Proceedings of the BioNLP Workshop on Linking Natural Language Processing and Biology at HLT-NAACL 06, pages 96–103,
New York City, June 2006. c©2006 Association for Computational Linguistics
Semi-supervised anaphora resolution in biomedical texts
Caroline Gasperin
Computer Laboratory,
University of Cambridge,
15 JJ Thomson Avenue,
Cambridge CB3 0FD, UK
cvg20@cl.cam.ac.uk
Abstract
Resolving anaphora is an important step
in the identification of named entities such
as genes and proteins in biomedical sci-
entific articles. The goal of this work
is to resolve associative and coreferential
anaphoric expressions making use of the
rich domain resources (such as databases
and ontologies) available for the biomed-
ical area, instead of annotated training
data. The results are comparable to ex-
tant state-of-the-art supervised methods in
thesamedomain. Thesystemisintegrated
into an interactive tool designed to assist
FlyBase curators by aiding the identifica-
tion of the salient entities in a given paper
as a first step in the aggregation of infor-
mation about them.
1 Introduction
The number of articles being published in biomedi-
cal journals per year is increasing exponentially. For
example, Morgan et al. (2003) report that more than
8000 articles were published in 2000 just in relation
to FlyBase1, a database of genomic research on the
fruit fly Drosophila melanogaster.
The growth in the literature makes it difficult for
researchers to keep track of information, even in
very small subfields of biology. Progress in the
field often relies on the work of professional cura-
tors, typically postdoctoral-level scientists, who are
1http://www.flybase.org
trained to identify important information in a sci-
entific article. This is a very time-consuming task
which first requires identification of gene, allele and
protein names and their synonyms, as well as sev-
eral interactions and relations between them.The in-
formation extracted from each article is then used to
fill in a template per gene or allele.
To extract all information about a specific
biomedical entity in the text and be able to fill in
the corresponding template, a useful first step is
the identification of all textual mentions that are re-
ferring to or are related with that entity. Linking
all these mentions together corresponds to the task
known as anaphora resolution in Natural Language
Processing.
Inthispaper, weareinterestedinlinkingautomat-
ically all mentions that refer to a gene or are related
to it (i.e. its ‘products’). For example, in the follow-
ing portion of text, we aim to link the highlighted
mentions:
‘‘... is composed of five proteins(1)
encoded by the male-specific lethal
genes(2) ... The MSL proteins(3)
colocalize to hundreds of sites ... male
animals die when they are mutant for any
one of the five msl genes(4).’’
In this work we use the output of a gene name
recogniser (Vlachos et al., 2006) and information
from the Sequence Ontology (Eilbeck and Lewis,
2004) to identify the entities of interest and the ge-
nomic relations among them. We also use RASP
(Briscoe and Carroll, 2002), a statistical parser, to
identify NPs (and their subconstituents) which may
beanaphoricallylinked. Oursystemidentifiescoref-
96
erential relations between biomedical entities (such
as (1) and (3), and (2) and (4) above) as well as as-
sociative links (relations between different entities,
e.g. the link between a gene and its protein as in (2)
and (3) above). A previous version of this system
was presented in (Vlachos et al., 2006); here we im-
prove its results due to refinements on some of the
steps previous to the resolution and to the anaphora
resolution process itself.
The large majority of the entities in biomedi-
cal texts are referred to using non-pronominal noun
phrases, like proper nouns, acronyms or definite de-
scriptions. Hence, we focus on these NPs and do
not resolve pronominal references (as pronouns rep-
resent only about 3% of the noun phrases in our do-
main).
In the following section, we detail the different
components of the anaphora resolution system. The
results are tested against hand-annotated papers, and
an extensive evaluation is provided in Section 3,
where the performance and errors are discussed.
2 The anaphora resolution system
Our system for anaphora resolution makes use of
lexical, syntactic, semantic and positional informa-
tion to link anaphoric expressions. The lexical infor-
mation consists of the words themselves. The syn-
tactic information consists of noun phrase bound-
aries and the distinction between head and pre-
modifiers (extracted from RASP output). The dis-
tance (in words) between the anaphoric expression
and its possible antecedent is taken into account as
positional information. The semantic information
comes from the named entity recognition (NER)
process and some extra tagging based on features
from the Sequence Ontology.
FlyBase is used as source of gene names, sym-
bols and synonyms, giving rise to training data for
the gene name recognition system detailed in Sec-
tion 2.1. The output of this system is tagged named
entities that refer to the fruit fly genes.
