Declaration of Rachel J. Watters on Authentication of Publication
`
`I, Rachel J. Watters, am a librarian, and the Director of Wisconsin TechSearch
`
`("WTS"), located at 728 State Street, Madison, Wisconsin, 53706. WTS is an
`
`interlibrary loan department at the University of Wisconsin-Madison. I have worked as
`
`a librarian at the University of Wisconsin library system since 1998. I have been
`
`employed at WTS since 2002, first as a librarian and, beginning in 2011, as the Director.
`
`Through the course of my employment, I have become well informed about the
`
`operations of the University of Wisconsin library system, which follows standard library
`
`practices.
`
`This Declaration relates to the dates of receipt and availability of the following:
`
`Beckwith, R. and Miller, G.A. (1990) Implementing a lexical
`network. International Journal of Lexicography, 3(4), 302-312.
`
`Standard operating procedures for materials at the University of Wisconsin(cid:173)
`
`Madison Libraries. When an issue was received by the Library, it would be checked in,
`
`stamped with the date of receipt, added to library holdings records, and made available
`
`to readers as soon after its arrival as possible. The procedure normally took a few days
`
`or at most 2 to 3 weeks.
`
`Exhibit A to this Declaration is true and accurate copy of the journal issue cover
`
`with library date stamp of International Journal of Lexicography (1990), from the
`
`University of Wisconsin-Madison Library collection. Exhibit A also includes an
`
`excerpt of pages 302 to 312 of that issue, showing the article entitled Implementing a
`
`1
`
`Page 1 of 17
`
`GOOGLE EXHIBIT 1028
`
`

`

`Declaration of Rachel J. Watters on Authentication of Publication
`
`lexical network (1990). Based on this information, the date stamp on the journal cover
`
`indicates Implementing a lexical network (1990) was received by the University of
`
`Wisconsin-Madison Libraries on January 4, 1991.
`
`Based on the information in Exhibit A, it is clear that the issue was received by
`
`the library on or before January 4, 1991 , catalogued and available to library patrons
`
`within a few days or at most 2 to 3 weeks after January 4, 1991.
`
`I declare that all statements made herein of my own knowledge are true and that
`
`all statements made on information and belief are believed to be true; and further that
`
`these statements were made with the knowledge that willful false statements and the like
`
`so made are punishable by fine or imprisonment, or both, under Section 1001 of Title 18
`
`of the United States Code.
`
`Date: March 3, 2020
`
`Wisconsin TechSearch
`Memorial Library
`728 State Street
`Madison, Wisconsin 53706
`
`Director
`
`2
`
`Page 2 of 17
`
`

`

`ISSN 0950-3846
`
`International Journal of
`
`Lexicography
`
`Volume 3 Number 4
`Winter 1990
`
`Page 3 of 17
`
`

