A Spreading Activation Model for Vocabulary Merging

A spreading activation model is proposed for the purpose of automatically merging databases with heterogeneous indexing systems. An overview of the connectionist approaches in information retrieval (IR) is discussed. Then, spreading activation models of IR are compared with the vector space models that have been widely adopted in the development of IR systems. Based upon the review of the previous vocabulary switching methods, the issue of compatibility in mapping terms from different indexing languages is discussed. And, I propose a spreading activation model of vocabulary merging in order to overcome some limitations of the previous approaches. The network representation and the term mapping process are explained. Finally, a research proposal is presented including test databases, implementation and evaluation of the system.