We then parse the text using RASP in order to ex-
tract the noun phrases and their subparts (head and
modifiers). Retagging gene names as proper names
before parsing improves the parser’s performance,
but otherwise the parser is used unmodified.
The Sequence Ontology (SO) can be used to iden-
tify words and phrases related to a gene: its sub-
types (e.g. oncogene, transposable element), parts
(e.g. transcript, regulatory region) and products (e.g.
polypeptide, protein). Subsection 2.3 details the in-
formation extracted from SO to type the non-gene
mentions.
2.1 Gene-name recognition
The NER system we use (Vlachos et al., 2006) is
a replication and extension of the system developed
by Morgan et al. (2004): a different training set and
software were used. For training data we used a
total of 16609 abstracts, which were automatically
annotated by a dictionary-based gene name tagger.
The dictionary consists of lists of the gene names,
symbolsandsynonymsextractedfromFlyBase. The
gene names and their synonyms that were recorded
by the curators from the full paper were annotated
automatically in each abstract, giving rise to a large
but noisy set of training data. The recognizer used
is the open source toolkit LingPipe2, implementing
a 1st-order HMM model using Witten-Bell smooth-
ing. A morphologically-based classifier was used
to deal with unknown gene names (that were not
present in the training data).
The performance of the trained recogniser on a
revised version of the test data used in Morgan et
al. (86 abstracts annotated by a biologist curator
and a computational linguist) was 80.81% recall and
84.93% precision.
2.2 Parsing and NP extraction
RASP is a pipelined parser which identifies sentence
boundaries, tokenises sentences, tags the tokens
with their part-of-speech (PoS) and finally parses
PoS tag sequences, statistically ranking the result-
ing derivations. We have made minor modifications
to RASP’s tokeniser to deal with some specific fea-
tures of biomedical articles, and manually modified
a small number of entries in the PoS tagger lexicon,
for example to allow the use of and as a proper name
(referring to a fruit fly gene). Otherwise, RASP uses
a parse ranking module trained on a generic tree-
bank and a grammar also developed from similar re-
sources.
The anaphora resolution system first tags genes
2http://www.alias-i.com/lingpipe/
97
using the gene recogniser. This means that identi-
fied gene mentions can be retagged as proper names
before the RASP parser is applied to the resulting
PoS sequences. This improves parser performance
as the accuracy of PoS tagging decreases for un-
known words, especially as the RASP tagger uses an
unknown word handling module which relies heav-
ily on the similarity between unknown words and
extant entries in its lexicon. This strategy works less
well on gene names and other technical vocabulary
from the biomedical domain, as almost no such ma-
terial was included in the training data for the tag-
ger. We have not evaluated the precise improvement
in performance as yet due to the lack of extant gold
standard parses for relevant text.
RASP can output grammatical relations (GRs) for
each parsed sentence (Briscoe, 2006). GRs are fac-
tored into binary lexical relations between a head
and a dependent of the form (GR-type head
dependent). We use the following GR-types to
identify the head-nouns of NPs (the examples of
GRs are based on the example of the first page un-
less specified otherwise):
• ncsubj encodes binary relations between
non-clausalsubjectsandtheirverbalheads; e.g.
(ncsubj colocalize proteins).
• dobj encodes a binary relation between ver-
bal or prepositional head and the head of the
NP to its immediate right; e.g. (dobj of
sites).
• obj2 encodes a binary relation between ver-
bal heads and the head of the second NP in a
double object construction; e.g. for the sen-
tence “Xist RNA provides a mark for specific
histones” we get (dobj provides mark)
(obj2 provides histones).
• xcomp encodes a binary relation between
a head and an unsaturated VP complement;
e.g. for the phrase “a class of regulators in
Drosophila is the IAP family” we get (xcomp
is family).
• ta encodes a binary relation between a head
and the head of a text adjunct delimited by
punctuation (quotes, brackets, dashes, com-
mas, etc.); e.g. for “BIR-containing proteins
(BIRPs)” we get (ta proteins BIRPs).
To extract the modifiers of the head nouns, we
search the GRs typed ncmod which encode binary
relations between non-clausal modifiers and their
heads; e.g (ncmod genes msl).
