Home Up Mail an Joachim Pimiskern Impressum

A semantic network

Last change: 05-Dec-2011   Deutsche Version

Here's a semantic network for download. Status: freeware. Please use the right mouse button, "Save as" or similar.

Download semantic network cluster_public.txt (text, ca. 15 MB)    cluster_public.zip (zipped text, ca. 2.5 MB)

You probably found this site since you got to cluster_public.txt via a search engine. This is a semantic network.

For the sake of simplicity the kinds of relationships between the symbols (the edges of the graph) are not distinguished. In many other semantic networks, edges are given a meaning, e.g. "is-a", "is-opposite-of" etc, but in this case this was not possible, since the network is too big.

In this context, the one and only relationship type means "has-something-to-do-with", which is a generalization of all other relationships.

File syntax

Records are sequences of non-empty lines. Between the records are one or more empty lines. Within a record, each line contains one word (or symbol, which means the same here). When you build up a graph from the file, all pairs symbols of a record have to be connected with each other.

Graphical representationTextual representation
A semantic network
FACE
NOSE

FACE
EAR

FACE
EYE

FACE
MOUTH

MOUTH
DRINK
EAT

DRINK
THIRST

EAT
HUNGER
FOOD

THIRST
HUNGER

THIRST
WATER

WATER
SOUP

SOUP
FOOD

FOOD
BREAD

Suggestion for programming: one object instance for each symbol Every symbol should get a list of pointers to the neighbour symbols. A hashtable would be appropriate to map the symbol strings to their object instances.

Philosophy

At the first glance, a symbol is nothing different from a number. "Water" is a list of five characters, just like xzjsd. To endow a symbol with meaning, it has to be embedded in an associative context of other symbols, for example liquid, wet etc.

Cultural aspects

Most of the words in the semantic network are in German. Although there are many foreign words contained, e.g. in English, it would be a good idea to translate a word you are searching for into German and then search in the graph for this translated word.

For example, the word Go is used in German only for the board game (aka weiqi, baduk). Someone who searches for words that are asociated with the English word "go" (the verb "to go") should translate this word into German first ("gehen") and then search for associations of gehen.



Related work

Wortschatz - Uni Leipzig
ConceptNet
Moby
Automatic Meaning Discovery Using Google (Rudi Cilibrasi, Paul Vitanyi)
WEBSOM - Self-Organizing Maps for Internet Exploration
OpenThesaurus
Finding words through spreading activation (Principia Cybernetica Web)
SEMAGER (Semantische Suchmaschine)
Automatisch berechnete Kollokationen aus dem DWDS Kerncorpus (lemmabasiert)
GlobalMind (MIT Medialab)
Garnoo (Stefan Ram)
Woxikon Synonyme
Some Ongoing KBS/Ontology Projects and Groups (Peter Clark)

Home Up Mail an Joachim Pimiskern Impressum