Sense Information for Disambiguation:
Confluence of Supervised and Unsupervised Methods
Kenneth C. Litkowski
CL Research
9208 Gue Road
Damascus, MD 20872
ken@clres.com
Abstract
For SENSEVAL-2, we disambiguated the lexical
sample using two different sense inventories.
Official SENSEVAL-2 results were generated
using WordNet, and separately using the New
Oxford Dictionary of English (NODE). Since
our initial submission, we have implemented
additional routines and have now examined the
differences in the features used for making sense
selections. We report here the contribution of
default sense selection, idiomatic usage, syntactic
and semantic clues, subcategorization patterns,
word forms, syntactic usage, context, selectional
preferences, and topics or subject fields. We also
compare the differences between WordNet and
NODE. Finally, we compare these features to
those identified as significant in supervised
learning approaches.
1 Introduction
CL Research’s official submission for SENSEVAL-2
used WordNet as the lexical inventory. Separately,
we also used a machine-readable dictionary (The
New Oxford Dictionary of English, 1998) (NODE),
mapping NODE senses automatically into WordNet
senses. We did not submit these results, since we
were not sure of the feasibility of using one
dictionary mapped into another. Our initial results
(Litkowski, 2001)  proved to be much better than
anticipated, achieving comparable levels of precision
although at lower levels of recall, since not all senses
in NODE mapped into WordNet senses.
Subsequently, we examined our results in more detail
(Litkowski, 2002), primarily focusing on the quality
of the mapping and its effect on our performance
using NODE. This led us to the conclusion that we
had likely performed at a considerably higher level
using the NODE inventory, with an opportunity for
even better performance as we were able to exploit
much more information available in NODE.
We have now identified what features (i.e., sense
information) were used in our disambiguation. In
particular, we have examined the role of (1) default
sense selection, (2) idiomatic usage, (3) typing (e.g.,
transitivity), (4) syntactic and semantic clues, (5)
subcategorization patterns, (6) word form (e.g.,
capitalization, tense, or number), (7) selectional
preferences (for verbs and adjectives), (8) syntactic
usage (e.g., nouns as modifiers), (9) context (in
definitions and in examples), and (10) topic area
(e.g., subject fields associated with definitions).
Our methodology enables us to compare the
features relevant to disambiguation in WordNet and
in NODE, allowing us to pinpoint differences
between the two sense inventories.1 In addition,
comparing our findings with those identified in
supervised machine learning algorithms, we can see
patterns of similarity with our features.
In the following sections, we describe our
methods of dictionary preparation, our
disambiguation techniques, our methodology for
analyzing features and our results. We discuss these
findings in terms of what they say about the
differences between the information available in each
of the two sense inventories, the possible
generalizability of our analysis technique, and how
our features relate to those used by other
SENSEVAL  participants who used supervised
learning techniques. Finally, we describe our future
plans of analysis, based on attempting to merge
1We have not yet determined how decisive these
features are in making correct sense selections. The
present study should be viewed as an examination of
sense distinctions in lexical resources.
                     July 2002, pp. 47-53.  Association for Computational Linguistics.
                 Disambiguation: Recent Successes and Future Directions, Philadelphia,
                             Proceedings of the SIGLEX/SENSEVAL Workshop on Word Sense
supervised and unsupervised word-sense
disambiguation.
2 Dictionary Preparation
CL Research’s DIMAP (Dictionary Maintenance
Programs) disambiguates open text against WordNet
or any other dictionary converted to DIMAP. The
dictionaries used for disambiguation operate in the
background (as distinguished from the foreground
development and maintenance of a dictionary), with
rapid lookup to access and examine the multiple
senses of a word after a sentence has been parsed.
DIMAP allows multiple senses for each entry, with
fields for definitions, usage notes, hypernyms,
hyponyms, other semantic relations, and feature
structures containing arbitrary information.
For SENSEVAL-2, WordNet was entirely
converted to alphabetic format for use as the
disambiguation dictionary. Details of this conversion
(which captures all WordNet information) and the
creation of a separate “phrase” dictionary for all
noun and verb multiword units (MWUs) are
described in Litkowski (2001). In disambiguation, the
phrase dictionary is examined first for a match, with
the full phrase then used to identify the sense
inventory rather than a single word.