When the head nouns take part in coordination, it
is necessary to search the conj GRs which encode
relations between a coordinator and the head of a
conjunct. There will be as many such binary rela-
tions as there are conjuncts of a specific coordinator;
e.g. for “CED-9 and EGL-1 belong to a large fam-
ily ...” we get (ncsubj belong and) (conj
and CED-9) (conj and EGL-1).
Last but not least, to identify definite descrip-
tions, we search the det GR for a definite speci-
fier, e.g. (det proteins The). By using the
GR representation of the parser output we were able
to improve the performance of the anaphora resolu-
tion system by about 10% over an initial version de-
scribed in (Vlachos et al., 2006) that used the RASP
tree output instead of GRs. GRs generalise more
effectively across minor and irrelevant variations in
derivations such as the X-bar level of attachment in
nominal coordinations.
2.3 Semantic typing and selecting NPs
To identify the noun phrases that refer to the entities
of interest, we classify the head noun as belonging
to one of the five following classes: “part-of-gene”,
“subtype-of-gene”, “supertype-of-gene”, “product-
of-gene” or “is-a-gene”. These classes are referred
to as biotypes.
Figure 1: SO path from gene to protein.
The biotypes reflect the way the SO relates en-
tities to the concept of the gene using the follow-
ing relations: derives from, member of, part of, and
is a, among others.3 We extracted the unique path
3We consider the member of relation to be the same as the
part of relation.
98
of concepts and relations which leads from a gene to
a protein. The result is shown in Figure 1.
Besides the facts directly expressed in this path,
we also assumed the following:4
1. Whatever is-a transcript is also part-of a gene.
2. Whatever is part-of a transcript is also part-of a
gene.
3. An mRNA is part-of a gene.
4. Whatever is part-of an mRNA is also part-of a
gene.
5. CDS is part-of a gene.
6. A polypeptide is a product (derived-from) of a
gene.
7. Whatever is part-of a polypeptide is also a
product of a gene.
8. A protein is a product of a gene.
We then used these assumptions to add new deriv-
able facts to our original path. For example, an exon
is a part of a transcript according to the SO, there-
fore, by the 2nd assumption, we add the fact that an
exon is a part of a gene. We also extracted infor-
mation about gene subtypes that is included in the
ontology as an entry called “gene class”. We con-
sider NPs as supertypes of a gene when they refer to
nucleotidesequencesthatarebiggerthanbutinclude
the gene.5
Finally, we tagged every NP whose head noun is
one of the items extracted from the SO with its bio-
type. For instance, we would tag “the third exon”
with “part-of-gene”.
The NPs whose head noun is a gene name tagged
in the NER phase also receive the “is-a-gene” bio-
type. Other NPs that still remain without biotype
info are tagged as “other-bio” if any modifier of the
head is a gene name.
This typing process achieves 75% accuracy when
evaluated against the manually annotated corpora
described in Section 3. The majority of the errors
4A curator from FlyBase was consulted to confirm the va-
lidity of these assumptions.
5In the SO a gene holds an is-a relation to “sequence” and
“region” entries.
(70%) are on typing NPs that contain just a proper
name, which can refer to a gene or to a protein. At
the moment, all of these cases are being typed as
“is-a-gene”.
The biotyped NPs are then selected and consid-
eredforanaphoraresolution. NPswiththesamebio-
type can be coreferent, as well as NPs with is-a-gene
and subtype-of-gene biotypes. The anaphoric rela-
tion between an is-a-gene NP and a part-of-gene or
product-of-gene NP is associative rather than coref-
erential.
2.4 Resolving anaphora cases
We take all proper namer (PNs) and definite de-
scriptions (DDs) among the filtered NPs as poten-
tial anaphoric expressions (anaphors) to be resolved.
As possible antecedents for an anaphor we take all
bio-typed NPs that occur before it in the text. For
each anaphor we look for its antecedent (the closest
previous mention that is related to it). For linking
anaphors to their antecedents we look at:
• headan: anaphor head noun
• heada: antecedent head noun
• modan: set of anaphor pre-modifiers
• moda: set of antecedent pre-modifiers
• biotypean: anaphor biotype
• biotypea: antecedent biotype
• d: distance in sentences from the anaphor
The pseudo-code to find the antecedent for the
DDs and PNs is given below:
• Input: a set A with all the anaphoric expres-
sions (DDs and PNs); a set C with all the possi-
ble antecedents (all NPs with biotype informa-
tion)
• For each anaphoric expression Ai:
– Let antecedent 1 be the closest preceding
NP Cj such that
head(Cj)=head(Ai) and
biotype(Cj)=biotype(Ai)
99
– Let antecedent 2 be the closest preceding
NP Cj such that
biotype(Cj)negationslash=biotype(Ai), but
head(Cj)=head(Ai) or
head(Cj)=mod(Ai) or
mod(Cj)=head(Ai) or
mod(Cj)=mod(Ai)
– Take the closest candidate as antecedent,
if 1 and/or 2 are found; if none is found,
the DD/PN is treated as non-anaphoric
• Output: The resolved anaphoric expressions in
A linked to their antecedents.
