AutomaticOntologyConstruction.ppt

Automatic Ontology Construction LU Qin Dept. of Computing, HK Polytechnic Univ. csluqin@.hk Collaborators: Ji Luning, Chen Yirong, Cui Gaoying, Yang Yuhang Outlines P1: Basic Concepts Background Definitions Relationship between Terminology and Ontology P2: Terminology Extraction P3: Ontology Construction Term Related Concepts Need for the appropriate use of words in many applications Knowledge acquisition Acquisition of terminology knowledge for domain relevant applications Examples: Document relevance with respect to “keywords”, summarization, event tracking, etc.. Automatic indexing Taking terms as a means to assign descriptors to documents, representation of information for information retrieval Need for the study of formation/variation of words Nature of text Texts are heterogeneous: need for different expository strategies: Difference in progression, rhetorical processes and means of expression Language dependent Knowledge/content may be domain dependent (ontological relations), yet, quite complex as free text are multi-domain in nature Deep understanding vs. shallow understanding Nature of domain specific information Information are revealed by the domain specific terms they represent: Example 1: “Bluetooth is a wireless communication protocol” Example 2: “Bluthooth is the name of a technology that is used for computers to talk to each other so that information can be exchanged by machines in short distances without the need for any physical connection” Question: What is “Bluetooth” Need for new word/phrase discovery and need for domain specificity discovery Need for prior-domain specific knowledge Identification of their relationships with existing domain term/knowledge Nature of terms Vehicles of domain knowledge Lexical representation of concepts Example: “protocol” definitions in Wikipedia “Protocol”: “a set of guidelines or rules” “Communication protocol”: “a set of rules governing communication between electronic devices” “Protocol (Computing)”: “