`

`International Journal of
`Lexicography
`
`Editor: Robert Ilson (58 Antrim Mansions, Antrim Road, London NW3 4XU, UK)
`Editorial Board
`European Association for Lexicography (EURALEX) - Executive Board:
`M Alvar Ezquerra, Ju D Apresjan, B T Atkins, H Bejoint, R R K Hartmann,
`F E Knowles, 0 Norling-Christensen, M Snell-Hornby, S Ter-Minasova, A Zampolli ,
`R F Ilson (co-opted)
`Dictionary Society of North America (DSNA)-Executive Board:
`J Algeo, W Frawley, D B Guralnik , V G McDavid , L T Milic, R J Steiner, L K
`Vandersall
`
`J Aitchison (London)
`J M Channell (Nottingham)
`COBUILD (Birmingham)
`Cordell Collection of Dictionaries
`(Terre Haute)- David E Vancil
`A P Cowie (Leeds)
`D Crystal (Bangor)
`Dictionaries Editorial Committee
`(OUP)- Lesley Burnett
`Dictionary Research Centre
`(Exeter)- R R K Hartmann
`F Dubois Charlier (Paris)
`Erlanger Zentrum fiir
`W orterbuch forsch ung (Erl a ngen(cid:173)
`N iirnberg)- F J Hausmann
`C J Fillmore (Berkeley)
`
`W Frawley (Newark, Delaware)
`Y Ikegami (Tokyo)
`Istituto di Linguistica
`Computazionale (Pisa)(cid:173)
`A Zampolli
`LADS (Liege)
`T McArthur (Cambridge)
`W Martin (Amsterdam)
`Igor Mel 'cuk (Montreal)
`MLA Discussion Group on Lexicography
`A W Read (New York)
`J Rey-Debove (Paris)
`SILEX (Lille)- P Corbin
`L Urdang (Old Lyme)
`H E Wiegand (Heidelberg)
`L Zgusta (Urbana-Champaign)
`
`Subscriptions
`Published quarterly, a t an annual subscription of £35 in the UK, US$68 in N. America, £42 elsewhere;
`single issues £10.50 in the UK, US$20 in N. America, £12.50 elsewhere. Prices include postage by
`surface mail or, for subscribers in the USA, Canada , India, Japan, Australia and New Zealand, by Air
`Speeded Post. Airmail rates are available on request.
`Contributors to the Journal are entitled to one year's subscription at half price.
`
`Order information
`Payment is required with all orders and subscriptions are accepted and entered by the volume. Payment
`may be made by cheque or Eurocheque (payable to Oxford University Press), National Girobank
`(account 500 1056), credit card (Access, Visa, American Express, Diners Club), or UNESCO coupons.
`Please send orders and requests for sample copies to the Journals Subscriptions Department , Oxford
`University Press, Pinkhill House, Southfield Road, Eynsham ox8 IJJ, UK. Telex 83147 OXPRESG.
`
`Advertising
`Advertisements are welcomed and rates will be quoted on request. Enquiries should be addressed to the
`Journals Advertising Coordinator, Journals Production Department, Oxford University Press, Pinkhill
`House, Southfield Road , Eynsham ox8 IJJ , UK.
`Authorization to photocopy items for internal or personal use, or the internal or personal use of
`specific clients, is granted by Oxford University Press for users in the USA registered with the
`Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee of
`$3.00 per copy is paid directly to CCC, 27 Congress Street, Salem, MA 01970. For those organizations
`that have been granted a photocopy licence by CCC a separate system of payment has been arranged.
`The fee code for users of the Transactional Reporting Service is: 0950-3846/90 $3.00.
`© 1990 Oxford University Press
`
`Page 4 of 17
`
`

`

`International Journal
`of Lexicography
`
`Volume 3 Number 4 Winter 1990
`
`SPECIAL ISSUE
`
`WordNet:
`An On-Line Lexical Database
`
`GUEST EDITOR
`GEORGE A. MILLER
`
`Page 5 of 17
`
`

`

`International Journal of
`Lexicography
`Volume 3 Number 4 Winter 1990
`
`Contents
`
`Introduction to WordNet: An On-line Lexical Database.
`George A. Miller, Richard Beckwith, Christiane Fellbaum,
`Derek Gross and Katherine J. Miller
`
`Nouns in WordNet: A Lexical Inheritance System.
`George A. Miller
`
`Adjectives in WordNet. Derek Gross and Katherine J. Miller
`
`English Verbs as a Semantic Net. Christiane Fellbaum
`
`Implementing a Lexical Network. Richard Beckwith and
`George A. Miller
`
`The EURALEX Bulletin
`
`235
`
`245
`
`265
`
`278
`
`302
`
`Oxford University Press
`
`Page 6 of 17
`
`