As naming conventions usually recommend gene
names to be lower-cased and protein names to be
upper-cased, our matching among heads and modi-
fiers is case-insensitive, allowing, for example, msl
gene to be related to MSL protein due to their
common modifiers.
Antecedent 1, if found, is considered coreferent
to Ai, and antecedent 2, associative. For example, in
the passage:
‘‘Dosage compensation, which ensures
that the expression of X-linked genes:Cj
is equal in males and females ... the
hypertranscription of the X-chromosomal
genes:Aj in males ...’’
the NP in bold font which is indexed as antecedent
Cj istakentobecoreferentialtotheanaphorindexed
as Aj. Additionally, in:
‘‘... the role of the roX genes:Ck
in this process ... which MSL proteins
interact with the roX RNAs:Ak ...’’
Ck meets the conditions to form an associative link
to Ak. The same is true in the following example
in which there is an associative relation between Cj
and Aj:
‘‘The expression of reaper:Cj has been
shown to be regulated by distinct stimuli
... it was shown to bind a specific
region of the reaper promoter:Aj ...’’
If we consider the example from the first page,
mention (1) is returned by the system as the corefer-
ent antecedent for (3), as they have the same biotype
and a common head noun. In the same example, (2)
is returned as a coreferent antecedent to (4), and (3)
as an associative antecedent to (4).
3 Evaluation
We evaluated our system against two hand-
annotated full papers which have been curated in
FlyBase and were taken from PubMed Central in
XML format. Together they contain 302 sentences,
in which 97 DDs and 217 PNs related to biomedical
entities (out of 418 NPs in total) were found.
For each NP, the following information was man-
ually annotated:
• NP form: definite NP, proper name, or NP.
• biotype: gene, part-of-gene, subtype-of-gene,
supertype-of-gene, product-of-gene, other-bio,
or a non-bio noun.
• coreferent antecedent: a link to the closest pre-
vious coreferent mention (if there is one).
• associative antecedent: a link to the closest pre-
vious associative anaphoric mention (if there is
one, and only if there is no closer coreferent
mention).
All coreferent mentions become linked together
as a coreference chain, which allows us to check for
previouscoreferentantecedentsofamentionbesides
the closest one.
Table 1 shows the distributions of the anaphoric
expressions according to the anaphoric relations
they hold to their closest antecedent.
coreferent associative no ant. Total
DDs 34 51 12 97
PNs 132 62 23 217
Total 166 113 35 314
Table 1: Anaphoric relation distribution
DDs and PNs in associative relations account for
27% of all NPs in the test data, which is almost dou-
ble the number of bridging cases (associative plus
coreferent cases where head nouns are not the same)
reported for newspaper texts in Vieira and Poesio
(2000).
Table 2 shows the distribution of the different bio-
types present in the corpus.
100
gene part subtype supertype product
67 62 1 7 244
Table 2: Biotype distribution
3.1 Results
The anaphora resolution system reaches 58.8% pre-
cision and 57.3% recall when looking for the clos-
est antecedent for DDs and PNs, after having been
provided with hand-corrected input (that is, perfect
gene name recognition, NP typing and selection). If
we account separately for coreference and associa-
tive relations, we get 59.47% precision and 81.3%
recall for the coreferent cases, and 55.5% precision
and 22.1% recall for the associative ones.
The performance of the system is improved if we
consider that it is able to find an antecedent other
than the closest, which is still coreferential to the
anaphor. These are cases like the following:
‘‘five proteins encoded by the
male-specific lethal genes ... The MSL
proteins ...’’
wherethesystemreturns“fiveproteins”asthecoref-
erent antecedent for “the MSL proteins”, instead
of returning “the male-specific lethal genes” as the
closest (in this case, associative) antecedent. Treat-
ing these cases as positive examples we reach 77.5%
precision and 75.6% recall6. It conforms with the
goal of adding the anaphor to a coreferential chain
rather than simply relating it to the closest an-
tecedent.
