1. Notes partly derived from the book "Dynamic Taxonomies and Faceted Search", Sacco, Giovanni Maria; Tzitzikas, Yannis (Eds.), Springer, 2009.
2. At a glance
   1. One of the major search paradigms on the web, which are:
      1. A) direct access through search engines
      2. B) navigational access via static taxonomies
      3. C) dynamic taxonomies or faceted search
         1. a definition of Dynamic Taxonomies
   2. Used in several application areas
      1. e-commerce, e-auctions, and e-catalogs
      2. e-government, human resources and job placement
      3. news portals, multimedia
      4. cultural heritage collections
      5. medical guideline and diagnostic systems
   3. Key Concepts
      1. Taxonomy
      2. Faceted taxonomy
         1. Materialized faceted taxonomy
      3. Compound Term Composition Algebra (CTCA)
         1. basic logical notions (picture)
      4. Focus
      5. Zoom points
         1. zoom-in point
         2. zoom-out point
3. Background & motivation
   1. Assumption: 2 main Information Access (IA) modes
      1. The start of it all: the information overload problem
   2. IA Mode #1: Focalized Search
      1. Eg queries on databases, information retrieval techniques (IR), search engines
         1. RATIONALE: focus on textual unstructured information bases
         2. CON: Semantic gap between the user model (concepts) and the model used by retrieval systems (words or strings of characters).
         3. CON: query services are either too simplistic (e.g. free text queries in IR systems or Web search engines), or too complex for casual users (e.g. SQL queries, or Semantic Web queries)
         4. CON: Demonstrated lack of precision in search engines
         5. CON: No exploration capabilities since results are presented as a flat list with no systematic organization
         6. CON: Poor user interaction because the user has to formulate her query with no or very little assistance
      2. Eg Static taxonomies (eg Yahoo)
         1. RATIONALE: Represent metadata, Often manually created and maintained
         2. PRO: support abstraction, thus are easily understood by users
         3. CON: Browsing is either too simplistic (e.g. “plain” Web links) or application specific (dynamic pages derived by specific application programs), and does not support conceptual exploration
         4. CON: Static taxonomies don't scale
      3. agent-mediated search
         1. RATIONALE: general semantic schemata (eg ontologies), since they are often difficult to understand and manipulate by the casual user, are mediated by specialized agents
         2. CON: the classic knowledge-based system paradigm, which does not take the user into account, but rather establishes a master-slave relationship between the system and the user.
   3. IA Mode #2: Exploratory search
      1. find the right object or object-seeking pattern
         1. quickly find all possibly relevant features
         2. freely focus on the most relevant feature according to his individual requirements and discard objects without that feature
         3. explore all the features correlated with the selected one
      2. knowledge-seeking pattern
         1. goal is to increase our knowledge
      3. wisdom-seeking patterns
         1. goal is to understand the inner laws of the information base
4. Main characteristics
   1. Interdisciplinary area, holistic approach
      1. Modeling issues
      2. User Interaction issues
      3. Taxonomy design
      4. Architecture, implementation, and performance issues
      5. Applications.
   2. DTs rely on a minimal taxonomic schema
      1. notion of Subsumption
      2. notion of concept, label
      3. notion of the extension of a concept
   3. Supports multidimensional taxonomies
      1. all possibly relevant features of an object are available
   4. Implements a User-centered approach
      1. incremental
      2. User friendly
      3. Hybrid approach, between querying and browsing
      4. Supports experimentation
      5. simplicity and minimality
      6. Support self-adapting exploration
   5. It's search effective
      1. dynamic taxonomies have an extremely fast convergence to small result sets
         1. empirical evidence
      2. no empty results
   6. It's a visual framework for conceptual exploration
      1. concept relationships other than subsumption (The Base Extensional Inference Rule)
      2. concepts relationships based on logical combinations of concepts (the Extensional Inference Rule)
      3. notion of a reduced taxonomy; genesis of the name 'dynamic taxonomy'
   7. Implements a shema-less schema design
      1. Concept relationships are dynamically discovered
      2. New concepts need not being represented explicitly
      3. avoid schema growth, combinatorial explosion
      4. highlights primitive relationships
   8. Dynamic taxonomies are more generic than faceted search systems
      1. faceted classification has more rigid rules (eg orthogonal organization)
      2. dynamic taxonomy model only requires a multidimensional taxonomy
      3. There are practical situations in which the violation of the orthogonal organization of facets is beneficial or required
   9. Dynamic Taxonomies Systems can be classified in various ways..:
      1. E.g.: attribute-value shallow taxonomies
      2. E.g.: objects classified under terminal concepts only.
      3. E.g.: 'AND' only refinement query
5. Example implementations
   1. Non RDF based
      1. Flamenco
      2. Qviz
      3. Delphi category browser
      4. Database of Mid-Victorian wood-engraved Illustration
      5. Charles Cushman Photograph Collection
         1. link to the technical implementation explanation
         2. link to the metadata schema used
      6. Elastic Lists Demo
      7. DJFacet
   2. RDF based
      1. Brownsauce
      2. Longwell (MIT-simile)
      3. Exhibit (MIT-simile)
      4. Kochief
      5. MSpace
      6. RDFbrowser
      7. SlashFacet
      8. KulturiSampo