`

`Implementing a Lexical Network*
`Richard Beckwith and George A. Miller, Princeton University
`
`Abstract
`
`The computer implementation of the lexical database described in the preceding papers
`falls naturally into two parts. WordNet is the lexical database; LexPert is the suite of
`software tools used to build and access the database. In addition to the lexical
`information contained in the source files written by the lexicographers, WordNet also
`contains an index of the familiarity of each word.
`LexPert includes routines for verifying the syntax of the lexical source files, for
`converting the source files into a database, for adding familiarity indices to the database,
`for searching the database, for dealing with inflectional morphology, and for displaying
`information for the user. The specifications for these routines are described.
`
`Lexicographers must be concerned with the presentation as well as the content
`of their work, and this concern is heightened when presentation moves from the
`printed page to the computer monitor. Printed dictionaries have become
`relatively standardized through many years of publishing (Vizetelly, 1915);
`expectations for electronic lexicons are still up for grabs. Indeed, computer
`technology itself is evolving rapidly; an indefinite variety of ways to present
`lexical information is possible with this new technology, and the advantages
`and disadvantages of many possible alternatives are still matters for experimen(cid:173)
`tation and debate. Given this degree of uncertainty, manner of presentation
`must be a central concern for the electronic lexicographer.
`WordNet is a pioneering excursion into this new medium. Considerable
`attention has been devoted to making it useful and convenient, but the
`solutions described here are unlikely to be the final word on these matters. It is
`hoped that readers will not merely note the shortcomings of this work, but will
`also be inspired to make improvements on it.
`One's first impression of WordNet is likely to be that it is an on-line
`thesaurus. It is true that sets of synonyms are basic building blocks, and with
`nothing more than synsets the system would have all the power of a thesaurus.
`When short glosses are added to the synsets, it resembles an on-line dictionary
`that has been supplemented with synonyms for cross referencing (Calzolari,
`1988). But as readers of the preceding articles will appreciate, WordNet
`includes much more information than that. In an attempt to model the lexical
`
`* Contributions to the computer programs in LexPert have been made by Marie Bienkowski,
`George Collier, Michael Colon, David Crabb, Andrew Gomory, Derek Gross, Brian Gustafson,
`Yana Kane, Benjamin Martin, Antonio Romero, Daniel Teibel, and Benjamin Wilkes. 'Sun
`Workstation' and 'Unix' are trademarks of Sun Microsystems and AT&T Bell Laboratories,
`respectively.
`
`International Journal of Lexicography, Vol. 3 No. 4
`
`© 1990 Oxford University Press
`0950-3846/90 $3.00
`
`Page 7 of 17
`
`

`

`Implementing a Lexical Network 303
`
`knowledge of a native speaker of English, WordNet has been given detailed
`information about relations between word forms and synsets. How this
`relational structure should be presented to the user raises questions that outrun
`the experience of conventional lexicography.
`In developing this on-line database, it has been convenient (even, perhaps,
`necessary) to divide the work into two interdependent tasks which bear a vague
`similarity to the traditional tasks of writing and printing a dictionary. One task
`was to write the source files that contain the basic lexical data described in the
`preceding articles; the contents of those files are the lexical substance of
`WordNet. The second task was to create a set of computer programs that
`would accept the source files and do all the work leading ultimately to the
`generation of a display for the user; those programs, collectively, are called
`LexPert. The purpose of this article is to provide a general description of the
`functions performed by the LexPert programs.
`Both W ordNet and LexPert were developed on Sun 3 and Sun 4 worksta(cid:173)
`tions under the Unix programming environment. It is planned to develop
`versions that will run on less powerful computers, but they are not yet available
`as this is written.
`
`Source Files
`
`The source files are the files written by lexicographers. They are the product of a
`detailed relational analysis of lexical semantics: a variety of lexical and semantic
`relations are used to represent the organization of lexical knowledge. Two kinds
`of building blocks are distinguished in the source files: word forms and word
`meanings. Word forms are represented in their familiar orthography; word
`meanings are represented by synsets - lists of synonymous word forms. Two
`kinds of relations are recognized, lexical and semantic. Lexical relations hold
`between word forms; semantic relations hold between word meanings, or
`between word forms and word meanings. The contents of the source files have
`been described in some detail in the preceding articles.
`The following strings in the source files are treated as word forms: (1) an
`orthographic representation of a word, (2) an orthographic representation
`followed by an integer for disambiguation, (3) an orthographic representation
`followed by a usage note, as for postnominal adjectives, or (4) an expression in
`square brackets, used to code word:word relations. Since the information
`associated with each word form necessarily includes the synset it belongs to, the
`word forms are semantically unambiguous and it is possible to associate with
`them such additional information as any direct antonyms they may have, the
`kinds of sentence frames they can enter into, the sets of inflections they can
`take, or usage notes.
`Word meanings are represented by synsets. Strings in the source files that
`conform to the following syntactic rules are treated as synsets:
`
`(1) Each synset begins with a curly bracket, '{'
`(2) Each synset is terminated with a curly bracket, '}'
`(3) Each element within a synset consists of
`
`Page 8 of 17
`
`

