An Ontology-Driven Approach to Extracting Descriptive Streams

摘要

As the amount of texts available online keeps rapidly growing, it becomes increasingly difficult for people to keep track of and locate the information they actually need. We can know "what is the text about" through current text processing tasks, such as information retrieval, text classification, information extraction, text mining, topic identification, and topic detection and tracking, and so on. However, we don't know "in what aspects the topic of a text is depicted and what orders those descriptive aspects follow". In this paper, we consider a task of representation of the contents of a document by threading descriptive aspects of topics besides the text's topics. This task is named as the descriptive steam extraction. A knowledge acquisition ontology-driven approach is proposed to extract descriptive streams of texts. The ontology mainly consists of three components: categories, their relationships, and linguistic vocabularies of these terms. Experiments show that the algorithm is encouraging, and the results can assist in locating text spans containing the knowledge, which users require.