Table 3 reports the number of coreferent and as-
sociative DDs and PNs that could be resolved. The
numbers on the left of the slash refer to relations
with the closest antecedent, and the numbers on the
right refer to additional relations found when links
with another antecedent are considered (all the new
positive cases on the right are coreferent, since our
evaluation data just contain associative links to the
closest antecedent).
Most of the cases that could be resolved are coref-
erent, and when the restriction to find the closest
antecedent is relaxed, the system manages to re-
solve 35 cases of DD coreference (64.7% recall).
6Weareabletocomputetheseratessinceourevaluationcor-
pus includes also a coreferent antecedent for each case where an
associative antecedent was selected.
coreferent associative no ant.
DDs 20/+2 14/+13 7
PNs 115/+9 11/+22 16
Table 3: Resolved anaphoric relations
It achieves very high recall (93.9%) on coreferen-
tial PNs. All the associative relations that are hand
annotated in our evaluation corpus are between an
anaphor and its closest antecedent, so when the re-
cency preference is relaxed, we get coreferent in-
stead of associative antecedents: we got 35 corefer-
ent antecedents for anaphors that had a closest asso-
ciative antecedent that could not be recovered. This
conforms to the goal of having coreference chains
that link all the mentions of a single entity.
The system could resolve around 27% of the as-
sociative cases of DDs, although fewer associative
antecedents could be recovered for PNs, mainly due
to the frequent absence of head-noun modifiers and
different forms for the same gene name (expanded
vs. abbreviated).
Although associative anaphora is considered to be
harder than coreference, we believe that certain re-
finements of our resolution algorithm (such as nor-
malizing gene names in order to take more advan-
tage of the string matching among NP heads and
modifiers) could improve its performance on these
cases too.
The anaphora resolution system is not able to
find the correct antecedent when there is no head or
modifier matching as in the anaphoric relation be-
tween ‘‘Dark/HAC-1/Dapaf-1’’ and ‘‘The
Drosophila homolog’’.
Theperformanceratesdropwhenusingtheoutput
of the NER system (presented in Section 2.1), RASP
parsing (Section 2.2) and SO-based NP typing (Sec-
tion 2.3), resulting in 63% precision and 53.4% re-
call.
When the NER system fails to recognise a gene
name, it can decrease the parser performance (as
it would have to deal with an unknown word) and
influences the semantic tagging (the NP containing
such a gene name won’t be selected as a possible an-
tecedent or anaphor unless it contains another word
that is part of SO). When just the NER step is cor-
rected by hand, the system reaches 71.8% precision
101
and 64.1% recall.
4 Related work
Previous approaches to solve associative anaphora
havemadeuseofknowledgeresourceslikeWordNet
(Poesio et al., 1997), the Internet (Bunescu, 2003)
and a corpus (Poesio et al., 2002) to check if there is
an associative link between the anaphor and a possi-
ble antecedent.
In the medical domain, Casta˜no et al. (2002)
used UMLS (Unified Medical Language System)7
as their knowledge source. They treat coreferential
pronominal anaphora and anaphoric DDs and aim
to improve the extraction of biomolecular relations
from MEDLINE abstracts. The resolution process
relies on syntactic features, semantic information
fromUMLS,andthestringitself. Theytrytoresolve
just the DDs that refer to relevant biotypes (corre-
sponding to UMLS types) such as amino acids, pro-
teinsorcells. Forselectingtheantecedents, theycal-
culate salience values based on string similarity, per-
son/number agreement, semantic type matching and
other features. They report precision of 74% and re-
call of 75% on a very small test set.
Yangetal.(2004)testasupervisedlearning-based
approach for anaphora resolution, evaluating it on
MEDLINE abstracts from the GENIA corpus. They
focus only on coreferent cases and do not attempt to
resolve associative links. 18 features describe the
relationship between an anaphoric expression and
its possible antecedent - their source of semantic
knowledge is the biotype information provided by
the NER component of GENIA. They achieved re-
call of 80.2% and precision of 77.4%. They also ex-
periment with exploring the relationships between
NPs and coreferential clusters (i.e. chains), select-
ing an antecedent based not just on a single candi-
date but also on the cluster that the candidate is part
of. For this they add 6 cluster-related features to
the machine-learning process, and reach 84.4% re-
call and 78.2% precision.
