1 / 12

Using activation spreading for ontology merging Miłosław L. Frey FGAN – FKIE Neuenahrer Str. 20

Using activation spreading for ontology merging Miłosław L. Frey FGAN – FKIE Neuenahrer Str. 20 53343 Wachtberg, Germany m.frey@fgan.de. Overview. Ontology Definition From ontology to network The network General idea Learning Creating hierarchy Taxonomy merging General Method

Download Presentation

Using activation spreading for ontology merging Miłosław L. Frey FGAN – FKIE Neuenahrer Str. 20

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Using activation spreading for ontology merging Miłosław L. Frey FGAN – FKIE Neuenahrer Str. 20 53343 Wachtberg, Germany m.frey@fgan.de RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  2. Overview • Ontology • Definition • From ontology to network • The network • General idea • Learning • Creating hierarchy • Taxonomy merging • General • Method • Summary RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  3. Ontology • Ontology (def.): In computational sciences an ontology is an explicit representation of knowledge in a given thematic domain. Ontology as a network From: Brachman, Schmolze, An Overview of the KL-ONE Knowledge Representation System, 1985 RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  4. Ontology • Ontology (def.): In computational sciences an ontology is an explicit representation of knowledge in a given thematic domain. Ontology as a network From: Brachman, Schmolze, An Overview of the KL-ONE Knowledge Representation System, 1985 RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  5. Internal structure of a node Spreading activation The node’s state depend on the other nodes’ activations, connections weights and time: each node defines a point in the multidimensional space described by features Learning (Sowa, 2002) Rote learning Connection weights change Restructuring Generalization Improves the taxonomy Allows for classification of unknown objects The network : general idea RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  6. The network: creating hierarchy by example (1) Input data One-dimensional example: 7 ellipses differentiated in the ratio of axes. RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  7. Just input data Final network: “discovery” and pruning The network: creating hierarchy by example (2) RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  8. Taxonomy merging: general • Taxonomy merging (by analogy to ontology merging) is a procedure of blending two or more taxonomies into a single one. • Two methods: • union (used in the method presented), • intersection. • For simplicity: • merging is regarded as completion: one taxonomy complements the other one. RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  9. (1) Starting taxonomies (2) Joining by features sharing Taxonomy merging: method (1) RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  10. (3) Restructuring (4) Pruning Taxonomy merging: method (2) RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  11. Summary Shown • Preliminary ideas • Connectionist method to join two taxonomies • Illustration by an artificial example Further work: • Apply to real-world data • Extend to other than is-a relations • Make the method symmetrical (no need to identify the main root node) RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

  12. Thank you for your Attention • Questions and Comments • are appreciated RESARCH INSTITUTE FOR COMMUNICATION, INFORMATION PROCESSING AND ERGONOMICS

More Related