`

`304 Richard Beckwith and George A. Miller
`
`(a) to code semantic relations: a word form, optionally preceded by a
`filename to indicate a word in a different file, followed by a comma,
`followed either by a space or a relational pointer and a space, or
`(b) to code lexical relations: an expression inside square brackets, '[' and ']',
`consisting of two or more elements of type (a), or
`(c) a gloss inside parentheses, '(' and ')'.
`This syntax is not inherently meaningful but adherence to it eases the
`computational burden of processing the source files. The information associ(cid:173)
`ated with each word meaning in the source files includes its semantic relations
`to other meanings. Table 1 collects those relations and their pointers in a single
`place; for elaboration, consult the preceding articles.
`In addition to the source files for nouns, verbs, and adjectives, WordNet also
`includes information about the familiarity of each word in the source files.
`
`Index of Familiarity
`One of the best known and most important psycholinguistic facts about the
`mental lexicon is that some words are much more fami liar than others. The
`familiarity of a word is known to influence a wide range of performance
`variables: speed of reading, speed of comprehension, ease of recall, probability
`of use. The effects are so ubiquitous that experimenters who hope to study
`anything else must take great pains to equate the words they use for familiarity .
`To ignore this variable in a lexical database that is supposed to reflect
`psycholinguistic principles would be unthinkable.
`In order to incorporate differences in familiarity into WordNet, a syntacti(cid:173)
`cally tagged index of familiarity is associated with each word form. This index
`does not reflect all of the consequences of differences of familiarity - some
`theorists would ask for strength indices associated with each relation - but
`accurate information on all of the consequences is not easily obtained. The
`present index is a first step.
`Frequency of use is usually assumed to be the best indicator of familiarity.
`The closed class words that play an important syntactic role are the most
`frequently used, of course, but even within the open classes of words there are
`large differences in frequency of occurrence that are assumed to correlate with -
`
`Table 1. The relations, R, and their pointers, p, in WordNet (x p-y iff
`Rp(x)= y).
`
`Nouns
`
`Synonym
`Antonym
`Hyponym
`Hypernym
`Meronym
`Holonym
`
`{ .. }
`[.!)
`
`@
`#
`%
`
`Verbs
`
`Adjectives
`
`Synonym
`Antonym
`Similar
`
`{ .. }
`(.!)
`&
`
`Synonym
`Antonym
`Troponym
`Hypernym
`Presupposition
`Inclusion
`Cause
`
`{ .. }
`[.!)
`
`@
`*
`*
`>
`
`Page 9 of 17
`
`

`

`Implementing a Lexical Network 305
`
`or to explain - the large differences in familiarity. The frequency data that are
`readily available in the technical literature, however, are inadequate for a
`database as extensive as WordNet. Thorndike and Lorge (1944) published data
`based on a count of some 5,000,000 running words of text, but they reported
`their results only for the 30,000 most frequent words. Moreover, they defined a
`'word' as any string of letters between successive spaces, so their counts for
`homographs are untrustworthy; there is no way to tell , for example, how often
`lead occurred as a noun and how often as a verb. Francis and Kucera (1982) tag
`words for their syntactic category, but they report results for only 1,014,000
`running words of text - or 50,400 word types, including many proper names -
`which is not a large enough sample to yield reliable counts for infrequently used
`words. (A comfortable rate of speaking is about 120 words/minute, so that
`1,000,000 words corresponds to 140 hours, or about two weeks of normal
`exposure to language.)
`Fortunately, an alternative indicator of familiarity is available. It has been
`known at least since Zipf ( 1945) that frequency of occurrence and polysemy are
`correlated. That is to say, on the average, the more frequently a word is used the
`more different meanings it will have in a dictionary. An intriguing finding in
`psycholinguistics (Jastrezembski, 198 I) is that polysemy seems to predict lexical
`access times as well as frequency does. Indeed, if the effect of frequency is
`controlled by choosing words of equivalent frequencies, polysemy is still a
`significant predictor of lexical decision times.
`Instead of using frequency of occurrence as an index of familiarity, therefore,
`WordNet uses polysemy. This measure can be determined from an on-line
`dictionary. If an index value of 0 is assigned to words that do not appear in the
`dictionary, and if values of I or more are assigned according to the number of
`senses the word has, then an index value can be made available for every word
`in every syntactic category. Associated with every word in WordNet, therefore,
`there is an integer that represents a count ( of the Collins English Dictionary) of
`the number of senses that word has when it is used as a noun, verb, or adjective.
`A simple example of how the familiarity index might be used is shown in
`Table 2. If, say, the superordinates of bronco are requested, WordNet can
`respond with the sequence of hypernyms shown in Table 2. Now, if all the terms
`
`Table 2. Hypernyms of bronco and their familiarity indices.
`
`Word
`
`Polysemy
`
`bronco
`@ -+horse
`@ -+equid
`@ -+odd-toed ungulate
`@ -+ herbivore
`@ -+mammal
`@ -+vertebrate
`@ -+anima l
`@ -+ organism
`
`14
`0
`0
`
`4
`2
`
`Page 10 of 17
`
`