Our system makes use of extant biomedical re-
sources focused on the relevant microdomain (fruit
fly genomics), and attempts to tackle the harder
problem of associative anaphora, as this constitutes
a significant proportion of cases and is relevant to
7http://www.nlm.nih.gov/research/umls/
the curation task. Our performance rates are lower
than the ones above, but did not rely on expensive
training data.
5 Concluding remarks
Our system for anaphora resolution is semi-
supervised and relies on rich domain resources: the
FlyBase database for NER, and the Sequence On-
tology for semantic tagging. It does not need train-
ingdata, whichisaconsiderableadvantage, asanno-
tating anaphora by hand is a complicated and time-
demanding task, requiring very precise and detailed
guidelines.
The resulting links between the anaphoric entities
are integrated into an interactive tool which aims to
facilitate the curation process by highlighting and
connecting related bio-entities: the curators are able
to navigate among different mentions of the same
entity and related ones in order to find easily the in-
formation they need to curate.
We are currently working on increasing our eval-
uation corpus; we aim to make it available to the
research community together with our annotation
guidelines.
We intend to enhance our system with additional
syntactic features to deal with anaphoric relations
between textual entities that do not have any string
overlap. We also intend to add different weights
to the features. The performance of the fully-
automated version of the system can be improved if
we manage to disambiguate between gene and pro-
tein names and infer the correct biotype for them.
The performance on associative cases could be im-
proved by normalizing the gene names in order to
find more matches among heads and modifiers.
Acknowledgements
This work is part of the BBSRC-funded FlySlip8
project. Caroline Gasperin is funded by a CAPES
award from the Brazilian government. Thanks to
Nikiforos Karamanis and Ted Briscoe for their com-
ments and help with this manuscript.
8http://www.cl.cam.ac.uk/users/av308/Project Index/
Project Index.html
102

References
Ted Briscoe and John Carroll. 2002. Robust accurate
statistical annotation of general text. In Proceedings
of LREC 2002, pages 1499–1504, Las Palmas de Gran
Canaria.
Ted Briscoe. 2006. Tag sequence grammars. Technical
report, Computer Laboratory, Cambridge University.
Razvan Bunescu. 2003. Associative anaphora resolu-
tion: A web-based approach. In Proceedings of EACL
2003 - Workshop on The Computational Treatment of
Anaphora, Budapest.
Jos´e Casta˜no, Jason Zhang, and James Pustejovsky.
2002. Anaphora resolution in biomedical literature. In
Proceedings of International Symposium on Reference
Resolution for NLP 2002, Alicante, Spain.
Karen Eilbeck and Suzanna E. Lewis. 2004. Sequence
ontology annotation guide. Comparative and Func-
tional Genomics, 5:642–647.
Alex Morgan, Lynette Hirschman, Alexander Yeh, and
Marc Colosimo. 2003. Gene name extraction using
FlyBaseresources. In Proceedings of ACL 2003 Work-
shop on Natural Language Processing in Biomedicine,
Sapporo, Japan.
Alex Morgan, Lynette Hirschman, Mark Colosimo,
Alexander Yeh, and Jeff Colombe. 2004. Gene
name identification and normalization using a model
organism database. J. of Biomedical Informatics,
37(6):396–410.
Massimo Poesio, Renata Vieira, and Simone Teufel.
1997. Resolving bridging descriptions in unrestricted
texts. In Proceedings of the Workshop on Operational
Factors in the Practical, Robust, Anaphora Resolution
for Unrestricted Texts, Madrid.
Massimo Poesio, Tomonori Ishikawa, Sabine Schulte
im Walde, and Renata Vieira. 2002. Acquiring lexical
knowledge for anaphora resolution. In Proceedings of
LREC 2002, Las Palmas De Gran Canaria.
Renata Vieira and Massimo Poesio. 2000. An
empirically-based system for processing definite de-
scriptions. Computational Linguistics, 26(4):525–
579.
Andreas Vlachos, Caroline Gasperin, Ian Lewin, and Ted
Briscoe. 2006. Bootstrapping the recognition and
anaphoric linking of named entities in Drosophila arti-
cles. In Proceedings of the PSB 2006, Hawaii.
Xiaofeng Yang, Jian Su, Gouodong Zhou, and Chew Lim
Tan. 2004. An NP-cluster based approach to coref-
erence resolution. In Proceedings of COLING 2004,
Geneva, Switzerland, August.