NODE was prepared in a similar manner, with
several additions. A conversion program transformed
the MRD files into various fields in DIMAP, the
notable difference being the much richer and more
formal structure (e.g., lexical preferences, grammar
fields, and subsensing). Conversion also
automatically created “kind” and “clue” regular
expression phrases under individual headwords, e.g.,
“(as) happy as a sandboy (or Larry or a clam)”
under happy was converted into a collocation pattern
for a sense under happy, written “(as|?) ~ as (a
sandboy | Larry | a clam)”, with the tilde marking the
target word. Further details on this conversion and
definition parsing to enrich the sense information are
also provided in Litkowski (2001). After parsing was
completed, a phrase dictionary was also created for
NODE.2
The SENSEVAL lexical sample tasks
(disambiguating one of 73 target words within a text
of several sentences) were run independently against
the WordNet and NODE sense inventories, with the
WordNet results submitted. To investigate the
viability of mapping for WSD, subdictionaries were
created for each of the lexical sample words. For
each word, the subdictionaries consisted of the main
word and all entries identifiable from the phrase
dictionary for that word. (For bar, in NODE, there
were 13 entries where bar was the first word in an
MWU and 50 entries where it was the head noun; for
begin, there was only one entry.)
The NODE dictionaries were then mapped into
the WordNet dictionaries (see Litkowski, 1999),
using overlap among words and semantic relations.
The 73 dictionaries for the lexical sample words gave
rise to 1372 WordNet entries and 1722 NODE
entries. Only 491 entries (of which, 418 were
MWUs) were common (i.e., no mappings were
available for the remaining 1231 NODE entries, all
of which were MWUs); 881 entries in WordNet were
therefore inaccessible through NODE. For the entries
in common, there was an average of 5.6 senses, of
which only 64% were mappable into WordNet, thus
creating our initial impression that use of NODE
would not be feasible.3
3 Disambiguation Techniques
Details of the disambiguation process are provided in
Litkowski (2001). In general, for the lexical sample,
the sentence containing the target word was first
parsed and the part of speech of the target word was
used to select the sense inventory. If the tagged word
was part of an MWU, the MWU's sense inventory
was used. The dictionary entry for the word was then
accessed. Before evaluating the senses, the topic area
of the context provided by the sentence was
“established”. Subject labels for all senses of all
content words in the context were tallied.
Each sense of the target was then evaluated,
based on the available information for the sense,
including type restrictions such as transitivity,
presence of accompanying grammatical constituents
such as infinitives or complements, selectional2WordNet definitions were not parsed. An experiment
showed the semantic relations identifiable through
parsing were frequently inconsistent with those in
WordNet.
3Note that a mapping from WordNet to NODE
generates similar mismatch statistics.
preferences for verbs and adjectives, form restrictions
such as number and tense, grammatical roles,
collocation patterns, contextual clue words,
contextual overlap with definitions and examples, and
topical area matches. Points were given to each sense
and the sense with the highest score was selected; in
case of a tie, the first sense was selected.
The top line of Table 1 shows our official results
using WordNet as the disambiguation dictionary,
with an overall precision (and recall) of 0.293 at the
fine-grained level and 0.367 at the coarse-grained
level. Disambiguating with NODE immediately after
the official submission and mapping its senses into
WordNet senses achieved comparable levels of
precision, with a coverage of 75% based on the
senses that could be mapped into WordNet, even
though the NODE coverage was 100%.
Since our original submission, we have
implemented many additional routines and improved
our NODE mapping to WordNet; our revised
precision shown in Table 1 are now 0.368 at the fine-
grained level and 0.462 at the coarse-grained level
using WordNet and 0.337 and 0.427 using NODE.
Of particular note are the facts that the mapping from
NODE to WordNet is now 89% and that precision is
comparable except for the verbs.
In Litkowski (2002), we examined the mapping
from NODE to WordNet in considerable detail.
Several of our findings are pertinent to our analysis
of the features affecting disambiguation. Table 1
reflects changes to the automatic mapping along with
hand changes. The automatic mapping changes
account for the change in coverage. The hand
mapping shows that the automatic mapping was
about 70% accurate. Interestingly, the hand changes
did not affect precision. In general, the fact that we
were able to achieve a level of precision comparable
to WordNet suggests the most frequent senses of the
lexical sample words were able to be disambiguated
and mapped correctly into WordNet.