`

`306 Richard Beckwith and George A. Miller
`
`with a familiarity index (polysemy count) of O or 1 are omitted, which are
`primarily technical terms, the hypernyms of bronco include simply: bronco
`@ -+horse@ -+animal@ -+organism. This shortened chain is much closer to
`what a layman would expect. The index of familiarity should be useful,
`therefore, when making suggestions for changes in wording. If a user requests a
`more familiar word, LexPert can search through WordNet for one with a
`higher index.
`WordNet would be a better simulation of human semantic memory if a
`familiarity index could be assigned to form-meaning pairs, rather than to word
`forms. The noun dog, for example, is used far more often with the meaning
`{dog, canine, pooch} than with the meaning {dog, detent, pawl, click}, yet both
`are presently assigned the same index, 16.
`
`Grinder
`
`In order to facilitate machine retrieval of information in WordNet, several
`stages of processing are applied to the source files. The program that does this
`processing is Grinder.
`The first step is to check that synsets in the source files are well-formed. This
`check is performed by a component routine, Verify, which is available for
`lexicographers to use prior to submitting their source files to Grinder. Verify
`also looks for two kinds of pointer errors in the noun and verb files: either a
`pointer is vacuous because it designates a non-existent synset, or it is
`ambiguous because it designates more than one synset. These errors must be
`corrected before a file can be incorporated into the database.
`Since each syntactic category has a different semantic structure, Grinder's
`next step depends on the syntactic category being processed. For nouns,
`Grinder combines the 25 source files into two: an index file and a data file. The
`index file is an alphabetical list of all the nouns in WordNet, along with the
`absolute addresses of their synsets in the data file and any information that is
`associated with undisambiguated word forms (e.g. , the index of familiarity).
`The data file contains the synsets, which in turn contain their lexical and
`semantic pointers. Grinder inserts any converse pointers that were omitted by
`the lexicographer (e.g., it inserts pointers to hyponyms), and then substitutes
`the addresses of the appropriate synsets in place of the lexicographers' semantic
`pointers. Each address bears a label that codes the kind of relation. If the
`relation is lexical, the address is for a specific word in a specific synset; if the
`relation is semantic, the address is for a synset. The synsets also contain the
`definitional gloss, if there is one in the source file.
`The source files for verbs are processed in much the same way as the noun
`files: they are combined into an index file and a data file. Associated with each
`verb form, however, is a sentence frame indicating something about the
`syntactic contexts in which the verb can be used - different verbs in the same
`synset may have different privileges of occurrence.
`Adjectives are also converted into an index and a data file, but adjectives
`demand special treatment. The two adjective source files are very different from
`each other. Pertainyms cannot be entered into the database until the addresses
`
`Page 11 of 17
`
`

