|
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.〔http://academic.csuohio.edu/kneuendorf/content/cpuca/qtap.htm〕 ==Design== Catpac is a self-organizing, i.e. unsupervised, interactive activation and competition (IAC) artificial neural network used for text analysis.〔 〕〔http://www.galileoco.com/literature/Wolfpak10a.pdf〕 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.〔 抄文引用元・出典: フリー百科事典『 ウィキペディア（Wikipedia）』 ■ウィキペディアで「Catpac」の詳細全文を読む スポンサード リンク
|