עידן ה- Web 3.0. ל- Web Semantic ו- Web 3.0. Info April 2008

Transcription

1 המעבר מט קסונו מיה ל אונ תו לו גיה וההבט חה של עידן ה- Web 3.0 ד"ר אורי חנני 1 April 2008 Info 2008 תו כ ן ה הרצאה 1.רקע לתחום ומושגי יס ו ד ד ר ך טקס ונ ומי ה ל אנתו לו ג י ה - מאינד ק ס, 2.עקרונו ת ומבנה מערכת אונ תול וגיה 3.המחשו ת ואירו ע ים מהשטח 4.כיו ונ י התפתחו ת של התחו ם והמע בר ל- Web Semantic ו- Web 3.0

2 Generalization from Catalog through Thesauri to Ontology What is an Ontology? Catalog/ ID Thesauri narrower term relation Formal is-a Frames General (properties) Logical constraints Terms/ glossary Informal is-a Formal instance Value Restrs. Disjointness, Inverse, partof 3 Deborah L. McGuinness. Ontologies Come of Age. in Dieter Fensel, Jim Hendler, Henry Lieberman, and Wolfgang Wahlster, editors. The Semantic Web: Why, What, and How, MIT Press, Information Representation 1. Terms keywords and phrases 2. Index An ordered list of terms paired with pointers to content within a collection 3. Glossary A list of terms with single definitions with respect to a particular domain 4. Vocabulary A list of terms, sometimes with definitions. A single level taxonomy. A vocabulary can be considered a very localized glossary 5. Dictionary An organized collection of definitions and uses of terms. A term may have more than one definition and more than one use 6. Thesaurus A collection of terms, their definitions (that

3 7. Taxonomy A structured vocabulary that embodies relationships among terms. The most common relationship is a hierarchy (tree structure). Some major taxonomies are: MARC, LCSH, MeSH 8. Ontology A particular theory about the nature of being or the kinds of existents 5 What is Taxonomy? Taxonomy is the Lingua Franca of bodies of content Taxonomies are structures that provide a way of classifying things into a series of hierarchical groups to make them easier to identify, study, or locate. Taxonomies impose order on content chaos and serve as the backbone through which end users access content. Montague Institute Review, Managing taxonomies strategically A taxonomy that is carefully built and truly representative of a domain is not just the foundation of an effective KM system, it is the interface and the platform through which information storage and retrieval are enabled.

4 selected standard Taxonomies typically have the following elements: Controlled vocabulary - Standard terms, Thesaurus Hierarchical relationships (Narrower than, Broader than) Cross references (See also, Related to) Logic schemes (organizational schemes, classification schemes such as: DDC, LC) Taxonomy may contain as few as two categories and as many as thousands of categories on one or more levels. 7 Montague Institute Review, Managing taxonomies strategically Questions related to Taxonomy Size - One taxonomy or Multiple taxonomies? Structure - Logic scheme Presentation - How to present it? Terminology Controlled or free? Different user perspectives - Marketing vs. Development, Managers vs. Employees Maintenance - Who maintains it? Evaluation criteria parameters in accordance with

5 To search via a computer without a taxonomy system is like trying to find your way around an unfamiliar country without a map. DelphiGroup, Taxonomy and Content Classification Browsing - Using Categories in a dynamic, interactive and iterative discovery process Search zooming Broadening / Narrowing searches Context surrounding finding relevant information with specific content filters are employed Project DESIRE The role of classification schemes in Internet resource description and discovery 9 Taxonomy Weaknesses Terrorism Hierarchical structure Type Redundancy to create multiple Nuclear perspectives Southeast Asia Middle East Logical manner of subdivided classes Biological Disability to connects different leaves Southeast Asia Middle East Chemical Southeast Asia Fixed relations types and impact Middle East Identical relations (BT/NT, RT) Religion Identical impact no weighting Islamic Christian differentiation Groups Al-Qaeda Jemaa Islamiah Taxonomy terms impact Key People Osama bin Laden Identical impact - no weighting

6 An Ontology is A formal definition of a body of knowledge A conceptual model of some portion of the world A collection of terms, relations and definitions relevant to a domain of knowledge A network of relationships or associations among concepts, augmented by ways that govern the behavior of the concepts 11 What is Ontology? An ontology is more than a taxonomy Terms importance differentiation Richer relationships between terms Variant relations definitions such as: Types Weights Multiple perspectives of the domain-specific knowledge

7 Taxonomy Ontology 13 Al-QAEDA REGIONS Pakistan Afghanistan Philippines Indonesia Singapore Malaysia ORGANIZATIONAL STRUCTURE Associates Membership Leaders AFFILIATED GROUPS MILF KMM Jemaah Islamiah Abu Sayaf FINANCIAL SUPPORT Fund raising Relief organizations Private donors Drug trafficking Money laundry Ontology in Citer Ontology organizes keywords and terms by capturing semantic relationships among the keywords in a specific context Ontology is the basis for the queries search on the targeted network Ontology terms are used to Locate/Gather/Summarize/Filter the harvested documents in a specific framework

