hand, the accuracy of our
indirect anaphora resolution system is not satis-
factory. In the future, we are planning to make
the case frames more wide-coverage, and im-
prove the indirect anaphora resolution by con-
sidering larger context and more various factors.

References

Ted Briscoe and John Carroll. 1997. Auto-
matic extraction of subcategorization from
corpora. In Proceedings of the 5th Confer-
ence on Applied Natural Language Process-
ing, pages 356–363.

Udo Hahn, Michael Strube, and Katja Markert.
1996. Bridging textual ellipses. In Proceed-
ings of the 16th International Conference on
Computational Linguistics, pages 496–501.

Daisuke Kawahara and Sadao Kurohashi. 2002.
Fertilization of case frame dictionary for ro-
bust Japanese case analysis. In Proceedings of
the 19th International Conference on Compu-
tational Linguistics, pages 425–431.

Daisuke Kawahara and Sadao Kurohashi. 2004.
Zero pronoun resolution based on automati-
cally constructed case frames and structural
preference of antecedents. In Proceedings of
the 1st International Joint Conference on
Natural Language Processing.

Daisuke Kawahara, Sadao Kurohashi, and Kˆoiti
Hasida. 2002. Construction of a Japanese
relevance-tagged corpus. In Proceedings of
the 3rd International Conference on Lan-
guage Resources and Evaluation, pages 2008–2013.

Sadao Kurohashi and Makoto Nagao. 1994. A
syntactic analysis method of long Japanese
sentences based on the detection of conjunc-
tive structures. Computational Linguistics,
20(4):507–534.

Sadao Kurohashi and Yasuyuki Sakai. 1999.
Semantic analysis of Japanese noun phrases:
A new approach to dictionary-based under-
standing. In Proceedings of the 37th Annual
Meeting of the Association for Computational
Linguistics, pages 481–488.

Masaki Murata, Hitoshi Isahara, and Makoto
Nagao. 1999. Pronoun resolution in Japanese
sentencesusingsurfaceexpressionsandexam-
ples. In Proceedings of the ACL’99 Workshop
on Coreference and Its Applications, pages
39–46.

Massimo Poesio, Tomonori Ishikawa,
Sabine Schulte im Walde, and Renata
Vieira. 2002. Acquiring lexical knowledge for
anaphora resolution. In Proceedings of the
3rd International Conference on Language
Resources and Evaluation, pages 1220–1224.

Michael Strube and Udo Hahn. 1999. Func-
tional centering – grounding referential coher-
ence in information structure. Computational
Linguistics, 25(3):309–344.

Jun-ichi Tajika, editor. 1997. Reikai Syogaku
Kokugojiten. Sanseido.

Renata Vieira and Massimo Poesio. 2000. An
empirically based system for processing defi-
nite descriptions. Computational Linguistics,
26(4):539–592.
