Catpac is a computer program that analyzes text samples to identify key concepts contained within the sample. It was developed chiefly by Joseph Woelfel, a University at Buffalo sociologist for the analysis of attitude formation and change in the sociological context. It uses text files as input and produces output such as word and alphabetical frequencies as well as various types of cluster analysis.
Contents
Design
Catpac is a self-organizing, i.e. unsupervised, interactive activation and competition (IAC) artificial neural network used for text analysis. The program generates a multidimensional scalar output organizing words throughout the text by creating a weighted word-by-word matrix that establishes the eigenvector centralities of concepts. The word-by-word matrix represents the relationship between one word and the occurrence of another. Catpac identifies important words and patterns based on the organization of the text. This process mimics the connections between neurons in a human brain, strengthening connections through conditioning to generate a pattern of similarities among all words within a body of text.
Use
Catpac has been used in academic scholarship to investigate massive textual data sets, as a strong semantic network analysis tool, for longitudinal analyses, for multilingual analyses, as a predictor of media usage and as a powerful content analysis tool.
Availability
Catpac is currently available in windows 32 bit format.