8 Richer relationships Types BT/NT Synonym Type/Type of Related to Reflected in Characteristic of Limitation of Aim of Disadvantage of Objectives of Negotiation Cooperation Weights Terms weigh: Relations weight: Ontology Example Multiple perspectives The Anthrax view

9 Multiple perspectives The Symptoms view 17 כש ירו ת אר גוני ת כשירות ארגונית ידע ארגוני מ פות ידע חקירות, דוו"חו תות, חוקרים צוותים טפסים מידע דואר אלקטרוני הצהרות,,

10 גילוי, מיפוי וניתוח Ontology Relations Type Relation types are user s definitions of the optional association between the ontology terms in a specific knowledge environment. Relation types can be: פרו פיל אישי וארגוני מילון מונחים אר גוני תמצות מידע - על מיפוי ידע סינון ודירוג ניטור דינמי אונתולוגיה ארגונית קטלוג ומיון איסוף ניתוח מידע חיפוש ואחזור 19 Ontology Editor Features Ontology Terms A single keyword or concept that is defined in the Ontology list and represented in the ontology map. The importance of the term is represented by its weight. Ontology terms are defined as source or target terms Ontology Associations The association is the interaction between the ontology terms within a specific knowledge environment. The importance of the association is represented by its weight.

11 The 4 Steps to Ontology Map 1. Insert Ontology Terms 2. Define Relation types (and Term Properties) 3. Create Ontology Associations 4. Enhance Ontology terms and associations 21 Ontology Build Up Process 1. Insert Ontology Terms Action User inserts or generates a list of most commonly used terms in the domain of interest. Terms can be synonyms, antonyms, associations, acronyms and abbreviations, keywords and phrases Terms should not contain Punctuations marks OR Transition/connecting words Sources User s understanding Key documents in the organization Thesaurus/Dictionary/Taxonomy schemes Leading articles/books/portals/guides in the fields Ontology library such as: DAML ( )

12 1. Insert Ontology Terms 23 Ontology Build Up Process 2. Define Relation Types Action User defines relation types. Relation types can be derived from Thesaurus/Taxonomy relations types such as: BT/NT, See Also, Related to Or to be defined according to organizational needs such as: Type of, Money Trail, Opportunity, Risk, Reflected in/reflected of Sources User s understanding Brain storming forum Thesaurus/Taxonomy schemes

13 2. Define Relation Types 25 Ontology Build Up Process 3. Create Ontology Associations Action User establishes Term-To-Term associations. The association can be one to one, one to many, one to many The association can be defined to only one direction or both directions (Reverse relation) User s understanding Sources Thesaurus/Taxonomy schemes Ontology library Citer s Clustering Methods

14 3. Create Ontology Associations 27 Ontology Build Up Process 3. Create Ontology Associations

15 4. Enhance Ontology terms and associations Action User enhances terms and associations using citer s clustering methods. 2 types of clustering are available: Matching Clustering and Roaming Clustering. Matching Clustering will reveal new associations between Ontology terms. Roaming Clustering will reveal both terms and associations. Sources Existing Articles/books/ guides Ontology library such as: DAML ( ) 29 Semantic Knowledge Map The Semantic Knowledge Map reveals Bulgaria's involvement in weapon/ nuclear materials trafficking, money transactions and laundering

16 31 Ontology Editor Citer Ontology Editor enables the full mapping & description of all known entities in the system (either from structured or unstructured sources). It provide another dimensional view of the organizational knowledge environment enabling concept exploration and knowledge formulation.

17 Mapping Terrorist activities: The Citer Ontology Server allows the mapping of entities and relationships. Disclaimer: The following example is used to illustrate KM Ideas, and is not meant nor represents expertise in Intelligence work. All data used in these examples are Internet open source materials. 33 Mapping Terrorist organizations Locations of Activities Operatives Detailed Description of the organization

18 Terrorist Profile Related Accomplices Membership Status 35 Mapping Terrorists Relationships Organizations Filtered & Focused View on Suspected terrorist s Affiliations Accomplices Activities Locations of Activities

19 37 Spain Terror Threat Ontology - Relations

20 39 Ontology Map Visualization South East Asia

21 האונ תו לוגיה מרחיבה את מו שג הסכ י מה של עולם ה- DB : הכללה של הסכימה והוספה של מרחב נתונים לא מובנים (כגון טקסט) מאפש רת הע בר ה ש ל נתונים מ רובים מובנים מב סיסי נתונים מייצ רת ה ל כה ל מ ע שה מ ער כ ת מידע ש ל מה ו ל א רק כ לי עז ר האונ תו לוגיה היא רשת משוכללת: המקור ה תפיסתי הוא ה מו ח האנושי מאפש ר מס ע בתוך מ ר חב אסוציא צי בי מ ערכת ח וקים מאפ שרת פ עולות י זומו ת: מע רכ ת מו מח ה, תיוג ידני/חברתי יציר ה או טומטית ש ל אונתולוגיות היסטורי ה: ה מ שקולות מאפ שרים כיוו נון ודינאמיקה שיתופיות ש ל אונתולוגיות ותחזוקה דינאמית הכללה של מודל התיזאורוס/מילון המידע מביאה למודל רחב יותר, Web Semantic וה- 3.0 ומכאן הדרך לעידן ה- Web Web 3.0

22 43 Web 3.0