The significant discrepancy between the entries
(all MWUs, 1231 entries in NODE not in WordNet
and 871 entries in WordNet not in NODE) in part
reflects the usual editorial decisions that would be
found in examining any two dictionaries. However,
since WordNet is not lexicographically based, many
of the differences are indicative of the idiosyncratic
development of WordNet. WordNet may identify
several types of an entity (e.g., apricot bar and
nougat bar), where NODE may use one sense (“an
amount of food or another substance formed into a
regular narrow block”) without creating separate
entries that follow this regular lexical rule.
For the most part, verb phrases containing
particles are equally present in both dictionaries (e.g.,
draw out and draw up), but NODE contains several
more nuanced phrases (e.g., draw in one's horns,
draw someone aside, keep one's figure, and pull
oneself together). NODE also contains many idioms
where a noun is used in a verb phrase (e.g., call it a
day, keep one's mouth shut, and go back to nature).
About 100 of our disambiguations using NODE were
to MWUs not present in WordNet (20% of our
coverage gap).
Of most significance to the sense mapping is the
classical problem of splitting (attaching more
importance to differences than to similarities,
resulting in more senses) and lumping (attaching
more significance to similarities than to differences,
resulting in fewer senses). Splitting accounts for the
remaining 80% gap in our coverage (where NODE
identified senses not present in WordNet). The effect
of lumping is more difficult to assess. When a NODE
definition corresponds to more than one sense in
WordNet, we may disambiguate correctly in NODE,
but receive no score since we have mapped into the
wrong definition; the WordNet sense groupings may
allow us to receive credit at the coarse grain, but not
at the fine grain. We have examined this issue in
more detail in Litkowski (2002), with the conclusion
that lumping reduces our NODE score since we are
unable to pick out the single WordNet sense answer.
More problematic for our mapping was the
absence of crucial information in WordNet. Delfs
Table 1. Lexical Sample Precision
Run
Adjectives Nouns Verbs Total
Items Fine Coarse Items Fine Coarse Items Fine Coarse Items Fine Coarse
WordNet Test 768 0.354 0.354 1726 0.338 0.439 1834 0.225 0.305 4328 0.293 0.367
NODE Test 420 0.288 0.288 1403 0.402 0.539 1394 0.219 0.305 3217 0.308 0.405
WordNet Test (R) 768 0.435 0.435 1726 0.430 0.535 1834 0.267 0.387 4328 0.368 0.462
NODE Test (R) 684 0.472 0.472 1567 0.429 0.537 1605 0.189 0.300 3856 0.337 0.427
(2001) described a sense for begin that has an
infinitive complement, but present only in an example
sentence and not explicitly encoded with the usual
WordNet verb frame. Similarly, for train, two
sentences were “tagged to transitive senses despite
being intransitive because again we were dealing with
an implied direct object, and the semantics of the
sense that was chosen fit; we just pretended that the
object was there.” In improving our disambiguation
routines, it will be much more difficult to glean the
appropriate criteria for sense selection in WordNet
without this explicit information than to obtain it in
NODE and map it into WordNet. Much of this
information is either not available in WordNet,
available only in an unstructured way, only implicitly
present, or inconsistently present.
4 Feature Analysis Methodology
4.1 Identifying Disambiguation Features
As indicated above, our disambiguation routines
assign point values based on a judgment of how
important each feature seems to be. The weighting
scheme is ad-hoc. For the feature analysis, we simply
recorded a binary variable for each feature that had
made a contribution to the final sense selection. In
particular, we identified the following features: (1)
whether the sense selected was the default (first)
sense (i.e., no other features were identified in
examining any of the senses), (2) whether the
identified sense was based on the occurrence of the
target word in an idiom, (3) whether a type
(specifically, transitivity) factored into the sense
selection, (4) whether the selected sense had any
syntactic or semantic clues, (5) whether a
subcategorization pattern figured into the sense
selection, (6) whether the sense had a specified word
form (e.g., capitalization, tense, or number), (7)
whether a syntactic usage was relevant (e.g., nouns
as modifiers or an adjective being used as a noun,
such as “the blind”), (8) whether a selectional
preference was satisfied (for verb subjects and
objects and adjective modificands), (9) whether we
were able to use a Lesk-style context clue from the
definitions or an example, and (10) topic area (e.g.,
subject fields, usage labels, or register labels
associated with definitions).
As the disambiguation algorithm proceeded, we
recorded each of the features associated with each
sense. After a sense was selected, the features
associated with that sense were written to a file (as a
hexadecimal number) for subsequent analysis. We
sorted the senses for each target word in the lexical
sample and summarized the features that were used
for all instances that had the same sense. We then
summarized the features over all senses and further
summarized them by part of speech. These results are
shown in Table 2.