`

`Implementing a Lexical Network 307
`
`of the pertinent noun or verb forms are determined. And, although the clusters
`of ascriptive adjectives are related only by antonymy and similarity, antonymy
`is a lexical relation between adjective forms, whereas similarity is a special
`relation between adjective meanings and adjective forms.
`Once the database is created by Grinder, there are two types of processed
`files: index files and processed versions of the original source files. A pair of
`these processed files exists for each syntactic category, resulting in a total of six
`such files.
`Grinder evolved progressively as work on WordNet proceeded, so the
`present account is, in effect, shooting at a moving target. Grinder was first
`written in Lisp, then C, and is now written in a combination of C and the Unix
`utility programs lex and yacc, thus enhancing the speed and portability of
`LexPert generally, and Grinder in particular.
`
`Retrieving Lexical Information
`
`In order to give a user access to information in the database, an interface is
`required. Interfaces enable end users to retrieve the lexical data and display it
`via a window-based tool or the command line. When considering the role of the
`interface, it is important to recognize the difference between a printed diction(cid:173)
`ary and a lexical database. LexPert creates its responses to a user's requests on
`the fly . Unlike an on-line version of a printed dictionary, where information is
`stored in a fixed format and displayed on demand, WordNet's information is
`stored in a format that would be meaningless to an ordinary reader. The
`interface provides a user with a variety of ways to retrieve and display lexical
`information. Different interfaces can be created to serve the purposes of
`different users, but all of them will draw on the same underlying lexical
`database.
`In the SunView interface, which is used on the Sun workstations, there is a
`window that provides a set of menus that make querying the database easy and
`intuitive. Users must initially indicate the word that they want to query -
`the
`target or seed word for the search - either by highlighting it with the mouse or
`entering it at a prompt. After the word is selected, buttons appear that indicate
`the syntactic categories WordNet has assigned to that word. The user clicks on
`one of the buttons, causing an appropriate menu to appear. This menu contains
`only choices that are relevant to the specific syntactic category that was
`indicated by the user; options that are relevant for that syntactic category but
`unavailable for that word are dimmed. Once the menu appears, the user selects
`one of the menu items, which causes the program to search for that information
`and display it in the window. The user can cancel a query by releasing the
`mouse button outside the menu area.
`The first step that the program takes in any search is to locate the word in the
`index files. If it fails to find an entry for the word, the program will search for a
`morphological variant of the word (see below). If it fails in this second attempt,
`an error message is printed in the window or (when using the command-line
`version) to the standard output (typically the terminal screen) indicating that
`no information is available for this word.
`
`Page 12 of 17
`
`

`

`308 Richard Beckwith and George A. Miller
`
`The search process is the same regardless of the type of search requested. The
`first step is to retrieve the index entry located in the appropriate index file - this
`will contain a list of addresses of the synsets in the data file in which the word
`appears. Then each of these synsets in the data file is searched for the requested
`information, which is then retrieved and formatted for output. Search is
`complicated by the fact that each synset containing the seed word also contains
`pointers to other synsets in the data file that may need to be retrieved and
`displayed, depending on the search type. For example, each synset in the
`hypernymic pathway points to the next synset in the hierarchy. If a user
`requests a recursive search on hypernyms (see Table 2), a hierarchical retrieval
`process is repeated until a synset is encountered that contains no further
`hypernymic pointers.
`All searches, regardless of the information they are searching for , are
`accomplished by a small set of flexible functions that were written in C + + .
`Table 3 contains a brief description of the different functions, which are used in
`various combinations to accomplish all of the searches available in WordNet.
`(Converse search primitives are available for presupposition, troponymy,
`cause, and inclusion, just as for hypernymy/hyponymy and meronymy/holo(cid:173)
`nymy, but appropriate names for those semantic relations have not yet been
`selected.) In order to maximize their flexibility, each function takes a small
`number of arguments that control its behavior. Thus, the same function may be
`used in recursive and non-recursive searches and in the retrieval of many
`different types of semantic information.
`The search code is a program built out of the search primitives in Table 3.
`Because LexPert uses a library of search primitives, it is possible to add new
`types of searches quite easily. The program just described has been developed in
`order to construct and test WordNet and contains a special set of searches that
`have been found useful for that purpose. That particular program does not
`exhaust the search capacity of the system; other workers might design different
`interfaces using a different search code. The present description will be limited
`to a few particular examples, simply to illustrate how search primitives can be
`combined to perform different searches.
`The simplest search is the search for synonyms, which merely calls the
`synonymy search primitive. The search is restricted to a single syntactic
`category and returns each of the synsets that the seed word is in. For example, if
`the user searches for synonyms of the noun dog, the search program returns two
`synsets: {dog, click, detent, pawl} and {pooch, dog, canine}. If the user searches
`for synonyms of the verb dog, the program returns one synset: {track, go after,
`dog, tag, trail, chase}.
`The antonym search involves more than the antonymy search primitive, in
`that it returns more than direct antonyms. Antonymy is a lexical relation, so the
`antonymy search primitive returns only words that are direct antonyms.
`LexPert's search code for antonyms reports both direct and indirect antonyms,
`so it must combine the antonymy search primitive with the synonymy search
`primitive or, in the case of adjectives, with the search primitive for similars. For
`example, if the user asks for antonyms of the noun weakness, LexPert returns
`strength. If the user asks for the antonyms of the verb weaken, LexPert returns
`
`Page 13 of 17
`
`

