THE NEURAL NET OF NEURAL NETWORK RESEARCH AN EXERCISE IN BIBLIOMETRIC MAPPING

In this paper we discuss the limits and potentials of bibliometric mapping based on a specific co-word analysis. The subject area is neural network research. Our approach is a 'simulation' of expert assessment by offering the reader a narrative of the field which can be used as background information when 'reading' the bibliometric maps. The central issue in the applicability of bibliometric maps is whether these maps may supply 'additional intelligence' to users. In other words, whether such a bibliometric tool has an epistemological value, in the sense that it ecriches existing knowledge by supplying 'unexpected' relations between specific 'pieces' of knowledge ('synthetic value') or by supplying 'unexpected' problems ('creative value'). We argue that sophisticated bibliometric mapping techniques are indeed valuable for further exploration of these 'epistemological' potentials. In particular, these techniques may open new avenues to study science as a self-organizing system in the form of a 'neural network like' structure of which the bibliometric map is a first-order approximation. In that sense, this paper deals with the 'neural net of neural network research' as our bibliometric techniques in fact mimic a connectionistic approach.