Reginald Ferber
Projects & Areas of Interest

Print View ] [ DEUTSCH ]

 Projects and Areas of Interest
 +Information Retrieval
 +Digital Libraries
 +Associative Word Nets
 +Knowledge Discovery in Text Collections
 +eLearning Search and Broker System
 +Simulation of Word Recognition
 +Language and Gender
 +Cellular Automata
 Contact and Disclaimer

 +JASIS: An Associative Model...
 +Simulation der Wortwahl
 +Isi'92: Vorhersage der Suchwortwahl ...
 +Associative Word Nets

Associative Word Nets

How humans learn, understand, store, and remember the "meaning" of words is an old and largely unanswered questions dealt with in many disciplines like philosophy, psychology, linguistics and other. Among the numerous explanations is one model that can be traced back to the Greek philosophers:

Association Theory

It is based on a simple observation: When subjects are asked to tell the words "that come to their mind" when they hear or read a given (stimulus) word the answers are from a limited set of (in our understanding) similar words. The associative model assumes that these associations between words (or other objects) are established (or learned) when the words (or objects) are frequently observed together.

Association theory does not give an answer to the above mentioned question of the representation and processing of the meaning of a word. But - going one step further - one can assume that the "network" of associations between words already constitute part of their meaning. This could mean that our memory is not that much a collection of isolated objects but that the relations between the objects or words are already a good part of that memory and play a crucial role in the processes of understanding and remembering.

Simulation of Word selection

I have followed up this line of arguments between 1990 and 1992 at the University of Paderborn in a project funded by the Heinz-Nixdorf-Institut. The goal was to simulate the selection of terms used in queries for a bibliographic database with an associative lexical net. The similarity values used in the net were calculated based on the co-occurrence of words in a large collection of about 250 000 psychological abstracts. I have implemented PC-programs to calculate the similarities and run the simulations.

The selections to simulate were taken from a sample of search records that professional searchers had performed following written requests of clients. The results of the simulations have been encouraging. More details can be found in the article An Associative Model of Word Selection in the Generation of Search Queries published 1995 in the Journal of the American Society for Information Science (JASIS) (Ferber, Wettler and Rapp 1995 [->]) and the German article Vorhersage der Suchwortwahl von professionellen Rechercheuren in Literaturdatenbanken durch assoziative Wortnetze (Ferber 1992 [->]).

HTML file generated 21. 10. 2003 by R. Ferber