`

`'D
`0
`<.,.)
`
`0 .... ;,;-
`?
`~ z 0
`)< ;s·
`r-0
`
`5·
`;?.
`0 a 0
`3
`
`'O
`
`Pl
`(IQ
`
`name of source file
`definition with synsets
`sense count of seed
`returns true if seed is adjective cluster head
`restrictions on relative location of head noun
`similars (near synonyms) of seed's cluster head
`frames that are permissible with seed
`
`verbs that seed properly temporally includes
`verbs that seed causes
`verbs that are a manner of seed
`verbs presupposed by seed
`things that seed is part of
`parts of seed
`children of seed
`parents of seed
`
`N V A
`N V A
`N V A
`A
`A
`A
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`V
`
`V
`V
`V
`V
`
`-
`
`-
`
`-
`
`-
`
`-
`
`N
`N
`N V
`N V
`
`-
`
`-
`
`source
`gloss
`familiarity
`head
`usage
`similar
`frame
`
`inclusion
`cause
`troponymy
`presupposition
`holonymy
`meronymy
`hyponymy
`hypernymy
`
`information
`Other
`
`relations
`Synset
`
`antonyms of seed
`synonyms of seed
`
`N V A
`N V A
`
`antonymy
`synonymy
`
`relations
`Word
`
`Information returned
`
`Category
`
`Function
`
`Table 3. Summary of primitive search commands.
`
`Page 14 of 17
`
`

`

`310 Richard Beckwith and George A. Miller
`
`strengthen, but then, because weaken is a synonym of de-escalate, the program
`also returns escalate. If the user asks for antonyms of the adjective weak,
`LexPert returns not only strong, but 15 indirect antonyms.
`Meronymy is coded only in the noun file, so part-oriented searches work only
`with nouns. The meronymy search primitive returns the synsets that are coded
`as parts in the synset containing the seed word. But meronymy is iQherited. In
`order to give a user all of the parts of the seed, not only the parts codea with the
`seed but all parts coded with hypernyms of the seed must be listed. LexPert's
`search code for this request must combine the meronymy search primitive with a
`recursive search for hypernyms. Moreover, since parts can have parts, the
`search must also look for meronyms of meronyms.
`Three options to the search code that can be invoked for words in any
`syntactic category are gloss, fami liarity, and source file. These options return
`the lexicographers' definitions of synsets, an index of the relative polysemy of
`the seed word, and the names of the WordNet raw source files that contain the
`seed word. The gloss and familiarity search primitives are meant to be used in
`combination with other searches. If a user asks for definitions, LexPert will
`return a definition with each synset that has one. If a user asks for familiarity,
`LexPert will return a sense count with every word that has one. The source file
`option can be used alone; it will report in which raw file the lexicographer
`included the seed word.
`
`Morphy
`
`Dictionaries ordinarily hang their information on head words, without separate
`listings for inflectional (or many derivational) forms of the word. In a printed
`dictionary, that practice causes little trouble; with a few highly irregular
`exceptions, morphologically related words are generally similar enough in
`spelling to the reference form that the eye, aided by boldface type, quickly picks
`them up. In an electronic dictionary, on the other hand, when an inflected form
`is requested, the response is likely to be a frustrating announcement that the
`word is not in the database. Users are required to know the reference form of
`every word they want to look up. In order to spare users the trouble of affix
`stripping, therefore, LexPert includes Morphy, a large program that gives
`WordNet some intelligence about English morphology.
`Morphy is designed as a stand-alone front end to the WordNet database. It
`includes inflectional morphological variants of English words, thus allowing the
`user (and automated software such as that described below) to find the root
`morphemes of inflectional variants as well as particular morphological variants
`of a root. For example, singular nouns are asso.ciated with their plurals;
`verb infinitives have pointers to (a) past and simple present tense, (b) the
`gerundive/progressive form, (c) past participles, and (d) third per