The first column shows the number of instances
for each part of speech and overall. The second
column shows the number of instances where the
disambiguation algorithm selected the default sense.
These cases indicate the absence of positive
information for selecting a sense and may be
construed as indicating that the sense inventory may
not make sufficient sense distinctions. The default
numbers are somewhat misleading for verbs, where
the mere presence of an object (recorded in the “with”
column) sufficed to make a selection “non-default”.
As well, the default selections may indicate that our
disambiguation does not yet make full use of the
distinctions that are available. As we make
improvements in our algorithm, we would expect the
number of default selections to decrease.
Table 2. Comparative Analysis of Features Used in WordNet and NODE Disambiguation
Instance Default Idiom Kind Clue Context Topics Form With As Prefs POS
WordNet
768 556 79 0 0 190 0 0 15 0 1 Adjectives
1754 1140 293 0 0 536 0 0 29 0 0 Nouns
1804 436 161 0 2 576 0 0 984 0 0 Verbs
4326 2132 533 0 2 1302 0 0 1028 0 1 Total
NODE
768 324 81 0 2 249 168 14 11 11 33 Adjectives
1754 456 269 14 94 546 364 317 28 136 3 Nouns
1804 175 105 61 124 564 285 353 573 187 108 Verbs
4326 955 455 75 220 1359 817 684 612 334 144 Total
The significant difference in the number of
default selections between WordNet and NODE is a
broad indicator that there is more information
available in NODE than in WordNet. In examining
the results for individual words, even in cases where
the “default” (or first) sense was being selected, the
decision was being made in NODE based on positive
information rather than the absence of information.
Generally (but not absolutely), the intent of the
compilers of both WordNet and NODE is that the
first sense correspond to the most frequent sense. The
relative importance of the default sense indicated by
our results suggests the importance of ensuring that
this is the case. In a few instances, the first NODE
sense did not correspond to the first WordNet sense,
and we were able to obtain a much better result
disambiguating in NODE than in WordNet by using
an appropriate mapping from NODE to a second or
third WordNet sense. The significance of the default
sense is important in the selection of instances in an
evaluation such as SENSEVAL; if the instances do
not reflect common usage, WSD results may be
biased simply because of the instance selection.
The “idiom” column indicates those cases where
a phrasal entry was used to provide the sense
inventory. As pointed out above, these correspond to
the MWUs that were created and account for over
10% of the lexical instances.
The “kind” and “clue” columns correspond to
either strong or slightly weaker collocational patterns
that have been associated with individual senses.
These correspond to similarly named sense attributes
used in the Hector database for SENSEVAL-1,
which was the experimental basis for NODE. As can
be seen in the table, these were relevant to the sense
selection for about 6.5 percent of the instances for
NODE. We converted several of WordNet’s verb
frames into clue format; however, they did not show
up as features in our analysis, probably because our
implementation needs to be improved. We expect that
further improvements will obtain some cases where
these are relevant in the WordNet disambiguation (as
well as increasing the number of cases where these
are relevant to NODE senses).
The context column reflects the significance of
Lesk-style information available in the definitions and
examples. In general, it appears that about a third of
the lexical instances were able to use this
information. This reflects the extent to which the
dictionary compilers are able to provide good
examples for the individual senses. Since space is
limited for such examples, our results indicate that
there will an inevitable upper limit of the extent to
which disambiguation can rely on such information
(a conclusion also reached by (Haynes 2001)).
The potential significance of subject or topic
fields associated with individual senses is indicated
by the number of cases where NODE was able to use
this information (nearly 20 percent of the instances).
NODE makes extensive use of subject labels,
particularly in the MRD. We included many subject
labels, usage labels, and register labels in our
WordNet conversion, but these did not surface in our
disambiguation with WordNet. They were very rare
for the lexical items used in SENSEVAL. The value
shown here is similar to the results obtained by
Magnini, et al. (2001), but their low recall suggests
that for more common words, there will be a lower
opportunity for their use.
The word form of a lexical item also emerged as
being of some significance when disambiguating with
NODE, slightly over 16 percent. In NODE, this is
captured by such labels as “often capitalized” or
“often in plural form”. No comparable information is
available in WordNet.
Subcategorization patterns (indicated under the
“with” column) were very important in both
WordNet (based on the verb frames) and NODE,
relevant in 55% and 32% of the sense selections,
respectively. As indicated, the “with” category is also
important for nouns. For the most part, this indicates
that a given noun sense is usually accompanied by a
noun modifier (e.g., “metal fatigue”).
