By Dmitri Popov
WordNet is a reference tool that lets you study the connections between words. In the words of its developers, WordNet is "... an online lexical reference system whose design is inspired by current psycholinguistic theories of human lexical memory. English nouns, verbs, adjectives and adverbs are organized into synonym sets, each representing one underlying lexical concept. Different relations link the synonym sets." (http://wordnet.princeton.edu/). In other words, since all the entries in WordNet are organized into synonym sets (synsets) and they contain definitions and examples, WordNet can be used both as a thesaurus and a conventional dictionary.
However, what makes WordNet a unique reference tool is that every synset is connected to other synsets via a number of relations. This means that for each word in WordNet, you can retrieve not only its synonyms, but also hypernyms, hyponyms, meronyms, and holonyms.
A hypernym describes the x is a kind of y relationship between words. For example, in the relationship an oak is a kind of tree, tree is a hypernym, or, in other words, tree is a superordinate of oak. Hyponym also describes the x is a kind of y relationship, but in reverse. In the previous example, oak is a hyponym, or a subordinate, of tree. Meronym denotes a constituent part of, or a member of something. For example, engine is a meronym of airplane. Holonym is a meronym in reverse. In the example above, airplane is a holonym of engine. There are a few other terms that are used in WordNet, but these four are enough to give you an indication that WordNet is more than an ordinary digital dictionary.
Most distributions provide a packaged version of WordNet, and you can install it using your package management tool. If your Linux distro doesn't include the WordNet package, you can download a tarball from WordNet's official website (http://wordnet.princeton.edu/obtain) and install it using the standard installation routine:
./configure make make install
The WordNet browser requires the Tcl/Tk packages, which must be installed before you build and install WordNet. Finally, if you use a Live CD Linux distro that supports klik, you can install WordNet from http://wordnet.klik.atekon.de/.
Once you have completed the installation, you are ready to explore WordNet. The man pages provide an exhaustive overview of WordNet's commands. See the box labeled "WordNet Commands" for a summary of command exmples.
-hype {n | v } and -hypo {n | v } display hypernyms and hyponyms respectively. For example, wn monkey -hypen returns the output shown in Listing 1.
| Listing 1: Monkey hypernyms |
01 Sense 1 02 monkey 03 => primate 04 => placental, placental mammal, eutherian, eutherian mammal 05 => mammal 06 => vertebrate, craniate 07 => chordate 08 => animal, animate being, beast, brute, creature, fauna 09 => organism, being 10 => living thing, animate thing 11 => object, physical object 12 => entity |
-tree {n | v} performs a recursive search that finds the hyponyms of each hyponym. For example, wn monkey -treen returns the output show in Listing 2.
In this particular case, this command displays a list of monkey species. If you want of different types of airplanes, run the wn airplane -treen command.
| Listing 2: Recursive Hyponymns for Monkey |
01 monkey 02 => Old World monkey, catarrhine 03 => guenon, guenon monkey 04 => talapoin, Cercopithecus talapoin 05 => grivet, Cercopithecus aethiops 06 => vervet, vervet monkey, Cercopithecus aethiops pygerythrus 07 => green monkey, African green monkey, Cercopithecus aethiops sabaeus 08 => mangabey 09 => patas, hussar monkey, Erythrocebus patas |
| WordNet Commands |
|
wn word -over provides an overview similar to a dictionary word article. The overview includes a number of senses, definitions, synonyms, and example sentences. wn word -syns {n | v | a | r} returns a list of synonyms for the specified word, where n=noun, v=verb, a=adjective, r=adverbs. For example, if you want to see synonyms for the noun monkey, use wn monkey -synsv, which returns: Sense 1
tamper, fiddle, monkey
=> manipulate
Sense 2
putter, mess around, potter, tinker, monkey, monkey around, muck about, muck around
=> work
|
If using the command line version of WordNet is not your cup of tea, you can opt for a graphical tool. WordNet includes its own graphical browser. Although the browser has a rather simplistic interface, it does allow you to access basic WordNet features. Looking up a word in the WordNet browser is a two-step process. First, enter a word into the Search Word field and press Enter. The application then returns an overview of the search term similar to the -over parameter. The Seaches for bar displays buttons for each syntactic category the found word belongs to, and you can use them to view more detailed information such as synonyms, coordinate terms, domains (for adjectives), etc.
Another graphical application based on WordNet, and a rather interesting one, is wnconnect (http://dingo.sbs.arizona.edu/~sandiway/wnconnect/). Enter two words, and wnconnect finds the shortest path or all possible connections between them. For example, enter the words apple and monkey, and wnconnect finds a connection between them and presents the final result as a graphical chart in PNG or PDF formats. Actually, finding connections between words can be quite addictive, and you can even turn it into a game. Just pick two random words and try to map a connection between them, then use wnconnect to see whether you've got it right.
If you have WordNet installed on your Linux server, you can access the application via Telnet or SSH. However, you can also install a full-blown local network dictionary server accessible via a web interface.
The easiest way to provide local network users with access to WordNet is to install a dictd server and a pre-formatted WordNet database on your local server. Both components are available at http://www.dict.org. Installing dictd is a rather standard process. Make sure that the flex, bison, and byacc packages are installed, then do:
./configure make make install
This installs the dictd server in the /usr/local/sbin directory. Next, download and unpack the WordNet tarball, which contains two files: wn.dict.dz and wn.index. Place the database files in any location on your system, for example, /usr/lib/dict. Create two configuration files: dict.conf for the dict client and dictd.conf for the dictd server. Put them into the /usr/local/etc directory. The dict.conf file should contain only the following line:
server localhost
The dictd.conf should look like:
database WordNet {data "/usr/lib/dict/wn.dict.dz" index "/usr/lib/dict/wn.index"}
To make sure that everything works properly, switch to the /usr/local/sbin directory and execute the dictd command as root. Then use the dict client to look up a word:
dict monkey
If everything works as it is supposed to, you can add a web interface to the dictionary server. Start with installing the Apache web server and the apache_mod_php module. Create a new text file, copy the PHP script from http://www.arachnoid.com/linux/dict.php.html and paste it into the file. In some cases (for example, on PCLinuxOS), you may need to enter the correct path to the dictd in the following line:
exec("/usr/bin/dict $equery 2>&1",$output,$error);
Save the file as wn.php in the /var/www/html directory. Now launch your browser and point it to the created page to check whether everything works properly.
Princeton University maintains a bare-bones online version of Wordnet. However, for the ultimate web-based version of WordNet, look no further than MultiWordNet On-line (http://multiwordnet.itc.it/online/). This is an implementation of WordNet for five different languages: English, Italian, Spanish, Hebrew, and Romanian. More impressive, however, is that you can view all these languages side by side, which makes MultiWordNet a quite unique language reference tool.
Born as an academic project, WordNet has become one of the most exciting and useful language reference tools available for the average user. WordNet's major advantage is its versatility: you can use it as a thesaurus and dictionary, but it also provides a fascinating insight into the world of language. This article gives you just a glimpse of WordNet's possibilities, and if you want to know more, WordNet's website (http://wordnet.princeton.edu/) and the WordNet book (http://www.amazon.com/gp/product/026206197X/) is a good starting point.