The “as” column corresponds to nouns used as
modifiers, verbs used as adjectives, and adjectives
used as nouns. These were fairly important for nouns
(7.7%) and verbs (10.3%).
The final column, “prefs”, corresponds to
selectional preferences for verb subjects and objects
and adjective modificands. In these cases, a match
occurred when the head noun in these positions either
matched literally or was a synonym or within two
synsets in the WordNet hierarchy. Although the
results were relatively small, this demonstrates the
viability of using such preferences.
Finally, anomalous entries in the table (e.g.,
nouns having subcategorization patterns used in the
sense selection) generally correspond to our parser
incorrectly assigning a part of speech (i.e., treating
the noun as a verb sense).
4.2 Variation in Disambiguation Features
Space precludes showing the variation in features by
lexical item. The attributes in NODE for individual
items varies considerably and the differences were
reflected in which features emerged as important.
For adjectives, idiomatic usages were significant
for free, green, and natural. Topics were important
for fine, free, green, local, natural, oblique, and
simple, indicating that many senses of these words
have specialized meanings. Form was important for
blind, arising from the collocation “the blind”. The
default sense was most prominent for colorless,
graceful (with only one sense in NODE), and
solemn. Context was important for blind, cool, fine,
free, green, local, natural, oblique, and simple,
suggesting that these words participate in common
expressions that can be captured well in a few choice
examples. Selectional preferences on the modificands
were useful in several instances.
For nouns, idioms were important for art, bar,
channel, church, circuit, and post. Clues (i.e., strong
collocations) were important for art, bar, chair, grip,
post, and sense. Topics were important for bar,
channel, church, circuit, day, detention, mouth,
nation, post, spade, stress, and yew (even though
yew had only one sense in NODE). Context was
important for art, authority, bar, chair, channel,
child, church, circuit, day, detention, facility,
fatigue, feeling, grip, hearth, lady, material, mouth,
nature, post, and restraint. The presence of
individual lexical items in several of these groupings
shows the richness of variations in characteristics,
particularly into specialized usages and collocations.
For verbs, idioms were important for call, carry,
draw, dress, live, play, pull, turn, wash, and work, a
reflection of the many entries where these words were
paired with a particle. Form was an important feature
for begin (over 50% of the instances), develop, face,
find, leave, match, replace, treat, and work.
Subcategorization patterns were important for all the
verbs. However, many verb senses in both WordNet
and NODE do not show wide variation in their
subcategorization patterns and are insufficient in
themselves to distinguish senses. Strong (“kind”) and
weak (“clue”) collocations are relatively less
important, except for a few verbs (collaborate, serve,
and work). Topics are surprisingly significant for
several verbs (call, carry, develop, dress, drive, find,
play, pull, serve, strike, and train), indicating the
presence of specialized senses. Context does not vary
significantly among the set of verbs, but it is a
feature in one-third of the sense selections. Finally,
selectional preferences on verb subjects and objects
emerged as having some value.
5 Generalizability of Feature Analysis,
Relation to Supervised Learning, and
Implications for Future Studies
The use of feature analysis has advanced our
perception of the disambiguation process. To begin
with, by summarizing the features used in the sense
selection, the technique identifies overall differences
between sense inventories. While our comments have
focused on information available in NODE, they
reflect only what we have implemented. Many
opportunities still exist and the results will help us
identify them.
In developing our feature analysis techniques, we
made lists of features available for the senses of a
given word. This gradually gave rise to the notion of
a “feature signature” associated with each sense. In
examining the set of definitions for each lexical item,
an immediate question is how the feature signatures
differ from one another. This allows us to focus on
the issue of adequate sense distinctions: what is it
that distinguishes each sense.
The notion of feature signatures also raises the
question of their correspondence to supervised
learning techniques such as the feature selection of
(Mihalcea & Moldovan, 2001) and the decision lists
used in WASPS (Tugwell & Kilgarriff 2001). This
raises the possibility of precompiling a sense
inventory and revising our disambiguation strategy to
identify the characteristics of an instance’s use and
then simply to perform a boolean conjunction to
narrow the set of viable senses.
The use of feature signatures also allows us to
examine our mapping functionality. As indicated
above, we are unable to map 10 percent of the senses
from NODE to WordNet, and of our mappings,
approximately 33 percent have appeared to be
inaccurate when examined by hand. When we
examine the instances where we selected a sense in
NODE, but were unable to map to a WordNet sense,
we can use these instances either to identify clear
cases where there is no WordNet sense.
In connection with the use of supervised learning
techniques, participants of other teams have provided
us with the raw data with which their systems made
their sense selections. The feature arrays from
(Mihalcea & Moldovan, forthcoming) identify many
features in common with our set. For example, they
used the form and part of speech of the target word;
this corresponds to our “form”. Their collocations,
prepositions after the target word, nouns before and
after, and prepositions before and after correspond to
our idioms, “clues”, and “with” features.
The array of grammatical relations used with
WASPS (Tugwell & Kilgarriff,  2001)  (such as
bare-noun, plural, passive, ing-complement, noun-
modifier, PP-comp) correspond to our “form”,
“clue”, “with”, and “as” features.
The data from these teams also identifies bigrams
and other context information. Pedersen (2001) also
provided us with the output of several classification
methods, identifying unigrams and bigrams found to
be significant in sense selection. These data
correspond to our “context” feature.
We have begun to array all these data by sense,
corresponding to our detailed feature analysis. Our
initial qualitative assessment is that there are strong
correspondences among the different data set. We
will examine these quantitatively to assess the
significance of the various features. In addition, while
several features are already present in WordNet and
NODE, we fully expect that these other results will
help us to identify features that can be added to the
NODE sense inventory.
6 Conclusions
Our analysis has identified many characteristics of
sense distinctions, but indicates many difficulties in
making such distinctions in WordNet (but also
NODE). It is questionable whether WSD has been
fully tested without a carefully drawn sense
inventory. A lexicographically-based sense inventory
shows considerable promise and invites the WSD
community to pool its resources to come up with
such an inventory.
Acknowledgments
I wish to thank Oxford University Press for allowing
me to use their data, and particularly to Rob Scriven,
Judy Pearsall, Glynnis Chantrell, Patrick Hanks,
Catherine Soanes, Angus Stevenson, Adam
Kilgarriff, and James McCracken for their invaluable
discussions, to Rada Mihalcea, Ted Pedersen, and
David Tugwell for making their data available, and
to the anonymous reviewers.

References
Delfs, L. (2001, 6 Sep). Verb keys. (Personal communication)
Haynes, S. (2001, July). Semantic Tagging Using WordNet
Examples. In Association for Computational Linguistics
SIGLEX Workshop (pp. 79-82). Toulouse, France.
Litkowski, K. C. (2002). SENSEVAL Word-Sense
Disambiguation Using a Different Sense Inventory and
Mapping to WordNet (CL Research No. 02-01). .
Damascus, MD.
Litkowski, K. C. (2001, July). Use of Machine Readable
Dictionaries for Word-Sense in SENSEVAL-2. In
Association for Computational Linguistics SIGLEX
Workshop (pp. 107-110). Toulouse, France.
Litkowski, K. C. (1999, June). Towards a Meaning-Full
Comparison of Lexical Resources. In Association for
Computational Linguistics SIGLEX Workshop (pp. 30-7).
College Park, MD.
Magnini, B., Strapparava, C., Pezzulo, G., & Gliozzo, A.
(2001, July). Using Domain Information for Word Sense
Disambiguation. In Association for Computational
Linguistics SIGLEX Workshop (pp. 111-4). Toulouse,
France.
Mihalcea, R., & Moldovan, D. (2001, July). Pattern Learning
and Active Feature Selection for Word Sense
Disambiguation. In Association for Computational
Linguistics SIGLEX Workshop (pp. 127-30). Toulouse,
France.
Mihalcea, R., & Moldovan, D. (Forthcoming). Word Sense
Disambiguation with Pattern Learning and Active Feature
Selection. Journal of Natural Language Engineering.
The New Oxford Dictionary of English. (1998) (J. Pearsall,
Ed.). Oxford: Clarendon Press.
Pedersen, T. (2001, July). Machine Learning with Lexical
Features: The Duluth Approach to SENSEVAL-2. In
Association for Computational Linguistics SIGLEX
Workshop (pp. 139-142). Toulouse, France.
Tugwell, D., & Kilgarriff, A. (2001, July). WASP-Bench: A
Lexicographic Tool Supporting Word Sense
Disambiguation. In Association for Computational
Linguistics SIGLEX Workshop (pp. 151-4). Toulouse,
France.
